==> picture [215 x 22] intentionally omitted <==

Single Technology Appraisal

Autologous anti-CD19-transduced CD3+ cells for treating relapsed or refractory B- cell acute lymphoblastic leukaemia in people 26 years and over [ID1494] Committee Papers

==> picture [215 x 22] intentionally omitted <==

NATIONAL INSTITUTE FOR HEALTH AND CARE EXCELLENCE

SINGLE TECHNOLOGY APPRAISAL

Autologous anti-CD19-transduced CD3+ cells for treating relapsed or refractory B-cell acute lymphoblastic leukaemia in people 26 years and over [ID1494]

Contents:

The following documents are made available to consultees and commentators:

Access the final scope and final stakeholder list on the NICE website.

Pre-technical engagement documents

1. Company submission from Kite, a Gilead company

2. Clarification questions and company responses:

• a. Main response

• b. Updated cost-effectiveness analyses appendix

3. Patient group, professional group, and NHS organisation submissions from:

• a. Leukaemia Care

• b. NCRI-ACP-RCP-RCR

• c. NHSE CAR-T tariff summary

• d. Gilead response to NHSE CAR-T tariff documents

4. Evidence Review Group report prepared by ScHARR

5. Evidence Review Group report – factual accuracy check

Post-technical engagement documents

6. Technical engagement response from company:

• a. Response form

• b. Additional evidence appendix

• c. Updated cost-effectiveness results

7. Technical engagement responses and statements from experts:

• a. Professor David Marks, Director Bristol BMT Unit/Head ALL program – clinical expert, nominated by Kite

• b. Sophie Wheldon, Advocacy Officer – patient expert, nominated by Leukaemia Care

8. Technical engagement responses from consultees and commentators: a. Leukaemia Care b. NCRI-ACP-RCP-RCR

==> picture [215 x 22] intentionally omitted <==

9. Evidence Review Group post-technical engagement documents prepared by ScHARR:

• a. ERG critique of company response to technical engagement

• b. ERG addendum post-committee meeting

• c. ERG additional analysis post-committee meeting, including NHSE CAR-T delivery costs

Any information supplied to NICE which has been marked as confidential, has been redacted. All personal information has also been redacted.

NATIONAL INSTITUTE FOR HEALTH AND CARE EXCELLENCE

Single technology appraisal

KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia [ID1494] Document B Company evidence submission

November, 2021

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Contents

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Tables

Table 1: The decision problem ................................................................................. 15 Table 2: Technology being appraised ...................................................................... 19 Table 3: Definitions of key treatment objectives in ALL ............................................ 25 Table 4: NICE treatment guidance for R/R adult ALL ............................................... 28 Table 5: Clinical effectiveness evidence ................................................................... 37 Table 6: Summary of trial methodology for ZUMA-3 ................................................ 41 Table 7: Summary of ZUMA-3 datasets ................................................................... 47 Table 8: Baseline demographics and characteristics at baseline (Phase 1 + 2 combined) ................................................................................................................ 48 Table 9: Baseline demographics and characteristics at baseline (Phase 2) ............. 49 Table 10: Summary of statistical analyses: ZUMA-3 ................................................ 51 Table 11: Summary of clinical effectiveness: ZUMA-3 ............................................. 56 Table 12: Overall survival (Phase 1 + 2 combined, data cut 23/07/21) ..................... 58 Table 13: DOR using investigator review (Phase 1 + 2 combined: data cut 23/07/21) ............................................................................................................................ 61 Table 14: RFS using investigator review (Phase 1 + 2 combined; data cut 23/07/21) ............................................................................................................................ 63 Table 15: Summary of efficacy endpoints (Phase 1 + 2 combined) .......................... 63 Table 16: Overall survival (Phase 1 + 2 combined) .................................................. 65 Table 17: RFS per investigator assessment (Phase 1 + 2 combined) ...................... 67 Table 18: Summary of overall complete response rates (Phase 2, mITT) ................ 69 Table 19: Summary of MRD status (Phase 2, mITT) ................................................ 70 Table 20: Duration of remission per central assessment (Phase 2, mITT) ............... 71 Table 21: Overall survival (Phase 2, mITT) .............................................................. 73 Table 22: RFS per central assessment (Phase 2, mITT) ......................................... 79 Table 23: Patient incidence of allo-SCT after treatment (Phase 2, mITT) ................ 81 Table 24: Summary of comparators ......................................................................... 85 Table 25: Summary of key ITCs used in the economic model .................................. 89 Table 26: Summary of ITC results (OS) ................................................................... 92 Table 27: Summary of ITC results (EFS) ................................................................. 93 Table 28: Overview of design for SCHOLAR-3 ...................................................... 103 Table 29: Overall summary of AEs ......................................................................... 111 Table 30: Subject incidence of AEs occurring in >10% of subjects by preferred term and worst grade (Phase 2, safety analysis set) ...................................................... 111 Table 31: Subject incidence of KTE-X19-related AEs occurring in ≥ 10% of subjects by preferred term and worst grade (Phase 2, safety analysis set) .......................... 113 Table 32: Subject incidence of SAEs occurring in ≥ 2 patients by preferred term and worst grade (Phase 2, safety analysis set) ............................................................. 114 Table 33: Subject incidence of KTE-X19-related SAEs occurring in ≥ 2 patients by preferred term and worst grade (Phase 2, safety analysis set) ............................... 115 Table 34: End-of-life criteria ................................................................................... 126 Table 35: Summary list of published cost-effectiveness studies ............................. 128 Table 36: Features of the economic analysis ......................................................... 140 Table 37: Summary of approach to modelling of clinical parameters...................... 145 Table 38: Patient baseline characteristics in the base case economic analysis ..... 147 Table 39: Summary of data sources adopted for different subgroups in the economic model – base case ................................................................................................. 149 Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 40: Distribution of patients in KTE-X19 arm (mITT analysis) ........................ 151 Table 41: Summary of curve selection – base case* .............................................. 156 Table 42: Adverse events rates included in the model ........................................... 169 Table 43: EQ-5D Indices by injection, relapse, and AE status (all observations in each time period) in ZUMA-3 trial phase 2 ............................................................. 174 Table 44: EQ-5D-5L Index (UK Crosswalk Value Set) by Injection, Relapse, and AE Status ..................................................................................................................... 174 Table 45: Base-case health state utilities: ZUMA-3 ................................................ 175 Table 46: Utility decrements associated with adverse events included in the model .......................................................................................................................... 177 Table 47: Health state utility values applied in scenario analyses .......................... 179 Table 48: Summary of utility values for cost-effectiveness analysis ....................... 179 Table 49: Drug administration costs used in the economic model .......................... 185 Table 50: Conditioning chemotherapy drug costs .................................................. 189 Table 51: Conditioning chemotherapy administration costs.................................... 189 Table 52: KTE-X19 bridging therapy costs ............................................................. 190 Table 53: KTE-X19 infusion and administration costs ............................................ 191 Table 54: Inotuzumab drug acquisition costs ......................................................... 192 Table 55: Inotuzumab administration costs ............................................................ 192 Table 56: Blinatumomab acquisition costs ............................................................. 194 Table 57: Blinatumomab administration costs ........................................................ 194 Table 58: Ponatinib acquisition costs applied in the model..................................... 195 Table 59: FLAG-IDA acquisition costs .................................................................... 196 Table 60: FLAG-IDA acquisition costs applied in the model ................................... 196 Table 61: FLAG-IDA administration costs applied in the model.............................. 197 Table 62: Health state costs for KTE-X19, EFS health state .................................. 199 Table 63: Health state costs for comparators, EFS health state ............................. 200 Table 64: Health state costs, all treatments, progressed disease health state ....... 201 Table 65: Health state costs, all treatments, progressed disease health state ....... 204 Table 66: Subsequent therapy one-off costs .......................................................... 204 Table 67: Subsequent allo-SCT distribution ........................................................... 205 Table 68: Subsequent allo-SCT cost ...................................................................... 205 Table 69: Subsequent allo-SCT follow-up cost breakdown .................................... 206 Table 70: CRS AE cost........................................................................................... 206 Table 71: VOD AE Cost.......................................................................................... 208 Table 72: Unit costs of adverse events included in the model ................................ 208 Table 73: Total one-off adverse event costs used in the model ............................. 212 Table 74: Terminal care costs ................................................................................ 212 Table 75: List of assumptions for the base case analysis ....................................... 213 Table 76: Base-case results (overall population) .................................................... 217 Table 77: Base-case results (Ph- population) ......................................................... 219 Table 78: Base-case results (Ph+ population) ........................................................ 222 Table 79: OWSA results, overall population, inotuzumab ....................................... 224 Table 80: OWSA results, overall population, FLAG-IDA ........................................ 225 Table 81: OWSA results, Ph- population, blinatumomab ....................................... 226 Table 82: OWSA results, Ph- population, FLAG-IDA ............................................. 227 Table 83: OWSA results, Ph- population, inotuzumab ........................................... 228 Table 84: OWSA results, Ph+ population, ponatinib ............................................... 229 Table 85: OWSA results, Ph+ population, FLAG-IDA ............................................ 230

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 86: OWSA results, Ph+ population, inotuzumab ........................................... 231 Table 87: Probability that KTE-X19 is the most cost-effective treatment ................ 233 Table 88: Probabilistic results - overall population ................................................. 234 Table 89: Probabilistic results - Ph- population ...................................................... 234 Table 90: Probabilistic results - Ph+ population ..................................................... 234 Table 91: Results of scenario analysis – overall population ................................... 239 Table 92: Results of scenario analysis – Ph- population ........................................ 241 Table 93: Results of scenario analysis – Ph+ population ....................................... 245 Table 94: Database searches in the clinical and economic SLR ............................ 323 Table 95: Clinical inclusion and exclusion criteria ................................................... 330 Table 96: Identified studies for KTE-X19 in r/r adult ALL ........................................ 333 Table 97: Studies evaluated for inclusion in the ITC............................................... 335 Table 98: Matching adjusted indirect comparison: included studies ....................... 336 Table 99: Summary of trials excluded from the MAIC ............................................ 338 Table 100: List of excluded studies at data extraction ............................................ 338 Table 101: List of studies excluded at eligibility screening...................................... 343 Table 102: Downs and Blacks checklist for quality assessment of ZUMA-3 .......... 358 Table 103: Quality assessment for TOWER ........................................................... 359 Table 104: Quality assessment for INO-VATE ....................................................... 360 Table 105: Economic inclusion and exclusion criteria ............................................ 377 Table 106: Overview of HRQoL and utility studies ................................................. 382 Table 107: Outcomes of studies reporting HRQoL and utilities .............................. 384 Table 108: Overview of Cost and Healthcare Resource Use studies ..................... 392 Table 109: Outcomes of studies reporting healthcare cost and resource use ........ 394 Table 110: Summary of model results compared with clinical data ........................ 414 Table 111: Summary of QALY gain by health state (overall population), KTE-X19 vs. inotuzumab ............................................................................................................. 415 Table 112: Summary of QALY gain by health state (overall population), KTE-X19 vs. inotuzumab ............................................................................................................. 415 Table 113: Summary of QALY gain by health state (Ph- population), KTE-X19 vs. blinatumomab ......................................................................................................... 415 Table 114: Summary of QALY gain by health state (Ph- population), KTE-X19 vs. FLAG-IDA ............................................................................................................... 416 Table 115: Summary of QALY gain by health state (Ph- population), KTE-X19 vs. inotuzumab ............................................................................................................. 416 Table 116: Summary of QALY gain by health state (Ph+ population), KTE-X19 vs. ponatinib ................................................................................................................. 417 Table 117: Summary of QALY gain by health state (Ph+ population), KTE-X19 vs. FLAG-IDA ............................................................................................................... 417 Table 118: Summary of QALY gain by health state (Ph+ population), KTE-X19 vs. Inotuzumab ............................................................................................................ 418 Table 119: Summary of costs by health state (overall population), KTE-X19 vs. inotuzumab ............................................................................................................. 418 Table 120: Summary of costs by health state (overall population), KTE-X19 vs. FLAG-IDA ............................................................................................................... 419 Table 121: Summary of costs by health state (Ph- population), KTE-X19 vs. blinatumomab ......................................................................................................... 419 Table 122: Summary of costs by health state (Ph- population), KTE-X19 vs. FLAGIDA ......................................................................................................................... 419

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 123: Summary of costs by health state (Ph- population), KTE-X19 vs. inotuzumab ............................................................................................................. 420 Table 124: Summary of costs by health state (Ph+ population), KTE-X19 vs. ponatinib ................................................................................................................. 420 Table 125: Summary of costs by health state (Ph+ population), KTE-X19 vs. FLAGIDA ......................................................................................................................... 420 Table 126: Summary of costs by health state (Ph+ population), KTE-X19 vs. inotuzumab ............................................................................................................. 421 Table 127: Summary of predicted resource use by category of cost (overall population), KTE-X19 vs. inotuzumab .................................................................... 421 Table 128: Summary of predicted resource use by category of cost (overall population), KTE-X19 vs. FLAG-IDA ...................................................................... 422 Table 129: Summary of predicted resource use by category of cost (Ph- population), KTE-X19 vs. blinatumomab .................................................................................... 422 Table 130: Summary of predicted resource use by category of cost (Ph- population), KTE-X19 vs. FLAG-IDA ......................................................................................... 424 Table 131: Summary of predicted resource use by category of cost (Ph- population), KTE-X19 vs. inotuzumab ........................................................................................ 424 Table 132: Summary of predicted resource use by category of cost (Ph+ population), KTE-X19 vs. ponatinib ............................................................................................ 425 Table 133: Summary of predicted resource use by category of cost (Ph+ population), KTE-X19 vs. FLAG-IDA ......................................................................................... 425 Table 134: Summary of predicted resource use by category of cost (Ph+ population), KTE-X19 vs. inotuzumab ........................................................................................ 426 Table 135: Baseline demographics and characteristics at baseline (Phase 1) ....... 428 Table 136: Summary of efficacy (Phase 1, target dose analysis set) ..................... 429 Table 137: Summary of variables applied in the economic model .......................... 434 Table 138: Summary of goodness-of-fit data for KTE-X19 EFS (naïve) – standard parametric and spline models ................................................................................. 455 Table 139: Summary of goodness-of-fit data for KTE-X19 EFS (naïve) – MCM .... 456 Table 140: Summary of goodness-of-fit data for KTE-X19 OS (naïve) – standard parametric and spline models ................................................................................. 458 Table 141: Summary of goodness-of-fit data for KTE-X19 OS (naïve) – MCM ...... 458 Table 142: Summary of goodness-of-fit data for inotuzumab EFS – standard parametric and spline models ................................................................................. 460 Table 143: Summary of goodness-of-fit data for inotuzumab EFS – MCM............. 461 Table 144: Summary of goodness-of-fit data for inotuzumab OS – standard parametric and spline models ................................................................................. 463 Table 145: Summary of goodness-of-fit data for inotuzumab OS – MCM .............. 463 Table 146: Summary of goodness-of-fit data for FLAG-IDA EFS – standard parametric and spline models ................................................................................. 465 Table 147: Summary of goodness-of-fit data for FLAG-IDA EFS – MCM .............. 466 Table 148: Summary of goodness-of-fit data for FLAG-IDA OS – standard parametric and spline models .................................................................................................. 468 Table 149: Summary of goodness-of-fit data for FLAG-IDA OS – MCM ................ 468 Table 150: Summary of goodness-of-fit data for KTE-X19 EFS (naïve, Phpopulation) – standard parametric and spline models ............................................ 470 Table 151: Summary of goodness-of-fit data for KTE-X19 EFS (naïve, Phpopulation) – MCM ................................................................................................. 471

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 152: Summary of goodness-of-fit data for KTE-X19 OS (naïve, Ph- population) – standard parametric and spline models ............................................................... 473 Table 153: Summary of goodness-of-fit data for KTE-X19 OS (naïve, Ph- population) – MCM .................................................................................................................... 473 Table 154: Summary of goodness-of-fit data for SCHOLAR-3 SCA-3 EFS – standard parametric and spline models ................................................................................. 475 Table 155: Summary of goodness-of-fit data for SCHOLAR-3 SCA-3 EFS – MCM476 Table 156: Summary of goodness-of-fit data for SCHOLAR-3 SCA-3 OS – standard parametric and spline models ................................................................................. 478 Table 157: Summary of goodness-of-fit data for SCHOLAR-3 SCA-3 OS – MCM. 478 Table 158: Summary of goodness-of-fit data for ponatinib PFS – standard parametric and spline models .................................................................................................. 480 Table 159: Summary of goodness-of-fit data for ponatinib PFS – MCM................. 481 Table 160: Summary of goodness-of-fit data for ponatinib OS – standard parametric and spline models .................................................................................................. 483 Table 161: Summary of goodness-of-fit data for ponatinib OS – MCM .................. 483

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figures

Figure 1: Overview of the CAR T-cell Administration Process .................................. 18 Figure 2: Enrichment and activation of T-cells via the XLP[TM ] or CLP process.......... 19 Figure 3: ALL sub-classifications .............................................................................. 22 Figure 4: 5-year relative survival of leukaemia types, stratified by age at diagnosis 23 Figure 5: ALL new cases and deaths, stratified by age group per year (2015–2017, UK) ........................................................................................................................... 24 Figure 6: Treatment algorithm for R/R adult B-cell ALL ............................................ 28 Figure 7: Proposed positioning of KTE-X19 in the adult ALL treatment pathway ..... 31 Figure 8: ZUMA-3 Phase 1 study design and dosing ............................................... 39 Figure 9: Subject Treatment Schema (Phase 1 and Phase 2) .................................. 41 Figure 10: Patient cohorts of ZUMA-3 ...................................................................... 47 Figure 11: Kaplan-Meier plot of OS (Phase 1 + 2 combined: data cut 23/07/21) ...... 59 Figure 12: Kaplan-Meier plot of OS for OCR subjects using investigator review by subsequent allogeneic SCT group (Combined Phase 1 + 2: data cut 23/07/21) ...... 60 Figure 13: Kaplan-Meier plot of DOR (Phase 1 + 2 combined: data cut 23/07/21) ... 61 Figure 14: Kaplan-Meier plot of RFS (Phase 1 + 2 combined; data cut 23/07/21) .... 62 Figure 15: Kaplan-Meier plot of DOR per investigator assessment (Phase 1 + 2 combined) ................................................................................................................ 65 Figure 16: Kaplan-Meier plot of OS (Phase 1 + 2 combined) ................................... 66 Figure 17: Kaplan-Meier plot of RFS per investigator assessment (Phase 1 + 2 combined) ................................................................................................................ 68 Figure 18: Kaplan-Meier plot of DOR per central assessment (Phase 2, mITT) ....... 73 Figure 19: Kaplan-Meier plot of overall survival (Phase 2, mITT Analysis Set) ........ 75 Figure 20: Kaplan-Meier plot of overall survival: CR versus CRi (Phase 2, mITT: patients with a CR or CRi) ........................................................................................ 76 Figure 21: Kaplan–Meier plot of overall survival: MRD negative versus MRD positive (Phase 2, mITT population) ...................................................................................... 77 Figure 22: Kaplan-Meier plot of OS stratified by subsequent SCT and OCR (Phase 2 mITT CR/CRi; data cut 23/07/21) ............................................................................. 78 Figure 23: Kaplan-Meier plot of RFS by central assessment (Phase 2, mITT) ......... 80 Figure 24: Overall survival for ZUMA-3 Phase 1+2 combined versus INO-VATE inotuzumab ............................................................................................................... 96 Figure 25: Overall survival for ZUMA-3 Phase 1+2 combined versus TOWER ........ 97 Figure 26: Overall survival for ZUMA-3 Phase 1+2 combined versus stacked IPD in INO-VATE and TOWER chemotherapy.................................................................... 98 Figure 27: Event-free survival for ZUMA-3 Phase 1+2 combined versus INO-VATE99 Figure 28: Event-free survival for ZUMA-3 Phase 1+2 combined versus TOWER. 100 Figure 29: Event-free survival for ZUMA-3 Phase 1+2 combined versus stacked IPD in INO-VATE and TOWER ..................................................................................... 101 Figure 30: Overall survival for ZUMA-3 Phase 1+2 combined versus ponatinib (45mg) .................................................................................................................... 102 Figure 31: Kaplan-Meier median OS (ZUMA-3 vs SCA-1) ..................................... 105 Figure 32: Kaplan-Meier median OS (ZUMA-3 vs SCA-3) ..................................... 106 Figure 33: Partitioned survival model structure ...................................................... 138 Figure 34: Modelled* KTE-X19 EFS, cure assumption at 3 years – lognormal ....... 157 Figure 35: Modelled* KTE-X19 OS, cure assumption at 3 years – lognormal ........ 158

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 36: Modelled inotuzumab EFS, cure assumption at 3 years – 1-knot hazard spline ...................................................................................................................... 159 Figure 37: Modelled inotuzumab OS, cure assumption at 3 years – 2-knot normal spline ...................................................................................................................... 160 Figure 38: Modelled FLAG-IDA EFS, cure assumption at 3 years – generalised gamma ................................................................................................................... 161 Figure 39: Modelled FLAG-IDA OS, cure assumption at 3 years – generalised gamma ................................................................................................................... 162 Figure 40: Modelled* KTE-X19 EFS (Ph-), cure assumption at 3 years – lognormal .......................................................................................................................... 163 Figure 41: Modelled* KTE-X19 OS (Ph-), cure assumption at 3 years – lognormal 164 Figure 42: Modelled blinatumomab EFS, cure assumption at 3 years – 1-knot hazard spline ...................................................................................................................... 165 Figure 43: Modelled blinatumomab OS, cure assumption at 3 years – lognormal . 166 Figure 44: Modelled ponatinib PFS, cure assumption at 3 years – lognormal 167 Figure 45: Modelled ponatinib OS, cure assumption at 3 years – lognormal .......... 168 Figure 46: OWSA results, overall population, inotuzumab ..................................... 224 Figure 47: OWSA results, overall population, FLAG-IDA ....................................... 225 Figure 48: OWSA results, Ph- population, blinatumomab ...................................... 226 Figure 49: OWSA results, Ph- population, FLAG-IDA ............................................ 227 Figure 50: OWSA results, Ph- population, inotuzumab .......................................... 228 Figure 51: OWSA results, Ph+ population, ponatinib ............................................. 229 Figure 52: OWSA results, Ph+ population, FLAG-IDA ........................................... 230 Figure 53: OWSA results, Ph+ population, inotuzumab ......................................... 231 Figure 54:Scatter plot, overall population ............................................................... 235 Figure 55: Scatter plot, Ph- population ................................................................... 235 Figure 56: Scatter plot, Ph+ population .................................................................. 235 Figure 57: CEAC, overall population ...................................................................... 236 Figure 58: CEAC, Ph- population ........................................................................... 236 Figure 59: CEAC, Ph+ population .......................................................................... 236 Figure 60: Clinical SLR PRISMA flow chart ............................................................ 333 Figure 61: Patient disposition ZUMA-3 (Phase 1) ................................................... 357 Figure 62: Patient disposition ZUMA-3 (Phase 2) ................................................... 358 Figure 63: Subgroup analyses of OCR rate based on investigator assessment (Phase 1 + 2 combined) ......................................................................................... 363 Figure 64: Subgroup analyses of OCR rate based on central assessment (Phase 2 mITT) ...................................................................................................................... 365 Figure 65: Economic SLR PRISMA flow chart ........................................................ 380 Figure 66: Kaplan-Meier plot of overall survival by dose (Phase 1, treated population) .......................................................................................................................... 433 Figure 67: Extrapolation of KTE-X19 EFS naïve – parametric survival models ...... 454 Figure 68: Extrapolation of KTE-X19 EFS naïve – spline models........................... 454 Figure 69: Extrapolation of KTE-X19 EFS naïve – MCM ........................................ 455 Figure 70: Extrapolation of KTE-X19 OS naïve – parametric survival models ........ 456 Figure 71: Extrapolation of KTE-X19 OS naïve – spline models ............................ 457 Figure 72: Extrapolation of KTE-X19 OS naïve – MCM ......................................... 457 Figure 73: Extrapolation of inotuzumab EFS – parametric survival models ............ 459 Figure 74: Extrapolation of inotuzumab EFS – spline models ................................ 459 Figure 75: Extrapolation of inotuzumab EFS – MCM ............................................. 460

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 76: Extrapolation of inotuzumab OS – parametric survival models ............. 461 Figure 77: Extrapolation of inotuzumab OS – spline models .................................. 462 Figure 78: Extrapolation of inotuzumab OS – MCM ............................................... 462 Figure 79: Extrapolation of FLAG-IDA EFS – parametric survival models .............. 464 Figure 80: Extrapolation of FLAG-IDA EFS – spline models .................................. 464 Figure 81: Extrapolation of FLAG-IDA EFS – MCM ............................................... 465 Figure 82: Extrapolation of FLAG-IDA OS – parametric survival models................ 466 Figure 83: Extrapolation of FLAG-IDA OS – spline models .................................... 467 Figure 84: Extrapolation of FLAG-IDA OS – MCM ................................................. 467 Figure 85: Extrapolation of KTE-X19 EFS naïve, Ph- population – parametric survival models .................................................................................................................... 469 Figure 86: Extrapolation of KTE-X19 EFS naïve, Ph- population – spline models . 469 Figure 87: Extrapolation of KTE-X19 EFS naïve, Ph- population – MCM ............... 470 Figure 88: Extrapolation of KTE-X19 OS naïve, Ph- population – parametric survival models .................................................................................................................... 471 Figure 89: Extrapolation of KTE-X19 OS naïve, Ph- population – spline models ... 472 Figure 90: Extrapolation of KTE-X19 OS naïve, Ph- population – MCM ................. 472 Figure 91: Extrapolation of blinatumomab EFS – parametric survival models ........ 474 Figure 92: Extrapolation of blinatumomab EFS – spline models ............................ 474 Figure 93: Extrapolation of blinatumomab EFS – MCM ......................................... 475 Figure 94: Extrapolation of blinatumomab OS – parametric survival models .......... 476 Figure 95: Extrapolation of blinatumomab OS – spline models .............................. 477 Figure 96: Extrapolation of blinatumomab OS – MCM ........................................... 477 Figure 97: Extrapolation of ponatinib PFS – parametric survival models ................ 479 Figure 98: Extrapolation of ponatinib PFS – spline models .................................... 479 Figure 99: Extrapolation of ponatinib PFS – MCM ................................................. 480 Figure 100: Extrapolation of ponatinib OS – parametric survival models ............... 481 Figure 101: Extrapolation of ponatinib OS – spline models .................................... 482 Figure 102: Extrapolation of ponatinib OS – MCM ................................................. 482

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

List of abbreviations

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.1 Decision problem, description of the technology and clinical care pathway

B.1.1 Decision problem

This submission appraises the clinical and cost-effectiveness of KTE-X19 within its anticipated European Medicines Agency (EMA) marketing authorisation, namely,

. This submission is consistent with the NICE final scope and the NICE reference case.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 1: The decision problem

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 14 of 270

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 15 of 270

Key: ALL, acute lymphoblastic leukaemia; FLAG-IDA, Fludarabine, cytarabine, granulocyte-colony stimulating factor, idarubicin; Ph-, Philadelphia chromosome-negative; Ph+, Philadelphia chromosome-positive; SCT, stem cell transplant.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 16 of 270

B.1.2 Description of the technology being appraised

The draft Summary of Product Characteristics (SmPC) is presented in Appendix C.

