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

Single Technology Appraisal

Axicabtagene ciloleucel for treating relapsed or refractory low-grade nonHodgkin lymphoma [ID1685]

Committee Papers

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

NATIONAL INSTITUTE FOR HEALTH AND CARE EXCELLENCE

SINGLE TECHNOLOGY APPRAISAL

Axicabtagene ciloleucel for treating relapsed or refractory low-grade nonHodgkin lymphoma [ID1685]

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

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

• a. Lymphoma Action

4. Evidence Review Group report prepared by Aberdeen HTA Group

5. Evidence Review Group report – factual accuracy check

Post-technical engagement documents

6. Technical engagement response from company

7. Technical engagement responses and statements from experts:

• a. Dr Graham Collins, Consultant Haematologist and lymphoma MDT lead – clinical expert, nominated by Gilead (company)

• b. Dr Tobias Menne, Consultant Haematologist – clinical expert, nominated by NCRI-ACP-RCP-RCR

8. Evidence Review Group critique of company response to technical engagement prepared by Aberdeen HTA Group

• 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 Axicabtagene ciloleucel for treating relapsed or refractory low-grade non-Hodgkin lymphoma [ID1685]

Document B

Company evidence submission

January 2022

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 1 of 155

Instructions for companies

This is the template for submission of evidence to the National Institute for Health and Care Excellence (NICE) as part of the single technology appraisal (STA) process. Please note that the information requirements for submissions are summarised in this template; full details of the requirements for pharmaceuticals and devices are in the user guide.

This submission must not be longer than 150 pages, excluding appendices and the pages covered by this template. If it is too long it will not be accepted.

Companies making evidence submissions to NICE should also refer to the NICE guide to the methods of technology appraisal and the NICE guide to the processes of technology appraisal.

In this template any information that should be provided in an appendix is listed in a box.

Highlighting in the template (excluding the contents list)

Square brackets and grey highlighting are used in this template to indicate text that should be replaced with your own text or deleted. These are set up as form fields, so to replace the prompt text in [grey highlighting] with your own text, click anywhere within the highlighted text and type. Your text will overwrite the highlighted section.

To delete grey highlighted text, click anywhere within the text and press DELETE.

Grey highlighted text in the footer does not work as an automatic form field, but serves the same purpose – as prompt text to show where you need to fill in relevant details. Replace the text highlighted in [grey] in the header and footer with appropriate text. (To change the header and footer, double click over the header or footer text. Double click back in the main body text when you have finished.)

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 2 of 155

Contents

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 3 of 155

Tables and figures

Table 1: The decision problem ................................................................................... 8 Table 2: Technology being appraised ...................................................................... 11 Table 3: Classification systems for follicular lymphoma ........................................... 14 Table 4: NICE recommendations for the treatment of symptomatic advanced-stage follicular lymphoma................................................................................................... 18 Table 5: Clinical effectiveness evidence ................................................................... 23 Table 6: Summary of trial methodology for ZUMA-5 ................................................ 25 Table 7: Baseline characteristics of FL patients in ZUMA-5 ..................................... 29 Table 8: Summary of statistical analyses for ZUMA-5 .............................................. 31 Table 9: Summary of ZUMA-5 cohorts and analysis sets presented ........................ 35 Table 10: Summary of response using central assessment per Lugano classification; FL patients with two or more lines of prior therapy, IAS ........................................... 37 Table 11: Summary of response using central assessment per Lugano classification; FL patients with three or more lines of prior therapy, IAS ......................................... 40 Table 12: Summary of response using central assessment per Lugano classification; FL patients with three or more lines of prior therapy, mITT ...................................... 42 Table 13: Natural history represented by response by line of therapy ...................... 49 Table 14: Baseline characteristics of patients pre- and post-weighting; FL patients with three or more lines of prior therapy, SCHOLAR-5 EES, ZUMA-5 mITT ............ 52 Table 15: Comparative analysis summary; FL patients with three or more lines of prior therapy, SCHOLAR-5, EES, ZUMA-5 mITT ..................................................... 55 Table 16: Safety summary for FL patients in ZUMA-5 .............................................. 59 Table 17: Common adverse events occurring in ≥ 30% of FL patients in ZUMA-5 .. 60 Table 18: Common treatment-related adverse events occurring in ≥ 20% of FL patients in ZUMA-5 ................................................................................................... 61 Table 19: Summary of CRS symptoms and events occurring in ≥ 4% of FL patients in ZUMA-5 ................................................................................................................ 63 Table 20: Summary of neurological events occurring in FL patients in ZUMA-5 ...... 64 Table 21: Incidence of cytopenia in FL patients in ZUMA-5 ..................................... 66 Table 22: End-of-life criteria ..................................................................................... 74 Table 23: Summary list of published NICE appraisals in r/r FL ................................ 76 Table 24: Data sources of clinical parameters used in the model ............................ 84 Table 25: Summary of base case approach used to model PFS, by treatment arm 90 Table 26: Axi-cel, PFS: standard parametric curve AIC and BIC statistics and landmark survival estimates ..................................................................................... 91 Table 27: Current 4L+ care: PFS: standard parametric curve AIC and BIC statistics and landmark survival estimates .............................................................................. 94 Table 28: Summary of base case approach used to model OS, by treatment arm .. 96 Table 29: Axi-cel, OS: standard parametric curve AIC and BIC statistics and landmark survival estimates ..................................................................................... 97 Table 30: Current 4L+ care: OS: standard parametric curve AIC and BIC statistics and landmark survival estimates ............................................................................ 100 Table 31: Summary of utility values identified in the literature ................................ 105 Table 32: Grade ≥ 3 adverse event data, axi-cel (ZUMA-5) ................................... 107 Table 33: Adverse event data, current 4L+ care (reported in TA627) .................... 107 Table 34: Adverse event durations ......................................................................... 109 Table 35: Summary of utility values for cost-effectiveness analysis ....................... 111 Table 36: Leukapheresis unit costs (NHS reference costs 2019/20)[65] ................... 114

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 4 of 155

Table 37: Conditioning chemotherapy unit costs (eMIT)[64] ..................................... 115 Table 38: Hospitalisation unit costs (NHS reference costs 2019/20)[65] ................... 117 Table 39: Axi-cel treatment costs per patient ......................................................... 118 Table 40: Distribution of current 4L+ care therapies ............................................... 119 Table 41: Current 4L+ care unit costs (list price) .................................................... 120 Table 42: Current 4L+ care – dosing schedules ..................................................... 120 Table 43: Administration costs (NHS reference costs 2019/20)[65] .......................... 122 Table 44: Drug and administration costs per 28-day model cycle .......................... 122 Table 45: Treatment durations (summary of product characteristics) .................... 123 Table 46: Subsequent treatment costs ................................................................... 124 Table 47: Resource use frequency ........................................................................ 125 Table 48: Resource use unit costs (NHS reference costs 2019/20) ....................... 125 Table 49: Diagnostic test costs .............................................................................. 125 Table 50: IVIG unit costs (MIMS)[63] ......................................................................... 127 Table 51: Summary of IVIG costs applied in the model .......................................... 127 Table 52: Adverse event management unit costs .................................................. 129 Table 53: End-of-life care costs (Round et al. [2015])[70] ......................................... 129 Table 54: Key model assumptions ......................................................................... 131 Table 55: Base case results (with PAS) ................................................................. 135 Table 56: Mean results of PSA (2,000 runs) and comparison with deterministic results (with PAS) ................................................................................................... 138 Table 57: OWSA results for the most influential model parameters on the ICER (with PAS) 142 Table 58: Scenario analysis results (with PAS) ...................................................... 144

Figure 1: Axi-cel anti-CD19 CAR construct and mode of action ............................... 13 Figure 2: Process of manufacturing and administering axi-cel ................................. 13 Figure 3: Clinical care pathway for patients with follicular lymphoma and proposed axi-cel positioning ..................................................................................................... 21 Figure 4: Study scheme for ZUMA-5 ........................................................................ 24 Figure 5: Patient disposition data for FL patients enrolled in the ZUMA-5 trial ......... 34 Figure 6: Kaplan–Meier plot for duration of response; FL patients with three or more lines of prior therapy, mITT ...................................................................................... 43 Figure 7: Kaplan–Meier plot for duration of response by best response; FL patients with three or more lines of prior therapy, mITT ......................................................... 43 Figure 8: Kaplan–Meier plot for progression-free survival; FL patients with three or more lines of prior therapy, mITT ............................................................................. 44 Figure 9: Kaplan–Meier plot for progression-free survival by best response; FL patients with three or more lines of prior therapy, mITT ........................................... 45 Figure 10: Kaplan–Meier plot for overall survival; FL patients with three or more lines of prior therapy, mITT ............................................................................................... 46 Figure 11: Kaplan–Meier plot for overall survival by best response; FL patients with three or more lines of prior therapy, mITT ................................................................ 46 Figure 12: (A) Real-world clinical site treatment patterns by line of therapy[a,b ] and (B) fourth line treatment patterns by region (US and Europe) ........................................ 50 Figure 13: Natural history represented by (A) progression-free survival (B) overall survival by line of therapy ......................................................................................... 51 Figure 14: Progression-free survival for axi-cel (ZUMA-5) and current 4L+ care (propensity score weighted SCHOLAR-5) ................................................................ 56

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 5 of 155

Figure 15: Overall survival for axi-cel (ZUMA-5) and current 4L+ care (propensity score weighted SCHOLAR-5) .................................................................................. 57 Figure 16: Model structure diagram ......................................................................... 80 Figure 17: Partitioned survival analysis; health state occupancy example ............... 81 Figure 18: Axi-cel progression-free survival: standard parametric curves ................ 91 Figure 19: Axi-cel PFS: selected curve (25% long-term survivorship) ...................... 93 Figure 20: Current 4L+ care PFS: standard parametric curves ................................ 94 Figure 21: Axi-cel versus current 4L+ care, PFS: selected ...................................... 95 Figure 22: Axi-cel overall survival: standard parametric curves ............................... 97 Figure 23: Axi-cel overall survival: selected curve (25% long-term survivorship) ..... 99 Figure 24: Current 4L+ care overall survival: standard parametric curves ............. 100 Figure 25: Axi-cel versus current 4L+ care, OS: selected ...................................... 101 Figure 26: Axi-cel versus current 4L+ care, OS and PFS: selected ....................... 102 Figure 27: Lifetime Markov trace for axi-cel ........................................................... 136 Figure 28: Lifetime Markov trace current 4L+ care ................................................. 136 Figure 29: Scatter plot of PSA with 2,000 iterations (with PAS) ............................. 139 Figure 30: Scatter plot of PSA with 2,000 iterations (with PAS; excluding survival analysis parameters) .............................................................................................. 139 Figure 31: Cost-effectiveness acceptability curves for PSA with 2,000 iterations (with PAS) 140 Figure 32: Tornado diagram showing OWSA results on ICER (with PAS) ............. 141

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 6 of 155

Decision problem, description of the technology and clinical care pathway

B.1.1. Decision problem

This submission covers the technology’s full anticipated marketing authorisation for this indication. Further details are provided in the decision problem summary presented in Table 1.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 7 of 155

Table 1: The decision problem

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 8 of 155

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 9 of 155

B.1.2. Description of the technology being appraised

A description of axicabtagene ciloleucel (Yescarta[®;] hereby referred to as axi-cel) is presented in Table 2.

The draft Summary of Product Characteristics (SmPC) for the relapsed or refractory (r/r) follicular lymphoma (FL [r/r FL]) indication is presented in Appendix C. The European Public Assessment Report for this indication can be provided on receipt.

Axi-cel was the first in a breakthrough class of chimeric antigen receptor (CAR) T- cell (CAR T-cell) therapies that are manufactured from patients’ own T-cells and engineered ex vivo to express antigen-specific CAR, enabling them to target and kill antigen-expressing tumour cells on return to the patient. The CAR construct used in axi-cel is a single-chain antibody fragment directed against CD19 linked to CD3ζ and CD28 T-cell activating domains; CD19 is a B-cell-specific cell surface antigen ubiquitously expressed in B-cell malignancies, including FL.[2]

Axi-cel is given as a single infusion treatment. The timescale from collection of the patient’s T-cells by leukapheresis, through transportation to the manufacturing facility, product manufacture, and delivery to the clinical centre in Europe is typically '''''' '''''''''''''.[3]

The axi-cel construct and mode of action is depicted in Figure 1. The manufacturing and administration process for axi-cel is depicted in Figure 2.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 10 of 155

Table 2: Technology being appraised

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 11 of 155

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 12 of 155

Figure 1: Axi-cel anti-CD19 CAR construct and mode of action

==> picture [430 x 314] intentionally omitted <==

Key: CAR, chimeric antigen receptors; scFv, single-chain variable region fragment.

Figure 2: Process of manufacturing and administering axi-cel

==> picture [452 x 114] intentionally omitted <==

Key: CAR, chimeric antigen receptors

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 13 of 155

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

B.1.3.1. Disease overview

Non-Hodgkin lymphoma (NHL) comprises a diverse group of cancers of the lymphatic system.[7] Indolent NHL (iNHL) describes slower growing lymphomas of this group; these lymphomas generally have longer survival times but are less likely to be cured compared with faster growing lymphomas.[7] FL is the most common type of iNHL that arises from B-lymphocytes (making it a B-cell malignancy) and accounts for approximately 19% of all cases of NHL.[8][9] It mainly affects adults over the age of 60, with no clear gender difference.[8, 9] Diagnosis and monitoring of FL includes assessment of several features that can help predict the likely disease course and thus inform treatment decisions.[8] These include grading and staging of disease, and risk categorisation based on demographic and basic disease characteristics; classification systems specific to FL are summarised in Table 3.

Table 3: Classification systems for follicular lymphoma

The FL disease course is one of interspersing remissions and relapses and nearly all patients will ultimately relapse. With each relapse the disease becomes more

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 14 of 155

resistant to treatment, leading to shorter remission periods over time.[11-14] UK clinical experts estimate approximately 10% of patients diagnosed with FL go on to receive ≥4 lines of therapy, with an estimated period of 10 years between the first and fourthline treatment.[1] These estimates are generally aligned to real-world evidence from the prospective observational LymphoCare study in the US, which reported that 9% of patients diagnosed with FL received fourth-line treatment after a median follow-up of 8 years (range: 0.02–10.34).[13]

Despite its commonly indolent nature, FL is highly heterogeneous and there is a subpopulation of approximately 10-20% of patients who experience a more aggressive disease trajectory, characterised by multiple relapses, treatment refractoriness, short remission periods and an overall shortened lifespan.[15, 16] Patients with FL who experience progression of disease within 2 years of receiving front-line chemoimmunotherapy (defined as ‘POD24’) have a particularly poor prognosis, with only a 50% overall survival (OS) estimate at 5 years, compared with 90% OS estimate at 5 years for those without POD24.[15, 17, 18] Other high-risk subpopulations of FL include patients who are chemoimmunotherapy resistant and fail to achieve a response within 6 months of completion of initial chemoimmunotherapy (treatment refractory) or have double-refractory disease (that is, patients with FL who are refractory to the first two lines of therapy, notably including both an alkylating agent and an anti-CD20 monoclonal antibody). Given that all patients with r/r FL who require ≥ 4 lines of active treatment are a heterogenous group for whom there is no established standard of care (see Section B.1.3.4) and who represent a significant unmet need, in this appraisal, all of the 4L+ patients are the population of interest, irrespective of them having any high-risk prognostic factors.[11-14] Furthermore, aligning to the patient population enrolled to the ZUMA-5 trial introduced in Section B.2, FL patients to be considered for treatment with axi-cel are expected to have Grade 1–3A, Stage III–IV disease of any risk category (after three or more lines of systemic therapy [4L+] as per the anticipated marketing authorisation outlined in Table 2).

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 15 of 155

B.1.3.2. Prognosis of patients with r/r FL after three or more lines of systemic therapy

An estimated 2,200 patients are newly diagnosed with FL each year in the UK[19] ; of these, an estimated 220 will receive at least four lines of therapy (approx.10% of all patients diagnosed).[1] Approximately 198 of these 220 patients are assumed to receive treatment in England or Wales (based on 90% of patients in the UK residing in England or Wales).[20]

The prognosis of patients with 4L+ r/r FL is generally poor and there is no established standard of care in this setting (see Section B.1.3.4). The LymphoCare study in the US recently reported median progression-free survival (PFS) of 0.69 years (95% confidence interval [CI]: 0.50, 0.97) in patients with FL receiving fourthline treatment (n = 229).[13] A further retrospective analysis in the US recently reported median PFS of 0.90 years (95% CI: 0.59, 1.10) and median event-free survival of 0.56 years (95% CI: 0.48, 0.84) in FL patients receiving fourth-line treatment (n = 198) (event-free survival defined as time to progression, change of treatment or death).[11] Clinical experts noted that median PFS on current fourth-line plus (4L+) care in the UK is estimated to be notably lower, at 0.31–0.46 years.[1]

Retrospective analysis in the US further reported median OS for FL patients receiving fourth-line treatment at 5.34 years (95% CI: 3.51, not reached).[11] This is higher than clinical experts expect for UK FL patients receiving fourth-line treatment, for whom they estimate a median OS of approximately 3 years.[1] The LymphoCare study did not report OS data by treatment line. The only other study identified that reported such data in the published literature base was a retrospective analysis of patients with r/r FL in Japan. This study reported a median OS of 1.01 years in patients with r/r FL after fourth-line chemotherapy.[12]

These varied data reported in the current published literature reinforce the heterogeneity of the disease course in FL.[11] It should also be acknowledged that the current evidence base is largely retrospective in nature, with small patient numbers and strong variability in the type and number of prior lines per country that had been received; taken together, these differences of the evidence base further reflect the heterogenous nature of r/r FL. In the SCHOLAR-5 study introduced in Section B.2.9,

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 16 of 155

patients receiving fourth-line treatment in real-world practice across Europe and the US had a median OS of '''''''''' months (Table 15).[21]

B.1.3.3. Burden of disease

Alongside reduced life expectancy, FL is associated with several physical and psychological symptoms that affect patients’ health-related quality of life (HRQL). The most common symptom of FL is a painless swelling in the lymph nodes of the neck, armpit or groin, but it can also be associated with ‘B-symptoms’ such as night sweats, erratic fever and weight loss.[8] Patients with FL presenting with multiple sites of lymphadenopathy can endure restricted movement, disfigurement, pain, and bone marrow disease that can result in anaemia, leukopenia and thrombocytopenia.[22, 23]

In addition to physical symptoms FL negatively affects the mental health of patients, with depression and stress commonly reported.[24-26] As a generally accepted chronic, incurable and progressive condition, this can also be emotionally unsettling to FL patients. Indeed, HRQL diminishes with each treatment relapse and UK clinical experts note that patients with r/r FL reaching the 4L+ treatment setting are likely to have a lower quality of life due to the fewer treatment options available.[1] In a UK cross-sectional study using a variety of patient-reported outcome instruments to assess HRQL, patients with relapsed FL were more likely to experience worse HRQL compared with FL patients who were newly diagnosed, in partial or complete remission or disease-free.[22] Patients with relapsed FL had lower mean physical, emotional, functional and social wellbeing scores and reported statistically significantly higher levels of anxiety, depression and activity impairment levels compared with disease-free patients.[22] As such, the burden of illness in patients with three or more lines of systemic therapy is expected to be particularly high (though data is limited, Section B.3.4)

HRQL is further affected by treatment toxicity effects; for example, chemotherapy has specifically been shown to worsen health functioning (p = 0.004), depressive symptoms (p = 0.005) and activity impairment (p = 0.009) compared with FL patients in remission but not on treatment.[22] Patients receiving active chemotherapy for disease progression displayed considerable impairment (daily activity impairment > 50%) including in overall work productivity.[27]

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 17 of 155

Alongside the burden on patients, FL also poses a substantial burden on carers. In a Canadian cross-sectional cohort of patients with iNHL, including FL, the majority of care (74%) was unpaid assistance from a partner or spouse, relative or friend.[27] This group of unpaid caregivers provided a mean of 9.8 days of care in the 30 days prior to data collection, with a mean of 11.3 days of absenteeism.

B.1.3.4. Clinical pathway of care

Table 4 summarises the National Institute for Health and Care Excellence (NICE) recommendations for the treatment of symptomatic advanced-stage FL. However, despite there being national guidance and recommendations in place, there is significant variability in how patients are managed therapeutically in the r/r FL setting, and notably, beyond the 3L setting there is no consensus or standard of care available in the clinical management pathway.[1, 28]

Table 4: NICE recommendations for the treatment of symptomatic advancedstage follicular lymphoma

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 18 of 155

Newly diagnosed FL patients are typically treated with rituximab plus a chemotherapy backbone (R-chemo) followed by rituximab maintenance for up to 2 years, in responding patients. At first relapse, patients who had a good response to initial R-chemo treatment are typically retreated with a different R-chemo regimen

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 19 of 155

(followed by rituximab maintenance in responding patients). Patients who did not have a good response to initial R-chemo treatment are typically treated with lenalidomide plus rituximab (R[2] ) or obinutuzumab with bendamustine. Although rituximab monotherapy is also considered as a treatment option for relapsed FL, it is rarely used outside of the maintenance setting in the UK, with an observed lack of clinical effect as a remission inducing treatment. This is reflected in its recommended use, that is, only when all alternative treatment options have been exhausted (i.e. if there is resistance to or intolerance of chemotherapy).[1, 30]

At second relapse, the same regimens are considered with treatment decisions made on a case-by-case basis. Consolidation with autologous stem cell transplantation (auto-SCT) is recommended at second or subsequent remission (complete or partial) for patients who have not already had a transplant and who are fit enough for transplantation.[29] Allogeneic stem cell transplant can also be considered at second or subsequent remission (complete or partial) for patients who are fit enough for transplantation and for whom a suitable donor can be found when auto-SCT has not resulted in remission or is inappropriate.[29] However, it should be noted that allogeneic stem cell transplant needs to be carefully considered due to the notable morbidity and mortality in the r/r setting.[1]

By the time patients reach third relapse, that is they have received three prior systemic therapies, they have typically exhausted recommended treatment options. It must also be noted that by the 4L+ setting FL patients will very likely have received multiple rituximab-based regimens throughout their treatment course and, therefore, it is expected (and as validated by UK clinical experts) that their response to further rituximab-based treatment will be suboptimal at best. In the absence of an established standard of care, current 4L+ therapy consists of recycling earlier-line treatment options for FL (Table 4) or resorting to generic haemato-oncology or experimental/compassionate use treatments. Treatment decisions are made on a case-by-case basis, considering factors such as patient fitness, treatment goals, response and durability of response to prior therapy.