KTE-X19 is a chimeric antigen receptor (CAR) T-cell therapy directed against CD19 – a B-cell-specific cell surface antigen that is expressed in most B-cell malignancies, including acute lymphoblastic leukaemia (ALL). KTE-X19 is manufactured from patients’ own T-cells, which are engineered ex-vivo to target the CD19-expressing tumour cells when they are returned to the patient. Figure 1 depicts the steps involved in the manufacturing and administration of CAR T-cell therapy.

Figure 1: Overview of the CAR T-cell Administration Process

==> picture [443 x 110] intentionally omitted <==

Key: CAR, Chimeric antigen receptor Source : adapted from (1).

The KTE-X19 CAR construct and mode of action is displayed in Table 2. Following engagement of KTE-X19 with CD19-expressing target cells, the CD3ζ domain activates the downstream signalling cascade that leads to T-cell activation, proliferation, and acquisition of effector functions, such as cytotoxicity. The intracellular signalling domain of CD28 provides a co-stimulatory signal that works in concert with the primary CD3ζ signal to augment the T-cell function, including IL-2 production (2). Together, these signals result in proliferation of KTE-X19 CAR T-cells and direct killing of target cells. Furthermore, the activated T-cells secrete cytokines and other molecules that can recruit and activate additional anti-tumour immune cells (3).

Unique to the production of KTE-X19 within the Kite CAR T-cell product franchise are the T-cell enrichment and activation stages within the manufacturing process, which is internally referred to as the XLP[TM ] process compared with the CLP process

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

used to manufacture axicabtagene ciloleucel (Yescarta[®] ). Differences between the two are depicted in Figure 2.

Figure 2: Enrichment and activation of T-cells via the XLP[TM] or CLP process

==> picture [426 x 127] intentionally omitted <==

Key: IL-2, interleukin-2; PBMC, peripheral blood mononuclear cell.

The XLP[TM ] process was introduced to minimise hypothetical product quality issues (and is the process used to manufacture KTE-X19 in the ZUMA-3 study). By positively selecting CD4+ and CD8+ cells during enrichment and activating the enriched T-cells with exogenous antibodies, circulating tumour cells are removed from the leukapheresis product, eliminating the risk of premature activation and exhaustion of CAR T-cells during the ex-vivo expansion step of the manufacturing process (which can occur if tumour cells are present in the leukapheresis product).

Table 2 provides summary information regarding the KTE-X19 technology. The draft SmPC is provided in Appendix C.

Table 2: Technology being appraised

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

• The CD3ζ signalling domain, working together with the CD28 domain to aid T-cell activation

• The KTE-X19 CAR construct and mode of action is depicted below.

==> picture [354 x 296] intentionally omitted <==

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Key: ALL, acute lymphoblastic leukaemia; CAR, chimeric antigen receptor; CHMP, Committee for Medicinal Products for Human Use; EMA, European Medicines Agency; MCL, Mantle cell lymphoma; PASLU, Patient Access Schemes Liaison Unit; R/R, relapsed/refractory

B.1.3 Health condition and position of the technology in the

treatment pathway

B.1.3.1 Disease overview

ALL is a rare haematological malignancy characterised by the abnormal proliferation and accumulation of lymphoblasts, and represents approximately 10% of all leukaemia cases (4,5).

Lymphoblasts are immature cells that normally differentiate into white blood cells (WBCs) including B lymphocytes (B-cells) and T lymphocytes (T-cells). In ALL, there is an accumulation of malignant, poorly differentiated lymphoblasts in the bone marrow, blood and extramedullary sites such as the lymph nodes, liver, spleen and central nervous system (CNS) (4).

ALL occurs in a bimodal age distribution and is most commonly diagnosed in people younger than 20 years of age; people over 20 years of age account for approximately 40% of ALL cases in the UK based on data from 2015-2017 (6).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Adult ALL cases normally develop from progenitors of the B-cell lineage with 82% of adults with ALL diagnosed with B-cell ALL in the UK; 87% of these are B-precursor cell ALL (Figure 3).

Figure 3: ALL sub-classifications

==> picture [446 x 281] intentionally omitted <==

Key: ALL, acute lymphoblastic leukaemia a Based on Cancer Research UK data 2015-2017 (6). b Calculated based on UK age-specific ALL incidence data reported by Cancer Research UK (2015-2017) (6). c Weighted average based on i) UK population-based cytogenetic study of 349 patients (≥15 years of age) with ALL diagnosed between 1983-2001 (7) ii) analysis of cytogenetic data from 1522 patients (15-65 years of age) with ALL encoded on the MRC UKALLXII/ECOG 2993 study (8).

d Based on data from a UK population-based cytogenetic study of 349 patients (≥15 years of age) with ALL diagnosed between 1983-2001. Of 240 B-cell ALL cases, 208 were precursor B-cell (7).

e Based on UKALLV12/ECOG data including 1508 newly diagnosed ALL patients (15-60 years of age). 136 patients were refractory to induction therapy, with a further 609 patients relapsing after achieving a remission (9).

The clinical presentation of patients with ALL can be non-specific, involving a

combination of constitutional symptoms and signs of bone marrow failure (anaemia, leukopenia and thrombocytopenia) (4). Many patients are diagnosed after an

emergency admission with symptoms that have developed quickly (10). Common B- precursor ALL symptoms include fever, weight loss and night sweats (collectively known as ‘B symptoms’), easy bleeding or bruising, fatigue, dyspnoea, dizziness, weakness, joint or bone pain, and frequent infection (11).

ALL cells are fast growing (hence the ‘acute’ nomenclature), and the disease has an aggressive course; leukaemic cells can quickly accumulate and if left untreated, ALL Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

would cause death within a few weeks or months (4,12). This aggressive disease results in low survival rates for ALL relative to chronic leukaemia types, particularly in older populations. Age at diagnosis has a striking impact on prognosis and most deaths occur in adults. Five-year survival rates in patients diagnosed before they reach 45 years of age are as high as 81%, whereas in patients diagnosed at or over 65 years of age, the 5-year survival rate is approximately 18% (Figure 4).

Whilst survival rates in children and younger adults have dramatically improved over time, older adults have not seen that same improvement. Pulte et al., (2009) found 5- year survival in 15–19-year-olds had increased from 41.0% to 61.1% between 19801984 and 2000-2004. Conversely, during that same time period, 5-year survival in adults ≥60 years of age only increased by 4.3%, from 8.4% to 12.7% (13).

Figure 4: 5-year relative survival of leukaemia types, stratified by age at diagnosis

==> picture [439 x 153] intentionally omitted <==

Key: ALL, acute lymphoblastic leukaemia; AML, acute myeloid leukaemia; CLL, chronic lymphocytic leukaemia; CML, chronic myelogenous leukaemia; SEER, Surveillance, Epidemiology, and End Results Program. The SEER databased reports survival rates between 2010 to 2016. Source: (14).

This reduced survival expectation is primarily driven by higher relapse rates following initial treatment for ALL in the adult population with cure rates estimated at 20‒40% for adult ALL compared with over 80% for paediatric ALL (15–17). Several factors are thought to feed into the higher relapse rates in adult ALL, including better tolerance of younger patients to more aggressive first-line treatment approaches (chemotherapy-based myeloablative treatments) (16–18). The reduced survival expectation is even further pronounced in older adults, primarily due to a prevalence of adverse-risk disease biology, comorbidities, and reduced tolerance that may

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

preclude intensive curative modalities, as well as increased prevalence of poor-risk disease factors in older patients (17).

One of the most prominent ALL subtypes in adults is Philadelphia-chromosomepositive (Ph+) ALL, an abnormality resulting from a t(9;22) (q34;q11) translocation that results in a BCR-ABL1 fusion gene. While the Ph+ genetic abnormality is rare in children, frequency increases with age, and it is the most common single genetic mutation in adult ALL (17). Ph+ ALL has historically been associated with poor disease prognosis, and is still considered a poor risk cytogenetic group despite the introduction of targeted treatment with tyrosine kinase inhibitors (TKIs) (16,19).

Summary survival data reported by the National Health Service (NHS) England estimate 5-year survival rates of 40% in adult ALL patients aged 25 to 64, reducing to 15% in adult ALL patients aged 65 or older (20). These data are depicted alongside ALL incidence data from Cancer Research UK in Figure 5.

Figure 5: ALL new cases and deaths, stratified by age group per year (2015– 2017, UK)

==> picture [431 x 187] intentionally omitted <==

Key : ALL, acute lymphoblastic leukaemia Source : (6,20).

B.1.3.2 Outcomes for adult ALL patients

The core goal of ALL therapy, as with any life-threatening disease, is to extend patients life expectancy while preserving their quality of life (QoL) and minimising the toxicity of treatment. This remains particularly exigent for the adult ALL population,

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

for whom survival outcomes are considerably worse than for the paediatric population (Figure 4).

Alongside survival extension, complete remission (CR) and complete remission with incomplete haematological recovery (CRi) provide a well-defined metric of disease control that can be used to monitor and determine response to treatment in patient management as well as in a clinical trial setting. Minimal residual disease (MRD) is also an important outcome measure considering its proven link to prognosis and patient QoL (21). These key treatment objectives are defined in Table 3.

Table 3: Definitions of key treatment objectives in ALL

Notes: a) the definition of CRi does vary across clinical trials. b) Blinatumomab NICE reimbursement criteria in Philadelphiachromosome negative adult ALL requires minimal residual disease of at least 0.1%. Source : (19,22,23).

Adult ALL has historically had a dismal prognosis, with limited treatment options and cure rates less than 40%, even with 1[st ] line treatment (17). Despite a high rate of response to first-line induction chemotherapy, only 30–40% of adult patients with ALL will achieve long-term remission (15). Following relapse to front-line therapy, prognosis is poor with most R/R adult ALL patients unlikely to live beyond a year (24). For example, median OS in the pivotal trials of blinatumomab and inotuzumab ozogamicin (hereafter inotuzumab) was 7.7 months (25,26). Considering the average age of adult R/R ALL adult patients is 40‒50 years, this disease is starkly reducing peoples life expectancy (25,26).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.1.3.3 Burden of disease

Due to the aggressive nature of R/R ALL combined with side effects of current treatments, adults with R/R ALL have a reduced QoL compared with both the general population and patients with other types of cancer (4).

In a comprehensive cancer centre survey conducted in Canada, adults with ALL reported the lowest QoL of all adult cancer patients surveyed, with a mean EQ-5D score of 0.70 and a mean visual analogue scale (VAS) score of 66.7, compared with EU5 population norms of 0.86-0.92 and 75-83 respectively (27,28).

B.1.3.4 Clinical Care Pathway

Formal treatment guidelines used to inform the most appropriate management of both newly diagnosed ALL and R/R ALL in Europe come from the European Society for Medical Oncology (ESMO) guidelines (19). Whilst NICE do not provide full guidelines on the treatment of ALL, they do provide a pathway for the treatment of lymphoid leukaemias (29).

B.1.3.4.1 ESMO Guidelines

Since the release of the ESMO guidelines in 2016, several new targeted therapies have been approved for use in the EU. The targeted therapy inotuzumab gained European approval for the treatment of adult ALL in June 2017 (30). Blinatumomab, another targeted therapy, gained approval in Europe in November 2015 (31). While blinatumomab was approved before the publication of the ESMO guidelines, the approval occurred during the development of the article; as such, it was listed as ‘under investigation’.

As described in section B.1.3.2, the goal of treatment is to induce CR whilst limiting toxicity of treatment, as both of these factors are correlated with overall survival (OS) (32). First line treatment begins with an induction phase, which generally consists of pegylated asparaginase (PEG-Asp) in combination with antineoplastic chemotherapy. Eligible patients with a suitable donor may receive a potentially curative stem cell transplant (SCT) at this stage. However, limitations with induction therapies (that is, the low CR rates) subsequently restrict the use of allo (allogeneic) -SCT, the main potentially curative treatment option available to adults with R/R ALL

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

in current practice. Of note is that in the UK, most patients are entered into clinical trials in first line, with the aim of evaluating different treatment approaches (33).

For those patients who do achieve CR with induction therapy, the potential benefits of allo-SCT still need carefully considering alongside the potential risks. ESMO guidelines recommend allo-SCT in first CR for Ph+ patients and all patients with poor early MRD response. However, the guidelines state that use of SCT in first CR is not defined in a satisfactory way and requires continuous update, with this partly due to the improving results with conventional and targeted chemotherapy regimens.

No specific recommendations have been made by the ESMO for the treatment of adult patients with R/R Ph- ALL, other than treatment with allo-SCT. Similarly, no specific recommendations have been made for the treatment of R/R Ph+ ALL in adults, except for the use of a different TKI to that given during the induction phase of treatment, preferably 2nd or 3rd generation TKIs. Where patients have relapsed post-allo-SCT or are ineligible/unlikely to achieve allo-SCT, ESMO provides no specific recommendations, perhaps in acknowledgement of the limited options available.

B.1.3.4.2 NICE lymphoid leukaemia treatment pathway

The NICE treatment pathway for lymphoid leukaemia has recently been updated (September 2020), providing an overview of recommendations for first-line treatment, treating complete remission with MRD, and treating R/R ALL in the UK (29).

Pegaspargase is recommended by NICE (TA408) as part of antineoplastic combination therapy for the treatment of newly diagnosed ALL (34). In Ph- patients in first CR with MRD (>0.1%), blinatumomab is recommended as a treatment option (TA589).

The NICE recommendations for the treatment of ‘fit’ R/R adult ALL are in part determined by the presence of the Philadelphia chromosome) as depicted in Figure 6.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 6: Treatment algorithm for R/R adult B-cell ALL

==> picture [451 x 196] intentionally omitted <==

Key: ALL, acute lymphoblastic leukaemia; BSC, best supportive care; CDF, Cancer Drugs Fund; Ph, Philadelphia chromosome; SCT, stem cell transplant

Source : adapted from the NICE lymphoid leukaemia pathway (29).

The reimbursement conditions for the four treatment options approved by NICE for R/R adult ALL are provided in Table 4. Of note is that tisagenlecleucel is recommended only in patients up to the age of 25 years who fulfil specific criteria, whilst blinatumomab is recommended only for Ph- patients and ponatinib is recommended only in specific Ph+ patients. Based on discussion with clinical experts, it is our understanding that ponatinib is not commonly given as a monotherapy in clinical practice in England, but instead as part of combination with chemotherapy (35).

Table 4: NICE treatment guidance for R/R adult ALL

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Key : ALL, acute lymphoblastic leukaemia; CDF, cancer drugs fund; CR, complete remission; NICE, National Institute for Clinical Excellence; PAS, patient access scheme; R/R, relapsed/refractory; TA, technology appraisal; TKI, tyrosine kinase inhibitor

Source : NICE lymphoid leukaemia pathway (29).

B.1.3.4.3 Unmet needs with current treatment

The introduction of novel treatment options such as biological targeted therapies (blinatumomab and inotuzumab) and TKI therapies has improved the prognosis of adult R/R ALL in recent years. However, in their pivotal clinical trials, blinatumomab and inotuzumab only yielded CR rates of 34% and 36%, respectively, and median OS times of 7.7 months (for both treatments) (25,26). Median OS for ponatinib in the pivotal Phase 2 trial (n=32) was 8.0 months (36). As a result, feedback received from UK clinical experts was that none of these options are considered curative, and that long-term outcomes for blinatumomab, inotuzumab, and ponatinib in UK clinical practice are largely contingent on subsequent SCT (35).

However, a significant proportion of patients cannot proceed to transplant, and posttransplant morbidity and mortality remain high, underlining a substantial unmet need. For example, in the pivotal blinatumomab TOWER study, only 24% of subjects in both arms went on to receive allo-SCT. Among these patients, 10/38 in the blinatumomab group (26%) and 3/12 in the chemotherapy group (25%) died during a median follow-up period of 206 and 279 days, respectively (25). In addition, high risk patients in the first line setting will have already received a SCT.

Five-year survival rates in adult ALL patients receiving SCT in the R/R setting remain below 25%, and allo-SCT can result in severe side effects such as graft versus host disease (GvHD), serious infection and veno-occlusive disease (VOD) (9,37,38). In a structured literature review of allo-SCT-related complications in leukaemia, long-term side effects included chronic GvHD (43% at 5 years post SCT), secondary tumour (21% at 20 years post SCT), hypothyroidism (11% at 15 years post-SCT), bronchiolitis obliterans (9.7% at 122 days post-SCT), cardiovascular disease (7.5%

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

at 15 years post-SCT) and avascular necrosis (5.4% at 10 years post-SCT). Not only do such side effects put patients’ lives at risk, but they can also be complicated and costly to manage, with a significant impact on QoL (27,38,39).

Tisagenlecleucel is approved only for the treatment of R/R B-ALL in paediatric and young adult patients up to 25 years of age. In the ELIANA trial, a CR/CRi rate of 83% was achieved. Notably, subgroup analysis in the subjects who were aged ≥ 10 years of age and < 18 years demonstrated a CR/CRi rate of 88% (95% CI: 69%, 97%), whereas the CR/CRi rate declined to 75% (95% CI: 43%, 95%) in subjects ≥ 18 years. The benefit-risk profile of tisagenlecleucel is not well established in the adult population (40).

B.1.3.4.4 Proposed positioning of KTE-X19 in the R/R adult ALL pathway

The proposed positioning of KTE-X19 aligned to the decision problem is displayed schematically in Figure 7.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 7: Proposed positioning of KTE-X19 in the adult ALL treatment pathway

==> picture [699 x 251] intentionally omitted <==

Key: ALL, Acute lymphoblastic leukaemia; BSC, best supportive care; CDF, cancer drug fund; Chemo, chemotherapy; Ph, Philadelphia chromosome; R/R, relapsed/refractory; SCT, stem cell transplant.

Notes: **where ineligible for stem cell transplant.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 30 of 270

Clinical experts felt that KTE-X19 was likely to be positioned in clinical practice for use in adults with R/R B-cell ALL who:

• Have relapsed post-SCT;

• Are ineligible for SCT (on the basis of age, frailty, comorbidities or other criteria);

• Are unlikely to achieve SCT via existing bridging therapies (primary refractory, relapsed within 12 months, failed ≥2 lines of prior therapy).

This is in line with the inclusion criteria of the pivotal trial evaluating KTE-X19 in R/R B-cell ALL: ZUMA-3 (41,42).

The addition of a potentially curative treatment option for these patients for whom allo-SCT is either not an option or not recommended would provide a valuable addition to the treatment armamentarium. The following text provides more detail on the unmet need that exists in the anticipated positioning of KTE-X19 in UK clinical practice.

R/R adult ALL: relapsed post-allo SCT:

As a potentially curative option, allo-SCT has 5-year OS rates of 23% in R/R adult ALL (9). Following relapse to allo-SCT, survival expectations remain poor. In a retrospective analysis of 465 ALL patients from European Group for Blood and Marrow Transplantation (EBMT) centres who had relapsed following allo-SCT, the median survival post-relapse was 5.5 months, and the estimated post-relapse 5-year survival rate was only 8% (with salvage treatments including chemotherapy, cytoreductive therapy, supportive care, donor lymphocyte infusion and second SCT) (43). Based on discussion with clinical experts practicing in England, although a second allo-SCT is permitted in certain circumstances (i.e. where the patient achieved a CR lasting ≥12 months with first SCT), clinicians do not perceive it to be a viable therapeutic option (35). Therefore, patients who relapse post-allo-SCT pose a significant unmet need. Treatment options for these patients include salvage chemotherapy, or blinatumomab/ inotuzumab (if these have not already been tried), outcomes of which are largely contingent on consolidation with SCT.

R/R adult ALL: ineligible for SCT

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

A number of factors affect eligibility for allo-SCT, including donor availability, remission status, depth of remission, and comorbidities (19). The EBMT group previously estimated between 5–30% of adults with R/R ALL would be eligible for allo-SCT consolidation due to the low CR rates observed with salvage chemotherapy (44). Eligibility has increased with the introduction of targeted treatments to the ALL treatment pathway and associated increase in CR rates for adult R/R ALL patients, but still remains low (25,26).

R/R adult ALL: unlikely to achieve SCT

Given the requirement to achieve CR prior to SCT, a number of groups are unlikely to achieve SCT, specifically:

• Adult R/R ALL that has relapsed within 12 months of first remission

The Spanish Programa Español de Tratamiento en Hematologia (PETHEMA) group analysed prognostic factors after first relapse in adult ALL patients enrolled in riskadapted PETHEMA trials (n = 263) Relapse within a year of first remission was associated with a particularly poor prognosis, with a 5-year survival probability of only 1.8% for these patients, compared with 15% for patients relapsing between 1–2 years of first remission and 31% for patients relapsing more than 2 years after first remission (45).

• Adults with primary refractory ALL

Adults with primary refractory ALL have a similarly poor prognosis, with median OS of 4-5 months and only ~30% CR on salvage chemotherapy (24). Primary refractory adult ALL has a 1-year survival rate of only 15% (46).

• Adults with R/R ALL that have failed ≥2 lines of prior therapy

Survival rates for patients with R/R B-ALL 1 year after the second, third, and fourth or higher lines of therapy are 26%, 18%, and 15%, respectively (24). Furthermore, CR rates decline with each subsequent line of treatment (CR rates of ≤ 47% with second ‑ line and higher chemotherapy vs ≤ 21% with third ‑ line and higher

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

chemotherapy), with associated reduction in likelihood of reaching allo-SCT (24,47– 52).

Therefore, the anticipated positioning of KTE-X19 in clinical practice addresses a considerable unmet medical need where no potentially curative option exists with current approved treatments associated with median OS of only 4-8 months (25,26).

B.1.3.5 Summary of Unmet Medical Need

R/R B-ALL in adult patients continues to represent an unmet medical need, as demonstrated by the low response rates and limited outcomes observed in recent studies of novel therapies. Complete response is required in order to receive a potentially curative SCT, which is associated with ~23% OS at 5 years (9,53). As described in section B.1.3.4.4, there are a number of groups within the R/R adult ALL population that have a particularly dismal prognosis. Among those with primary relapsed ALL, patients with short first remissions (< 12 months) have worse outcomes than patients who relapse after a longer first remission (CR rates of 22% vs 41%, respectively) (52). CR rates decline with each subsequent line of treatment ‑ (CR rates of ≤ 47% with second line and higher chemotherapy vs ≤ 21% with third ‑ line and higher chemotherapy) (24,47–52). Furthermore, survival rates for patients with R/R B-ALL 1 year after the second, third, and fourth or higher lines of therapy are 26%, 18%, and 15%, respectively (24).

Novel treatment strategies are also needed for older patients with R/R B-ALL, a population who remain challenging due to the high morbidity and mortality associated with intensive chemotherapy regimens and SCT and many are too frail to withstand these treatments, as well as the increased incidence of high-risk factors such as Ph+ disease among older patients with ALL (24,54).

Collectively, these results highlight the need for additional therapies such as KTE-X19 that can induce more durable responses with potentially long-term survival in adult patients with R/R B-ALL. This is particularly true in those with the highest unmet need, including those who have relapsed post-SCT or are ineligible for SCT, or those with particularly poor prognostic indicators such as primary refractory

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

disease or first relapse within 12 months that make them unlikely to be able to achieve SCT.

B.1.4 Equality considerations

No equality issues related to the use of KTE-X19 have been identified.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.2 Clinical effectiveness

B.2.1 Identification and selection of relevant studies

See appendix D for full details of the process and methods used to identify and select the clinical evidence relevant to the technology being appraised.

A systematic literature review (SLR) was conducted to identify all relevance clinical evidence associated with the decision problem outlined in section B.1.1. Full details are provided in Appendix D.

B.2.2 List of relevant clinical effectiveness evidence

See Appendix D1.1 for full details of the process and methods used to identify and select the clinical evidence relevant to KTE-X19.