Axi-cel would offer a new treatment option in this 4L+ setting, as depicted in Figure 3.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 20 of 155

Figure 3: Clinical care pathway for patients with follicular lymphoma and proposed axi-cel positioning

==> picture [446 x 209] intentionally omitted <==

Key: 1L, first-line; 2L, second-line; 4L, fourth-line; alloSCT, allogeneic stem cell transplantation; ASCT, autologous stem cell transplant; Benda, bendamustine; CHOP, cyclophosphamide, doxorubicin, vincristine and prednisolone; CHVPi, cyclophosphamide, doxorubicin, etoposide, prednisolone and interferon-α; CVP, cyclophosphamide, vincristine and prednisolone; FLIPI, Follicular Lymphoma International Prognostic Index; MCP, mitoxantrone, chlorambucil and prednisolone; NICE, National Institute for Health and Care Excellence; O, obinutuzumab; R, rituximab; R-B, rituximab with bendamustine; R[2] , lenalidomide with rituximab. Source: NICE Pathways – Treating follicular lymphoma.[29]

B.1.3.5. Unmet need

Patients with r/r FL who have received three or more lines of systemic therapy represent a difficult-to-treat patient group who often follow an aggressive, chemotherapy-resistant disease course, with poor prognosis and no established standard of care.

Current 4L+ treatment options in England are limited to recycling earlier-line treatments to which patients may have reduced tolerance, as well as reduced effectiveness; or resorting to generic haemato-oncology treatments with little expectation of effect or experimental treatments with no proven benefit. Patients in England actually have fewer treatment options at later lines than those in Wales and Scotland, where idelalisib (a licensed Pi3Kδ inhibitor) is an additional treatment available through routine baseline commissioning to patients with FL that is refractory to two prior lines of treatment.[31, 32] This is in contrast to England, where

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 21 of 155

NICE, in its final appraisal document (FAD), did not recommend idelalisib for treating FL that has not responded to two prior lines of treatment in adults.[33]

Although survival expectations depend on several factors, including previous treatment regimens received and response to previous treatment, patients are generally not expected to survive beyond approximately 3 years with current 4L+ care options that are currently available in England.[1]

B.1.4. Equality considerations

No equality issues for axi-cel are foreseen, but as noted above there are existing inequalities in current non-immunochemotherapy treatment options available in England compared with Wales and Scotland. Through provision of a nonimmunochemotherapy treatment option to patients in England (as well as Wales and Scotland), recommendation of axi-cel could help reduce this current inequality across the devolved nations of the UK.[31, 32]

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 22 of 155

==> picture [29 x 13] intentionally omitted <==

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 axi-cel.

B.2.2. List of relevant clinical effectiveness evidence

Table 5 summarises the clinical effectiveness evidence supporting axi-cel for the treatment of adult patients with r/r FL on the 4L+ treatment pathway.

Table 5: Clinical effectiveness evidence

As ZUMA-5 is a single-arm study, comparator data are provided by an international,

multicentre, external control cohort study, SCHOLAR-5, designed to provide

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 23 of 155

comparative evidence in patients with r/r FL meeting the ZUMA-5 eligibility criteria. SCHOLAR-5 is described in more detail in Section B.2.9.

B.2.3. Summary of methodology of the relevant clinical effectiveness evidence

Table 6 provides a summary of the trial methodology for ZUMA-5.

ZUMA-5 is a Phase II, multicentre, open-label study evaluating the efficacy of axi-cel in adults with either r/r B-cell iNHL of the histological subtypes FL or marginal zone lymphoma (MZL).[34] This submission focuses solely on patients with r/r FL who have received three prior therapies (aligning with the anticipated market authorisation).

Up to approximately 125 patients with FL were to be enrolled and treated with axi-cel at a target dose of 2 x 10[6] anti-CD19 CAR T-cells/kg body weight.[34] Each patient was to proceed through the study periods depicted in Figure 4.

Figure 4: Study scheme for ZUMA-5

==> picture [402 x 180] intentionally omitted <==

Key: CAR, chimeric antigen receptor. Source: ZUMA-5 Clinical Study Protocol.[34]

The primary endpoint of the ZUMA-5 trial was the overall response rate, defined as the incidence of patients achieving complete response (CR) or partial response (PR) as determined by independent central review per Lugano classification (hereafter referred to as central assessment).[34] Secondary endpoints included CR rate, ORR and CR rate in patients who received three or more lines of prior therapy, ORR by

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 24 of 155

investigator assessment, duration of response (DOR), PFS, OS, and safety (Table 6).

Table 6: Summary of trial methodology for ZUMA-5

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 25 of 155

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 26 of 155

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 27 of 155

B.2.3.1. Baseline characteristics

Table 7 provides a summary of baseline characteristics, including demographic and clinical characteristics, for FL patients.

The FL populations presented here are:

• The full analysis set: all patients who were enrolled in the trial

• The safety analysis set (SAS): all patients treated with any dose of axi-cel

• The inferential analysis set (IAS): the first ≥80 patients enrolled into the study who met the eligibility criteria for the pivotal cohort were treated with any dose of axi-cel, and had the opportunity to be followed for 18 months post axi-cel infusion

Baseline characteristics are presented for patients with two or more lines of prior therapy (as per ZUMA-5 eligibility criteria) and patients with three or more lines of prior therapy (as per anticipated marketing authorisation).

The demographics and clinical characteristics of patients were largely consistent across all analysis sets. A high proportion of patients had high-risk Follicular Lymphoma International Prognostic Index (FLIPI) scores (that is FLIPI Scores ≥ 3), and the majority of patients were refractory to prior therapy and had received at least three prior lines.[35] The majority of patients had progression of disease within 24 months of completion of first anti-CD20-based chemoimmunotherapy (i.e. POD24) and approximately a third had double-refractory disease. Clinical experts in the UK confirmed the baseline characteristics of patients enrolled to ZUMA-5 were generally

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 28 of 155

representative of patients who would typically be considered for axi-cel within its anticipated marketing authorisation in clinical practice.[1]

Table 7: Baseline characteristics of FL patients in ZUMA-5

==> picture [452 x 600] intentionally omitted <==

----- Start of picture text -----
FL patients with two or FL patients with three or more
more lines of prior therapy lines of prior therapy
Characteristics
FAS (n = SAS (n IAS (n = FAS (n = SAS (n = IAS (n =
127) = 124) 86) 80) 78) 60)
Median age, years '''''''''''' ''''''''' ''''''''''''''' ''''' '''''''''' '''''''''' '''''''''' '''''''''' ''''''''' '''''
''''''''''
(range) ''''''''' '''''''' ''''''' ''''''' '''''''''
Aged ≥65 years, n (%) ''''''' '''''''''' ' ''''''''' ''''''''' ''''''' ''''''''' '''''' ''''''''' ' ''''''''''
Male, n (%) '''''' ''''''''' ''''' '''''''''' ' '''''''''' '''''' '''''''''' '''''' '''''''''' '''''''''''
ECOG performance
status, n (%)
0 ''''' ''''''''''' ''''' '''''''''' ' ''''''''' ''''''' ''''''''' ''''''' ''''''''' '''''''''''
1 '''''' ''''''''' ''''''''''' '''''' ''''''''' '''''' ''''''''' '''''' '''''''''' '''''''''''
FL histological category
at trial entry, n (%)
Grade 1 '''''' '''''''''' '''''''''' '''''''''' '''''' '''''''''' '''''' ''''''''' '' '''''''''
Grade 2 '''''' '''''''''' ''''''' '' ' '''''''''' ''''''' '''''''''' '''''' ''''''''' '''''''''''
Grade 3a ''''' '''''''''' '''''' ''''''''' '''''' ' ''''''' '''''''''' ''''''' ''''''''' ''''''
FLIPI total score, n (%)
Low risk (0–1) ''' ''''''' ' '''''''''' ''' '''''''''' ''''''' '''''''''' '''''' ''''''''' ''' '''''''''
Intermediate risk (2) '''''' ''''''''' ' '''''''''' '''''''''' '''''' ''''''''' '''''' ''''''''' ''''''''''
High risk (3–5) '''''' ''''''''' ' '''''''''' '''''' ''''''''' '''''' '''''''''' '''''' '''''''''' '''''''''
Relapsed/refractory
disease [a] , n (%)
Relapsed '''''' ''''''''' '' ''''''''' ''''''''' ''''''' ''''''''' ''''''' '''''''''' '' '''''''''''
Refractory ''''''' '''''''''' '''''''''' ''''''''''' '''''' '''''''''' '''''' '''''''''' ' '''''''''''
Double-refractory ''''''' '''''''''' '''''' ''''''''' ''''''''''' '''''' ''''''''' '''''' ''''''''''' ' '''''''''
subgroup [a] , n (%)
Median no. of prior ''''''' '''''''''''''''''''''' '''''''' ''''''' '''''''''''' '''''''' ''''''' '''''''
'''''''''''' ''''''''''''
therapies (range) ''''''''' ''''''''''''''''''''''''' '''''''''''''''''''''''''
''''''''''' '''''''''
Number of prior lines of
therapy, n (%)
1 '''' ''''''' ''' ''''''' ''' '' '' '''
2 ''''''' '''''''''' '' '''''''''' '''''' ''''''''' '' '' '''
3 '''''' '''''''''' '''''''''' '''''' ''''''''' '''''' ''''''''' ''''''' '''''''''' ' '''''''''
4 '''''' ''''''''' ' '''''''''' ' ''''''''''' '''''' ''''''''' ''''''' '''''''''' '' '''''''''
≥5 '''''' ''''''''' ' '''''''''' '''''' ''''''''' ''''' '''''''''' '''''' '''''''''' ''''''''''
Prior auto-SCT, n (%) '''''' ''''''''' ' '''''''''' '''''' ''''''''' '''''' '''''''''' '''''' '''''''''' '''''''''''
Prior PI3K inhibitor, n (%) ''''''' '''''''''' ''''' '''''''''' ''''''''''' '''''' ''''''''' '''''' '''''''''' ' '''''''''''
----- End of picture text -----

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 29 of 155

B.2.4. Statistical analysis and definition of study groups in the relevant clinical effectiveness evidence

Table 8 provides a summary of the statistical analysis and definitions of analysis sets in ZUMA-5.

Approximately 160 patients were to be enrolled and treated with axi-cel, including 125 FL patients, with at least 80 patients with FL in the IAS.[34] The trial used a single-

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 30 of 155

arm design to investigate the ORR in patients with r/r B-cell iNHL treated with axi-cel with a hypothesised target response rate in patients with FL of 60%.

The primary analysis was performed when at least 80 patients with FL were enrolled into the IAS, and patients had had the opportunity to be followed up for at least 12 months following the first disease assessment. After the primary analysis, additional follow-up analyses of efficacy and safety were to be performed, as needed to satisfy regulatory requirements and to perform long-term efficacy, safety and OS follow-up.

• Follow-up Analysis 1 was planned when at least 80 patients with FL in the IAS had had the opportunity to be followed up for 18 months

• Further analyses are planned (follow-up Analysis 2) when at least 80 patients with FL in the IAS have had the opportunity to be followed up for 24 months

This submission presents data from follow-up Analysis 1 (i.e. the 18-month analysis). Although all patients would have had the opportunity to be followed for 18 months after axi-cel infusion, some might have <18 months of actual follow-up because they were unable to attend the 18-month study visit.[35] For any patient unable to attend the 18-month study visit, an additional patient was added to the IAS to ensure that at least 80 patients had ≥18 months of follow-up. The IAS, therefore, included the first 86 patients to account for patients who were unable to reach the 18-month study visit per protocol.

Table 8: Summary of statistical analyses for ZUMA-5

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 31 of 155

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 32 of 155

B.2.4.1. Patient disposition data

Figure 5 provides a summary of patient disposition data for FL patients in ZUMA-5.

In total, '''''''''' patients with FL were enrolled into the ZUMA-5 trial (that is, underwent leukapheresis).[35] Axi-cel was successfully manufactured for all '''''''''' (100%) FL patients; there were no delays in infusion due to manufacturing issues. For patients who received axi-cel, the median time from leukapheresis to delivery of axi-cel to the study site was '''''''''' days (range: ''''''''''''''), and the median time from leukapheresis to administration was ''''''''''' days (range: '''''''''''''''''''' days).

At the data cut-off date for follow-up Analysis 1 (the 18-month analysis), ''''''''' patients with FL had received lymphodepleting chemotherapy and proceeded to axi-cel infusion.[35] '''''''''''''' patients did not receive either lymphodepleting chemotherapy or axi-cel infusion for the following reasons:

''''  Patient death ('''' '''')

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 33 of 155

• Ineligible for treatment due to low platelet levels and was subsequently treated in a compassionate use study ('''' ''' '''')

• Partial withdrawal from the study at data cut-off; however, this patient may be treated in the future ('''' ''' '''')

Figure 5: Patient disposition data for FL patients enrolled in the ZUMA-5 trial

==> picture [416 x 121] intentionally omitted <==

==> picture [416 x 120] intentionally omitted <==

Key: FL, follicular lymphoma. Source: Adapted from the ZUMA-5 CSR 18-Month Addendum.[35]

Bridging therapy was received by four FL patients between enrolment and

lymphodepleting chemotherapy; all four had documented measurable disease after bridging therapy.[35]

Of the '''''''''' FL patients enrolled, ''''' had received three or more prior lines of therapy; '''''' of whom were treated with axi-cel and ''''' of whom had at least 18 months of follow-up at the time of 18-month analysis.

B.2.5. Quality assessment of the relevant clinical effectiveness evidence

Quality assessment of ZUMA-5 was conducted using the Downs and Black checklist, full details of which are provided in Appendix D.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 34 of 155

Within the context of a single-arm study design, the overall risk of bias in ZUMA-5 is thought to be low. The primary endpoint (ORR) was determined by independent central review as per the Lugano classification and provides an objective estimate of treatment effect of relevance to clinical practice (where response to treatment is the primary measure of effect). The single-arm design does, however, necessitate a need for an indirect treatment comparison to provide relative treatment-effect estimates required for decision-making that is associated with higher uncertainty than a controlled trial would otherwise stipulate. This is further discussed in Section B.2.13. In terms of intervention, the axi-cel treatment schedule adopted in ZUMA-5 reflects the administration and dosing practice of axi-cel in clinical practice. Of note, there would be no additional impact or requirement for further FL-specific site qualification or referrals/monitoring and related processes, over and above what already exists for DLBCL/PMBCL patients in England.

B.2.6. Clinical effectiveness results of the relevant trials

ZUMA-5 patient groups and analysis sets for which data are presented are summarised in Table 9.

Efficacy data are presented for the IAS of FL patients with two or more lines of prior therapy as this was the predefined primary efficacy analysis set. However, to align with the anticipated marketing authorisation and target population for reimbursement, efficacy data are also presented for the IAS and SAS of FL patients with three or more lines of prior therapy; for efficacy analyses this latter population is referred to as the modified intent-to-treat (mITT) population from hereon in.

This submission presents data from follow-up Analysis 1 (the 18-month analysis).

Table 9: Summary of ZUMA-5 cohorts and analysis sets presented

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 35 of 155

B.2.6.1. FL patients with two or more lines of prior therapy

B.2.6.1.1. Response and duration of response (IAS)

The primary efficacy endpoint of ORR in patients with r/r FL who had the opportunity to be followed for at least 18 months from first disease assessment date following axi-cel treatment was '''''''''''' patients (''''''%). This ORR was statistically significantly greater than the pre-specific historical control rate of '''''''% ('''' ''' '''''''''''''''').[35] The CR rate was ''''''''''''' patients ('''''''%). This CR was statistically significantly greater than the ''' ''' '''''''''''''''' pre-specified control rate of ''''''% ( ).

Among the ''''''' patients who achieved a CR or PR, the median time to first objective response was <1 month.[35] Of patients who initially had a PR (n = ''''''), '''''' later achieved a CR. The median DOR in all responders (n = '''''') was not reached and '''''' patients ('''''''%) had an ongoing response at data cut-off (minimum follow-up of 18 months from first disease assessment; median follow-up of '''''''''' months for DOR).

The median DOR in CRs (n = '''''') was not reached and ''''''' patients ('''''''%) had an ongoing response at data cut-off.

A summary of response and duration of response data are presented in Table 10.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 36 of 155

Table 10: Summary of response using central assessment per Lugano classification; FL patients with two or more lines of prior therapy, IAS

N = '''''' Objective response rate (CR + PR), n (%) '''''' '''''''''' [95% CI] ''''''''' ''''''' p-value vs historical control rate ''''''''''''''''''' Best objective response Complete response rate, n (%) '''''' '''''''''' [95% CI] ''''''''' ''''''' Partial response, n (%) '''''' ''''''''' [95% CI] '''''' '''''''' Stable disease, n (%) '''' ''''''' [95% CI] ''''''' ''''''' Progressive disease, n (%) '''' ''''''' [95% CI] ''''''' ''''' Time to response Median time to response in all responders, ''''''''''' '''''''''''' ''''''''' months (range) Median time to response in CRs, months ''''''''''' '''''''''''' '''''''''''' (range) Duration of response Median duration of response in all responders, '''''''' '''''''''' '''''''''''' months (range) Median duration of response in CRs, months ''''''' ''''''''''' ''''''''''''' (range) Key: CI, confidence interval; CR, complete response; CSR, clinical study report; FL, follicular lymphoma; IAS, inferential analysis set; NE, not evaluable; PR, partial response. Source: ZUMA-5 CSR 18-Month Addendum.[35]

The Kaplan–Meier plots for DOR and DOR by best response for all patients in the IAS are provided in Appendix L.

B.2.6.1.2. Progression-free survival (IAS)

Median PFS ''''''''' '''''''' ''''''''''' '''''''''''''''''' '''''''''''''' ''''''' ''''''''''' ''''''' '''''''''''''''''''''' ''''''''''''' in patients

with r/r FL who had the opportunity to be followed for at least 18 months from first disease assessment date, and had a median follow-up time for PFS of '''''''''' months (reverse Kaplan–Meier approach).[35]

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 37 of 155

At the time of analysis, ''''''' patients ('''''''%) had progressed or died.[35] The estimated 12-month PFS rate was ''''''''''''% (95% CI: ''''''''''', ''''''''''') and the estimated 18-month PFS rate was ''''''''''% (95% CI: '''''''''', '''''''''''').

Among the ''''''' patients who achieved a CR, ''''''' ('''''''''''%) had progressed or died.[35] The estimated 12 and 18-month PFS rates in CRs were '''''''''''% (95% CI: ''''''''''%, ''''''''''%) and ''''''''''''% (95% CI: ''''''''''%, '''''''''''%), respectively.

The Kaplan–Meier plots for PFS and PFS by best response for all patients in the IAS are provided in Appendix L.

B.2.6.1.3. Overall survival (IAS)

Median OS '''''''' '''''''' '''''''''''' '''''''''''''''''''' ''''''''''''' '''''''' ''''''''' ''''''''' in patients with r/r FL who had the opportunity to be followed for at least 18 months from first disease assessment date following axi-cel treatment, and had a median follow-up time for OS of '''''''''' months (reverse Kaplan–Meier approach).[35]

At the time of analysis, '''''' patients ('''''%) had died. The estimated 12-month OS rate was ''''''''''% (95% CI: '''''''''', ''''''''''') and the estimated 18-month OS rate was ''''''''''% (95% CI: '''''''''''', ''''''''''').

Among the ''''''' patients who achieved a CR, '''''''''''''''' patients (''''''%) had died. The estimated 12- and 18-month OS rates in CRs were '''''''''''% (95% CI: ''''''''''%, '''''''''''%) and '''''''''''% (95% CI: '''''''''''%, '''''''''''%), respectively.

The Kaplan–Meier plots for OS and OS by best response for all patients in the IAS are provided in Appendix L.

B.2.6.1.4. Retreatment data (IAS)

Overall, '''''' FL patients were retreated with axi-cel after disease progression. '''''''''''' of these patients achieved a CR to retreatment and the remaining ''''''''''' achieved a

PR.[35] At the time of analysis, ''''''''''' patents (''''''''''''''' CRs and ''''''''' PR) had an ongoing response to retreatment and neither the median DOR nor the median PFS had been reached.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 38 of 155

B.2.6.1.5. Subsequent therapy (SAS)

''''''''''''''''''''''''''' ('''''%) patients with FL in the safety analysis set had subsequent anticancer therapy.[35] No single regimen was used to treat more than ''''''''' ''''''''''''''''''''', further demonstrating the lack of established standard of care at later-line settings. A summary of the regimens that were given is provided in Appendix L.

'''''''''''' (''''%) patients with FL in the safety analysis set had a stem cell transplant after treatment with axi-cel (''''''''' auto-SCT and ''''''''''''' allogeneic stem cell transplant). ''''''''''''' of these patients initially had a PR but then progressed; the '''''''''''''' was a nonresponder whose disease progressed.

B.2.6.2. FL patients with three or more lines of prior therapy

B.2.6.2.1. Response and duration of response (IAS)

The secondary efficacy endpoint of ORR in patients with FL who had received three or more lines of prior therapy and had the opportunity to be followed for at least 18 months from first disease assessment date following axi-cel treatment was ''''''''''''''' patients ('''''%). This ORR was significantly greater than the pre-specific historical control rate of ''''''% ('''' '''' ''''''''''''''''').[35] The CR rate was ''''''''''''' patients (''''''%), significantly greater than the pre-specified control rate of ''''''% ('''' ''' '''''''''''''''').

The median DOR in all responders (n = '''''') was not reached and '''''' patients ('''''''%) had an ongoing response at data cut-off (minimum follow-up of 18 months from first disease assessment, median follow-up of '''''''''''' months for DOR).[35] The median DOR in CRs (n = '''''') was not reached and '''''' patients (''''''%) had an ongoing response at data cut-off.

A summary of response and duration of response data are presented in Table 11.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 39 of 155

Table 11: Summary of response using central assessment per Lugano classification; FL patients with three or more lines of prior therapy, IAS

N = ''''' Objective response rate (CR + PR), n (%) '''''' ''''''''' [95% CI] '''''''''' ''''''' p-value vs historical control rate '''''''''''''''''' Best objective response Complete response rate, n (%) '''''' '''''''''' [95% CI] ''''''''' ''''''' Partial response, n (%) ''' '''''''''' [95% CI] '''''' '''''''' Stable disease, n (%) '''' '''''''' [95% CI] ''''''' '''' Progressive disease, n (%) ''' '''''' [95% CI] '''''' ''''' Duration of response Median duration of response in all responders, ''''''' '''''''''''' '''''''''''' months (range) Median duration of response in CRs, months ''''''' '''''''''''' '''''''''''''' (range) Key: CI, confidence interval; CR, complete response; CSR, clinical study report; FL, follicular lymphoma; IAS, inferential analysis set; NE, not evaluable; PR, partial response. Source: ZUMA-5 CSR 18-Month Addendum.[35]

The Kaplan–Meier plots for DOR and DOR by best response for FL patients with three or more lines of prior therapy in the IAS are provided in Appendix L.