One Phase 1/2 trial was identified in the clinical SLR that provides direct clinical evidence for the efficacy and safety of KTE-X19 for the treatment of adult R/R B-cell ALL: ZUMA-3 (NCT02614066) (55). Eight records were retrieved relating to ZUMA-3, including two publications relating to the Phase 1 results, one publication relating to the Phase 2 results, and 5 conference abstracts (Table 96).

ZUMA-3 is an ongoing Phase 1/2, multicentre, open-label study evaluating the safety and efficacy of KTE-X19 in adult subjects with R/R B-ALL (55). Phase 1 results were published by Shah et al., (2021) (41). Phase 2 results are provided by the Shah et al ., (2021) publication in The Lancet (42). The data cut-off for the Phase 2 publication is 9[th ] September 2020, and this is the same as the data cut-off for the clinical study report (CSR) (56). Therefore, where possible, the publicly available Shah et al ., (2021) publications for Phase 1 and Phase 2 published in Blood and The Lancet , respectively, will be the primary sources for information in this section, with data from the CSR used to supplement where deemed appropriate (41,42,56).

Patients from both Phase 1 and Phase 2 will be followed up to 15 years after the last patient received KTE-X19. Preliminary results from the most recent analysis with data cut-off 23/07/21 provide longer-term data on the durability of KTE- X19. Whilst

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

more detail will be made available through the evaluation process, key results of this most recent data cut-off are presented in Section B.2.6.

Table 5: Clinical effectiveness evidence

Key: ALL, acute lymphoblastic leukaemia; N/A, not applicable; R/R, relapsed/refractory. Notes: outcomes in bold are those directly used in the economic modelling.

B.2.3 Summary of methodology of the relevant clinical

effectiveness evidence

B.2.3.1 Trial design

ZUMA-3 is a Phase 1/2 multicentre, open-label study evaluating the safety and efficacy of KTE-X19 in adult subjects with R/R B-ALL In this study, R/R was defined as 1 of the following:

• Primary refractory

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

• First relapse following a remission lasting ≤12 months

• R/R after second-line or higher therapy

• R/R after allo-SCT (provided the transplant occurred ≥100 days prior to enrolment and that no immunosuppressive medications were taken ≤4 weeks prior to enrolment)

The rationale for these definitions was based on the historically poor outcomes observed in these patient populations (18,24,57,58). In particular, CR rates to salvage treatments have been shown to be lower for patients with first remission durations < 12 months compared with those who relapse after a longer first remission, and decrease with each subsequent line of treatment (24,52).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 8: ZUMA-3 Phase 1 study design and dosing

==> picture [443 x 250] intentionally omitted <==

Key: CAR, chimeric antigen receptor; DLT, dose-limiting toxicity; DOR, duration of remission; MRD, minimal residual disease; RFS, relapse-free survival. Source: (41) .

During Phase 1, approximately 3 to 12 subjects with high burden R/R B-ALL disease (defined as > 25% leukaemia blasts in the bone marrow [M3 marrow] or ≥ 1,000 blasts/mm[3 ] in the peripheral blood) who were evaluable for dose limiting toxicities (DLTs) were to be assessed to evaluate the safety of KTE-X19, with rate of DLTs within 28 days the primary endpoint (41). Additionally, around 40 subjects with high or low disease burden R/R B-cell ALL were enrolled to further assess safety and were also evaluated for secondary efficacy endpoints. A Safety Review Team (SRT), which comprised representatives of the sponsor together with at least 1 study investigator, was to review safety data and make recommendations regarding further enrolment in Phase 1 or proceeding to Phase 2 based on the incidence of DLTs and overall safety profile of KTE-X19 (56) (Figure 8).

The initial dose of KTE-X19 investigated in Phase 1 was 2 x 10[6 ] anti-CD19 CAR+ T- cells/kg based on the DLT, and maximum tolerated dose observed in first use studies of KTE-X19 at the National Cancer Institute. Other doses explored in Phase 1 were 0.5 x 10[6 ] anti-CD19 CAR+ T-cells/kg and 1 x 10[6 ] anti-CD19 CAR+ T-cells/kg.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

An analysis of Phase 1 was conducted when 41 subjects treated with KTE-X19 had had the opportunity to be followed for at least 2 months. The results of this analysis were reported in the ZUMA-3 End-of-Phase 1 Summary and are presented in B.2.6 (41). On the basis of these results the SRT recommended initiating the Phase 2 portion of the study at the target dose of 1x10[6 ] anti-CD19 CAR+ T-cells/kg dose.

During Phase 2, approximately 50 subjects in the modified intent-to-treat (mITT) analysis set were to be assessed to evaluate the efficacy and safety of KTE-X19 at the target dose (1x10[6 ] anti-CD19 CAR T-cells/kg). The mITT analysis set was defined as all subjects enrolled and treated with KTE-X19 in Phase 2.

In total, 125 subjects were enrolled and treated with KTE-X19 in phases 1 and 2 combined. The primary analysis was to occur when the overall study enrolment had been completed and the last treated subject in the mITT analysis set had the opportunity to complete the Month 6 disease assessment. At the time of the data cutoff date for the primary analysis (09/09/20), all subjects in the mITT analysis set had had the opportunity to be followed for at least 10 months after the KTE-X19 infusion (56).

In the Phase 2 portion, adult patients with R/R cell ALL who met the criteria listed in Table 6 were enrolled and treated with KTE-X19 at a target dose of 1 x 10[6 ] antiCD19 CAR+ T cells/kg (hereafter referred to as target dose). Phase 2 was designed to evaluate the efficacy and safety of KTE-X19 at target dose. First remission assessment was conducted at Day 28 but patients from both Phase 1 and Phase 2 will be followed up to 15 years after the last patient received KTE-X19. On this basis, the final study completion date is estimated to be September 2035 (55).

Once deemed eligible and enrolled into the study, subjects in both phases were to follow the same treatment schedule (Figure 9) and procedural requirements and proceed through the following study periods:

• Screening

• Enrolment/leukapheresis

• Bridging chemotherapy and cerebrospinal fluid (CSF) prophylaxis

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

• Lymphodepleting chemotherapy

• KTE-X19 treatment

• Post-treatment assessment

• Long-term follow-up

Figure 9: Subject Treatment Schema (Phase 1 and Phase 2)

==> picture [452 x 171] intentionally omitted <==

Key : CSF, cerebrospinal fluid; IP, investigational product; IV, intravenous. Source: Adapted from ZUMA-3 CSR (56).

a) CSF prophylaxis (administered any time during screening through 7 days prior to KTE-X19 infusion): All subjects were to receive CSF prophylaxis consisting of an intrathecal regimen according to institutional or national guidelines. CSF prophylaxis could be administered with the screening lumbar puncture.

b) Bridging chemotherapy (administered after leukapheresis and completed at least 7 days or 5 half-lives, whichever was shorter, prior to initiating lymphodepleting chemotherapy): Bridging chemotherapy was recommended for all subjects, particularly those subjects with high disease burden at screening (M3 marrow [> 25% leukemic blasts] or ≥ 1,000 blasts/mm3 in the peripheral circulation).

B.2.3.2 Trial methodology

Table 6: Summary of trial methodology for ZUMA-3

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

immunosuppressive medications for at least 4 weeks prior to enrolment

2. Morphological disease in the bone marrow (>5% blasts) 3. Subjects with Ph+ disease were eligible if they were intolerant to TKI therapy or if they had R/R disease despite treatment with at least 2 different TKIs 4. Aged 18 years or older 5. ECOG performance status of 0 or 1

3. Absolute neutrophil count ≥ 500/µL unless, in the opinion of the principal investigator, cytopenia was due to underlying leukaemia and was potentially reversible with leukaemia therapy

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

• Relapsed/refractory post SCT

• • Relapsed/refractory after ≥2 lines of prior therapy

Key: AE, adverse event; ALL, acute lymphoblastic leukaemia; AST, aspartate aminotransferase; ALT, alanine aminotransferase; CNS, central nervous system; CR, complete remission; CRi, complete remission with incomplete haematological recovery; CRS, cytokine release syndrome CT, computed tomography; CTCAE, Common Terminology Criteria for Adverse Events DOMP, dexamethasone, 6-mercaptopurine, methotrexate, and vincristine; ECOG, Eastern Cooperative Oncology Group; FLAG-IDA, fludarabine, cytarabine, granulocyte-colony stimulating factor; GVHD, graft-versus-host disease; IL, interleukin; IV, intravenous; MRD, minimal residual disease; Ph, Philadelphia chromosome; R/R, relapsed/refractory; SCT, stem cell transplant; TKI, tyrosine kinase inhibitor; VAD, vincristine, doxorubicin, and dexamethasone; WBC, white blood cell. Note: for a full list of eligibility criteria please refer to the supplementary materials of Source: Shah et al., (2021) (42); ZUMA-3 CSR (56).

B.2.3.3 Patient datasets and baseline characteristics

Whilst there are multiple datasets within the ZUMA-3 Phase 1/2 trial (Figure 10) the clinical effectiveness section focuses on the Phase 1 + 2 combined dataset, defined as all subjects in ZUMA-3 to receive KTE-X19 at target dose (n=78) (Table 7). Treated patients align to the costing framework proposed for KTE-X19, where only treated patients are paid for by the NHS.

The Phase 2 mITT population – the results of which were published in The Lancet - is also presented to provide further support for the clinical effectiveness of KTE-X19 (42). Data from the Phase 2 intent-to-treat (ITT) population (n=71) is included for baseline characteristics in section B.2.6, with further information available in the CSR (56). Information on the Phase 1 target dose (n=23) population is provided in Appendix L.

As described in Section B.2.10, toxicity management recommendations were revised during Phase 1, with 9 of 23 subjects treated at target dose managed under the revised adverse event (AE) guidelines, which were then carried through to Phase 2.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 10: Patient cohorts of ZUMA-3

==> picture [460 x 284] intentionally omitted <==

Key: CAR T-cell, chimeric antigen receptor T-cell; ITT, intent-to-treat; mITT, modified intent-to-treat. Notes: Datasets highlighted in red are presented in the main body of form B, boxes highlighted in black are presented in the appendix. Other data is available in the clinical study report. Target dose is defined as 1x10[6] anti-CD19 CAR T-cells/kg. Source: Adapted from ZUMA-3 publications (41,42).

Table 7: Summary of ZUMA-3 datasets

Note: Phase 1 + 2 combined is defined as all patients who received KTE-X19 at the target dose of 1 x 10[6] anti-CD19 CAR+ T- cells/kg.

Table 8 presents key baseline characteristics for the Phase 1 + 2 combined dataset. Almost half ( ) of subjects treated had prior treatment with blinatumomab, and

had prior treatment with inotuzumab. Outcomes in the setting of blinatumomab and inotuzumab failure have not been well studied to date, although limited reports indicate that once patients fail blinatumomab, responses to subsequent lines of therapy deteriorate, leaving patients with very limited options (59,60).

In addition, had relapsed post-SCT. As discussed in section B.1.3.2, outcomes

for patients who relapse post-SCT are especially dire. The median survival postCompany evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

relapse is 5.5 months, and the estimated post-relapse 5-year survival rate is only 8%, even with second SCT as a salvage treatment option, which is unlikely to be the case in the UK (35,43). Approximately 1 in 5 subjects ( ) had Ph+ disease, a similar percentage to that reported by Fielding et al ., (2007) in a UK adult R/R ALL population (9).

Table 9 provides a summary of key baseline characteristics for the Phase 2 mITT and ITT datasets, the populations of which are comparable to the combined population. The median age of all treated patients was 40 years (range: 19-84) and 15% were aged 65 years or over. This average age is in line with published data for adult ALL, and given the poor outcomes in this group, emphasises the stark reduction in life expectancy for adults with R/R ALL (24).

Table 8: Baseline demographics and characteristics at baseline (Phase 1 + 2 combined)

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Key : CNS, central nervous system; CR, complete remission; CRh, complete remission with partial hematologic recovery; CRi, complete remission with incomplete hematologic recovery; ECOG, Eastern Cooperative Oncology Group; LVD, longest vertical dimension; MLL, mixed lineage leukaemia; NR, no response; PD, progressive disease; PR, partial remission; SCT, stem cell

transplant; SPD, sum of the products of diameters; STDEV, standard deviation.

Notes : Excludes information collected after retreatment. Baseline is defined as the last assessment prior to the start of conditioning chemotherapy. *a number of these categories are co-incident, hence the groups combined add up to >100%. Source: Table 14.1.4.6, ZUMA-3 CSR (56). Combined results from Phase 2 mITT and Phase 1 target dose.

Table 9: Baseline demographics and characteristics at baseline (Phase 2)

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Key : ALL, acute lymphoblastic leukaemia; allo-SCT, allogenic stem cell transplantation; BM, bone marrow; CNS, central nervous system; ECOG, Eastern Cooperative Oncology Group; ITT, intent-to-treat; mITT, modified intention to treat; MLL, mixed lineage leukaemia; NR, not reported; Ph+, Philadelphia chromosome-positive; R/R, relapse/refractory; SCT, stem cell transplant. Note : Baseline is defined as the last assessment prior to the start of the lymphodepleting chemotherapy. Source : (42).

B.2.4 Statistical analysis and definition of study groups in the

relevant clinical effectiveness evidence

B.2.4.1 Analysis population

The Phase 2 mITT analysis set was considered to include all patients who received a dose of KTE-X19; this analysis set was used for the hypothesis testing of the primary endpoint and other efficacy analyses, as well as safety analyses.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

The Phase 2 ITT dataset comprised all enrolled patients. The Phase 1 + 2 combined dataset consisted of all patients treated in Phase 1 and Phase 2 at the recommended Phase 2 dose of KTE-X19 (Table 7) (42).

During Phase 1, the SRT was chartered to review safety data and make recommendations on further study conduct and progression of the study from Phase 1 to Phase 2 based on the incidence of DLTs and serious adverse events (SAEs). The DLT-evaluable cohort included the first 3 patients treated at the 2 x 10[6 ] antiCD19 CAR T cells/kg dose level (41).

B.2.4.2 Sample size

ZUMA-3 used a single-arm design to test for an improvement in overall complete remission (OCR) (defined as achieving CR/CRi) rate. A sample size of 50 subjects in Phase 2 was to provide approximately 93% power to distinguish between an active therapy with a 65% true OCR rate from a therapy with a response rate of ≤ 40%, with a 1-sided alpha level of 0.025.

The rationale for a prespecified 40% OCR historical control rate was informed by rates observed in published studies of second-line or later chemotherapy and SCT regimens and in pivotal studies of blinatumomab. The blinatumomab studies, which included subject populations similar to those who were to be enrolled in ZUMA-3, resulted in CR/complete response with incomplete haematologic recovery (CRh) rates of approximately 42%; the CR rates were 32.4% in the Phase 2 trial (Study MT103-211) and 33.6% in the Phase 3 TOWER study (25,61).

A step-down test of the secondary endpoint of MRD[- ] rate was to be performed against an MRD[- ] rate of 30% only if the testing of the OCR rate reached statistical significance, so that the family-wise type I error would be controlled at a 1-sided 2.5% level under the hierarchical testing scheme (42).

B.2.4.3 Statistical analysis

A summary of the statistical analyses for ZUMA-3 is available in Table 10.

Table 10: Summary of statistical analyses: ZUMA-3

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Key: CR/CRi, complete response/complete response with incomplete haematological recovery; SCT, stem cell transplant. Source: (56).

Primary efficacy endpoint

Per protocol, the primary efficacy analysis was carried out when all KTE-X19-treated patients had completed at least the 6-month disease assessment. An exact binomial test was used to compare the observed rate of CR/CRi to the historical control rate. Two sided 95% CIs were calculated using the Clopper-Pearson method. This test assumed the independence of the individual subject responses. CIs were provided about the CR/CRi rate, as well as the CR rate and CRi rate separately (42).

Secondary efficacy endpoints

Hypothesis testing of the secondary endpoint of MRD[- ] rate was to be performed against an MRD[- ] rate of 30% if the testing of the OCR rate was significant. The control rate was selected based on the MRD[- ] rates of approximately 30% that were observed among all subjects treated with blinatumomab in the pivotal studies (25,62) Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

• MRD[-] remission : rate and 95% CIs estimated for treated subjects, subjects with a CR, subjects with a CRi, and subjects with either a CR or CRi combined

• Duration of remission (DOR) : the primary analysis of DOR used the KaplanMeier (KM) method, considering all relapses and deaths as events for DOR. The reverse Kaplan-Meier approach was to be used to estimate the follow-up time for DOR (63)

• Allo-SCT rate : subject incidence rate of on-study allo-SCT was to be summarised overall, and by subjects achieving CR, CRi, or CR/CRi

• Relapse-free survival : KM plots, estimate of the median RFS, and 2-sided 95% CIs were generated

• Overall survival : KM plots, estimates of median OS, and 2-sided 95% CIs were generated.

Subgroup analyses

The CR/CRi rate with 95% CIs was generated for subgroups of the mITT analysis set defined by the selected covariates listed in Table 6.

Safety analyses

Safety analyses were conducted on the safety analysis set. The primary analysis of safety data summarised all AEs and laboratory values with an onset on or after KTEX19 infusion.

B.2.4.4 Participant flow

Details of participant flow in Phase 1 and Phase 2 of ZUMA-3 in the form of a CONSORT diagram are provided in Appendix D1.2.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.2.5 Quality assessment of the relevant clinical effectiveness evidence

Quality assessment of ZUMA-3 was conducted using the Downs and Black checklist, full details of which are provided in Appendix D1.3.

Within the context of a single-arm study design, the overall risk of bias is considered to be low. The primary endpoint of OCR was determined independently by central assessment and provides an objective assessment of treatment effect that is directly relevant to clinical practice, where response to treatment is the primary measure of effect.

The single-arm nature of ZUMA-3 does however necessitate a need for an indirect treatment comparison (ITC) to provide relative effect estimates required for decision making. Use of ITCs is associated with higher uncertainty compared to a controlled trial, with this discussed in more detail in section B.2.9. In terms of intervention, all patients in the combined Phase 1 + 2 dataset (N = 78) treated with KTE-X19 reflect the administration and dosing of KTE-X19 expected in clinical practice, and that of the anticipated marketing authorisation.

With regard to generalisability to clinical practice in England, ZUMA-3 included subjects who had received a number of prior therapies considered as standard of care (SoC) in the R/R adult ALL treatment pathway. These include of subjects receiving prior blinatumomab, and receiving prior inotuzumab. In addition, the percentage of Ph+ patients were comparable between ZUMA-3 ( ) and UK R/R clinical practice (22%) (section B.2.13).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.2.6 Clinical effectiveness results of the relevant trials

Summary of clinical effectiveness results

• The efficacy and safety of KTE-X19 in the treatment of adults with R/R ALL has been demonstrated in the open-label, multi-centre, ZUMA-3 trial.

• Patients with R/R disease were defined as primary refractory, in first relapse following a remission lasting ≤ 12 months, R/R after second-line or higher therapy, or R/R after allo-SCT.

• At the most recent data cut-off (23/07/21), with median follow-up of months, KTE-X19 demonstrated an unprecedented median OS of months in the Phase 1 + 2 combined dataset.

• KM estimates of OS at 6 and 12 months were and , respectively, with estimated to be alive at 24 months

• The OCR rate per investigator assessment was 74.4%, with 58 of 78 subjects treated with KTE-X19 at target dose achieving OCR. The CR rate was 62.8% (50 of 78 subjects).

• 79.5% (62 of 78 subjects) treated with KTE-X19 achieved MRD negativity, including all but one patient – for whom data was not available - to achieve CR/CRi.

• KTE-X19 induced durable remission in patients achieving OCR, with a median duration of remission of months.

• A sensitivity analysis of median OS stratified by censoring at allo-SCT demonstrate that survival appeared to be independent of subsequent SCT

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 11: Summary of clinical effectiveness: ZUMA-3

Key: DOR, duration of remission, ITT, intent-to-treat; mITT, modified intent-to-treat, MRD, minimal residual disease, OCR, overall complete remission; RFS, relapse-free survival. Notes: ITT includes all patients enrolled to the relevant phase of the study. mITT refers to subjects who received treatment with KTE-X19, or with regard to the Phase 1 portion the subjects who received KTE-X19 at the target dose of 1 x 10[6] CAR T-cells/kg. *, based on data cutoff 09/09/20. **, based on data cutoff 23/07/21.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 55 of 270

KTE-X19 cohorts and analysis sets are summarised in Table 7, including the location within the submission that data is presented.

The primary analysis was planned when the overall study enrolment was complete and the last treated patient in the mITT population had had the opportunity to complete the Month 6 disease assessment. This occurred on 9[th ] September 2020, with a median actual follow-up from KTE-X19 infusion of months in Phase 1 and months in Phase 2 (56).

Preliminary results from the most recent interim analysis with data cutoff 23/07/21 provide longer-term data on the durability of all patients treated with KTE-X19 at target dose. Whilst more detail will be made available through the evaluation process, key results are provided.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.2.6.1 ZUMA-3: (Phase 1 + 2 combined)

B.2.6.1.1 Data cut off 23/07/21:

Initial data from the 23/07/21 data cutoff is presented to provide evidence on the long-term efficacy of KTE-X19 where available. Median actual follow-up for all treated subjects in Phase 1 + 2 at this data cutoff was months (95% CI:

). Further data (including a CSR) will become available during technical engagement.

Overall survival:

Data from the most recent data cutoff of ZUMA-3 demonstrates the durable effect of KTE-X19 on OS (Figure 11). At a median actual follow-up of months (95% CI: ) in all treated subjects, the KM median OS was months (95% CI: ) (Figure 11). Notably, in a sensitivity analysis of median OS stratified by censoring at allo-SCT, survival in responders appeared to be independent of subsequent SCT at the most recent data analysis (Figure 12). KM estimates of OS at 6 months and 12 months were (95% CI: ) and (95% CI: ), respectively (Table 12).

Table 12: Overall survival (Phase 1 + 2 combined, data cut 23/07/21)

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

42 months 48 months 54 months Median (95% CI) follow-up time (months) (reverse KM approach)

==> picture [75 x 49] intentionally omitted <==

Data cutoff date = 23/07/2021.

Key : CI, confidence interval; DCO, data cutoff date; KM, Kaplan-Meier; NE, not estimable; OS, overall survival. Notes : Overall survival for subjects treated with KTE-X19 is defined as the time from KTE-X19 infusion date to the date of death from any cause. '+' indicates censoring. Source: (64).

Figure 11: Kaplan-Meier plot of OS (Phase 1 + 2 combined: data cut 23/07/21)

==> picture [456 x 281] intentionally omitted <==

Key: CI, confidence interval; NE, not estimable. Source: (64).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 12: Kaplan-Meier plot of OS for OCR subjects using investigator review by subsequent allogeneic SCT group (Combined Phase 1 + 2: data cut 23/07/21)

==> picture [458 x 269] intentionally omitted <==

Data cutoff date = 23/07/21.

Key : CI, confidence interval; NE, not evaluable; NR, not reached; OCR, overall complete remission; SCT, stem cell transplant. Source: (64).

Duration of remission

At the most recent data cutoff (23/07/21), among the 58 subjects who achieved a CR or CRi, the KM median duration of remission (DOR) was months (95% CI: ), with a reverse KM median follow-up time for DOR of months (95% CI: ). Overall, subjects were censored: subjects were in ongoing remission as of the data cut off date, 14 subjects had an allo-SCT, subjects started new anticancer therapy, and subject was lost to follow-up. subjects relapsed, and died. The KM estimates of the proportion of responders who remained in remission at 6 and 12 months from first response were (95% CI: ) and (95% CI: ), respectively (Figure 13) (Table 13).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 13: Kaplan-Meier plot of DOR (Phase 1 + 2 combined: data cut 23/07/21)

==> picture [464 x 277] intentionally omitted <==

Data cutoff date: 23/07/21

Key: CI, confidence interval; CR, complete remission; Cri, complete remission with incomplete haematologic recovery; NE, note evaluable.

Source: (64).

Table 13: DOR using investigator review (Phase 1 + 2 combined: data cut 23/07/21)

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

15 months 18 months 24 months Median (95% CI) follow-up time (months) (reverse KM approach)

Data cutoff date = 23/07/21.

Key: CI, confidence interval; CR, complete remission; CRi, complete remission with incomplete hematologic recovery; DOR, duration of response; KM, Kaplan-Meier; mITT, modified intent-to-treat; NE, not estimable; OCR, overall complete remission; SCT, stem cell transplant.

Notes : Investigator review is presented in this table. Percentages are based on number of all dosed subjects in Phase 1 two 1e6 cohorts and Phase 2 with overall complete remission (CR + CRi). DOR is defined as the time from the first complete remission (CR or CRi) to relapse or death from any cause in the absence of documented relapse. '+' indicates censoring. Source: (64).

Relapse-free survival:

KM estimates of relapse-free survival (RFS) rates at 6 and 12 months were (95% CI: ) and (95% CI: ), respectively. The KM median RFS was months (95% CI: ), with a reverse KM median follow-up time for RFS of months (95% CI: ) (Figure 14).

It is important to note that the rate of censoring is high due primarily to patients either being in remission at time of data cut-off or receiving a SCT.

Figure 14: Kaplan-Meier plot of RFS (Phase 1 + 2 combined; data cut 23/07/21)

==> picture [460 x 277] intentionally omitted <==

Data cutoff date: 23/07/21. Key: CI, confidence interval; NE, not estimable. Source: (64).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 14: RFS using investigator review (Phase 1 + 2 combined; data cut 23/07/21)

Data cutoff date = 23/07/21.

Key : CI, confidence interval; CR, complete remission; CRi, complete remission with incomplete hematologic recovery; KM, Kaplan-Meier; RFS, relapse-free survival; SCT, stem cell transplant.

Notes : Percentages are based on the number of subjects in safety analysis set.