B.2.6.2.2. Progression-free survival (IAS)

Median PFS '''''''' ''''''' ''''''''''' '''''''''''''''''' '''''''''''' ''''''' ''''''''''''' '''''''''' in patients with FL who had

received three or more lines of prior therapy and had the opportunity to be followed for at least 18 months from first disease assessment date following axi-cel treatment, and a median follow-up time for PFS of '''''''''' months (reverse Kaplan–Meier approach).[35]

At the time of analysis, '''''' patients (''''''%) had progressed or died.[35] The estimated 12-month PFS rate was ''''''''''% (95% CI: ''''''''''', '''''''''') and the estimated 18-month PFS rate was ''''''''''% (95% CI: '''''''''', '''''''''''').

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 40 of 155

Among the ''''' patients who achieved a CR, '''''' ('''''''%) had progressed or died.[35] The estimated 12 and 18-month PFS rates in CRs were '''''''''''% (95% CI: ''''''''''''%, ''''''''''''%) and '''''''''''% (95% CI: '''''''''''%, ''''''''''%), respectively.

The Kaplan–Meier plots for PFS and PFS by best response for FL patients with three or more lines of prior therapy in the IAS are provided in Appendix L.

B.2.6.2.3. Overall survival (IAS)

Median OS '''''''' '''''''' '''''''''''' ''''''''''''''''''' ''''''''''''' '''''''' ''''''''''''' ''''''''' in patients with FL who had received three or more lines of prior therapy and had the opportunity to be followed for at least 18 months from first disease assessment date following axi-cel treatment, and had a median follow-up time for OS of '''''''''' months (reverse Kaplan–Meier approach).[35]

At the time of analysis, '''''' patients (''''''%) had died.[35] The estimated 12-month OS rate was ''''''''''% (95% CI: ''''''''''', '''''''''') and estimated 18-month OS rate was ''''''''''% (95% CI: ''''''''''', '''''''''''').

Among the '''''' patients who achieved a CR, ''''''' (''''''%) had died. The estimated 12 and 18-month OS rates in CRs were ''''''''''% (95% CI: '''''''''''%, ''''''''''%) and '''''''''''% (95% CI: ''''''''''%, '''''''''''%), respectively.

The Kaplan–Meier plots for OS and OS by best response for FL patients with three or more lines of prior therapy in the IAS are provided in Appendix L.

B.2.6.2.4. Response and duration of response (mITT)

Post-hoc analyses of ORR in patients with FL who had received three or more lines of prior therapy and were treated with axi-cel was ''''''''''''''' patients (''''''%). This ORR was significantly greater than the pre-specific historical control rate of ''''''% ('''' '''' '''''''''''''''). The CR rate was ''''''''''''''' patients (''''''%), significantly greater than the prespecified control rate of ''''''% ('''' '''' ''''''''''''''''). The median DOR in all responders (n = '''''') was not reached with a median follow-up time of '''''''''' months (reverse Kaplan– Meier approach).

A summary of response and duration of response data is presented in Table 12.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 41 of 155

Table 12: Summary of response using central assessment per Lugano classification; FL patients with three or more lines of prior therapy, mITT

''' ''' ''''' Objective response rate (CR + PR), n (%) '''''''''''''''''''''''' ''''''' [95% CI] Best objective response Complete response rate, n (%) ''''''''''''''''''''''''' ''''''' [95% CI] Partial response, n (%) ''''''' '''''''''' [95% CI] '''''' ''''''' Stable disease, n (%) ''' ''''''' [95% CI] ''''''' ''''''' Progressive disease, n (%) '''' ''''''' [95% CI] '''''' ''''' Duration of response Median duration of response in all responders, ''''''' '''''''''''' '''''''''''''' months (range) Key: CI, confidence interval; CR, complete response; CSR, clinical study report; FL, follicular lymphoma; mITT, modified intent-to-treat; NE, not evaluable; PR, partial response. Notes: mITT includes all patients treated with any dose of axi-cel. Source: ZUMA-5 CSR 18-month addendum.[35]

The Kaplan–Meier plots for DOR and DOR by best response for FL patients with three or more lines of prior therapy in the mITT are provided in Figure 6 and Figure 7.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 42 of 155

Figure 6: Kaplan–Meier plot for duration of response; FL patients with three or more lines of prior therapy, mITT

==> picture [429 x 119] intentionally omitted <==

==> picture [429 x 119] intentionally omitted <==

Key: CI, confidence interval; FL, follicular lymphoma; mITT, modified intent-to-treat; NE, not evaluable Notes: mITT includes all patients treated with any dose of axi-cel. Source: ZUMA-5 CSR 18-month addendum[35] and data on file.

Figure 7: Kaplan–Meier plot for duration of response by best response; FL

patients with three or more lines of prior therapy, mITT

==> picture [427 x 115] intentionally omitted <==

==> picture [427 x 115] intentionally omitted <==

Key: CI, confidence interval; CR, complete response; FL, follicular lymphoma; mITT, modified intentto-treat; NE, not evaluable; PR, partial response. Notes: mITT includes all patients treated with any dose of axi-cel. Source: ZUMA-5 CSR 18-month addendum[35] and data on file.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 43 of 155

B.2.6.2.5. Progression-free survival (mITT)

Median PFS ''''''''' ''''''' '''''''''''' ''''''''''''''''' ''''''''''''' ''''''' '''''''''''' '''''''''' in patients with FL who had received three or more lines of prior therapy and were treated with axi-cel, after a median follow-up time for PFS of '''''''''' months (reverse Kaplan–Meier approach).

At the time of analysis, '''''' patients ('''''''%) had progressed or died. The estimated 12month PFS rate was ''''''''''% (95% CI: '''''''''''', '''''''''''') and the estimated 18-month PFS rate was '''''''''''% (95% CI: ''''''''''', '''''''''').

Among the '''''' patients who achieved a CR, ''''''' ('''''%) had progressed or died. The estimated 12 and 18-month PFS rates in CRs were ''''''''''''''' (95% CI: ''''''''''''''''' '''''''''''''''''' and '''''''''''% (95% CI: ''''''''''''''''' '''''''''''''''''' respectively.

The Kaplan–Meier plots for PFS and PFS by best response for FL patients with three or more lines of prior therapy treated with axi-cel are provided in Figure 8 and Figure 9.

Figure 8: Kaplan–Meier plot for progression-free survival; FL patients with three or more lines of prior therapy, mITT

==> picture [453 x 116] intentionally omitted <==

==> picture [453 x 116] intentionally omitted <==

Key: CI, confidence interval; FL, follicular lymphoma; mITT, modified intent-to-treat; NE, not evaluable.

Notes: mITT includes all patients treated with any dose of axi-cel. Source: ZUMA-5 CSR 18-month addendum[35] and data on file.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 44 of 155

Figure 9: Kaplan–Meier plot for progression-free survival by best response; FL patients with three or more lines of prior therapy, mITT

==> picture [453 x 115] intentionally omitted <==

==> picture [453 x 115] intentionally omitted <==

Key: CI, confidence interval; CR, complete response; FL, follicular lymphoma; mITT, modified intentto-treat; NE, not evaluable; PR, partial response. Notes: mITT includes all patients treated with any dose of axi-cel. Source : ZUMA-5 CSR 18-month addendum[35] and data on file.

B.2.6.2.6. Overall survival (mITT)

Median OS ''''''''' ''''''' ''''''''''''' '''''''''''''''''''' '''''''''''' ''''''' '''''''''''' ''''''''' in patients with FL who had received three or more lines of prior therapy and were treated with axi-cel, after a median follow-up time for OS of ''''''''''' months (reverse Kaplan–Meier approach).

At the time of analysis, '''''' ('''''''%) had died. The estimated 12-month OS rate was '''''''''''% (95% CI: '''''''''''''''', ''''''''''''''''') and the estimated 18-month OS rate was ''''''''''''% (95% CI: '''''''''''''', ''''''''''''''''').

Among the '''''' patients who achieved a CR, ''''''' patients (''''''%) had died. The estimated 12 and 18-month OS rates in CRs were ''''''''''% (95% CI: '''''''''''%, '''''''''''%) and ''''''''''% (95% CI: '''''''''''''''''' ''''''''''''''''' respectively.

The Kaplan–Meier plots for OS and OS by best response for FL patients with three or more lines of prior treated with axi-cel are provided in Figure 10 and Figure 11.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 45 of 155

Figure 10: Kaplan–Meier plot for overall survival; FL patients with three or more lines of prior therapy, mITT

==> picture [455 x 110] intentionally omitted <==

==> picture [455 x 110] intentionally omitted <==

Key: CI, confidence interval; FL, follicular lymphoma; mITT, modified intent-to-treat; NE, not evaluable. Notes: mITT includes all patients treated with any dose of axi-cel. Source: ZUMA-5 CSR 18-month addendum[35] and data on file.

Figure 11: Kaplan–Meier plot for overall survival by best response; FL patients

with three or more lines of prior therapy, mITT

==> picture [453 x 115] intentionally omitted <==

==> picture [453 x 114] intentionally omitted <==

Key: CI, confidence interval; CR, complete response; FL, follicular lymphoma; mITT, modified intentto-treat; NE, not evaluable; PR, partial response. Notes: mITT includes all patients treated with any dose of axi-cel. Source: ZUMA-5 CSR 18-month addendum[35] and data on file.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 46 of 155

B.2.7. Subgroup analysis

Subgroup analyses demonstrated consistent ORR and CR rates across subgroups, with rates that were generally comparable with the those observed in the overall population. Subgroup analyses were only conducted on the total FL population.

A summary of results for subgroups analysed is provided in Appendix E.

B.2.8. Meta-analysis

Not applicable.

B.2.9. Indirect and mixed treatment comparisons

A systematic literature review identified a paucity of clinical evidence for patients with FL who have received at least three prior therapies (Appendix D). Three potentially relevant studies that may have been used for comparative efficacy data were identified:

• Batlevi et al. (2020)[11]

• Link et al. (2019)[13]

• Fuji et al. (2020)[12]

However, all three studies were considered highly limited in their suitability as a source of comparative efficacy data for axi-cel in patients with 4L+ r/r FL in England. Firstly, none of the studies reported baseline characteristics in the 4L+ FL setting. Secondly, none of the studies were in a European setting, which may have led to unobserved biases due to both differences in the natural history of disease and potentially differing treatment patterns. As a result, no matching comparison for treatment-effect modification has been conducted for these three studies.

B.2.9.1. SCHOLAR-5

B.2.9.1.1. Methods

An international, multicentre, external control cohort study, SCHOLAR-5, was designed to provide comparative evidence for axi-cel in patients with r/r FL meeting ZUMA-5 eligibility criteria.[21] SCHOLAR-5 was also designed to help characterise the natural history of FL and current treatment patterns.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 47 of 155

The SCHOLAR-5 cohorts were created from multiple data sources:

• Cohort A – retrospective cohort created from electronic medical records of six sites, including university hospitals and cancer centres with two sites based in the UK and other sites based in France, Spain, Portugal and the US

• Cohort B – retrospective cohort created from the Vanderbilt University Medical Center’s Synthetic Derivative: a fully de-identified database derivative of electronic medical records from the university

• Cohort C – prospective cohort created from an open-label Phase II study, DELTA, that enrolled patients with r/r FL who had not responded to or were refractory to rituximab and an alkylating agent and were treated with idelalisib

Adult patients with r/r FL or MZL starting third or later-line therapy on or after April 2011 to July 2014 (depending on source) were eligible for the SCHOLAR-5 cohort.[21] Before analysis set construction, cohorts were restricted to FL patients who had received at least three prior lines of therapy, aligning to the anticipated marketing authorisation for axi-cel.

Two analysis sets were initially constructed to meet the study objectives. The realworld analysis set contained Cohort A and B patients (n = 58) and was used to characterise the natural history of FL and current treatment patterns.[21] The effectiveness analysis set (EAS) contained Cohort A, B and C patients with FL (n = 82) and was used to provide primary comparative analysis. A modified EAS was also constructed that contained EAS patients who had a recorded Eastern Cooperative Oncology Group score and age range within the patients of ZUMA-5 (n = 67).

The primary effectiveness endpoint was predefined as ORR within a line of treatment, but a variety of methods were used to assess response across cohorts.[21] Secondary effectiveness endpoints of interest included CR rate, DOR, PFS and OS.

Propensity scoring methods, specifically standardised mortality ratio (SMR) weighting, were applied to account for imbalances of confounders between the ZUMA-5 and SCHOLAR-5 populations prior to comparative analyses. Full details of the propensity scoring methods and outcomes are provided in the SCHOLAR-5 technical report in the reference pack.[21] Sensitivity analyses using propensity score

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 48 of 155

matching rather than SMR weighting are also described in the SCHOLAR-5 technical report.

B.2.9.1.2. Natural history and treatment pattern outcomes

Natural history and real-world treatment pattern data from SCHOLAR-5 are summarised in Table 13, Figure 12 and Figure 13.

The lack of standard of care in later-line settings is clear to see in real-world treatment pattern data, with fourth-line treatments consisting of a varied mix of immunochemotherapy, chemotherapy and experimental treatments.[36] In post-hoc analyses of the UK cohort of patients (''' '''' '''''), a similar mix of

immunochemotherapy, chemotherapy and experimental treatments were reported, along with what we assume is compassionate use of PI3Ki-based treatment[21] (as such treatment is not routinely reimbursed in National Health Service [NHS] England[33] ).[21]

Natural history data from SCHOLAR-5 further demonstrate the reduced effectiveness of treatment over time, as observed in previous studies.[11-13] ORR reduced from 66% with third-line treatment to 53% and 37% with fourth and fifth or later-line treatments, respectively (Table 13).[36] Survival outcomes similarly reduced with each line of treatment (Figure 12 and Figure 13). The proportion of patients alive at 18-months reduced from 83% with third-line treatment to 72% and only 36% with fourth and fifth or later-line treatment, respectively (Table 13).[36]

Table 13: Natural history represented by response by line of therapy

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 49 of 155

Figure 12: (A) Real-world clinical site treatment patterns by line of therapy[a,b ] and (B) fourth line treatment patterns by region (US and Europe)

==> picture [28 x 23] intentionally omitted <==

==> picture [344 x 237] intentionally omitted <==

==> picture [452 x 203] intentionally omitted <==

Key: Benda, bendamustine; Chemo, chemotherapy; CHOP, cyclophosphamide, doxorubicin, vincristine and prednisolone; CVP, cyclophosphamide, vincristine and prednisolone; LoT, line of therapy; R[2] , lenalidomide with rituximab; SCT, stem cell transplant. Note :[a] , All SCHOLAR-5 patients (i.e. excludes DELTA trial data);[b] , percentage values represent proportion of patients with line of treatment Source: Ghione et al. (2021a)[37] ; Ghione et al. (2021b).[36]

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 50 of 155

Figure 13: Natural history represented by (A) progression-free survival (B) overall survival by line of therapy

==> picture [452 x 451] intentionally omitted <==

Source: Ghione et al. (2021).[36]

B.2.9.1.3. Comparative analyses

Baseline characteristics of patients pre- and post-weighting are provided in Table 14. Both the IAS and the mITT of ZUMA-5 were used for weighting and comparative analyses; IAS data are provided in Appendix M and mITT data are provided below. Before weighting, significant differences were observed in POD24 status, the number of prior lines of therapy and the proportion of patients >65 years of age (''' '''

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 51 of 155

''''''''''' ''''''' ''''').[21] After applying SMR weights, the effective sample size for SCHOLAR5 reduced from 82 patients to 77, but there were no significant differences remaining in baseline characteristics.

Note: It was not feasible to compare patients from SCHOLAR-5 to ZUMA-5 based on double-refractory status as the SCHOLAR-5 protocol did not include doublerefractory status as a variable. It is acknowledged that this is a limitation as doublerefractory status is a recognised poor prognostic indicator in the r/r FL treatment setting. However, UK clinical experts agreed that POD24, tumour bulk, time since last treatment and age are highly appropriate prognostic factors to which comparative analyses between SCHOLAR-5 and ZUMA-5 have been conducted.

Table 14: Baseline characteristics of patients pre- and post-weighting; FL patients with three or more lines of prior therapy, SCHOLAR-5 EES, ZUMA-5 mITT

==> picture [454 x 359] intentionally omitted <==

----- Start of picture text -----
Pre-weighting Post-weighting
SC-5
Characteristics
SC-5 Z-5 p-value Z-5 p-value
(EES =
(n = 82) (n = 78) [SMD] (n = 78) [SMD]
77)
POD24, n (%)
Yes ''''''
'''''' '''''''''''''''''''' '''''' ''''''''''''''''''''
''''''''''''' '''''''''''''
'''''''''''''''''''' ' '''''''''''''''''''' ''''''
No
'''''' '''''' '''''''''''''''''' ''''''' '''''''''''''''''''' ''''''''''''''''
''''''''''''''''''' '''''''''''''''''''' '''''''''''''''''' ''''
Missing ''' '''''''''''' ''' ''''''''''' ''' '''''''''''' ''' ''''''''''''
Prior lines of therapy
Mean (SD) '''''''''''' '''''''''' '''''''''''''''''''' ''''''''''''
''''''''''''' ''''''''''''''' ''''''''''''' ''' '''''''''''''' '''''''''''''
Median (range) ''''''' '''''''''''''''''''' '''''''' '''''''''''''''''''' ''''''''''''''' '''''''' '''''''''''''''''''' ''''''' '''''''''''''''''''' '''''''''''''''''
' '''''' ' ''''''
Relapsed/refractory
to prior line of
therapy
Relapsed ''''''' '''''' ''''''''''''' '''''' '''''' ''''''''''''''
'''''''''''''''''''' ''''''''''''''''''' '''''''''''''''' '''''''''''''''''' '''''''''''''''''''' ''''''''''''''''
Refractory '''''' '''''' '''''' '''''
''''''''''''''''' '''''''''''''''''''' '''''''''''''''''' '''''''''''''''''''
Missing ''' ''''''''''' '''' '''''''''' '''' '''''''''''' '''' ''''''''''''
----- End of picture text -----

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 52 of 155

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 53 of 155

Comparative analysis results are summarised in Table 15, and Kaplan–Meier curves for PFS and OS are shown in Figure 14 and Figure 15.

ORR and CR rates were markedly higher in ZUMA-5 than SCHOLAR-5, with odds ratios of '''''''''''' and '''''''''''''' respectively, which were statistically significant at the '''''''''' '''''''''' ''''''''''' '''' ''' ''''''''''''''''[21] Although median PFS, OS and DOR were not reached for ZUMA-5, first quartile estimates exceeded the medians reached in SCHOLAR-5 for all time to event outcomes.

Cox model hazard ratios predict an ''''''% reduction in risk of disease progression or death ('''' '''' '''''''''''''') and a '''''% reduction in risk of death alone ('''' '''' '''''''''''''') with axicel treatment versus current care.[21] For DOR, cox model hazard ratios predict a ''''''% reduction in the risk of relapse after response (''' '''' ''''''''''''').

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 54 of 155

Table 15: Comparative analysis summary; FL patients with three or more lines of prior therapy, SCHOLAR-5, EES, ZUMA-5 mITT

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 55 of 155

Figure 14: Progression-free survival for axi-cel (ZUMA-5) and current 4L+ care (propensity score weighted SCHOLAR-5)

==> picture [484 x 109] intentionally omitted <==

==> picture [484 x 110] intentionally omitted <==

==> picture [484 x 109] intentionally omitted <==

Key: 4L+, fourth-line plus; mITT, modified intent-to-treat. Source: Gilead data on file

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 56 of 155

Figure 15: Overall survival for axi-cel (ZUMA-5) and current 4L+ care

(propensity score weighted SCHOLAR-5)

==> picture [484 x 109] intentionally omitted <==

==> picture [484 x 110] intentionally omitted <==

==> picture [484 x 109] intentionally omitted <==

Key: 4L+, fourth-line plus; mITT, modified intent-to-treat. Source: Gilead data on file

Sensitivity analyses across different analysis sets from SCHOLAR-5 and ZUMA-5 are provided in the appendices of the SCHOLAR-5 technical report.[21] Irrespective of the analysis set used, axi-cel demonstrated a significant response and survival benefit compared with current care for the treatment of FL patients who had received at least three lines of prior therapy.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 57 of 155

B.2.9.2. Uncertainties in the indirect and mixed treatment comparisons

Comparative analyses clearly indicate improved response and survival outcomes with axi-cel compared with current care, but the wide CIs around point estimates indicate that the exact magnitude of improvement is unclear. Robustness of comparative analyses are, however, supported with consistent findings across primary and sensitivity analyses exploring different weighting methods and analysis sets across the two studies.

All comparisons of real-world data to clinical trial data are subject to common limitations relating to differences in disease assessment approaches across these settings. Real-world data response and progression outcomes from SCHOLAR-5 are likely inflated compared with ZUMA-5 owing to significant differences in the way such data are sourced and collected in the real-world versus a controlled trial environment. Response and progression outcomes in clinical practice are typically derived and interpreted on a centre-by-centre basis by an individual treating physician based on a heterogenous schedule of when that patient is available for follow-up, and typically without bone marrow confirmation. Further inherent limitations with the real-world data set of SCHOLAR-5 included missing baseline characteristic data for some potentially confounding factors, such as FLIPI score, particularly at later lines of therapy, and variation in line of therapy definitions. The latter resulted in a loss of sites and patients in the feasibility phase of the study.

Finally, the highly variable treatment approach at fourth and later-line prevents any treatment-specific comparisons to be conducted. However, the blended comparator approach to comparative analyses offers a high degree of external validity and generalisability considering the lack of standard of care in current 4L practice.

In the absence of clinical evidence for patients with FL who have received at least three prior therapies in the existing evidence base, SCHOLAR-5 provides the strongest evidence on which to form comparative effectiveness conclusions for axicel versus current care.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 58 of 155

B.2.10. Adverse reactions

B.2.10.1. Safety summary

Table 16 presents an overview of safety outcomes from ZUMA-5 at the time of the 18-month analysis.