Relapse-free survival for subjects who received KTE-X19 is defined as the time from the KTE-X19 infusion date to the date of relapse or death from any cause. Subjects who received KTE-X19 but did not achieve CR or CRi as the best overall response are counted as events on KTE-X19 infusion date. Source: (64).

B.2.6.1.2 Data cut off 09/09/20:

Key efficacy endpoint data for the Phase 1 + Phase 2 combined dataset is presented in Table 15, with results then described in greater detail.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 15: Summary of efficacy endpoints (Phase 1 + 2 combined)

Data cutoff date = 09/09/2020. Key : BFBM, blast-free hypoplastic or aplastic bone marrow; CI, confidence interval; CR, complete remission; CRh, complete remission with partial haematological recovery; CRi, complete remission with incomplete haematological recovery; DOR, duration of remission; OCR, overall complete remission; OS, overall survival; RFS, relapse-free survival. Source : Table S12 (42).

OCR rate by investigator assessment:

Among the 78 subjects who were treated with the target dose across Phase 1 + 2, the OCR rate per investigator assessment was 74.4% (58 of 78 subjects; 95% CI: ), with a CR rate of 62.8% (49 of 78 subjects, 95% CI: ) (56).

DOR by investigator assessment:

Among the 58 subjects who achieved a CR or CRi, the KM median DOR was 13.4 months (95% CI: ), with a reverse KM median follow-up time for DOR of months (95% CI: ). Overall, subjects were censored: 19 subjects were in ongoing remission as of data cutoff ( of those patients with CR/CRi), 13 had an allo-SCT, started a new anti-cancer therapy, and was lost to follow-up. subjects relapsed, and died. The KM estimates of the proportion of responders who remained in remission at 6 and 12 months from first response were (95% CI: ) and (95% CI: ), respectively (56).

The KM median DOR was months (95% CI: ) for subjects with CR, and months (95% CI: ) for subjects with CRi (56).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 15: Kaplan-Meier plot of DOR per investigator assessment (Phase 1 + 2 combined)

==> picture [451 x 266] intentionally omitted <==

Data cutoff date = 09/09/2020.

Key: CI, confidence interval; CR, complete remission; Cri, complete remission with incomplete haematological recovery; NE, not estimable. Source: ZUMA-3 clinical study report Figure 14.2.8.7 (56).

OS:

KM estimates of OS at 12 and 18 months were (95% CI: ) and

(95% CI: ), respectively. The KM median OS was 22.4 months (95% CI: 15.9 months, NE), with a reverse KM median follow-up time for OS at months (95% CI: ) (Table 16).

Table 16: Overall survival (Phase 1 + 2 combined)

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Survival free rates (%) (95% CI) by KM estimation at 3 months 6 months 9 months 12 months 15 months 18 months 24 months Reverse KM median follow-up time for OS, months (95% CI)

Data cutoff date = 09/09/2020.

Key : CI, confidence interval; DCO, data cutoff date; KM, Kaplan-Meier; Max, maximum; Min, minimum; NE, not estimable; OS, overall survival.

Notes : 1e6 = 1 x 10[6] anti-CD19 CAR T cells/kg. OS for subjects treated with KTE-X19 is defined as the time from KTE-X19 infusion date to the date of death from any cause. Subjects who had not died by the analysis data cutoff date were censored at their last contact date prior to the data cutoff date, with the exception that subjects known to be alive or determined to ha ve died after the data cutoff date were censored at the data cutoff date. '+' indicates censoring. Source : ZUMA-3 clinical study report Table 14.2.7.5 (56).

Figure 16: Kaplan-Meier plot of OS (Phase 1 + 2 combined)

==> picture [468 x 275] intentionally omitted <==

Data cutoff date = 09/09/2020. Key: CI, confidence interval; NE, not estimable; OS, overall survival. Source: ZUMA-3 clinical study report Figure 15 (56).

MRD:

79.5% (62 of 78 subjects) treated with KTE-X19 achieved MRD negativity, including

all but one patient – for whom data was not available - to achieve CR/CRi.

RFS by investigator assessment:

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

KM estimates of RFS rates at 6 and 12 months were

(95% CI:

)

and (95% CI: ), respectively. The KM median RFS was 10.3 months (95% CI: 5.6, 14.4 months), with a reverse KM median follow-up time for RFS of months (95% CI: ) (56).

It is important to note that the rate of censoring is high due primarily to patients either being in remission at time of data cut-off or receiving a SCT.

Table 17: RFS per investigator assessment (Phase 1 + 2 combined)

==> picture [449 x 338] intentionally omitted <==

----- Start of picture text -----
RFS Phase 1 + 2 combined (n=78)
Number of subjects, n 78
Events, n (%)
Censored, n (%)
KM median (95% CI) RFS (months) 10.3 (5.6, 14.4)
Min, Max RFS (months)
Events
Relapse, n (%)
Death, n (%)
Subject's best overall response not CR or CRi, n (%)
Censoring reason
Ongoing remission, n (%)
Allogeneic SCT, n (%)
Started new anti-cancer therapy, n (%)
Lost to follow up, n (%)
Withdrawal of consent, n (%)
Event-free rates % (95% CI) by KM estimation at
3 months
6 months
9 months
12 months
Median (95% CI) follow-up time (months) (reverse KM
approach)
----- End of picture text -----

Data cutoff date = 09/09/2020.

Key : CI, confidence interval; CR, complete remission; CRi, complete remission with incomplete hematologic recovery;KM, Kaplan-Meier; Max, maximum; Min, minimum; NE, not estimable; RFS, relapse-free survival; SCT, stem cell transplant. Notes: 1e6 = Percentages are based on the number of subjects in safety analysis set. RFS for subjects who received KTEX19 is defined as the time from the KTE-X19 infusion date to the date of relapse or death from anycause. Subjects who received KTE-X19 but did not achieve CR or CRi as the best overall response are counted as events on theKTE-X19 infusion date. '+' indicates censoring. Source : Table 32 (56)

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 17: Kaplan-Meier plot of RFS per investigator assessment (Phase 1 + 2 combined)

==> picture [465 x 285] intentionally omitted <==

Data cutoff date = 09/09/2020. Source: ZUMA-3 clinical study report Figure 16 (56).

Rate of allo-SCT

In total, 17.9% (14 of 78 subjects) of the Phase 1 + 2 combined dataset went on to receive an allo-SCT during ZUMA-3 at the investigators discretion. This included 18.1% (10 of 55 subjects) treated with KTE-X19 at Phase 2, and 17.4% (4 of 23 subjects) treated with KTE-X19 at target dose at Phase 1.

Of those to receive a subsequent allo-SCT; 9 subjects had achieved CR, and 3 subjects had achieved a CRi with KTE-X19 treatment. One subject treated at Phase 2 had inconsistent assessment between investigator and central assessment (CRi by investigator assessment vs. blast-free hypoplastic/aplastic bone marrow by central assessment) (56). A further subject treated at Phase 1 (who had extramedullary disease at baseline) had achieved a best overall response of PR to KTE-X19 treatment based on the investigator assessment of disease response.

B.2.6.2 ZUMA-3 Phase 2 mITT

Primary efficacy endpoint: OCR

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

The primary efficacy endpoint was met, with an OCR rate of 70.9% (39 of 55 subjects) in the phase 2 mITT population; significantly greater than the prespecified historical control rate of 40% (p < 0.0001) (95% CI: 57%, 82%). Among the 70.9% who achieved a CR or CRi, the median time to response was 1.1 months (range: months). A summary of OCR and best overall response per central assessment for the mITT population is provided in Table 18.

Table 18: Summary of overall complete response rates (Phase 2, mITT)

Key: BFBM, blast-free hypoplastic or aplastic bone marrow; CI, confidence interval; CR, complete remission; CRh, complete remission with partial haematological recovery; CRi, complete remission with incomplete haematological recovery; mITT, modified intent-to-treat; NR, no response; OCR, overall complete remission; PR, partial remission. Notes: 95% confidence interval is based on Clopper-Pearson method. Data cutoff date = 09/09/20. Source: ZUMA-3 clinical study report Table 14.2.1.1 and Table 14.2.2.1 (56). Shah et al (2021) (42).

Secondary efficacy endpoints

Minimal residual disease:

A summary of MRD negative status as determined by the central laboratory for the mITT population is provided in Table 19.

The overall MRD negative rate for patients treated in Phase 2 (mITT) was 76% (42 of 55 patients; 95% CI: ), significantly higher than the prespecified control rate of 30%, therefore the secondary efficacy endpoint was met (p < 0.0001). MRD[- ] rate increased to 97% in patents achieving CR or CRi (38 of 39 patients; 95% CI:

), with 1 subject who achieved CR not having samples available for MRD Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

assessment (42). In total, 30 subjects with CR, 8 subjects with CRi, and 4 subjects with blast-free hypoplastic or aplastic bone marrow (BFBM), hence the higher MRD negative rate compared to OCR rate.

The survival advantage of achieving MRD negativity in both adults and children has been demonstrated by Berry et al ., (2017) in a meta ‑ analysis of 39 studies (albeit following induction therapy), and was further re-enforced by recent long-term blinatumomab data (65,66). This supports the clinical value of every evaluable subject to achieve CR/CRi achieving MRD[- ] remission in the ZUMA-3 mITT population.

Of the 13 subjects who were not considered MRD[- ] overall, 9 subjects were nonresponders, 3 subjects were not evaluable for disease response, and as mentioned above 1 subject with a CR did not have MRD assessments performed (56).

Table 19: Summary of MRD status (Phase 2, mITT)

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Key: BFBM, blast-free hypoplastic or aplastic bone marrow; CI, confidence interval; CR, complete remission; CRh, complete remission with partial haematological recovery; CRi, complete remission with incomplete haematological recovery; mITT, modified intent to treat; MRD, minimal residual disease; OCR, overall complete remission.

Notes: MRD status is determined by the central laboratory. 95% confidence interval is based on Clopper-Pearson method. a, The numerator for MRD negative rate overall is based on an MRD-negative finding at any post infusion visit. Percentage for MRD negative rate overall is based on the number of patients in the mITT population;[b] , 30 patients with CR, eight patients with CRi, and four patients with BFBM achieved MRD negativity at any post infusion visit;[c] , The numerator for MRD negative rate is based on an MRD-negative finding at any post infusion visit. Percentage is based on the number of patients with OCR (CR or CRi). Disease response is based on central assessment;[d] , The numerator for MRD negative rate is based on an MRD-negative finding at any post infusion visit. Percentage is based on the number of patients with the corresponding best overall response. Disease response is based on central assessment. Source: ZUMA-3 clinical study report Table 14.2.8.1 (56).

DOR by central assessment:

Among the 39 subjects who achieved OCR (CR or CRi), the median time to response was 1.1 months (range: ).

The KM median DOR for all patients treated in Phase 2 was 12.8 months (95% CI: 8.7, NE), with a reverse KM median follow-up time for DOR of months (95% CI: ). Overall, 26 patients were censored including 12 patients in ongoing remission at data cut-off and nine patients who had subsequent allo-SCT. Only one patient who achieved remission with KTE-X19 had subsequently died at the time of data cut-off. The longest DOR to date is months (censored at data cut-off). The proportion of patients still in remission with KTE-X19 at 6 and 12 months from first remission was 76% and 56%, respectively (42,56).

At data cut-off, 31% (12 of 39) with CR/CRi were in ongoing remission, 23% (9 of 39) had proceeded to subsequent allo-SCT, 13% (5 of 39) proceeded to other anticancer therapies, 31% (12 of 39) had relapsed, and 3% (1 of 39) had died (Table 20).

A sensitivity analysis was conducted in which disease assessments obtained after allo-SCT were included in the derivation of DOR. Notably, in this analysis the median DOR was also 12.8 months (95% CI: 9.4 months, NE), consistent with the main analysis, with a reverse KM median follow-up time for DOR of months (95% CI: ), suggesting that KTE-X19 has the potential to be used as a standalone therapy (42,56).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 20: Duration of remission per central assessment (Phase 2, mITT)

Key: CI, confidence interval; CR, complete remission; CRi, complete remission with incomplete haematological recovery; DOR, duration of remission; KM, Kaplan–Meier; mITT, modified intent-to-treat; NE, not estimable; OCR, overall complete remission; SCT, stem cell transplant.

Notes: Percentages are based on the number of patients in the mITT population with OCR (CR or CRi). DOR is defined as the time from the first complete remission (CR or CRi) to relapse or death from any cause in the absence of documented relapse. Patients not meeting the criteria by the analysis data cut-off date were censored at their last evaluable disease assessment date prior to the data cut-off date, new anticancer therapy (excluding resumption of a TKI) start date, or SCT date, whichever was earlier.[a] , KM estimates represent the proportion of responders remaining in remission by time from first response. Source : ZUMA-3 clinical study report Table 14.2.5.1.1 (56).

The KM median DOR was 14.6 months (95% CI: 9.6, NE) for patients with CR and

8.7 months (95% CI 1.0, 12.8) for patients with CRi (Figure 18).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 18: Kaplan-Meier plot of DOR per central assessment (Phase 2, mITT)

==> picture [448 x 262] intentionally omitted <==

Data cutoff date: 09/09/2020.

Key: CI, confidence interval; CR, complete remission; CRi, complete remission with incomplete haematological recovery; DOR, duration of remission; KM, Kaplan–Meier; mITT, modified intent to treat; NE, not estimable; NR, not reached. Source : Adapted from Figure 3a (42).

Overall survival

At the time of data cut-off, 32 patients (58%) in the mITT population were alive; the proportion of patients estimated to be alive at 12 and 18 months was 71% and 59%, respectively (42,56).The KM median OS was 18.2 months (95% CI: 15.9, NE), with a reverse KM median follow-up time for OS of months (95% CI: ) (Table 21).

Table 21: Overall survival (Phase 2, mITT)

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Key: CI, confidence interval; CR, complete remission; CRi, complete remission with incomplete haematological recovery; DCO, data cut-off date; KM, Kaplan–Meier; mITT, modified intent-to-treat; MRD, minimal residual disease; NE, not estimable; OS, overall survival.

Notes: Overall survival for patients treated with KTE-X19 is defined as the time from KTE-X19 infusion date to the date of death from any cause. Patients who had not died by the analysis data cut-off date were censored at their last contact date prior to the data cut-off date, with the exception that patients known to be alive or determined to have died after the data cut-off date were censored at the data cut-off date. '+' indicates censoring.

Source : ZUMA-3 clinical study report Table 14.2.7.1 (56).

As detailed in Table 21, KM estimates of OS at 12 and 18 months were 71.4% (95% CI: ) and 58.6% (95% CI: ), respectively (Figure 19). The KM median OS was not reached (95% CI: ) for subjects with CR or CRi and was 2.4 months (95% CI: ) for all other subjects in the mITT

analysis set (Figure 20). The KM median OS was not reached (95% CI:

) for subjects with CR and was months (95% CI: ) for subjects with CRi. Almost all patients who achieved CR with KTE-X19 were estimated to be alive at 12 months ( ). At the time of primary data cut-off,

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

providing a median actual follow-up of months, OCR patients ( ) including CR patients ( ) were known to be alive (56).

The KM median OS was (95% CI: ) for subjects who were MRD[- ] and was (95% CI: ) for subjects who were MRD[+ ] and (95% CI: ) for subjects with missing MRD assessments (Figure 21). The proportion of patients who achieved MRD[- ] estimated to be alive at 12 and 18 months was and respectively (56).

Figure 19: Kaplan-Meier plot of overall survival (Phase 2, mITT Analysis Set)

==> picture [448 x 261] intentionally omitted <==

==> picture [105 x 7] intentionally omitted <==

----- Start of picture text -----
Data cutoff date: 09/09/2020.
----- End of picture text -----

Key: CI, confidence interval; CR, complete remission; CRi, complete remission with incomplete haematological recovery; KM, Kaplan–Meier; mITT, modified intent to treat; NE, not evaluable; NR, not reached; OS, overall survival. Source : Adapted from Figure 3d (42)

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 20: Kaplan-Meier plot of overall survival: CR versus CRi (Phase 2, mITT: patients with a CR or CRi)

==> picture [459 x 272] intentionally omitted <==

Data cutoff date: 09/09/2020.

Key: CI, confidence interval; CR, complete remission; CRi, complete remission with incomplete haematological recovery; KM, Kaplan–Meier; mITT, modified intent to treat; NE, not evaluable; NR, not reached; OS, overall survival. Source : ZUMA-3 clinical study report Figure 14.2.10.1.1 (56).

Figure 21: Kaplan–Meier plot of overall survival: MRD negative versus MRD positive (Phase 2, mITT population)

==> picture [459 x 272] intentionally omitted <==

Data cutoff date: 09/09/2020.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Key: CI, confidence interval; KM, Kaplan–Meier; mITT, modified intent to treat; MRD, minimal residual disease; NE, not evaluable; NR, not reached; OS, overall survival.

Source : ZUMA-3 clinical study report Figure 14.2.10. (56).

Data from the most recent data cutoff (23/07/21) provides longer-term evidence on the effect of allo-SCT consolidation of KTE-X19 (Figure 22). Of note is that sensitivity analysis of median OS stratified by censoring at allo-SCT demonstrate that survival appeared to be independent of subsequent SCT based on the Phase 2 mITT population (56). This supports the curative, standalone potential of KTE-X19.

Figure 22: Kaplan-Meier plot of OS stratified by subsequent SCT and OCR (Phase 2 mITT CR/CRi; data cut 23/07/21)

==> picture [459 x 295] intentionally omitted <==

Key: CI, confidence interval; NE, not estimable. Source: (64).

Relapse-free survival:

KM estimates of RFS rates at 6 and 12 months were 57.6% (95% CI: and 44.3% (95% CI: ), respectively. The KM median RFS was 11.6 months (95% CI: 2.7, 15.5 months), with a reverse KM median follow-up time for RFS of months (95% CI: ) (Figure 23).

)

Among subjects with CR or CRi, the KM median RFS was 14.2 months (95% CI:

11.6 months, NE). The proportion of patients achieving CR/CRi estimated to be Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

relapse-free and alive at 6 and 12 months was 81% and 63%, respectively. The KM median RFS was months (95% CI: ) for subjects with CR and months (95% CI: ) for subjects with CRi (56).

In patients achieving MRD negativity with KTE-X19, the KM median RFS was (95% CI: ), and the proportion estimated to be relapse-free and alive at 6 and 12 months was and , respectively (Table 22).

Table 22: RFS per central assessment (Phase 2, mITT)

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

12 months KM estimates of RFS rates in patients with CRi, % (95% CI) 6 months 12 months KM estimates of RFS rates in MRD negative patients, % (95% CI) 6 months 12 months

Key: CI, confidence interval; CR, complete remission; CRi, complete remission with incomplete haematological recovery; KM, Kaplan–Meier; mITT, modified intent-to-treat; NE, not estimable; RFS, relapse-free survival; SCT, stem cell transplant. Notes: Percentages are based on the number of patients in the mITT population. RFS for patients who received KTE-X19 is defined as the time from the KTE-X19 infusion date to the date of relapse or death from any cause. Patients who received KTEX19 but did not achieve CR or CRi as the best overall response are counted as events on the KTE-X19 infusion date. '+' indicates censoring. Source : ZUMA-3 clinical study report Table 14.2.6.1 (56).

Figure 23: Kaplan-Meier plot of RFS by central assessment (Phase 2, mITT)

==> picture [436 x 256] intentionally omitted <==

==> picture [105 x 7] intentionally omitted <==

----- Start of picture text -----
Data cutoff date: 09/09/2020.
----- End of picture text -----

Key : CI, confidence interval; CR, complete remission; CRi, complete remission with incomplete haematological recovery; KM, Kaplan–Meier; mITT, modified intent to treat; NE, not evaluable; NR, not reached; RFS, relapse free survival. Source : Adapted from Figure 3c (42).

Rate of allo-SCT

The incidence of allo-SCT after KTE-X19 infusion in the mITT analysis set is

summarised in Table 23. 18% (10 of 55 subjects) in the mITT analysis set received allo-SCT while in remission after the initial KTE-X19 infusion; of these, 7 subjects had achieved a CR and 2 subjects had achieved a CRi to KTE-X19 based on central assessment of disease response. One subject received an allo-SCT after achieving

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

a CRi per investigator assessment but was considered to have a best response of BFBM per central assessment. Overall, the median time from KTE-X19 infusion to allo-SCT was 98 days (range: days).

Data from the most recent data cutoff (23/07/21) provides longer-term evidence on the effect of allo-SCT consolidation of KTE-X19. Of note is that sensitivity analysis of median OS stratified by censoring at allo-SCT demonstrate that survival appears to be independent of subsequent SCT (Figure 22). This supports the curative, standalone potential of KTE-X19.

Of the 10 subjects who received allo-SCT after KTE-X19 infusion, ( subjects) died within 100 days after allo-SCT. The remaining ( subjects) were in ongoing remission 100 days after the transplant.

Table 23: Patient incidence of allo-SCT after treatment (Phase 2, mITT)

Key: allo-SCT, allogeneic stem cell transplant; CI, confidence interval; CR, complete remission; CRi, complete remission with incomplete haematological recovery; mITT, modified intent to treat; SCT, stem cell transplant. Notes: Only transplants received while in remission after KTE-X19 infusion and before retreatment are included. Transplants that were received after subsequent anticancer therapy are not included. Response of CR or CRi is based on central assessment. a, the overall patient incidence includes one patient who had CRi per investigator assessment but was assessed as BFBM (blast free hypoplastic or aplastic bone marrow) as per central assessment; b, mortality rate and ongoing response 100 days after allo-SCT were calculated using the number of patients who received an allo-SCT as the denominator. Source : ZUMA-3 clinical study report Table 14.2.9.1 (56).

When comparing the 9 of 39 subjects to achieve OCR who were consolidated with

allo-SCT with the 30 of 39 subjects who were not, those receiving consolidation with allo-SCT were less heavily pre-treated (median vs prior lines of therapy), none of them were Ph+, none had prior allo-SCT (vs in the OCR without consolidation),

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

and subjects consolidated with allo-SCT had a lower % blasts in the bone marrow after bridging chemotherapy ( vs ) (56).

Notably, all treated subjects at Phase 2 who went on to receive an allo-SCT after KTE-X19 had not received a previous allo-SCT. Based on discussion with clinical experts, Kite understand that a second allo-SCT is not seen as a viable treatment option in the UK (35). Additionally, in the UK allo-SCT would not be considered as an option to consolidate remission, as was the case for all subjects to receive allo-SCT post-KTE-X19 in the ZUMA-3 Phase 2 mITT population. Given the anticipated positioning of KTE-X19 in adults with R/R ALL who have relapsed post-allo-SCT, are ineligible for allo-SCT, or are unlikely to achieve allo-SCT due to poor prognostic factors, it is highly unlikely that KTE-X19 would be used as a bridge to allo-SCT, instead being considered as a standalone treatment option in UK clinical practice.

Other outcome measures: EQ-5D-5L

Across all 5 dimensions of the EQ-5D-5L, the proportion of evaluable subjects who reported no health problems at screening (baseline) ranged from (pain/discomfort) to (self-care). Shortly after KTE-X19 treatment, the proportion of subjects reporting no problems ranged from (usual activities) to

(self-care) at Day 28. By month 3, the proportion of subjects reporting no problems rebounded (mobility and pain/discomfort) or reached higher levels (selfcare, usual activities, and anxiety/depression) as compared to proportions at baseline. By month 12, the proportions of subjects reporting no problems were higher than those proportions at screening across all 5 domains, ranging from

(pain/discomfort) to (self-care), suggesting a trend of recovery or improvement over time. Additionally, the proportion of subjects reporting severe or extreme problems on each domain was consistently low ( ) at each time point after KTEX19 treatment (56).

The median VAS score was 70.0 (range: 5 to 100) at screening and increased over time, with higher median scores of 80.0 (range: 20 to 100) at Day 28, 80.0 (range: 50 to 100) at Month 3, 85.0 (range: 40 to 100) at Month 6, and 87.5 (range: 70 to 100) at Month 12. The vast majority of subjects maintained stable VAS scores (absolute

change of <7 points) or demonstrated clinically meaningful improvement (increase of Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

≥7 points) relative to their scores at screening over time: at Day 28, at Month 3, at Month 6, and at Month 12 (56,67).

In summary, patient-reported outcomes as measured by the EQ-5D-5L VAS remained stable or improved relative to values at baseline for the majority of subjects following treatment with KTE-X19 (≥70% of evaluable subjects considered stable or improved across time points from Day 28 through Month 12).

B.2.6.3 Summary of KTE-X19 clinical effectiveness

The efficacy and safety of KTE-X19 for the treatment of adult R/R ALL has been demonstrated in the single-arm Phase 1/2 ZUMA-3 trial (55). Patients with R/R disease were defined as primary refractory, in first relapse following a remission lasting ≤ 12 months, R/R after second-line or higher therapy or R/R after allo-SCT, representing the heterogenous adult R/R ALL population presenting in clinical practice. Definitions were based on the historically poor outcomes observed in these patient populations (18,24,57,58).

The ZUMA-3 Phase 1 + 2 combined population represents an adult R/R ALL group with an especially poor prognosis, including almost half ( ) with prior blinatumomab treatment, and a similar proportion receiving ≥3 prior treatments at baseline ( ) (Table 8). In addition, had relapsed post-SCT, a group with an especially dire prognosis, where median OS is 5.5 months (43).

Despite having received multiple prior therapies to which they had experienced suboptimal response, 74.4% of adult R/R ALL patients treated with KTE-X19 achieved OCR, and 62.8% achieved CR.

Furthermore, 79.5% had no detectable cancer cells remaining as demonstrated by MRD negativity, including all but one patient – for whom data was not available – to achieve CR/CRi. The survival advantage of achieving MRD negativity in both adults and children has been demonstrated by Berry et al ., (2017) in a meta ‑ analysis of 39 studies (albeit following induction therapy), and was further re-enforced by recent long-term blinatumomab data (65,66). This supports the clinical value of every KTEX19-treated evaluable subject to achieve CR/CRi also achieving MRD[- ] remission.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

At the most recent data cutoff with actual median follow-up of months, and , were estimated to be alive at 12 and 18 months, respectively. Survival measurements from time of KTE-X19 treatment report an unprecedented median OS estimate of KM median OS was months (95% CI: ). Notably, survival in responders appeared to be independent of subsequent SCT.

B.2.7 Subgroup analysis

Subgroup analyses based on baseline disease and treatment covariates were prespecified and conducted for selected efficacy and safety endpoints. These subgroups were explored to better characterise patient populations for whom KTEX19 may provide the most benefit.

The OCR rate with 95% CIs was generated for subgroups of the mITT analysis set defined by selected treatment covariates. A forest plot of proportions (and 95% CI) of subjects achieving an OCR for each subgroup was generated. Full results are presented in Appendix E .

The OCR rate was largely consistent across most pre-planned subgroups, including those defined by baseline demographics, clinical characteristics and treatment history. Whilst OCR in the Phase 1 + 2 combined dataset was highest in those with 1 prior line of therapy ( ) ( ), the OCR rate in the subjects with ≥4 prior lines of therapy was , supporting the effectiveness of KTE-X19 in heavily pretreated subjects. Of note is that OCR rate was actually higher in those with a prior SCT ( ) than it was in those without prior SCT ( ) ( vs respectively). This OCR rate for subjects with prior SCT supports one of the proposed positionings of KTE-X19, as a treatment option for those who have relapsed post-allo-SCT.

While subgroup analyses must be interpreted with caution given the small sample sizes involved, clinical benefit was observed compared to historical controls irrespective of patient demographics, disease characteristics or treatment history. It should also be noted that while relatively small on face value, the sample size in ZUMA-3 is representative of the rarity of adult R/R ALL.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.2.8 Meta-analysis

Meta-analysis is not required for KTE-X19 as a single study provides data for this intervention.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.2.9 Indirect and mixed treatment comparisons

Table 24: Summary of comparators

Key: FLAG-IDA, Fludarabine, cytarabine, granulocyte-colony stimulating factor, idarubicin; Ph+, Philadelphia chromosomepositive; Ph-, Philadelphia chromosome-negative.

In the absence of head-to-head clinical trial evidence of KTE-X19 versus either inotuzumab, blinatumomab, ponatinib, or salvage chemotherapy (FLAG-IDA), an SLR was conducted to identify relevant evidence on the comparator treatments for the purposes of conducting a possible indirect treatment comparison.

A total of 68 publications were included, of which 17 were RCT publications related to the TOWER and INO-VATE trials and the remaining 51 publications reported on non-randomised studies (i.e., single-arm trials and observational studies). The SLR was conducted on June 12, 2019, and subsequently updated in November 2020 to ensure all relevant literature was captured. For methods and results of the SLR please refer to Appendix D1.1.

Details of the 12 studies included in the SLR that were further evaluated for eligibility to be included in an ITC are listed in Table 97.

In the context of the evidence base available (single-arm trial data), it was not feasible to perform an anchored indirect treatment comparison to evaluate the comparative effectiveness of KTE-X19 versus relevant comparators. As such, both naïve ITCs and matching-adjusted indirect comparisons (MAICs) were conducted in line with the NICE decision support unit (DSU) technical support document (TSD) 18 (68). MAICs use individual patient-level data (IPD) from trials of one treatment to match baseline summary statistics reported from trials of another treatment (69,70). Matching baseline characteristics in this way enables the comparison of treatment outcomes across balanced trial populations. Full details of the ITC methodology and results are available in the separate ALL MAIC report (71).

In addition, a retrospective cohort study (SCHOLAR-3) was conducted, utilising a matched cohort derived from IPD sampled from historical clinical trials to further contextualise the results of ZUMA-3. A post-hoc analysis was conducted which Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

matched patients from ZUMA-3, irrespective of whether they had prior treatment with blinatumomab and inotuzumab to patients from historical trials who had not previously received blinatumomab therapy (SCHOLAR-3 synthetic control arm [SCA]-3). As the treatment assignments in SCA-3 were either blinatumomab ( ) or SoC chemotherapy ( ), the study population was further restricted within the analyses to include only patients who had received blinatumomab in the SCA-3 arm (i.e., excluding SoC chemotherapy patients) matched to patients from ZUMA-3.

The SCHOLAR-3 analysis meant that two data sources were available for the comparison with blinatumomab: the matched SCHOLAR-3 SCA-3 cohort and the pseudo-IPD recreated from TOWER. The key difference between these two analyses can be summarised as follows:

• When using the SCHOLAR-3 SCA-3 cohort, the ZUMA-3 patient characteristics remain unadjusted while the IPD of patients who received blinatumomab was matched to that of patients in the ZUMA-3 phase 2 mITT cohort based on their individual propensity score. In addition, use of SCA-3 ensures the comparison retains almost all of the ZUMA-3 mITT dataset ( ) compared to 37-39 of 55 subjects for the MAIC comparison vs

• blinatumomab

• Conversely, when using the MAIC analyses, the ZUMA-3 IPD was weighted to match the reported average characteristics reported in TOWER for the intervention arm, and adjusted event-free survival (EFS) and OS KM for KTEX19 were provided based on the weighted data. As TOWER enrolled Phpatients only, the MAIC excluded ZUMA-3 patients that were Ph+

For our base case economic analyses, SCHOLAR-3 SCA-3 was considered a more appropriate source for comparison with blinatumomab than the MAIC or naïve comparison because the target population to which characteristics were matched was that of ZUMA-3. As explained in B.2.6.2, 100% of patients recruited to ZUMA-3 matched the anticipated positioning of KTE-X19 in UK clinical practice, that is, patients who have either failed or are unlikely to achieve SCT. Conversely,

blinatumomab requires consolidation with SCT to be curative, therefore the patients recruited to TOWER are unlikely to be generalisable to the population treated with Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

KTE-X19 in UK clinical practice. Furthermore, the point estimates for relative efficacy of KTE-X19 versus blinatumomab from SCHOLAR-3 and the naïve comparison were aligned, whereas that from the MAIC diverged, although are unlikely to be statistically different (Table 26).

On the same principle, the naïve comparisons against inotuzumab and salvage chemotherapy (FLAG-IDA) underpin our base case economic analyses for these comparators, as the MAICs involve re-weighting of the ZUMA-3 patient characteristics to match those of INO-VATE and/or TOWER. INO-VATE notably excluded patients who had failed two or more prior therapies.

While ponatinib is considered a relevant comparator for adults with Ph+ R/R B-cell ALL disease, performing a MAIC was not deemed feasible. This is due to the small number of patients with Ph+ ALL in ZUMA-3 phase 1/2. Therefore, a naive comparison was conducted using data from the mITT ZUMA-3 Phase 1/2 and overall PACE population (note that in the economic model, the overall mITT ZUMA-3 phase 1/2 survival analysis is used for the ponatinib analysis, rather than the Ph+ subgroup). Although there are limitations of a naïve approach, clinical advisors felt that Ph expression is not expected to have an impact on the efficacy of CAR T-cells, including KTE-X19, and that Ph status was a low-rank prognostic factor (35). Unadjusted analyses for EFS were not conducted given different progression-related time-to-event outcomes were reported in the TKI studies (i.e. progression-free survival [PFS] in PACE), and RFS in ZUMA-3).

In summary, three categories of ITC were carried out against the various comparators that are of particular relevance to the economic analysis:

• i. MAIC (vs inotuzumab, blinatumomab, FLAG-IDA)

• ii. SCHOLAR-3 SCA-3: a matched cohort derived from IPD data from historical clinical trials (vs blinatumomab)

• iii. Naïve (unadjusted) comparison (vs ponatinib, inotuzumab, blinatumomab, FLAG-IDA)

A summary of the key ITCs of particular relevance to the economic analysis is presented in Table 25. The outcomes of focus (EFS and OS) are those needed for

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

the economic analysis (see Section B.3). Results of the ITC for OS are summarised in Table 26, with EFS summarised in Table 27.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 25: Summary of key ITCs used in the economic model

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 88 of 270

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 89 of 270

Key: AD, aggregate data; ALL, acute lymphoblastic leukaemia; CR, complete remission; EFS, event-free survival; HR, hazard ratio; IPD, individual patient data; ITT, intention to treat; KM, KaplanMeier; MAIC, matching-adjusted indirect comparison; mITT, modified intention to treat; OS, overall survival; R/R, relapsed or refractory The mITT phase 1+2 dataset comprises 55 phase 2 patients and the 23 phase 1 patients treated with the target dose of 1 x 10[6] cells/kg.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 90 of 270

Table 26: Summary of ITC results (OS)

Key: CI, confidence interval; ESS, effective sample size; HR, hazard ratio; mITT, modified intention-to-treat; NE, not estimable

Note: 3-level salvage: first salvage, second salvage vs. rest, 2-level salvage: first salvage vs. rest. SCHOLAR-3 is a retrospective cohort study utilizing data from the Phase 2 ZUMA-3 investigational trial (mITT) and IPD sampled from historical clinical trials in relapsed or refractory adult ALL contained within the Medidata Enterprise Data Store (MEDS) database to create a matched synthetic control arm.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 91 of 270

Table 27: Summary of ITC results (EFS)

Key: CI, confidence interval; EFS, event-free survival; HR, hazard ratio; MAIC, Matching-adjusted indirect comparison.

Note: 3-level salvage: first salvage, second salvage vs. rest, 2 -level salvage: first salvage vs. rest. SCHOLAR-3 is a retrospective cohort study utilizing data from the Phase 2 ZUMA-3 investigational trial (mITT) and IPD sampled from historical clinical trials in relapsed or refractory adult ALL contained within the Medidata Enterprise Data Store (MEDS) database to create a matched synthetic control arm. For the naïve comparison with ponatinib and SCHOLAR-3 comparison versus blinatumomab, no data on EFS is available. *range based on salvage status (2-level, 3-level).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 92 of 270

B.2.9.1 MAIC:

For the MAIC, of the 12 studies identified in the SLR to be evaluated for eligibility, two studies were included in the final comparisons: INO-VATE and TOWER Table 98. Rationale behind the exclusion of the remaining 10 studies is provided in Table 99.

The MAIC was conducted in several steps. The first step was to conduct a feasibility assessment to determine the degree of overlap in study designs and populations and the extent that it is possible to generate unbiased comparisons. In the next step, we redefined outcomes in the IPD for ZUMA-3 to match the outcomes definitions of the aggregate data from comparator trials. A logistic propensity score model was used to estimate weights for the IPD such that the weighted mean baseline characteristics of interest for the population in ZUMA-3 matched those reported in the comparator trials. The choice of covariates for the propensity score models was based on clinician interviews regarding prognostic factors of significance in R/R ALL, as well as potential effect modifiers (71).

These above steps resulted in a ZUMA-3 IPD dataset with a weighted trial population that matched those of the comparator trial(s) of interest for the included covariates. Using these weights, outcomes for KTE-X19 were predicted for the population in the comparator trial by reweighting the observed outcomes from ZUMA-3. Treatment comparisons were then conducted across the balanced trial populations. Full details of the methodology and results for OS and RFS/PFS/EFS of KTE-X19 versus interventions considered to represent SoC are presented in the MAIC report (71).

After exploring different models and examining the effective sample size (ESS) for the four analysis populations, the most inclusive model that achieved convergence was selected for the MAIC comparisons. The MAIC with INO-VATE matched on duration of first remission <12 months, prior SCT, age, Eastern Cooperative Oncology Group (ECOG) (0 vs. rest), salvage status, bone marrow blast at screening, complex karyotype and Philadelphia chromosome status. For TOWER, the MAIC matched on primary refractory, duration of first remission < 12 months,

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

prior SCT, age, ECOG (0 vs. rest), salvage status, and Philadelphia chromosome status. Covariates reported by both trials (INO-VATE and TOWER) were matched for the comparison with chemotherapy. Given INO-VATE only enrolled patients who were due to receive their first or second salvage treatment, additional scenario analyses were conducted for salvage status: a) 3-level salvage (i.e., first salvage, second salvage vs. rest), and b) 2-level salvage (first salvage vs. rest).

Results

Comparisons were performed for each of the four analysis population sets of ZUMA3 (mITT Phase 2, mITT Phase 1+2, ITT Phase 2, ITT Phase 1+2) with INO-VATE, TOWER and the pooled chemotherapy arms in INO-VATE and TOWER for OS and EFS. Results for the Phase 1 + 2 combined dataset are presented here. For the results of other populations please refer to the MAIC report (71).

Overall survival

Note that for the ZUMA-3 Phase 1 + 2 combined population, OS was calculated from date of KTE-X19 infusion.

Inotuzumab:

The estimated HRs for the Phase 1 + 2 combined population including the two salvage status scenario analyses were all in favour of KTE-X19; after adjustment, these differences were significant. Given the overall small sample size in Phase 2 ZUMA-3 populations, and the matching to INO-VATE resulting in very small ESS, the results of the comparisons with INO-VATE should be interpreted with caution (Figure 24).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 24: Overall survival for ZUMA-3 Phase 1+2 combined versus INO-VATE inotuzumab

==> picture [458 x 253] intentionally omitted <==

==> picture [452 x 5] intentionally omitted <==

Key : ESS, effective sample size; HR, hazard ratio; MAIC, matching-adjusted indirect comparison; m, median; OS, overall survival. 3-level salvage: first salvage, second salvage vs. rest, 2-level salvage: first salvage vs. rest. Source: (71).

Blinatumomab:

Following adjustment, the KM curves shifted downwards; however, the estimated HRs were all in favour of KTE-X19 (range: ). Similar to the comparisons with INO-VATE, differences were statistically significant for the Phase 1+2 combined population after adjustment (Figure 25).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 25: Overall survival for ZUMA-3 Phase 1+2 combined versus TOWER

==> picture [462 x 255] intentionally omitted <==

Key : ESS, effective sample size; HR, hazard ratio; MAIC, matching-adjusted indirect comparison; m, median; OS, overall survival. 3-level salvage: first salvage, second salvage vs. rest, 2-level salvage: first salvage vs. rest. Source: (71).

FLAG-IDA:

In the MAICs with chemotherapy, reconstructed IPD for INO-VATE and TOWER were combined to create a single chemotherapy arm. Results suggested that KTEX19 was superior to the combined chemotherapy arm in terms of OS, including for both salvage status scenarios (Figure 26).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 26: Overall survival for ZUMA-3 Phase 1+2 combined versus stacked IPD in INO-VATE and TOWER chemotherapy

==> picture [458 x 262] intentionally omitted <==

Key : ESS, effective sample size; HR, hazard ratio; MAIC, matching-adjusted indirect comparison; m, median; OS, overall survival. 3-level salvage: first salvage, second salvage vs. rest, 2-level salvage: first salvage vs. rest. Source: (71).

Event-free survival

Note that for ZUMA-3 mITT populations, EFS was calculated from date of infusion.

Inotuzumab:

Unlike the results from the OS analysis, although the estimated HRs for the Phase 1 + 2 combined dataset and for the two salvage status scenarios were all in favour of KTE-X19 after adjustment, these differences were not statistically significant (Figure 27).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 27: Event-free survival for ZUMA-3 Phase 1+2 combined versus INOVATE

==> picture [458 x 260] intentionally omitted <==

Key: EFS, event-free survival; ESS, effective sample size; HR, hazard ratio; m, median; MAIC, matching-adjusted indirect comparison. 3-level salvage: first salvage, second salvage vs. rest, 2-level salvage: first salvage vs. rest. So urce: (71).

Blinatumomab:

Following adjustments, the EFS KM curves shifted minimally, and the HRs for the unadjusted and adjusted comparisons were very similar. Although the 95% CIs became wider after applying weights, the results showed a statistically significant difference in EFS with KTE-X19 compared to blinatumomab for all four populations, both before and after adjustment. It should be noted that the proportional hazards assumption was violated for the mITT Phase 1+2 population when the 3-level salvage was matched, and therefore these results should be interpreted with caution (Figure 28).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 28: Event-free survival for ZUMA-3 Phase 1+2 combined versus TOWER

==> picture [462 x 263] intentionally omitted <==

Key: EFS, event-free survival; ESS, effective sample size; HR, hazard ratio; m, median; MAIC, matching-adjusted indirect comparison. 3-level salvage: first salvage, second salvage vs. rest, 2-level salvage: first salvage vs. rest. Source : (71).

FLAG-IDA:

Overall, the EFS KM curves shifted minimally after adjustment, but the number of patients at risk dropped significantly from 0 to 6 months. It should be noted that the proportional hazards assumption was violated for the mITT Phase 1+2 population (Figure 29).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 29: Event-free survival for ZUMA-3 Phase 1+2 combined versus stacked IPD in INO-VATE and TOWER

==> picture [457 x 261] intentionally omitted <==

Key: EFS, event-free survival; ESS, effective sample size; HR, hazard ratio; m, median; MAIC, matching-adjusted indirect comparison. 3-level salvage: first salvage, second salvage vs. rest, 2-level salvage: first salvage vs. rest. Source : (71).

Conclusions of the MAIC

Findings from the MAICs suggested KTE-X19 had a favourable effect on OS and EFS compared to inotuzumab, blinatumomab, and chemotherapy regimens in R/R ALL patients. The methods used for the MAIC in this analysis aligned with recommendations from the NICE guidance for population-adjusted indirect comparisons (68). Limitations of the MAIC are discussed in section B.2.9.4.

Given the anticipated positioning of KTE-X19 in patients who have relapsed postSCT, or are unlikely to achieve/ineligible for SCT, it was concluded that the population of ZUMA-3 was most reflective of anticipated use in clinical practice. Therefore, rather than adjusting ZUMA-3 for comparisons to the populations from the TOWER and INO-VATE trials, it was considered that the most appropriate comparison for the cost-effectiveness analysis were the unadjusted ones.

B.2.9.2 Naïve comparison

For each outcome of interest in each patient population, a model without individual weights provides a ‘naïve’ estimate of the treatment effect of KTE-X19 versus each

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

comparator where the relative treatment effect was estimated based on the observed outcomes of interest from each trial without adjusting for any between-study differences.

HRs for OS and EFS were estimated by means of a Cox proportional hazards model based on the (unadjusted) IPD from ZUMA-3 and the reconstructed IPD from the published KM curves from each external study. Treatment effects of interest were expressed with point estimates and 95% CIs.

Results from the naïve comparisons with inotuzumab and FLAG-IDA are presented in Figure 24 and Figure 26, respectively for OS, and Figure 27 and Figure 29, respectively, for EFS.

Results from the naïve comparison between Ph+ population in ZUMA-3 and PACE (ponatinib 45mg) for OS are presented in Figure 30 (36,42). As a result of the small sample size in ZUMA-3, there were minimal changes in the number of patients at risk across the different time points, resulting in relatively flat KM curves for ZUMA-3. Across the comparisons, all naïve analyses suggested KTX-19 was favourable to ponatinib.

Figure 30: Overall survival for ZUMA-3 Phase 1+2 combined versus ponatinib (45mg)

==> picture [456 x 248] intentionally omitted <==

Key: HR, hazard ratio; m, median; N, number; OS, overall survival.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Source : (71).

B.2.9.3 SCHOLAR-3:

SCHOLAR-3 is a retrospective cohort study that selected patients from the Medidata Enterprise Data Store (MEDS) database of historical clinical trials that matched the inclusion/exclusion criteria of ZUMA-3. From the resulting pool a matched cohort of patients with similar baseline characteristics to ZUMA-3 was analysed for clinical outcome to quantify the relative effectiveness of KTE-X19.

The primary objective of SCHOLAR-3 was to describe the OCR rate in patients sampled from historic clinical trials that were previously naïve to blinatumomab or inotuzumab therapy. Baseline characteristics for matched populations are available in the SCHOLAR-3 CSR (72).

There were three subsets for SCHOLAR-3, SCA-1 (blinatumomab/inotuzumab naïve) and SCA-2 (blinatumomab/inotuzumab experienced) were pre-specified, with SCA-3 added post-hoc. As clinical outcomes beyond OS were not available for patients exposed to blinatumomab or inotuzumab in the MEDS database, SCA-3 only included blinatumomab and inotuzumab naïve patients for whom all clinical outcomes were available and matching to ZUMA-3 was done on the other baselines characteristics. This might have led to bias against KTE-X19 given that prior

treatment with blinatumomab has been shown to impact effectiveness of subsequent CD19 CAR-T treatment in ALL (60). SCA-3 was considered worthwhile to obtain two relatively large, matched cohorts to quantify the clinical effectiveness of KTE-X19 compared to blinatumomab.

A brief summary of the SCHOLAR-3 study is provided in Table 28.

Table 28: Overview of design for SCHOLAR-3

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Key: OCR, overall complete remission; r/r, relapsed/refractory; SCA, synthetic control arm. Source: (72) .

Full details of the methodology and results for SCHOLAR-3 are available in the SCOLAR-3 CSR (72).

Results

SCA-1

For the primary objective, it was estimated that (95% CI ) of patients in SCA-1 (blinatumomab/ inotuzumab naïve) achieved OCR at week 24. For the first secondary objective, the comparison of OCR rate between matched ZUMA-3 and SCA-1 arms, matched patients from ZUMA-3 had an OCR rate of (95% CI: ). When compared to SCA-1 patients, matched ZUMA-3 patients had a significantly higher odds of achieving OCR (95% CI: ) (p= ).

The comparison of OS between matched ZUMA-3 and SCA-1 patients showed that ZUMA-3 patients had a higher median OS in comparison to SCA-1 patients, months (95% CI: ) versus months (95% CI: ) respectively. A hazard ratio (HR) derived through a univariate Cox regression showed that ZUMA-3 patients were less likely to die than patients in the SCA-1 group, HR (95% CI: ) (p = ).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 31: Kaplan-Meier median OS (ZUMA-3 vs SCA-1)

==> picture [459 x 309] intentionally omitted <==

Key: OS, overall survival; SCA, synthetic control arm. Source: Figure 6 (72)

SCA-2

The comparison of OS between matched ZUMA-3 and SCA-2 patients showed that ZUMA-3 patients had a higher median OS in comparison to SCA-2 patients, months (95% CI: ) versus months (95% CI:

), respectively. A HR derived through a multivariate Cox regression adjusted for percentage bone marrow blasts and prior lines of therapy did not indicate a statistically significant difference, HR (95% CI ).

SCA-3

In the non-prespecified post hoc analysis, (95% CI: ) of patients in the ZUMA-3 arm achieved OCR at Week 24 while (95% CI: ) from SCA-3 achieved OCR at the same timepoint. ZUMA-3 patients had times higher odds of achieving OCR in comparison to SCA-3 patients (95% CI: ) (pvalue ).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

A further post-hoc analysis comparing OS between matched ZUMA-3 and SCA-3 patients showed ZUMA-3 patients to have a median OS of months (95% CI: ) and SCA-3 patients to have a median OS of months (95% CI: ). A univariate Cox regression showed that ZUMA-3 patients were less likely to die in comparison to SCA-3 patients; HR (95% CI: ) (p = ).

Figure 32: Kaplan-Meier median OS (ZUMA-3 vs SCA-3)

==> picture [466 x 255] intentionally omitted <==

Key: OS, overall survival; SCA, synthetic control arm. Source: Figure 13 SCHOLAR-3 CSR (72).

B.2.9.4 Uncertainties in the indirect and mixed treatment

comparisons

MAIC:

The methods used for the MAIC in this analysis aligned with recommendations from the NICE guidance for population-adjusted indirect comparisons (70). However, it is important to highlight the limitations of this type of cross-study comparison. Specifically, the analysis is limited to study-level aggregate data from the

publications of the comparator studies. In the absence of IPD from the comparator studies, it is challenging to evaluate the extent of bias in the treatment effect estimates and it is likely that some confounding variables remained unbalanced. For example, unlike ZUMA-3, which only enrolled patients with an ECOG score of 0 or 1,

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

both INO-VATE and TOWER also enrolled patients with an ECOG score of 2. While the proportion of patients with ECOG 0 was matched, outcomes had to be assumed to be comparable for those with ECOG 1 or 2.

In contrast, some of the patients in ZUMA-3 would not have been eligible for the comparator studied. INO-VATE required patients to have received no more than one prior salvage therapy, while there was no limit on the number of salvage therapies received in ZUMA-3. TOWER only enrolled patients with Ph- ALL, whereas of patients had Ph+ ALL in the ZUMA-3 combined Phase 1+2 population. Differences in the proportions of other key clinical characteristics were found to be influential on the associated weighting of ZUMA-3. For instance, in ZUMA-3 the proportion of patients with prior SCT ranged from to in the four analysis populations, whereas in INO-VATE only 18% of patients received prior SCT. Therefore, patients with prior SCT were down-weighted whereas patients without prior SCT were up weighted in ZUMA-3. A closer examination of some of the patients with extreme weights showed that in general patients who were in first salvage status, with a duration of remission less than < 12 months, with no prior SCT, or with bone marrow blast >50% at screening tend to be up-weighted in the model.

Despite these limitations, the HRs for EFS and OS versus blinatumomab, inotuzumab and FLAG-IDA from the Phase 1+2 MAICs used in the economic model largely remained stable when compared with the naïve comparisons. The largest difference was observed with OS versus blinatumomab (naïve ; MAICs [3salvage status] and [2-salvage status]), indicating that these treatment effects are likely robust despite the statistical uncertainty associated with individual MAICs.