In total, '''''''''' ('''''''%) of all FL patients in the SAS experienced at least one adverse event (AE); '''''''' (''''''%) patients experienced worst Grade 3 or higher AEs, and ''''''' ('''''%) had serious AEs (SAEs).[35] As of the 18-month analysis data cut-off date, '''''''''''

''''''''' '''''''''''''''''' '''''''' '''''''''''' '''''''''' ''''' '''''' ''''''' '''''''''''''''''' '''''''''''''''''' ''''' ''''''''''''''' ''''''''''''' ''''''''' ''''''''' '''''

''''''''''''''''''' ''''''''''''''' '''''''''''''''''''''''''''''' '''''''''''''''''''''' '''''''''' '''''''''' ''''''''''''''''''''''''''' ''''' ''''''''''''''''''''''''''' '''''

''''''''''''''''''''''''''' ''''''' ''''''''' '''' Similar safety outcomes were observed in FL patients with at least three prior lines of therapy in the SAS (Table 16).

Table 16: Safety summary for FL patients in ZUMA-5

B.2.10.2. Common adverse events

Table 17 presents the most common AEs (occurring in ≥ 30% of patients with FL) following treatment with axi-cel.

The most common any grade AEs of all FL patients in the SAS were pyrexia ('''''''''' patients [''''''%]), hypotension ('''''' patients ['''''''%]) and headache ('''''' patients ['''''''%]); whereas the most common Grade 3 or above AEs were neutropenia ('''''' patients

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 59 of 155

[''''''%]) and anaemia ('''''' patients [''''''%]).[35] A similar safety profile was observed in FL patients with at least three prior lines of therapy in the SAS (Table 17).

Table 17: Common adverse events occurring in ≥ 30% of FL patients in

ZUMA-5

SAEs that occurred in ≥ 2% of patients are summarised in Appendix N.

Among all FL patients in the SAS, the most common SAEs were pyrexia (''''' patients [''''''%]), pneumonia (''''''''''''' patients [''''%]), encephalopathy ('''''''''''' patients [''''%]) and confusional state ('''''''''''''' patients ['''%]).[35] The most common Grade 3 or higher SAEs were encephalopathy (''''''''''' patients [''''%]), pneumonia ('''''''''''''''' patients [''''%]) and confusional state ('''''''''' patients [''''%]).

Among FL patients with at least three prior lines of therapy in the SAS, the most common SAEs were pyrexia ('''''' patients [''''''%]), pneumonia (''''''' patients ['''%]), confusional state (''''''''' patients [''''%]) and encephalopathy (''''''''' patients ['''%]).[35] The most common Grade 3 or higher SAEs were encephalopathy ('''''''''' patients [''''%]), pneumonia ('''''''' patients [''''%]) and confusional state (''''''''''''' patients ['''%]).

B.2.10.3. Treatment-related adverse events

Table 18 presents the most common treatment-related AEs (occurring in ≥ 20% of patients with FL) following treatment with axi-cel.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 60 of 155

The most common any grade treatment-related AEs of all FL patients in the SAS were pyrexia (''''' patients [''''''%]), hypotension (''''''' patients ['''''%]) and headache ('''''' patients [''''''%]); whereas the most common Grade 3 or above treatment-related AEs were hypoxia (''''''''''' patients [''''''%]) and neutropenia ('''''' patients [''''''%]).[35] A similar safety profile was observed in FL patients with at least three prior lines of therapy in the SAS (Table 18).

Table 18: Common treatment-related adverse events occurring in ≥ 20% of FL patients in ZUMA-5

B.2.10.4. Adverse events of special interest

B.2.10.4.1. Cytokine release syndrome

CRS is an AE induced by the activated T-cells upon engagement with the CD19 target, and therefore is generally considered to be related to treatment with CAR T- cell therapy. In ZUMA-5, the severity of CRS was graded according to a modification of the grading system proposed by Lee et al.[38]

Table 19 presents CRS rates and the most common symptoms of CRS (occurring in ≥ 4% of patients with FL) following treatment with axi-cel.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 61 of 155

CRS was experienced by '''''' (''''''%) FL patients, of which '''''''''''' (''''%) had worst Grade 3 or higher CRS and '''''''''' ('''%) had worst Grade 5 CRS.[35] The most common symptoms of CRS Grade 3 or higher were hypoxia ('''''' patients ['''%]), pyrexia (''''''' patients ['''%]) and hypotension ('''''''''''' patients [''''%]). The most common serious CRS symptoms by any grade were pyrexia (''''' patients ['''%]), hypoxia ('''''''''''' patients [''''%]) and hypotension ('''''''' patients [''''%]).

The median time to onset of CRS in FL patients was ''''''''' days (range: ''''''''''') following axi-cel infusion.[35] At the 18-month analysis data cut-off date, CRS had

resolved in ''''' ''''''' '''''''' '''''''''''''''''' in ZUMA-5; '''''''' '''''''''''''''' '''''''''' ''''''' '''''''''' '''' '''''''''' '''''

''''''''''''''''' ''''''''''''''' ''''''''''''''''''''''''''''' ''''''''''''''''''''''''' '''''''' '''''''''. For the ''''''' patients with FL whose CRS had resolved, the median duration of CRS was ''''''' days (range: '''''''''''').

A similar CRS profile was observed in FL patients with at least three prior lines of therapy in the SAS (Table 19).

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 62 of 155

Table 19: Summary of CRS symptoms and events occurring in ≥ 4% of FL patients in ZUMA-5

B.2.10.4.2. Neurological events

Neurological events are a commonly reported AE of CAR T-cell therapy, manifesting through a diverse range of symptoms. It is strongly recommended that patients are systematically monitored and undergo neurological assessments both prior to and following CAR T-cell infusion. In ZUMA-5, neurological events were identified based on a modification of the search strategy by Topp et al.[39]

Table 20 presents neurological event rates and symptoms of neurological events following treatment with axi-cel.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 63 of 155

In total, '''''' patients (''''''%) had at least one neurological event of any grade, '''''' (''''''%) had worst Grade 3 neurological events and ''''''''' (''''%) patients had worst Grade 4 neurological events.[35] '''''''' ''''''''''''''' had a Grade 5 neurological event. The most common Grade 3 or higher neurological events were encephalopathy ('''''' patients ['''%]), confusional state ('''''' patients ['''%]) and aphasia ('''''''''''' patients [''''%]). The most common serious neurological events by any grade were encephalopathy (''''''''''''' patients ['''%]), confusional state ('''''''''''''' patients [''''%]), somnolence, aphasia, agitation and immune effector cell-associated neurotoxicity syndrome (''''''''' patients each ['''%]).

The median time to onset of neurological event in FL patients was ''''''' days (range: '''''''''''''''); ''''''' ''''''''''''''''''''' had neurological events with an onset >80 days after the axicel infusion.[35] Of note, consulted clinical experts noted that the delayed/late-onset, low-grade neurological events observed in the handful of FL patients, was not likely to have any considerable impact.[1] At the 18-month analysis data cut-off date, neurological events had resolved in ''''''' patients (''''''%). Of the '''''''''''' '''''''''''''''''' with

unresolved neurological events, '''''''''' ''''''''' ''' '''''''''''''''''''''''''' ''''' '''''''''''''''''' '''''''''''' ''''''''' ''''''''''''''''''' '''''''''''''''''' ''''' ''''''''''''''''' '''''''' '''''''''''' '''''''''' '''''''' '''' '''''' '''''''''''''''''''' ''''''''''''''' . For the 67

patients with FL whose neurological event had resolved, the median duration of the event was '''''''''' days (range: ''''''''''''''').

A similar neurological event profile was observed in FL patients with at least three prior lines of therapy in the SAS (Table 20).

Table 20: Summary of neurological events occurring in FL patients in ZUMA-5

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 64 of 155

==> picture [452 x 216] intentionally omitted <==

----- Start of picture text -----
N (%) FL patients with two or more FL patients with three or more
lines of prior therapy SAS (n = lines of prior therapy SAS (n =
124) 78)
Any Grade Grade 5 Any Grade Grade 5
grade ≥ 3 grade ≥ 3
Agitation '''''' ''''''' '''' '''''''' '''' '''''' ''' '''''' ''' ''''''' ''' '''''''
Disturbance in '''' ''''''' ''' ''''''' ''' ''''''' ''' ''''''' ''' ''''''' ''' ''''''
attention
Dysarthria ''' ''''''' ''' '''''' ''' ''''''' ''' '''''' '''' '''''' ''' '''''''
Paraesthesia '''' ''''''' '''' '''''' ''' ''''''' '''' ''''''' '''' ''''''' '''' ''''''
Delirium '''' '''''' '''' ''''''' '''' ''''''' ''' ''''''' ''' '''''' ''' '''''''
Hallucination ''' ''''''' '''' ''''''' ''' ''''''' '''' '''''' ''' ''''''' ''' '''''''
Key: FL, follicular lymphoma; SAS, safety analysis set.
Source: ZUMA-5 CSR 18-Month Addendum. [35]
----- End of picture text -----

Cytopenia

Table 21 presents the incidence of cytopenia following treatment with axi-cel. Cytopenias were consistent with the known toxicities of the conditioning regimen of cyclophosphamide and fludarabine. Grade ≥ 3 neutropenia, thrombocytopenia and anaemia occurred in ''''''%, ''''''% and ''''''% of FL patients, respectively. A similar cytopenia profile was observed in FL patients with at least three prior lines of therapy in the SAS (Table 21). Prolonged (duration >30 days) Grade ≥ 3 neutropenia, thrombocytopenia and anaemia occurred in ''''''%, '''''''% and '''% of FL patients, respectively. For FL patients whose events had resolved, the mean (standard deviation) and median (range) times to onset of cytopenias were '''''''''' ('''''''''''') and ''''''' (''''''''') days after axi-cel infusion. The median duration of cytopenias were '''''''''' (range: ''''''''''''''') days. Similar findings were observed in FL patients with at least three prior lines of therapy in the SAS.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 65 of 155

Table 21: Incidence of cytopenia in FL patients in ZUMA-5

B.2.10.4.3. Infections

Normal B-cells can be eliminated by CAR T-cell therapy alongside malignant B-cells, leading to B-cell aplasia that can put patients at increased risk of infection and hypogammaglobulinaemia (immunoglobulin <4g/L).

Infections were experienced by ''''' (''''''%) of the FL patients, of which ''''' (''''''%) had worst Grade 3 infections, and ''''''''' ('''%) had worst Grade 4 infection.[35] '''''''' '''''''''''''''' ''''''''' '''' ''''''''''''' '''''''''''''''' ''' ''''''''''''''''''''' The most common infections were pneumonia ('''''' patients [''''''%]), upper respiratory tract infection (''''''' patients, [''''''%]), and oral candidiasis, sinusitis and urinary tract infection ('''''''''' patients each, ['''%]). Worst Grade 3 events included pneumonia ('''''''''''''' patients [''''%]) and urinary tract infection

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 66 of 155

(''''''''''''' patients, ['''%]). The single worst Grade 4 event was sepsis (''''''''' patient,

['''%]). ''''''' ''''''''''''''' ''''' ''''''''''''''''''''''''' ''''''''''''' '''''''''''''''''''''' '''''' ''''' ''''''' '''''''''''''''''''''' '''''''''''''''''''' ''''''''''

'''''''''''''' '''''''''''' '''' ''''''''''''''' ''''''''''''''''''' ''''''''''''''' ''''' ''''''''''''''''''''''' '''''''''' '''''''''''''''''''''' ''''' '''''' ''''''''''''''''''' '''''''''

''''' ''''''''''' '''''''''''' '''''''''''' '''''''''''' ''''' ''''''''''''''''' ''''' ''''''' '''''''''''''

B.2.10.5. Concomitant medication

Among FL patients, '''''' patients (''''''%) were treated with corticosteroids (with or without tocilizumab), '''''' (''''''%) were treated with tocilizumab (with or without corticosteroids) and '''''' patients (''''''%) were treated with corticosteroids and tocilizumab.[35] '''''''''''''''''' patients ('''''%) were treated with vasopressors and '''''' patients ('''''''%) were treated with immunoglobulins.