Of note, the point estimate of the naïve OS HR for blinatumomab ( ) was identical to that from the SCHOLAR-3 analysis, whereas that from the MAIC ( ) diverged. Thus, the naïve ITC appears to have produced more valid results than the MAIC, given the SCHOLAR-3 analysis involved matching individual patients to the correct target population. This also supports, by inference, use of a naïve comparison for inotuzumab in the economic analysis.

SCHOLAR-3:

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Despite the study using IPD from historical clinical trials that have been captured using electronic case report forms (eCRFs) and verified through originating sponsors’ verification processes, the risk of misclassification bias can’t be completely ruled out. To mitigate this, plausibility checks for the data from historical clinical trials were carried out if the historical trials had published results. To limit further misclassification bias during the derivation of the cohorts all statistical procedures were double programmed.

Variation in the definition of various endpoints across historical clinical trials could lead to information bias in this study. To mitigate against this, baseline variables and all endpoints were defined based on their constituent variables as per definitions in the ZUMA-3 study.

In order to minimise selection bias, ZUMA-3 inclusion criteria were used for this study, with Kite blinded to patient selection, treatment and outcomes until analysis was complete. In addition, propensity matching was used to minimise heterogeneity.

In order to account for confounding a robust strategy was developed for this study and while double robust multivariate models were considered for comparative analysis, limitations due to the final sample sizes led to issues of convergence.

This study seeks to emulate a “physicians’ choice” arm from a randomised experiment. Furthermore, as this study is building matched cohorts from historic clinical trials treatment effects may be over-estimated in comparison to real world practice. This may affect the external validity of the study design.

The consistently superior efficacy of KTE-X19 was demonstrated in all three cohorts. The SCA-3 cohort analysis represented a comparison potentially biased against KTE-X19, given that the ZUMA-3 cohort included a large number of patients who had previously failed targeted therapies, whereas the SCA-3 cohort included only those naïve to inotuzumab and blinatumomab. Given that failure of targeted therapies is generally considered a poor prognostic factor, it is notable that a significant OS benefit HR ( ) was observed, demonstrating the value of KTE-X19 in its proposed position in the pathway.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.2.10 Adverse reactions

The safety and tolerability of KTE-X19 for the treatment of adult patients with R/R B- cell ALL was evaluated as a secondary outcome in ZUMA-3. The Phase 2 safety analysis set was defined as all subjects treated with KTE-X19 (N = 55).

Of note is that, based on the safety and efficacy observations at Phase 1, the SRT decision was to explore the safety profile of the 1 x 10[6 ] anti-CD19 CAR T cells/kg dose level with the implementation of modified toxicity management

recommendations. Under the revised AE management guidelines, patients were not administered tocilizumab for neurological events unless in the context of CRS, and steroid use was initiated for Grade 2 neurological events in comparison to Grade 3 neurological events in the previous guidelines.

Data on AEs for the Phase 1 target dose population (N = 23) is not reported separately from the safety analysis set (N = 45), except for on a few instances, such as exposure to KTE-X19. Given almost half (22 of 45 subjects) of those in the safety analysis set did not receive KTE-X19 at target dose, and 36 of 45 subjects received KTE-X19 prior to the revised toxicity management guidelines, data from the Phase 1 safety analysis set is not presented here.

AEs were coded with the Medical Dictionary for Regulatory Activities (MedDRA) version 23.0.

B.2.10.1 Exposure to KTE-X19:

In Phase 2, the median weight-adjusted dose of KTE-X19 was x 10[6 ] anti-CD19 CAR T-cells/kg (range: x 10[6 ] anti-CD19 CAR T-cells/kg. the median total number of anti-CD19 CAR T-cells in the KTE-X19 infusion was x 10[6 ] cells (range: x 10[6 ] cells – x 10[6 ] cells), and the median total number of T cells infused was x 10[6 ] cells (range: x 10[6 ] cells – x 10[6 ] cells). Of the 55 subjects treated, ( ) received KTE-X19 within 10% of the planned target dose (56).

Among all subjects treated at target dose at Phase 1, the median weight-adjusted dose of KTE-X19 was x 10[6 ] anti-CD19 CAR T-cells/kg (range: x 10[6 ] cells/kg). The median total number of anti-CD19 CAR T cells in the KTE-X19 infusion Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

was x 10[6 ] cells (range: x 10[6 ] to x 10[6 ] cells), and the median total number of T cells infused was x 10[6 ] cells (range: x 10[6 ] to x 10[6] ). Of the 23 subjects, ( ) received within 10% of the planned total dose (56).

B.2.10.2 Duration of hospitalisation for KTE-X19 infusion

For subjects in Phase 2, the median duration of hospitalisation during the KTE-X19 infusion and until discharge following the infusion was days (range: days). For subjects treated at target dose in Phase 1, the median duration of hospitalisation during the KTE-X19 infusion until discharge following the infusion was days (range: days) (56).

B.2.10.3 Safety summary

The safety profile of KTE-X19 in ZUMA-3 was generally similar to that observed in other indications, although a higher incidence of Grade 3 or higher cytokine release storm (CRS) was observed. This was also the case for tisagenlecleucel, where CRS was more commonly observed in ALL compared to other indications (73).

All treated patients had at least one AE, and of 55 patients ( ) had KTE-X19 related AEs, with patients ( ) experiencing KTE-X19 related AEs that were worst Grade 3 or higher (Table 29). The most common worst Grade 3 or higher KTEX19 related AEs were ,

and

.

Forty-one of 55 patients (75%) experienced an SAE, while of 55 patients ( ) had at least one SAE related to KTE-X19, the most frequently reported of which were , , and . There were two deaths observed due to AEs that were considered related to KTE-X19 (brain herniation [day 8] and septic shock [day 18]) (42).

An overview of AEs experienced by subjects in the Phase 2 safety analysis set, as well as the Phase 1 + 2 combined dataset are presented in Table 29, demonstrating the consistency of KTE-X19’s safety profile, with slightly more favourable results in the Phase 2 safety analysis set potentially as a result of the revised AE management plan described earlier on in this section.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 29: Overall summary of AEs

Data cutoff date = 09Sep2020. Key : mITT, modified intent-to-treat. Notes : TEAEs include all AEs with an onset on or after initiation of the KTE-X19 infusion. For subjects who underwent retreatment with KTE-X19, the AEs occurring during the retreatment period are not included. Subjects were summarized at their highest grade per the Common Terminology Criteria for Adverse Events version 4.03. Source : ZUMA-3 clinical study report Table 47; Table 14.3.1.1.4 (56).

B.2.10.4 Common adverse events

AEs that occurred in ≥ 10% of patients in the Phase 2 mITT population are summarised in Table 30.

Table 30: Subject incidence of AEs occurring in >10% of subjects by preferred term and worst grade (Phase 2, safety analysis set)

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Data cutoff date = 09/09/2020.

Key: AE, adverse event; CTCAE, Common Terminology Criteria for Adverse Events; TEAE, treatment emergent adverse event.

Notes: Preferred terms are sorted in descending order of total frequency in the 'Any' column. Adverse events are coded using MedDRA version 23.0 and graded using CTCAE 4.03. Multiple incidences of the same AE in one patient are counted once at the highest grade for that patient. TEAEs include all AEs with an onset on or after initiation of the KTE-X19 infusion. For patients who underwent retreatment with KTE-X19, the AEs occurring during the retreatment period are not included. a, four patients had Grade 5 acute lymphocytic leukaemia, and six patients had other Grade 5 AEs. Source : Shah et al., 2021; ZUMA-3 clinical study report Table 14.3.3.1.1. (42,56).

A summary of AEs related to KTE-X19 that occurred in ≥ 10% of subjects in Phase 2 is provided in Table 31. The most common KTE-X19-related AEs of any grade were

, , and

. The most common KTE-X19-related AEs that were worst Grade 3 or higher were , , and

.

Table 31: Subject incidence of KTE-X19-related AEs occurring in ≥ 10% of subjects by preferred term and worst grade (Phase 2, safety analysis set)

==> picture [468 x 331] intentionally omitted <==

----- Start of picture text -----
MedDRA preferred term, n (%) Any Worst Worst Worst Worst Worst
Grade 1 Grade 2 Grade 3 Grade 4 Grade 5
----- End of picture text -----

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

==> picture [467 x 132] intentionally omitted <==

----- Start of picture text -----
MedDRA preferred term, n (%) Any Worst Worst Worst Worst Worst
Grade 1 Grade 2 Grade 3 Grade 4 Grade 5
----- End of picture text -----

Data cutoff date = 09/09/2020. Key : AE, adverse event; Medical Dictionary for Regulatory Activities. Notes : AEs that occurred during retreatment period are not included. Preferred terms are sorted in descending order of total frequency in the ‘Any’ column. AEs are coded using MedDRA version 23.0 and graded using the Common Terminology Criteriafor Adverse Events version 4.03. Multiple incidences of the same AE in 1 subject are counted once at the highest grade for this subject.

Source : ZUMA-3 clinical study report Table 50 (56).

The most common SAEs at Phase 2 were hypotension (

), pyrexia

( ) and hypoxia ( ). The most common worst Grade 3 or higher SAEs were hypotension ( ), hypoxia ( ) and pyrexia ( ) (Table 32).

) (Table 32).

Table 32: Subject incidence of SAEs occurring in ≥ 2 patients by preferred term and worst grade (Phase 2, safety analysis set)

==> picture [468 x 330] intentionally omitted <==

----- Start of picture text -----
MedDRA preferred term, n (%) Any Worst Worst Worst Worst Worst
Grade 1 Grade 2 Grade 3 Grade 4 Grade 5
Subjects with any serious TEAE 41 (75) )
Hypotension
Pyrexia
Hypoxia
Acute lymphocytic leukaemia
Encephalopathy
Aphasia
Confusional state
Dyspnoea
Pneumonia
Respiratory failure
Seizure
Tachycardia
Fatigue
Febrile neutropenia
Haemophagocytic
lymphohistiocytosis
----- End of picture text -----

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

==> picture [467 x 102] intentionally omitted <==

----- Start of picture text -----
MedDRA preferred term, n (%) Any Worst Worst Worst Worst Worst
Grade 1 Grade 2 Grade 3 Grade 4 Grade 5
Paraparesis
Sepsis
Septic shock
Sinus tachycardia
----- End of picture text -----

Key: AE, adverse event; CTCAE, Common Terminology Criteria for Adverse Events; SAE, serious adverse event; TEAE, treatment emergent adverse event.

Notes : Preferred terms are sorted in descending order of total frequency in the 'Any' column. Adverse events are coded using MedDRA version 23.0 and graded using CTCAE 4.03. Multiple incidences of the same AE in one patient are counted once at the highest grade for that patient. TEAEs include all AEs with an onset on or after initiation of the KTE-X19 infusion. For patients who underwent retreatment with KTE-X19, the AEs occurring during the retreatment period are not included. The safety analysis set compromises 55 patients.

Source : ZUMA-3 clinical study report Table 14.3.4.1.1. (56).

At Phase 2, t patients (

) had at least 1 SAE related to KTE-X19, the

most frequently reported of which were

, and

. The most common worst Grade 3 or higher SAEs related to KTE-X19

were , and

(Table 33)

Table 33: Subject incidence of KTE-X19-related SAEs occurring in ≥ 2 patients by preferred term and worst grade (Phase 2, safety analysis set)

==> picture [467 x 307] intentionally omitted <==

----- Start of picture text -----
MedDRA preferred term, n (%) Any Worst Worst Worst Worst Worst
Grade 1 Grade 2 Grade 3 Grade 4 Grade 5
Subjects with any KTE-X19-related
serious AE
----- End of picture text -----

Data cutoff date = 09/09/2020.

Key : AE, adverse event; CTCAE, Common Terminology Criteria for Adverse Events; SAE, serious adverse event. Notes : AEs that occurred during retreatment period are not included. Preferred terms are sorted in descending order of total frequency in the 'Any' column. Adverse events are coded using MedDRA version 23.0 and graded using CTCAE 4.03. Multiple incidences of the same AE in one patient are counted once at the highest grade for that patient. Source : ZUMA-3 clinical study report Table 14.3.13.1.1. (56).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.2.10.5 Adverse events of special interest

Cytokine release syndrome

In the Phase 2 safety analysis set, 89% (49 of 55 subjects) had CRS, and 24% (13 of 55 subjects) had CRS that was worst Grade 3 or higher. No subject had Grade 5 CRS.

Of the 49 subjects with CRS, the most common CRS symptoms of any grade were pyrexia (46 subjects, 94%), hypotension (33 subjects, 67%), and sinus tachycardia (18 subjects, 37%). The most common worst Grade 3 or higher CRS symptoms were pyrexia (19 subjects, 39%), hypotension (16 subjects, 33%) and hypoxia (11 subjects, 22%) (42).

Among subjects who had CRS, the median time to onset was 5.0 days (range: ) after KTE-X19 infusion. As of the data cut off, CRS had resolved in 46 of 49 subjects. For the remaining 3 subjects, CRS was ongoing at the time of death due to progressed disease (PD) on Day 21 in 1 subject, brain herniation on Day 8 in 1 subject, and pneumonia on Day 15 in 1 subject.

For subjects whose CRS had resolved, the median duration of CRS was days (range: ). Two subjects had CRS with a total duration : 1 subject had CRS for days with a prolonged CRS symptom of Grade 2 nonserious nausea for days, and 1 subject had CRS for days with a prolonged CRS symptom of Grade 1 nonserious increased C-reactive protein (CRP) for days.

Neurological events

In Phase 2, 60% (33 of 55 subjects) had at least 1 neurologic AE of any grade, including 25% (14 of 55 subjects) with worst Grade 3 or higher neurologic AEs. One subject had a Grade 5 neurologic AE of brain herniation (42).

The most common neurologic AEs of any grade were tremor (15 subjects, 27%), confusional state (14 subjects, 25%), and encephalopathy (12 subjects, 22%) (42). The most common worst Grade 3 or higher neurologic AEs were

, encephalopathy (4 subjects, 7%), and

.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Among subjects who had neurologic AEs, the median time to onset was 9.0 days (range: days) after KTE-X19 infusion. As of the data cutoff date, neurologic AEs had resolved in 29 of 33 subjects. Among the 29 subjects whose neurologic AEs had resolved, the median duration of neurologic AEs was days (range: ).

The four subjects with unresolved neurologic AEs either at data cutoff date or time of death are described below:

• One subject had Grade 4 serious encephalopathy, Grade 3 nonserious agitation, and Grade 1 nonserious confusion, which all started on Day 5; Grade 4 cerebral oedema, which started on Day 6; and a fatal event of Grade 5 serious cerebral herniation on Day 8. All events were ongoing at the time of death and were deemed related to KTE-X19.

• One subject had Grade 3 serious paralysis of the lower extremity, which started on Day 10 and was ongoing at the time of death due to PD on Day 553. The event was deemed related to KTE-X19.

• One subject had Grade 3 serious paraparesis, which started on Day 9 and was ongoing at the time of death due to PD on Day 483. The event was deemed unrelated to KTE-X19.

• One subject had Grade 1 nonserious finger numbness, which started on Day 29 and was ongoing as of the data cutoff date. The event was deemed unrelated to KTE-X19.

Other AEs of interest:

• Cytopenias: 49% (27 of 55 subjects) had thrombocytopenia in Phase 2, including 44% (24 of 55 subjects) with worst Grade 3 or higher thrombocytopenia, 49% (27 of 55 subjects) had neutropenia, all of which were worst Grade 3 or higher, and 53% (29 of 55 subjects had anaemia at Phase 2, including 49% (27 of 55 subjects) with worst Grade 3 or higher anaemias.

• Infections: 40% (22 of 55 subjects) had infections in Phase 2, including 25% (14 of 55 subjects) with worst Grade 3 or higher infections.

• Hypogammaglobulinaemia: 7% (4 of 55 subjects) had

hypogammaglobulinaemia in Phase 2, none of which was Grade 3 or higher

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Tocilizumab was given to 80% (44 of 55 subjects), steroids were given to 75% (41 of 55 subjects) and vasopressors were given to 40% (22 of 55 subjects) (42).

B.2.10.6 Safety overview

The safety profile observed in ZUMA-3 is similar to that observed with other CAR T- cell therapies, and the risk management protocol for KTE-X19 is well established, typified by CRS and neurological events that are the most prominent toxicities of cellular immunotherapy. The safety profile of KTE-X19 in ZUMA-3 was generally similar to that observed in other indications, although a higher incidence of Grade 3 or higher CRS was observed. This was also the case for tisagenlecleucel, where CRS was more commonly observed in ALL compared to other indications (73).

ALL is clinically associated with cytopenias of 1 or more lineage, with cytopenias and infections among the most common AEs observed with SoC therapies. Across TOWER and INO-VATE, grade ≥ 3 neutropenia occurred in 38%-58% of subjects, with Grade 3 or higher infections occurring in 34% of subjects treated with blinatumomab and 52% treated with chemotherapy in TOWER (25,74). The rate of these identified risks in ZUMA-3 was therefore similar to those observed across different studies and treatment modalities, consistent with the underlying disease.

Notably, health-related quality of life (HRQoL) data presented in Section B.2.6.2 suggest no long-term impact on QoL, with results either stabilising or improving versus baseline from Day 28 to Month 12.

In addition, in a long-term analysis of patients treated at the pivotal dose level in Phase 1, no new safety signals were observed after the median follow-up time of 39.9 months, indicating favourable long-term safety in R/R B-precursor ALL (42).

Of the 55 subjects treated with KTE-X19 at Phase 2 in ZUMA-3, there were 2 deaths considered related to treatment. This compares favourably with SCT, where treatment-related mortality rates of 20-40% are typically observed, even with reduced-intensity conditioning (75,76).

Since the approval of tisagenlecleucel, axicabtagene ciloleucel, and KTE-X19 (in mantle cell lymphoma) through the Cancer Drugs Fund (CDF) in NHS England,

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

clinicians are increasingly experienced and familiar with toxicity management for CAR T-cell therapy in UK clinical practice. Real-world data of CD19 CAR T-cell therapy - albeit in high-grade lymphoma – in England showed lower rates of Grade ≥ 3 CRS or neurological events, with increased use of tocilizumab and steroids compared to pivotal trials (77). When taken in the context of the growing familiarity and knowledge of the CAR T-cell safety profile, as well as better patient selection and overall management, we may therefore expect a similar translation of ZUMA-3 safety data into clinical practice.

As recommended in the draft SmPC for KTE-X19, patients should be monitored daily for the first 10 days following infusion of KTE-X19 for signs and symptoms of potential CRS, neurological events, and other toxicities. Physicians should consider hospitalisation for the first 10 days post infusion or at the first signs/symptoms of CRS and/or neurological events; after the first 10 days following the infusion, the patient should be monitored at the physician’s discretion. Patient should be instructed to remain within proximity (within 2 hours of travel) of a qualified treatment centre for at least 4 weeks following infusion.

B.2.11 Ongoing studies

ZUMA-3 is ongoing and will provide additional evidence of KTE-X19 for the treatment of adults with r/r ALL. Patients will be followed up to 15 years after the last patient received KTE-X19. On this basis, the final study completion date is estimated to be September 2035.

Preliminary results from the most recent analysis with data cut-off 23/07/21 provides longer-term data on the durability of all patients treated with KTE-X19 at target dose. Whilst more detail will be made available through the evaluation process, key results are presented in Section B.2.6. The CSR for this data cutoff is expected to be available in January 2022.

B.2.12 Innovation

KTE-X19 is a personalised medicine in which the patient’s own T cells are collected and engineered ex-vivo to express a chimeric antigen receptor which programmes them to target and kill the cancer cells when they are returned to the patient in a

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

single infusion. The production of KTE-X19 includes a specific step designed to remove tumour cells from the leukapheresis harvest and enrich for mature T-cells. This is unique to the production of KTE-X19, distinguishing it from axicabtagene ciloleucel.

KTE-X19 represents a breakthrough treatment in the R/R adult ALL setting. Whilst SCT offers a potentially curative treatment option with 5-year OS rates of 23% in R/R ALL, a significant portion of R/R adult ALL patients relapse post-SCT, are ineligible for SCT, or are unlikely to achieve a CR required for SCT (9). As discussed in Section B.1.3.4.3, in the absence of consolidation with SCT, current treatment options are life-extending but not curative and are associated with OS of 4-8 months. Demonstrating an unprecedented median OS estimate of months at latest data cutoff, with and of subjects estimated to be alive at 12 and 18 months respectively, and OS appearing independent of subsequent SCT, KTE-X19 represents a significant advancement for this patient population. The hope KTE-X19 could offer to patients, carers and healthcare professionals should not be undervalued.

Collectively, the outcomes achieved with current treatments (see section B.1.3.5) highlight the need for additional therapies such as KTE-X19 that can induce deeper and more durable responses and potentially achieve long-term survival in adult patients with R/R B-ALL, particularly in patients who have relapsed post-SCT or are ineligible for SCT, or those with particularly poor prognostic indicators that mean they are unlikely to achieve SCT, such as primary refractory disease or first relapse within 12 months. In these patient populations, KTE-X19 represents a paradigm shift as a potentially curative treatment option.

While the main health-related benefits will have been captured in the QALYs for KTE-X19, it is difficult to capture true innovation in such a calculation, and the significant difference this treatment choice could make to patients, carers and healthcare services is such that KTE-X19 access would represent a step change in management of R/R adult ALL.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.2.13 Interpretation of clinical effectiveness and safety evidence

B.2.13.1 Principal findings from the clinical evidence

The efficacy and safety of KTE-X19 in the treatment of adults with B-precursor R/R ALL has been demonstrated in the open-label, multi-centre, ZUMA-3 trial.

The ZUMA-3 Phase 1 + 2 combined population represents an adult R/R ALL group with an especially poor prognosis, including almost half ( ) with prior blinatumomab treatment, and a similar proportion receiving ≥3 prior treatments at baseline ( ) (Table 8). In addition, had relapsed post-SCT, a group with a particularly dire outlook, where median OS is 5.5 months (see Section B.1.3.5) (43).

Among all patients to receive KTE-X19 at target dose, almost three-quarters of subjects (74.4%) achieved an OCR, including 62.8% who achieved CR.

Additionally, 79.5% (62 of 78 subjects) had no detectable cancer cells remaining as demonstrated by MRD negativity, including all but one patient – for whom data was not available – to achieve CR/CRi. The survival advantage of achieving MRD negativity in both adults and children has been demonstrated in a previous 39-study meta-analysis (albeit following induction therapy), and was further re-enforced by long term blinatumomab data (65,66). The KM median DOR for the 58 subjects who achieved CR or CRi was months.

At the time of primary data cut-off, providing a median actual follow-up of months for all treated subjects, of subjects were known to be alive. KM estimates of OS at 12 and 18 months were and , respectively. This OS is unprecedented in the adult R/R ALL population, particularly in the context of the heavily pre-treated population recruited to ZUMA-3, with currently approved treatments demonstrating median OS of 4-8 months in pivotal trials (25,26).

B.2.13.2 Strengths and limitations of the evidence base

The autologous cellular therapy nature of KTE-X19 necessitates open-label treatment.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

The response rates achieved in ZUMA-3 compare favourably to those observed with currently available treatment options for adults with R/R ALL. In a meta-analysis of OCR rate using the ZUMA-3 Phase 2 mITT dataset, only the Phase 3 TOWER study, which compared blinatumomab to SoC chemotherapy, used the same stringent haematologic recovery criteria for OCR as used in ZUMA-3 (78). The pooled estimates of OCR rate in the TOWER study was 32%. By contrast, ZUMA-3 demonstrated an OCR rate of 70.9% in the mITT analysis set. Furthermore, a further meta-analysis of 12 studies focusing on CR yielded a pooled CR rate of 30%, compared to the 56.4% rate observed in ZUMA-3. Additionally median OS in ZUMA3 Phase 2 mITT (18.2 months) compared favourably to the pooled estimate for SoC (6.9 months) (78).

It should also be noted that almost all of the studies included in the 12-study metaanalysis enrolled subjects who were naïve to blinatumomab and inotuzumab. Outcomes in the setting of blinatumomab or inotuzumab failure have not been well studied to date, although limited reports indicate that once patients fail blinatumomab, responses to subsequent lines of therapy deteriorate, leaving patients with very limited options (59,60). In the ZUMA-3 study, the OCR rate in subjects even after prior blinatumomab treatment (accounting for nearly half of those in the Phase 1 + 2 combined dataset [ of 78 subjects]) remained high at . For the of 78 subjects with prior inotuzumab, the OCR rate was (Figure 63).

In addition to the observed survival benefits, results of the EQ-5D-5L and VAS suggest that long-term HRQoL is not negatively impacted by KTE-X19 therapy.

Given the high CR rate and magnitude of improvements in DOR, OS, and RFS observed with KTE-X19 when compared with currently available treatments, as well as the high unmet need in this patient population, including elderly patients, KTEX19 represents an important new therapeutic option for patients with R/R B-ALL. Overall, the results of ZUMA-3 demonstrate a clinically meaningful benefit over currently available therapies and a positive benefit-risk profile of KTE-X19 for the treatment of R/R B-ALL.

To address the evidence gap regarding long-term benefit, a series of survival

scenarios have been modelled within the cost-effectiveness analyses presented in Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

section B.3. Kite, a Gilead Company, are also open to KTE-X19 being a CDF candidate to facilitate timely patient access alongside longer-term data collection.

B.2.13.3 Applicability of clinical evidence to practice

B.2.13.3.1 Patient characteristics

The combined Phase 1 + 2 population of ZUMA-3 represents a heavily pre-treated patient group, with having received ≥3 prior treatments at baseline. Subjects had failed a range of standard treatment options, including allo-SCT ( ), blinatumomab ( ), and inotuzumab ( ). NICE guidelines recommend the latter two targeted therapies for the treatment of Ph- R/R adult ALL, and R/R adult ALL, respectively. Blinatumomab is also recommended for Ph- R/R adult ALL patients who achieve remission with MRD+ (>0.1%) if the disease is in first remission. These recommendations combined with clinical feedback inform us that these two therapies are SoC for R/R adult ALL, and therefore prior use in ZUMA-3 can be considered representative of UK clinical practice (35). In addition, of subjects enrolled in ZUMA-3 Phase 2 were Ph+, consistent with the 22% Ph+ in UK R/R clinical practice (9).

Whilst the population enrolled to ZUMA-3 was heterogeneous, this is representative of the R/R adult ALL population, which is heterogeneous in nature. Although ZUMA3 did not include any UK sites, there were study sites in France, Germany, and the Netherlands, where treatment and management of R/R adult ALL is likely to be influenced by ESMO guidelines (see Section B.1.3.4.1).

Based on feedback received from clinical experts, we understand that KTE-X19 would be positioned for patients who have relapsed post-SCT, are ineligible for SCT, or unlikely to be able to achieve an SCT via current SoC (35). It is noteworthy that 100% of those treated in ZUMA-3 meet this positioning criteria. The patient characteristics at baseline can therefore be considered generalisable to those who are likely to receive KTE-X19 in clinical practice.

It should also be acknowledged that 14 of 78 subjects (17.9%) with a KTE-X19 induced-remission received a subsequent allo-SCT; 9 subjects had achieved CR, 3 subjects had achieved a CRi. In addition, 1 subject had achieved CRi by investigator Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

assessment but BFBM per central assessment, while a further subject with extramedullary disease had achieved partial remission (PR). We understand that consolidating CAR T-cell induced remission with an allo-SCT is not considered an appropriate option in the UK (35). If a patient suffers a relapse following KTE-X19 treatment, a subsequent allo-SCT may theoretically be an option, but clinical expert opinion suggests that the number of people who would be candidates for a subsequent allo-SCT at this stage would be negligible (35,79).

Of note is that 13 of the 14 of the subjects that received an allo-SCT in ZUMA-3 had not received a prior allo-SCT at baseline. In the anticipated positioning of KTE-X19 in the high unmet need population who have relapsed post-SCT or are

ineligible/unlikely to achieve SCT, all patients will have either had prior SCT, or be ineligible. Based on clinical feedback, we understand that a second allo-SCT is not considered a viable treatment approach in the UK, even though it is permitted in certain cases (35). KTE-X19 can therefore be considered a standalone treatment option.

Notably, the median OS at the most recent data cutoff for the Phase 1+2 combined dataset was comparable when subjects were censored (main analysis) or were not censored (sensitivity analysis) at the time of allo-SCT, further supporting use of KTEX19 as a standalone therapy (Figure 12) (Figure 22).

B.2.13.3.2 Analysis sets

In consideration of the most appropriate analysis set for decision making, the KTEX19 Phase 1 + 2 combined dataset (N = 78) is presented and used in the subsequent cost-effectiveness analysis. This analysis set provides data on all subjects treated with KTE-X19 at the anticipated dose of the EU marketing authorisation and provides the longest follow-up on treated subjects. This analysis set provides data for all treated patients, irrespective of follow-up.

B.2.13.3.3 Service provision

The manufacturing process of KTE-X19 has a unique step whereby tumour cells are removed from the leukapheresis harvest prior to ex-vivo expansion of patient T-cells. This should help KTE-X19 manufacturing attempts to be successful first-time and

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

facilitate prompt delivery of KTE-X19 to the patient. This is particularly important with a rapidly progressing disease where being able to manufacture the CAR T product successfully during the first attempt is important for ensuring the patient receives therapy as promptly as possible. In June 2020, a European CAR T-cell manufacturing facility was approved for use by the EMA, with the objective of substantially reducing time from apheresis to Qualified Person release, while avoiding transatlantic transport, thus enabling a faster product manufacturing period for European patients. In ZUMA-3, KTE-X19 was successfully manufactured for 53 of 54 subjects at Phase 1, and 65 of 71 subjects at Phase 2. The median time between leukapheresis and KTE-X19 infusion was days (range: days) for patients in the US, and days (range: days) for patients in the EU. All KTEX19 product administered within the ZUMA-3 trial was manufactured in the US; the manufacturing times are expected to reduce in the EU when the manufacturing facility in the Netherlands is able to manufacture KTE-X19, currently anticipated to be .

Importantly, KTE-X19 does not have additional or different infrastructure and personnel needs compared with other CAR T-cell therapies and therefore would fit into current service provisions for such treatment, already set up within NHS England.

B.2.13.3.4 End-of-life

KTE-X19 satisfies the criteria to be considered an effective end-of-life therapy. Previous pivotal trials in R/R adult ALL demonstrate a median OS with current SoC of 4-8 months (25,26). Taken in the context of a median OS of months demonstrated for treated patients in ZUMA-3, KTE-X19 is expected to extend this life expectancy by far more than the requisite 3 months.

It should also be noted that blinatumomab, inotuzumab, and ponatinib were all considered to meet the criteria for end-of-life (80–82). KTE-X19 eligibility for end-oflife is presented in Table 34.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 34: End-of-life criteria

Key : ITT, intention-to-treat; OS, overall survival

Note that the clinical data refers to patients that receive KTE-X19 infusion in ZUMA-3 only, while the model output accounts for the survival of patients that did not receive KTE-X19 infusion

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.3 Cost effectiveness

B.3.1 Published cost-effectiveness studies

An SLR was conducted to identify economic evidence within the relapsed/refractory B-precursor ALL indication (adult population patients of ≥18 years) to inform a health-economic model. A detailed description of the methods underpinning the SLR is provided in Appendix G. Following full text screening, 14 economic evaluation studies were identified from the SLR.

Two of the studies presented economic evaluations based on TOWER, a randomized, open-label, phase III clinical trial which compared blinatumomab with standard of care chemotherapy in adult patients with relapsed/refractory ALL. Four studies were economic evaluations based on INO-VATE, a randomized, open-label, phase III clinical trial in which inotuzumab ozogamicin was compared to investigator’s choice of chemotherapy regimen in adult patients with relapsed/refractory CD22-positive ALL. Furthermore, 3 studies based their economic evaluations on both the TOWER and the INO-VATE trials. One study was based on the ALCANTARA study, which was a Phase 2 study of blinatumomab in adult subjects with relapsed/refractory Ph+ B-precursor ALL (83). The remaining publications did not provide information regarding whether the evaluations utilised outcomes from clinical trials. A summary of the identified studies is provided in Table 35.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 35: Summary list of published cost-effectiveness studies

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 127 of 270

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 128 of 270

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 129 of 270

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 130 of 270

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 131 of 270

Key : ICER, incremental cost-effectiveness ratio; QALYs, quality-adjusted life years

*Three separate analyses were conducted: (1) InO OS obtained by applying the anchored MAIC-adjusted hazard ratio (HR) for InO vs. blinatumomab (1.40, 95%CI 0.87-2.24) to blinatumomab MAIC-adjusted TOWER OS (Gompertz distribution); (2) InO and SoC OS based on Weibull mixture cure distribution fit to INO-VATE-ALL OS; blinatumomab OS obtained by applying MAIC-adjusted HR for blinatumomab vs. SoC (0.55, 95%CI 0.38-0.80) to INO-VATE-ALL SoC OS; and (3) unanchored comparison of MAIC-adjusted Weibull mixture cure fit to blinatumomab OS from TOWER and InO OS from INO-VATE-ALL.

** Three analyses were conducted based on alternative approach for the MAIC (anchored through standard of care [SoC] vs unanchored), proportional hazards (PH) assumptions, and reference overall survival (OS) distributions. Complete remission rates, utilities, duration of therapy, and use of subsequent therapies also were MAIC-adjusted.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.3.2 Economic analysis

None of the identified cost-effectiveness studies addressed the decision problem. A de novo cost-effectiveness model was thus developed for the economic analysis. The model structure was informed by the previous NICE appraisals of CAR T-cell therapies in diffuse large B-cell lymphoma (DLBCL) (94) and in R/R mantle cell lymphoma (MCL) (95) (TA559 and TA677).

B.3.2.1 Patient population

The patient population considered in the analysis is adults (≥18 years old) with relapsed or refractory (R/R) B-precursor ALL for whom SCT is not indicated. This is in line with the anticipated market authorisation for KTE-X19 in R/R B-cell ALL and the pivotal trial evaluating KTE-X19 in R/R B-cell ALL: ZUMA-3 (41,42). This population is considered generalisable to the anticipated positioning of KTE-X19 in the UK according to clinical expert opinion (see section B.1.3.4.4):

• Relapsed post SCT

• Ineligible for SCT (on the basis of age, frailty or other exclusion criteria)

• Unlikely to achieve SCT via existing bridging therapies (primary refractory, relapsed within 12 months, failed ≥2 lines of prior therapy)

The economic analysis was performed for three different patient populations:

• overall population (mITT population of ZUMA-3)

• Ph- population

• Ph+ population

All three populations are considered to be of clinical relevance to decision makers in the R/R ALL treatment landscape since the comparator regimens differ based on Ph expression. The ZUMA-3 trial was not powered to detect outcome differences by Ph status. However, unlike with some targeted therapies (TKIs), clinical experts did not expect the effectiveness of KTE-X19 to differ based on the Ph status of the patients.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.3.2.2 Model structure

The cost-effectiveness model was built as a three-state partitioned survival model (Figure 33). The partitioned survival model comprises three mutually exclusive health states: EFS, PD, and death.

All patients enter the model in the EFS state. Patients who achieve CR or CRi remain in the EFS state, while those who do not achieve CR or who relapse or progress transition to the PD state. From the EFS health state, patients can transition to either the PD or death health state. Following progression, patients can only transition to the death state, an absorbing health state. The model uses a weekly cycle length. As is common in partitioned survival models, transitions across health states were not explicitly modelled. The health state occupancy at each model cycle was determined from the cumulative survival probabilities derived from independently modelled EFS and OS curves, for both the intervention and comparators:

• The EFS curve enabled the modelling of patients in the EFS health state at each cycle (patients’ event-free and alive)

• The proportion of patients occupying the PD health state at each cycle was estimated by subtracting the proportion of patients that were event-free and alive (EFS curve) from the proportion of patients alive (OS curve)

• Patients occupying the death state at each cycle were estimated by subtracting the proportion of patients alive (OS curve) from the total cohort.

The partitioned survival model structure reflects the clinical pathway of the disease; once patients progress, they cannot return to the EFS health state.

In the partitioned survival model, patients alive at 3 years are assumed to be ‘cured’ and are thus considered to be long-term survivors. This assumption is applied for both KTE-X19 and the comparators.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

The possibility of achieving cure is an accepted outcome in ALL, usually following allo-SCT (96) and has been accepted as an outcome following CAR T-cell therapy in other indications appraised by NICE (94), (97), (95). The factors contributing to defining it are mainly centered on deep and durable eradication of cancer cells linked with prolonged survival. Although there were various discussions on the time of cure and the estimated cure fraction in the technology appraisals for blinatumomab (NICE TA450) and tisagenlecleucel (NICE TA554) (79,80), the committees were in favour of the assumption that patients who survive beyond 2 to 5 years are effectively cured. Figure 16 (2020 September data cut, see section B.2.6.1.2) shows that KTEX19 was associated with a survival plateau where no death events were reported for a period of 25 months among patients remaining in the study (mITT, Phase 1 and Phase 2 combined). The latest ZUMA-3 data cut (2021 July, too recent to be incorporated into the model) demonstrates that around 40% of the ZUMA-3 mITT Phase 1 and Phase 2 combined population is alive from 3 years (see section B.2.6.1.1, Figure 11). This indicates that among patients achieving a response, a proportion of R/R B-cell ALL patients may achieve long-term remission following treatment with KTE-X19. Furthermore, clinical experts were supportive that KTE-X19 could be positioned as a potentially curative, standalone therapy in R/R ALL (35).

In previous NICE appraisals of CAR T-cell therapies in diffuse large B-cell lymphoma (DLBCL) and in R/R mantle cell lymphoma (MCL) (TA559, TA567, and TA677) (94), (97), (95), it was assumed that cured patients, albeit having heightened risk of death versus the age-equivalent general population, incur lower resource use and have improved HRQoL compared to non-cured patients. Similarly in the current analysis, patients assumed to be cured (those alive beyond the 3-year time-point) incur an increased risk of death (excess mortality) compared to the general population. A standardised mortality ratio (SMR) of 1.09 was applied to the background mortality (section B.3.3.3). The SMR of 1.09 has been sourced from a study in DLBCL, Maurer et al., 2014 (98), which was used by the company in the most recent NICE appraisal for KTE-X19 in mantle cell lymphoma (TA677) and was the ERG’s preferred SMR in TA567 (Tisagenlecleucel in R/R DLBCL). Although no long-term data are available that compare outcomes post allo-SCT in R/R DLBCL vs. those in R/R ALL, short-term outcomes (up to 2 years) on current SoC for DLBCL are very

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

similar to those observed in the blinatumomab and inotuzumab R/R ALL clinical studies (99) (Table 110). Furthermore, a recent study from Australia and New Zealand (Kliman et al. 2020) showed that recipients of allo-SCT who survived to at least 2 years and were disease-free continued to experience long-term outcomes close to the general population (100). Survival ratios ranged from 96% to 99% of the age matched population per year (lowest CI above 94%) (100). 25% of the cohort included in the Kilman et al., 2020 study had a diagnosis of ALL (100).

Patients alive at the 3-year time-point in the partitioned survival model then incur general population utility. With the current ZUMA-3 data cut, RFS (used as a proxy for EFS, see section B.3.3.2.1) modelled curves appear to plateau at 20-25% whereas the OS curves suggest a plateau around 40%. This would result in a proportion of patients in the PD health state being alive for a long time which is not compatible with the pathology. This is because the way RFS KM are derived does not allow for robustly informative extrapolation:

• The curves start from lower probability of survival (excluding the nonresponders, looking at RFS curves for only CR/CRi patients the RFS at 2-3 year is more aligned to the plateau seen for OS (~35-40%)

• There is also a high level of censoring 40% consisting mainly of patients in ongoing remission (15%) and patients who received a subsequent allo-SCT (18%), representing a proportion without progression of 33% again much more aligned with the OS plateau ~40%.

This assumption is applied for both intervention and comparators as it is not a ZUMA-3 specific issue but is seen also in other studies:

• The INO-VATE modelled OS curve plateau around 16% at 3 years, while the EFS modelled curve plateau around 8% at 3 years

• The SCHOLAR-3 SCA-3 modelled OS curve plateau at 11% at 3 years, while the EFS modelled curve plateau at 0% at 3 years.

The structure and the health states are in line with the primary objectives of the

treatment in R/R ALL: avoiding disease progression, avoiding worsening in quality of Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

life and extending survival. In addition, this model structure is in line with previous technology appraisals submitted to NICE in R/R B-cell ALL (79,80).

While the survival impact of allo-SCT is not explicitly considered in the model structure (as no SCT-related health state is included), the impact of allo-SCT from a costing and quality of life perspective is accounted for. For each treatment in the model, the impact of allo-SCT is accounted for upfront by weighting the cost and utility impact associated with subsequent allo-SCT by the reported proportion of patients receiving allo-SCT, as observed in the different studies (TOWER, INOVATE, PACE). Although 14 patients received SCT in the ZUMA-3 mITT Phase 1 and Phase 2 combined trial (14/78=17.9%), no KTE-X19 patients are assumed to receive allo-SCT in the model. KTE-X19 expected positioning is for patients who have either relapsed following SCT or are considered unlikely to be able to achieve an SCT. According to UK clinical experts, no patients would receive a second allo-SCT and allo-SCT is not expected to be given as consolidation following a CAR T-cell therapy (see section B.3.5.3.1). Sensitivity analyses which censored patients who had received an SCT in ZUMA-3 showed no difference in survival outcomes in both the earlier and later data cut (Figure 22).

In the de novo model, for the patients in the KTE-X19 arm who underwent leukapheresis but did not go on to receive KTE-X19 infusion in ZUMA-3, rather than modelling this as an initial decision tree, this was instead accounted for by using cost multipliers. This is consistent with the approach used in TA559 and in TA677.

Figure 33: Partitioned survival model structure

==> picture [332 x 174] intentionally omitted <==

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.3.2.2.1 Features of the de novo analysis

The analysis was conducted from the perspective of the NHS and personal social services (PSS) in England. The base-case analysis thus considered only direct healthcare costs, whilst broader societal costs were explored in a scenario analysis. The model spans a lifetime horizon. Given that the mean age of the patients at the start of the analysis was 43 years, a time horizon of 57 years was deemed sufficient to align with the maximum life expectancy of patients. A 20-year time horizon was explored in scenario analysis. Costs and outcomes were discounted at an annual rate of 3.5%, in line with the NICE reference case (101). Different discount rates were explored in the scenario analysis given that KTE-X19 has the potential to restore patients with a short life expectancy to near-full health over an extended period.

A summary of the key features of de novo economic analysis and their justification is provided in Table 36. A comparison is provided with the features of previous appraisals in R/R ALL including both adult and paediatric populations.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 36: Features of the economic analysis

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 139 of 270

Key : TA, technical appraisal

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 140 of 270

B.3.2.3 Intervention technology and comparators

B.3.2.3.1 Intervention: KTE-X19

KTE-X19 is a single-infusion product, for autologous and intravenous use only, administered at a target dose of 1 x 10[6 ] anti-CD19 CAR T-cells/kg. Prior to infusion, patients are treated with a nonmyeloablative conditioning regimen consisting of fludarabine 25 mg/m[2] /day administered intravenously (IV) for 3 days starting 4 days before planned infusion and cyclophosphamide 900 mg/m[2] /day administered intravenously (IV) for 1 day 2 days before planned infusion. Bridging chemotherapy (administered after leukapheresis and before conditioning chemotherapy) is recommended for all patients, particularly those with high disease burden at baseline.

The dosing for both the pre-infusion conditioning chemotherapy and KTE-X19 is based on the doses used in ZUMA-3 Phase 1 and Phase 2 combined.

B.3.2.3.2 Comparators

The comparators considered in the cost-effectiveness model represent the current SoC for patients with R/R ALL in the UK. The comparators align with the most recent (July 2021) NICE pathway for treating relapsed or refractory acute lymphoblastic leukaemia (29), clinical expert opinion and the final NICE scope for KTE-X19. The comparator technologies were differentiated for the 3 different patient populations:

• Overall population (irrespective of Ph expression): inotuzumab ozogamicin and FLAG-IDA

• Ph- population: blinatumomab, inotuzumab ozogamicin, and FLAG-IDA

• Ph+ population: ponatinib, inotuzumab ozogamicin, and FLAG-IDA.

1 Inotuzumab ozogamicin

As per NICE TA541 (81), inotuzumab ozogamicin is recommended as an option for treating R/R CD22 positive B-cell precursor ALL in adults. Patients who have Ph+ disease should have received at least 1 tyrosine kinase inhibitor.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Inotuzumab dosing reflected in the model was in line with INO-VATE study (26). Inotuzumab is administered IV at a dose of 0.8 mg/m[2 ] on day 1, 0.5mg/m[2 ] on day 8 and day 15 in cycle 1 (21-day cycle). From cycle 2 onwards (28-day cycles) it is administered 0.8 mg/m[2 ] or 0.5 mg/m[2 ] on day 1, 0.5 mg/m[2 ] on day 8 and day 15. Treatment may continue up to 6 cycles. In the model it was assumed that patients on inotuzumab would receive on average 3 cycles of therapy, in line with the INO-VATE study results (the median number of cycles received in INO-VATE for inotuzumab patients was 3).

2 Blinatumomab

As per NICE TA450 (80), blinatumomab is recommended as an option for treating Ph- R/R B-cell ALL in adults. Blinatumomab dosing in the model was in line with the TOWER study (25). Blinatumomab is administered IV at a dose of 9 μg/day during week 1 of cycle 1 then 28 μg/day for the remainder of the cycle and during subsequent cycles (28-day cycles) followed by a treatment-free interval of 2 weeks. In the model, it was assumed that patients on blinatumomab would receive on average 1.45 cycles, in line with Von Stackelberg et al. (2016) and TA554 (79,109).

3 Ponatinib

As per NICE TA451, ponatinib is recommended as an option for treating Ph+ R/R ALL when either:

• the disease is resistant to dasatinib

• the subject is intolerant to dasatinib (or)

• when a T315I gene mutation is present.

Ponatinib dosing reflected in the model was in line with PACE study (110). Ponatinib is administered orally at a daily dose of 45 mg/day. In the model, it was assumed that patients on ponatinib would be treated until disease progression (the ponatinib EFS curve was used as proxy to reflect patients on treatment) or for a maximum of 3 months, in line with the ponatinib SmPC (111). Based on UK clinical expert opinion, ponatinib was assumed to be given in combination with chemotherapy. In the absence of clinical data to inform outcomes, this was accounted for within the costs

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

only. As no information regarding the precise regimen could be identified, we assumed the costs of FLAG-IDA in the economic model

4 FLAG-IDA

According to clinical expert opinion, FLAG-IDA is the most commonly used salvage chemotherapy regimen for treating adults with R/R B-cell ALL and this regimen was used as the basis for drug costs. The dosing of each FLAG-IDA therapy is provided below:

• Fludarabine: 30 mg/m[2 ] for 5 consecutive days per 28-day cycle

• Cytarabine: 2 g/m[2 ] for 6 consecutive days per 28-day cycle

• Filgrastim: 0.005 mg/kg for 9 total days

• Idarubicin: 8 mg/m[2 ] for 3 days.

In the model, it was assumed that patients on FLAG-IDA would be treated until disease progression (FLAG-IDA EFS curve used as proxy to reflect patients on treatment) or for a maximum of 4 28-day cycle, in line with UK clinical practice.

B.3.3 Clinical parameters and variables

The model has been constructed to analyse a number of different combinations of population, comparator, ITC data and survival analyses. An overview of the different types of analyses available in the model and the choice of base case and sensitivity analyses is provided in Table 37.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 37: Summary of approach to modelling of clinical parameters

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 144 of 270

Key : IPD: individual patient-level data; ITC, indirect treatment comparison; MCM, mixture-cure model; mITT, modified intention-to-treat; SPM, standard parametric model

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 145 of 270

B.3.3.1 Baseline characteristics

The baseline characteristics for the modelled population are provided in Table 38. These parameters were informed by the baseline characteristics of the patients who received CAR T-cell infusion: mITT in the ZUMA-3 Phase 2 trial and in the subgroup of patients who received the same CAR T-cell dose in the ZUMA-3 Phase 1 trials of KTE-X19 (Phase 1 and Phase 2 combined dataset). The mITT population was used in the model base case in order to predict outcomes specific to those patients receiving KTE-X19.

The patients enrolled in Phase 1 and Phase 2 combined ZUMA-3 were considered appropriate to support the analysis, as discussed in sections B.2.3.2. The patient baseline characteristics adopted in the base case economic analysis are reported in Table 38.

Table 38: Patient baseline characteristics in the base case economic analysis

Key : BSA, body surface area; mITT, modified intention-to-treat Source: ZUMA-3 CSR (56)

The mean age, proportion of females and England life tables (2018-2020) were used to calculate general population background mortality (112). The mean age was also used to calculate age-related utility decrements. The mean weight and mean body surface area were used to calculate treatment dosage for those treatments whose posology is based on weight or body surface area.

Baseline characteristics from the overall ZUMA-3 population, irrespective of Ph expression, were adopted for all the subgroups included in the economic analysis: overall population, Ph- subgroup, and Ph+ subgroup. Clinical experts had confirmed that Ph status was unlikely to affect the clinical effectiveness of KTE (35,71).

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

B.3.3.2 Clinical efficacy inputs

The primary efficacy outcomes adopted within the economic model were EFS and OS. These endpoints allowed us to define health state occupancy in the partitioned survival model.

The ZUMA-3 trial was designed as a single-arm trial and was thus used to inform EFS and OS for KTE-X19 only. An SLR was conducted to identify relevant published data for the comparators in adult patients with R/R B-cell ALL (see section B.2.1). For each of the modelled comparators, pseudo IPD for EFS and OS were generated using the algorithm by Guyot et al. (2012) (113), after having digitised the published KM plots and using associated event information.

As explained in section B.2.9, a number of ITC approaches were used to compare the efficacy of KTE-X19 with the comparators in the scope, including naïve ITCs, MAICs and a synthetic control arm, SCHOLAR-3 SCA-3, for blinatumomab. KTEX19 is expected to be positioned for patients who have either failed or are unlikely to achieve SCT, whereas the comparators are used as bridging therapy to allow patients to receive a potentially curative allo-SCT. They are therefore expected to be used in subtly different populations in clinical practice. As 100% of the ZUMA-3 patients are generalisable to its anticipated UK positioning, ZUMA-3 should provide the target population for any adjustments. As explained in section B.2.9, naïve comparisons were therefore considered to be the most representative base case for all comparators other than blinatumomab, for which an adjusted comparison to the ZUMA-3 population was available from SCHOLAR-3 SCA-3. In the latter analysis the blinatumomab IPD had been matched to the appropriate target population, that of ZUMA-3. As the OS HR point estimate in the naïve ITC against blinatumomab was identical to that from the SCHOLAR-3 adjusted comparison, this provided further justification that the results from naïve comparisons might also be more reliable than the MAICs for the other comparators.

Analysis of the IPD informing the naïve ITC and MAIC demonstrated the proportional hazards assumption to be violated in both the naïve and adjusted comparisons (71), therefore all survival analyses for the economic model were carried out by fitting survival curves independently to each comparator.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 39 provides an overview of the sources used to inform intervention and comparators’ clinical efficacy in each subgroup. Rationale for these choices, summarised above, has also previously been provided in section B.2.9.

Table 39: Summary of data sources adopted for different subgroups in the economic model – base case

Key: mITT, modified intention to treat; SCA, synthetic control arm

B.3.3.2.1 KTE-X19

Consistent with the patient baseline characteristics, EFS and OS inputs for KTE-X19 were based on the analysis of patients who received CAR T-cell infusion, mITT (n=78: ZUMA-3 Phase 2 n=55 and ZUMA-3 Phase 1 n=23). The IPD from ZUMA-3 Phase 1 and Phase 2 combined were directly adopted to inform KTE-X19 EFS and OS. Relapse-free survival (RFS) was used as a proxy for EFS as this was the endpoint in ZUMA-3.

In the base case, only data for patients who successfully received CAR T-cell infusion (mITT dataset) was used to provide EFS and OS inputs in the economic

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

model. This approach accurately reflected the clinical benefits of KTE-X19 without the confounding of those patients that did not receive CAR T-cell infusion (i.e., received other treatments).

Different assumptions were applied to estimate EFS and OS for patients who failed to receive CAR T-cell infusion based on the reasons for failing to receive KTE-X19 as observed in ZUMA-3:

• Proportion of patients who received CAR T-cell infusion followed KTE-X19 EFS and OS curves (incurring corresponding treatment costs and outcomes)

• Proportion of patients who failed to received CAR T-cell infusion due to adverse events followed FLAG-IDA EFS and OS curves (incurring corresponding treatment costs and outcomes)

• Proportion of patients who failed to received CAR T-cell infusion due to other reasons (e.g. manufacturing failure) followed relevant comparators’ EFS and OS curves based on the subgroup under evaluation (incurring corresponding treatment costs and outcomes).

The assumption of modelling patients who failed to receive CAR T-cell infusion due to adverse events as being treated with FLAG-IDA was the same suggested by the ERG in NICE TA554. It allowed representation of the poor prognosis of patients failing to receive CAR T-cell infusion due to adverse events.

Two scenario analyses were performed to test the above assumption:

• Proportion of patients who failed to received CAR T-cell infusion (regardless of the reason) followed FLAG-IDA EFS and OS curves (incurring corresponding treatment costs and outcomes)

• Proportion of patients who failed to received CAR T-cell infusion (regardless of the reason) followed relevant comparators’ EFS and OS excluding FLAG-IDA (incurring corresponding treatment costs and outcomes).

Table 40 provides the distribution of patients in KTE-X19 arm.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 40: Distribution of patients in KTE-X19 arm (mITT analysis)

Key : mITT, modified intention-to-treat; NA, not applicable Source: ZUMA-3 CSR (56)

Although Ph expression is not expected to affect efficacy of KTE-X19, efficacy data for the Ph- subgroup in ZUMA-3(n=61, 78% of the total cohort of n=78) were implemented to inform KTE-X19 EFS and OS. The same approach was not taken for the Ph+ subgroup as the sample size of the corresponding subgroup from mITT ZUMA-3 Phase 1 and Phase 2 combined (n=17, 22% of the total cohort of n=78) was considered too small to inform KTE-X19 EFS and OS data.

B.3.3.2.2 Inotuzumab ozogamicin

INO-VATE PFS and OS published KM curves were used to reproduce inotuzumab pseudo-IPD (74). INO-VATE enrolled R/R ALL patients irrespective of Ph expression and did not provide disaggregated efficacy results for Ph- and Ph+ subgroups that could be utilized for survival analyses. Therefore, the overall INO-VATE population was used to reflect inotuzumab EFS and OS absolute outcomes in the economic analysis for all three subgroups.

B.3.3.2.3 Blinatumomab

Two sources of clinical data were available for blinatumomab in the economic model; matched IPD from the SCA-3 cohort of the SCHOLAR-3 study and published data from the TOWER (25) study:

• The SCHOLAR-3 methodology has been described in section B.2.9.3 and in the separate SCHOLAR-3 CSR (72). Briefly, the SCA-3 cohort consisted of

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

patients from historical trials who were blinatumomab-naive. The rationale for this analysis was that SCA-3 patients would generally have better outcomes than patients who had previously failed blinatumomab, leading to a bias against the matched ZUMA-3 patients.

• TOWER EFS and OS published KM curves were used to reproduce blinatumomab pseudo-IPD (25). TOWER enrolled R/R ALL patients who were Ph-.

As explained in section B.2.9, the unadjusted ZUMA-3 population in the SCHOLAR-3 analysis is the most generalisable to its intended positioning in UK clinical practice. Therefore, SCA-3 was used to inform EFS and OS absolute outcomes in the economic analysis for the Ph- subgroup. Note that, although the SCHOLAR-3 ITC utilised the Phase 2 mITT dataset for the analysis of KTE-X19 against blinatumomab (see section B.2.9.3), in the economic model, for consistency with other comparators, the KTE-X19 Phase 1 and 2 combined mITT dataset is used for the comparison with blinatumomab, with a sensitivity analysis using the Phase 2 mITT.

B.3.3.2.4 Ponatinib

PACE PFS and OS published KM curves were used to reproduce ponatinib pseudoIPD as described in the MAIC report (71,110). PACE enrolled R/R ALL patients who were Ph+, thus, the overall PACE population was used to inform EFS and OS absolute outcomes in the economic analysis for the Ph+ subgroup.

While in ZUMA-3 RFS was used (see section B.3.3.2.1), in PACE PFS was defined as the interval from the first dose of study treatment until progression or death. Thus, the PACE PFS curve was used as a proxy for EFS. This was deemed to be an appropriate approach by the health economics and clinical experts in the one-to-one interviews, given the publicly available comparative data.

B.3.3.2.5 FLAG-IDA

Comparator data from INO-VATE and TOWER EFS and OS published KM curves were used to reproduce pseudo-IPD (25,74). The pseudo-IPD from the two datasets was pooled to inform FLAG-IDA EFS and OS in the economic analysis for the overall population, the Ph- subgroup, and the Ph+ subgroup.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

In the INO-VATE and TOWER trials, the chemotherapy regimen given to patients in the control arms was based on investigator’s choice. Out of the 162 patients allocated to the INO-VATE control arm, 102 (63%) were treated with FLAG, 38 (23%) were treated with MXN/Ara-C, and 22 (14%) were treated with HIDAC. Out of the 109 patients treated in the TOWER (25) control arm, 49 (45%) were treated with FLAG with or without idarubicin, 19 (17%) were treated with HIDAC, 22 (20%) were treated with high-dose methotrexate-based regimens, and 19 (17%) were treated with clofarabine-based regimens. Consequently, the source used to inform FLAGIDA in the economic model reflected a blend of different salvage chemotherapy regimens. However, there are no clear superior salvage chemotherapy regimens used in the treatment of R/R B-cell ALL in adults. The preference for one course of therapy over others depends on several factors such as safety profile, local clinical practice, and physician preference. For the economic analysis, it was thus assumed that the salvage chemotherapies (investigated in INO-VATE and TOWER and used to reflect FLAG-IDA EFS and OS) were no different in terms of treatment effect.

The combined INO-VATE and TOWER population was thus used to inform the FLAG-IDA EFS and OS absolute outcome in the economic analysis for the overall population, the Ph- subgroup, and the Ph+ subgroup.

B.3.3.3 Survival inputs and assumptions

As the follow-up periods of the studies used to inform intervention and comparators’ EFS/PFS and OS curves (ZUMA-3, INO-VATE, TOWER, and PACE) were shorter than the model time horizon, extrapolation of EFS/PFS and OS observed data was required. A range of models were fitted to the KTE-X19 and comparator arm EFS/PFS and OS, in line with NICE DSU TSD 14 and 21 (114,115). Since the proportional hazard assumption was violated in the comparison of KTE-X19 versus relevant comparators (71), independent models were adopted.

EFS was primarily defined as time from randomisation until the date of relapse after achieving CR, CRi or CRh within a set number of weeks post treatment initiation. Patients that progressed in this period were assumed to have an EFS duration of 1 day, resulting in a significant number of patients being assigned an event at day 1. To avoid convergence issues due to the shape of the EFS KM curve, survival

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

models were only fit amongst responders. The estimated survival curves were later weighted in the economic model to account for patients that did not achieve a response following treatment initiation at cycle 1. This does not apply for ponatinib, since PFS curves were used.

The modelled parametric curves included standard parametric models such as exponential, Weibull, log-logistic, lognormal, Gompertz, and generalised gamma functions and flexible models, such as spline models. The spline models were fitted based on the algorithm by Royston and Parmar (2002) (116). A series of one-, two-, and three-knot restricted cubic spline models using hazard, odds and normal scales were explored. The location of the knots was chosen based on the quantiles of log uncensored survival times, as per the default settings in the flexsurv package in R. Although flexible models have good statistical fit to observed KM data with complex hazard functions, such models often fail to accurately reflect the clinical mechanisms underlying the observed hazard functions and predict unrealistic long-term data. Therefore, in addition to the above models, mixture cure models (MCMs) were fitted using the flexsurvcure package in R where the cure fraction is calculated using a logarithmic model. This type of model represents the population as a combination of two subpopulations: one reflects non-cured patients, who have a risk of experiencing an event defined by a standard parametric function, whilst the other reflects cured patients, who have a risk of experiencing an event as per the general population.

In line with NICE DSU TSD 14 (114) and TSD 21 (115), parametric survival models, splines models and MCMs were considered for all treatments arms and were compared and assessed using the below goodness-of-fit criteria:

• Akaike information criterion (AIC) and Bayesian information criterion (BIC), where smaller AIC/BIC values indicate a better statistical fit;

• A visual inspection of the fitted curves, where the fitted models were overlaid on the KM curves assessing how closely the model data matches reported trial survival data;

• Whether the predicted cure fractions for the comparators were in line with the proportion of patients reported to have survived following receipt of an allo-SCT

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

• Clinical plausibility of long-term extrapolations beyond the trial period based on clinical experts’ opinion and relevant published external data where possible

As stated in NICE DSU TSD 21 (115), MCMs are used in cases where there is supporting evidence that a proportion of patients treated with the intervention will be effectively ‘cured’, and as such, are subject to background mortality. However, NICE DSU TSD 21 also affirms that performing MCM on small datasets raises issues around the practicability and plausibility of being able to estimate cure fraction. A sufficient number of patients at risk are thus needed in the tail of the distribution. The results using MCMs showed a wide variation in cure rates in this analysis (see Appendix N), therefore the data were considered too immature to provide robust results using MCMs and these will only be referred to briefly henceforth.

As discussed in detail in B.3.2.2, patients alive from the 3-year time-point are assumed to be cured and incur an increased risk of death (excess mortality) compared to the general population. An SMR of 1.09 was applied to the background mortality (section B.3.3.3). The SMR of 1.09 has been sourced from a study in DLBCL, Maurer et al., 2014 (98).

Although the comparators are not standalone curative treatments, given their intended use as bridging therapies to allo-SCT, a well-established curative treatment in ALL (96), the use of a cure assumption (parametric model followed by adjusted general population mortality from 3 years) was also deemed appropriate for the comparators. To accommodate this assumption, the economic model enables the combination of a parametric model (standard or spline) for a period of time defined by the user followed by general population mortality with an excess risk of death applied. This approach enabled the inclusion of a clinically validated cure assumption after fitted parametric models that did not require the estimation of a proportion of cured patients from immature data.

Therefore, in the base case, the hybrid approach was adopted for all the endpoints and comparators from 3 years onwards. The survival of patients considered to be effectively ‘cured’ was reflected through age- and gender- matched background mortality calculated using the England life tables (2018-2020) (112). In the base Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

case, a mortality adjustment was applied using a SMR of 1.09 based on the literature (Maurer et al., 2014 (98)) and in line with previous NICE TA in DLBCL (97) (see section B.3.2.2). An alternative assumption regarding the SMR (SMR of 2.5) from the NICE STA for inotuzumab in R/R ALL (81) was explored in a scenario analysis.

Age- and gender-specific general population mortality, modelled through England life tables (2018-2020) (112), was also used in the model to ensure that the estimated and extrapolated risk of death (OS) of the modelled cohort at any timepoint was not inferior to the risk of death of the matched general population. In addition, it was ensured that the modelled event-free patients would not overestimate the modelled alive patients by directly capping the EFS curves with the OS curve.

B.3.3.3.1 Summary of curve selection

Table 41 provides a summary of the survival functions adopted in the base case up to the timepoint of cure (3 years in the base case). The details of the survival analyses performed for each comparator/endpoint are provided in the following sections.

Table 41: Summary of curve selection – base case*

Key : EFS, event-free survival; OS, overall survival; PFS, progression -free survival * Note : in the base case, all the treatment arms’ EFS/PFS and OS follow a hybrid approach (cure assumption from 3 years onwards)

B.3.3.3.2 Overall R/R B-cell ALL population

For the overall population, unadjusted KTE-X19 KM curves were used to inform the

comparison versus inotuzumab and FLAG-IDA in the cost-effectiveness model. The

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

INO-VATE intervention arm informed inotuzumab EFS and OS, while pooled INOVATE and TOWER comparator arms informed FLAG-IDA EFS and OS.

The base case approach combined a parametric model with adjusted general population mortality from 3 years onwards. Note that in the NICE TA for inotuzumab in R/R ALL (TA541) cure was conditional on having received SCT and MRDnegativity and cure fractions were not reported. Therefore a 3-year cure applied to all patients receiving inotuzumab in this model can be considered optimistic.

1 KTE-X19, EFS

Figure 67 through Figure 69 in Appendix N present the various models fitted to the EFS patient level data for KTE-X19 (naïve). The AIC/BIC statistics are presented in Table 138 and in Table 139, Appendix N.

Based on the smaller AIC/BIC and the visual fit to the KM data, the lognormal standard parametric model followed by adjusted general population mortality from 3 years was adopted in the base case.

Figure 34: Modelled KTE-X19 EFS, cure assumption at 3 years – lognormal*

==> picture [467 x 234] intentionally omitted <==

Key : EFS: event-free survival; KM, Kaplan Meier

*Note that the modelled KTE-X19 EFS curve include those patients who did not receive CAR T-cell infusion and are assumed to receive one of the comparators

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

2 KTE-X19, OS

Figure 70 through Figure 72 in Appendix N present the various models fitted to the OS patient level data for KTE-X19 (naïve). The AIC/BIC statistics are presented in Table 140 and Table 141, Appendix N.

Based on the smaller AIC/BIC and the visual fit to the KM data, the lognormal standard parametric model followed by adjusted general population mortality from 3 years was adopted in the base case.

Figure 35: Modelled KTE-X19 OS, cure assumption at 3 years – lognormal*

Key : KM, Kaplan Meier; OS, overall survival

*Note that the modelled KTE-X19 OS curve include those patients who did not receive CAR T-cell infusion and are assumed to receive one of the comparators.

3 Inotuzumab, EFS

Figure 73 through Figure 75 in Appendix N present the various models fitted to the

EFS patient level data from the INO-VATE intervention arm. The AIC/BIC statistics are presented in Table 142 and in Table 143, Appendix N.

Based on the smaller AIC/BIC and the visual fit to the KM data, the 1-knot hazard spline model followed by adjusted general population mortality from 3 years was adopted in the base case.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 36: Modelled inotuzumab EFS, cure assumption at 3 years – 1-knot hazard spline

==> picture [444 x 222] intentionally omitted <==

----- Start of picture text -----
100%
90%
80% Inotuzumab modelled EFS
70%
60% Inotuzumab KM EFS
50%
40%
30%
20%
10%
0%
0 2 4 6 8 10
Time (years)
Survival %
----- End of picture text -----

Key : EFS, event-free survival; KM, Kaplan Meier

4 Inotuzumab, OS

Figure 76 through Figure 78 in Appendix N present the various models fitted to the OS patient level data from the INO-VATE intervention arm. The AIC/BIC statistics are presented in Table 144 and Table 145, Appendix N.

Based on the smaller AIC/BIC and the visual fit to the KM data, the 2-knot normal spline model followed by adjusted general population mortality from 3 years was adopted in the base case.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 37: Modelled inotuzumab OS, cure assumption at 3 years – 2-knot normal spline

==> picture [452 x 226] intentionally omitted <==

----- Start of picture text -----
100%
90%
80% Inotuzumab modelled OS
70%
60% Inotuzumab KM OS
50%
40%
30%
20%
10%
0%
0 2 4 6 8 10
Time (years)
Survival %
----- End of picture text -----

Key : KM, Kaplan Meier; OS, overall survival

5 FLAG-IDA, EFS

Figure 79 through Figure 81 in Appendix N present the various models fitted to the EFS patient level data for INO-VATE and TOWER pooled control arms. The AIC/BIC statistics are presented in Table 146 and Table 147, Appendix N.

Based on the smaller AIC/BIC and the visual fit to the KM data, the generalised gamma standard parametric model followed by adjusted general population mortality from 3 years was adopted in the base case.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 38: Modelled FLAG-IDA EFS, cure assumption at 3 years – generalised gamma

==> picture [452 x 226] intentionally omitted <==

----- Start of picture text -----
100%
90%
FLAG-IDA modelled EFS
80%
70%
60% FLAG-IDA KM EFS
50%
40%
30%
20%
10%
0%
0 2 4 6 8 10
Time (years)
Survival %
----- End of picture text -----

==> picture [175 x 8] intentionally omitted <==

----- Start of picture text -----
Key : EFS, event-free survival; KM, Kaplan Meier
----- End of picture text -----

6 FLAG-IDA, OS

Figure 82 through Figure 84 in Appendix N present the various models fitted to the OS patient level data for INO-VATE and TOWER pooled control arms. The AIC/BIC statistics are presented in Table 148 and Table 149, Appendix N.

Based on the smaller AIC/BIC and the visual fit to the KM data, the generalised gamma standard parametric model followed by adjusted general population mortality from 3 years was adopted in the base case.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 39: Modelled FLAG-IDA OS, cure assumption at 3 years – generalised gamma

==> picture [452 x 226] intentionally omitted <==

----- Start of picture text -----
100%
90%
FLAG-IDA modelled OS
80%
70%
60% FLAG-IDA KM OS
50%
40%
30%
20%
10%
0%
0 2 4 6 8 10
Time (years)
Survival %
----- End of picture text -----

Key : KM, Kaplan Meier; OS, overall survival

B.3.3.3.3 Ph- R/R B-cell ALL population

For the Ph- population, unadjusted KTE-X19 KM curves from the ZUMA-3 Phsubgroup (mITT Phase 1 and Phase 2 combined dataset) were used to inform the comparison versus inotuzumab, FLAG-IDA, and blinatumomab in the costeffectiveness model. Since inotuzumab and FLAG-IDA outcomes were not differentiated by Ph expression, the same data and curve selection as the overall population was used (see section B.3.3.3.2). The SCHOLAR-3 SCA-3 dataset, with the exclusion of patients receiving salvage chemotherapy, was used to inform the blinatumomab EFS and OS in the economic model.

Consistent with the overall population, the base case approach combined a parametric model with adjusted general population mortality from 3 years onwards for all comparators in this subgroup.

7 KTE-X19, EFS (Ph-)

Figure 85 through Figure 87 in Appendix N present the various models fitted to the EFS patient level data for KTE-X19 (naïve). The AIC/BIC statistics are presented in Table 150 and in Table 151, Appendix N.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Based on the smaller AIC/BIC and the visual fit to the KM data, the lognormal standard parametric model followed by adjusted general population mortality from 3 years was adopted in the base case.

Figure 40: Modelled KTE-X19 EFS (Ph-), cure assumption at 3 years – lognormal*

==> picture [475 x 243] intentionally omitted <==

----- Start of picture text -----
Key : EFS: event-free survival; KM, Kaplan Meier
----- End of picture text -----

*Note that the modelled KTE-X19 EFS curve include those patients who did not receive CAR T-cell infusion and are assumed to receive one of the comparators

8 KTE-X19, OS (Ph-)

Figure 88 through Figure 90 in Appendix N present the various models fitted to the OS patient level data for KTE-X19 (naïve). The AIC/BIC statistics are presented in Table 152 and Table 153 , Appendix N.

Based on the smaller AIC/BIC and the visual fit to the KM data, the lognormal standard parametric model followed by adjusted general population mortality from 3 years was adopted in the base case.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 41: Modelled KTE-X19 OS (Ph-), cure assumption at 3 years – lognormal*

==> picture [467 x 232] intentionally omitted <==

Key : KM, Kaplan Meier; OS, overall survival *Note that the modelled KTE-X19 OS curve include those patients who did not receive CAR T-cell infusion and are assumed to receive one of the comparators

9 Blinatumomab, EFS

Figure 91 through Figure 93 in Appendix N present the various models fitted to the EFS patient level data from SCHOLAR-3 SCA-3. The AIC/BIC statistics are presented in Table 154 and Table 155, Appendix N.

Based on the smaller AIC/BIC and the visual fit to the KM data, the 1-knot hazard spline model followed by adjusted general population mortality from 3 years was adopted in the base case.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 42: Modelled blinatumomab EFS, cure assumption at 3 years – 1-knot hazard spline

==> picture [460 x 228] intentionally omitted <==

Key : EFS, event-free survival; KM, Kaplan Meier

10 Blinatumomab, OS

Figure 94 through Figure 96 in Appendix N present the various models fitted to the OS patient level data for SCHOLAR-3 SCA-3. The AIC/BIC statistics are presented in Table 156 and Table 157, Appendix N.

Based on the smaller AIC/BIC and the visual fit to the KM data, the lognormal model followed by adjusted general population mortality from 3 years was adopted in the base case.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 43: Modelled blinatumomab OS, cure assumption at 3 years – lognormal

==> picture [458 x 228] intentionally omitted <==

Key : KM, Kaplan Meier; OS, overall survival

B.3.3.3.4 Ph+ R/R B-cell ALL population

For the Ph+ population, unadjusted KTE-X19 KM curves were used to inform the comparison versus inotuzumab, FLAG-IDA, and ponatinib in the cost-effectiveness model. Since the ZUMA-3 Ph+ subgroup was deemed too small to be used for survival analyses, the same data (mITT ZUMA-3 Phase 1 and Phase 2 combined, n=78) and curve selection as the overall population was used (see section B.3.3.3.2). Since inotuzumab and FLAG-IDA outcomes were not differentiated by Ph expression, the same data and curve selection as the overall population was used (see section B.3.3.3.2). PACE informed ponatinib PFS and OS absolute outcomes.

Although ponatinib is not a curative treatment, given its intended use as a bridging therapy to allo-SCT, a well-established curative treatment in ALL, the use of a cure assumption (parametric model followed by adjusted general population mortality from 3 years) was deemed appropriate.

11 Ponatinib, PFS

Figure 97 through Figure 99 in Appendix N present the various models fitted to the PFS patient level data for PACE arm. The AIC/BIC statistics are presented in Table 158 and Table 159, Appendix N.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Based on the smaller AIC/BIC and the visual fit to the KM data, the lognormal standard parametric model followed by adjusted general population mortality from 3 years was adopted in the base case.

Figure 44: Modelled ponatinib PFS, cure assumption at 3 years – lognormal

==> picture [452 x 231] intentionally omitted <==

----- Start of picture text -----
100%
90%
Ponatinib modelled PFS
80%
70%
Ponatinib KM PFS
60%
50%
40%
30%
20%
10%
0%
0 2 4 6 8 10
Time (years)
Survival %
----- End of picture text -----

Key : KM, Kaplan Meier; PFS, progression-free survival

12 Ponatinib, OS

Figure 100 through Figure 102 in Appendix N present the various models fitted to the OS patient level data for PACE arm. The AIC/BIC statistics are presented in Table 160 and Table 161, Appendix N.

Based on the smaller AIC/BIC and the visual fit to the KM data, the lognormal standard parametric model followed by adjusted general population mortality from 3 years was adopted in the base case.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Figure 45: Modelled ponatinib OS, cure assumption at 3 years – lognormal

==> picture [452 x 231] intentionally omitted <==

----- Start of picture text -----
100%
90%
Ponatinib modelled OS
80%
70%
Ponatinib KM OS
60%
50%
40%
30%
20%
10%
0%
0 2 4 6 8 10
Time (years)
Survival %
----- End of picture text -----

Key : KM, Kaplan Meier; OS, overall survival

B.3.3.4 Adverse events

The incidence of adverse events (AEs) for individual treatments was informed by individual clinical trials. For KTE-X19, these included grade 3 or 4 AEs occurring post-infusion in ≥5% of the population (ZUMA-3 mITT Phase 1 and Phase 2 combined, with Phase 1 target dose obtained from the publication) (42). For blinatumomab, grade ≥3 AEs observed during TOWER that occurred in ≥5% of adults in the first cycle of therapy were used in the model (25). Since the INO-VATE study reported serious AEs that occurred in ≥2% of the safety population (26), these figures were used in the model to reflect inotuzumab safety profile. Treatmentemergent AEs of any grade occurring in ≥20% of the total population were taken from the phase 2 PACE trial for the ponatinib arm (110). In order to be consistent with the efficacy data, the adverse event rates for the FLAG-IDA arm was pooled from the control arms of the INO-VATE (inotuzumab) and TOWER (blinatumomab) trials. AEs rates for each treatment arm are reported in Table 42.

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia

© Kite, a Gilead Company (2021) All rights reserved

Table 42: Adverse events rates included in the model

Company evidence submission template for KTE-X19 for previously treated B-precursor adult acute lymphoblastic leukaemia © Kite, a Gilead Company (2021) All rights reserved Page 168 of 270