B.2.10.6. Safety overview

The safety profile observed in ZUMA-5 is like that observed with axi-cel in diffuse large B-cell lymphoma (DLBCL) and other CAR T-cell therapies, for which there are established risk management guidelines.

Since the approved access of axi-cel and other CAR T-cell therapies through the Cancer Drugs Fund (CDF) in NHS England, clinicians are increasingly comfortable with toxicity management for this CD19-directed CAR-T therapy class. Real-world data for patients with high-grade lymphoma treated with CD19 CAR T-cell therapy in NHS England showed lower rates of Grade ≥ 3 CRS and Grade ≥ 3 neurological events, with increased use of tocilizumab and steroid use, than reported across the pivotal clinical trials of tisagenlecleucel-T (JULIET) and axi-cel in this indication (ZUMA-1).[40, 41] Of note, in ZUMA-5 '''''' '''''''''''''''''' had low-grade neurological events with an onset >80 days after the axi-cel infusion. However, UK clinical experts confirmed that they did not expect this to have much impact on patient management nor was it of particular concern in terms of any detriment this could have on the quality of life of patients specifically within the scope of this appraisal.[1]

Further real-world data from King’s College Hospital and, as evidenced through wider national UK clinical practice, has showed that most patients treated with CD19 CAR T-cell therapy in practice do not appear to have an increased risk of early or late infections, despite CRS and administration of tocilizumab and steroids.[42] Immunoglobulin levels also remained above 4g/L in most patients, suggesting that

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 67 of 155

the humoral immune system can be restored despite initial B-cell aplasia, which was significant both pre and post-CAR T-cell therapy. Only three patients (6%) required regular immunoglobulin replacement post-CAR T-cell therapy; two for recurrent chest infections. This is also consistent with European guideline recommendations around the use of intravenous immunoglobulins following CAR T-cell therapy only for those patients with ongoing/recurrent infections, to which this represents a small minority of patients in clinical practice.

As recommended in the SmPC for axi-cel (Appendix C), patients should be monitored for the first 10 days following infusion for signs and symptoms of potential CRS, neurological events and other toxicities. After the first 10 days, the patient is to be monitored at the physician’s discretion, but patients should remain within proximity of a qualified clinical facility for at least 4 weeks following infusion. At least one dose of tocilizumab in the event of CRS and emergency equipment must be available prior to axi-cel infusion, and the treatment centre must have access to an additional dose of tocilizumab within 8 hours of each previous dose.[43]

Blood counts should be monitored after axi-cel infusion and patients should also be monitored for signs and symptoms of infection before, during and after axi-cel infusion (and treated appropriately). Prophylactic antimicrobials should be administered according to standard institutional guidelines. Immunoglobulin levels should also be monitored after treatment with axi-cel and managed using infection precautions, antibiotic prophylaxis and immunoglobulin replacement.

B.2.11. Ongoing studies

The ZUMA-5 study is ongoing; data from the follow-up Analysis 2 (that is, 24-month data cut) is expected to become available in December 2021.

B.2.12. Innovation

Axi-cel is a personalised, transformative medicine in which the patients’ own T-cells are engineered ex vivo to target and kill cancer cells, upon return to the patient via a single infusion.

Axi-cel was the first of the breakthrough class of CAR T-cell therapies to receive European Medicines Agency and US Food and Drug Administration approval; it now

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 68 of 155

represents a breakthrough treatment for patients with r/r FL, providing a novel treatment option in later-line settings where there is currently no established standard of care. This was also recognised by NICE, which noted that there is no established clinical care in this setting when issuing the final scope for this appraisal following amendments made at draft scope (see Table 1).[1] Instead, current care options are limited to recycling earlier-line treatments to which patients may now have a reduced tolerance, as well as reduced effectiveness; or resorting to generic haemato-oncology treatments with little expectation of effect or experimental treatments with no proven benefit.[1]

Despite the more indolent nature of FL compared with previous CAR T-cell indications, it can still be an aggressive disease in a not insignificant proportion (approximately 10–20%[15, 16] ) of patients and prognosis by the time of third relapse is typically poor, with patients unlikely to survive beyond approximately 3 years.[1, 21] Axicel is expected to offer a significant extension to this life expectancy and while survival benefit will be captured in the quality-adjusted life years (QALY), the truly innovative nature of axi-cel and the difference it could make to lives is difficult to truly reflect in such a calculation. FL is an emotionally unsettling disease due to its chronic and progressive nature and the hope that axi-cel could offer is likely to provide positive value independent of expected survival and health status.[44] There are also clear administration benefits of a single treatment infusion with CAR-T versus the recurrent cyclic nature of conventional treatments that are unlikely to be fully captured in the QALY. Indeed, the potential benefits of this “one-time” infusion versus patients returning regularly for recurrent therapy until progression, should not be underestimated for patients and their quality of life (especially so in the 4L+ FL setting), and this has been expressed by practising UK CAR-T physicians as potentially preferable both for patient care and long-term healthcare resourcing/capacity (Gilead data on file).

The innovation of axi-cel has been previously recognised by NHS England and NICE, and its introduction was considered a step change in the DLBCL pathway.[45] A similar step change could be achieved with the introduction of axi-cel to the FL pathway.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 69 of 155

B.2.13. Interpretation of clinical effectiveness and safety evidence

B.2.13.1. Principal findings from the clinical evidence

The vast majority ('''''''''''''''''% depending on analysis set) of FL patients who had received at least three prior therapies and were treated with axi-cel in ZUMA-5 responded to treatment, and the CR rate was ''''''''''''''% (depending on analysis set). ORR and CR rates across analysis sets were significantly greater than pre-specified control rates. At the time of the 18-month analysis data cut-off, ''''''% of responders and ''''''% of complete responders with a minimum follow-up of 18 months, had an ongoing response (IAS data).

Median PFS and OS are yet to be reached in any analysis set. At the time of the 18month analysis data cut-off, ''''''% of FL patients who had received at least three prior therapies in the IAS were alive, and '''''% of patients were alive and progression-free; in the mITT, ''''''''''' were alive and '''''''''''' of patients were alive and progression-free. These proportions increased to ''''''% and ''''''%, respectively, in complete responders in the IAS; in the mITT the proportions increased to '''''% and ''''''%, respectively

ORR and CR rates were markedly higher with axi-cel compared with current care in SCHOLAR-5, with odds ratios of '''''''''''''' and ''''''''''''', respectively. DOR, PFS and OS analyses showed a clinically meaningful and statistically significant reduction in the risk of relapse post response, as well as the risk of disease progression or death, and of death alone of ''''''%, ''''''% and ''''''%, respectively ('''''' '''' '''' ''''''''''''') with axi-cel compared with current care.

B.2.13.2. Strengths and limitations of the evidence base

ZUMA-5 provides prospective clinical trial data demonstrating the clinical benefit of axi-cel treatment for FL patients who have had three prior therapies: a difficult-totreat group of patients who often have an aggressive, chemotherapy-resistant disease course and for whom there is no current standard of care.

Although ZUMA-5 does not provide head-to-head clinical trial data, an attempt was made to address this evidence gap through the utilisation of data from SCHOLAR-5, an external control cohort study which used established, widely recognised methodology. Along with comparative effectiveness analyses, SCHOLAR-5 provides

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 70 of 155

further data supporting a lack of standard of care in r/r FL and a diminished treatment effect with each line of therapy in FL, as previously reported.[11-13, 28]

A limitation of both ZUMA-5 and SCHOLAR-5 is the lack of HRQL data collection. Quality-of-life outcomes as recorded by direct documentation in electronic medical records were to be defined as exploratory endpoints in SCHOLAR-5, but there were insufficient data for analysis as most sites did not collect any patient-reported outcome data. However, '''' '''''''''''''''''' ''''''''''''''''' ''''' '''''''''''''''''''''''''' '''''''''''''''''''''''' ''''''''''''' ''''''''''' '''''''''''''''''''''''' ''''''''' '''''''''''''''''' '''''''''''''''''''''''''''' '''''''''''''''' '''''''''' '''''' '''''''''''''''' ''''''''''''''''''''' ''''''' ''''''''''''''''''''' '''''''''' '''''''''' '''''''''''''''' '''''''''''' '''''' '''''''''''

A further limitation of ZUMA-5 is the immaturity of data with median DOR, PFS or OS yet to be reached within the current 18-month follow-up data. While a positive signal for the longer-term benefits of axi-cel treatment for r/r FL, there is uncertainty around the magnitude of this benefit in clinical practice. In other lymphomas, axi-cel survival curves are starting to plateau, representing the possibility of healthy long-term survival for a proportion of patients.[46, 47] Although this is yet to be demonstrated in an indolent lymphoma setting, it is expected that this long-term survival benefit will translate to the r/r FL setting, based on the unique mechanism and transformative nature of CAR T-cell therapy. This expectation is supported by clinical experts, who noted that it is reasonable to assume a proportion of patients with r/r FL (25%) who are treated with axi-cel may have mortality hazards that behave more in line with the general population after a given time point (see Section B.3).[1] However, in recognition of the current uncertainty, we acknowledge that axi-cel is likely to be a CDF candidate. As a CDF drug, axi-cel would offer clinicians with a much-needed effective treatment option for patients who otherwise face a very uncertain outcome and poor quality of life. It would also allow time for further data collection needed to robustly assess the cost-effectiveness before a final decision on routine reimbursement is made. '''''''''''''''''''''''''''' '''' '''' '''''''''' '''''''''''''''''''''''' '''''''''' ''''''' '''''''''''''''''''''''''''' ''''''''''''''''''''' ''''''''' ''''''' '''''''''''''''' ''''''''''''''''''''''''' '''''''''''''''''' '''''''' '''''''' ''''''''''''''''' ''''''''''''''''' ''''' ''''''''''''''' ''''' ''''

''''''''''''''''' ''''' ''''''''''' '''''''''''''''''

B.2.13.3. Applicability of clinical evidence to practice

The trial population of ZUMA-5 represents a patient group with an aggressive disease course, with a high proportion of the FL patients who had received at least

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 71 of 155

three prior lines of therapy having chemo resistant (''''''% refractory; '''''% doublerefractory) and early relapse disease (''''''% were POD24) (IAS data).[35] Clinical validation confirms that the patients enrolled into ZUMA-5 are generally representative of patients in the UK with r/r FL on the 4L+ treatment pathway who would be considered for CAR-T treatment[1] , despite the lack of UK sites involved in this trial.

The total trial population is broader than the target population for reimbursement, but predefined and post-hoc analyses provide a complete data set for adult patients with r/r FL on the 4L+ treatment pathway; this is the anticipated marketing authorisation indication. There are also several analysis sets explored in ZUMA-5; data are presented for patients treated with axi-cel who have at least 18 months of follow-up in the latest data cut-off (IAS), and for all patients treated with axi-cel (mITT) in preceding clinical sections. mITT data are used in the economic model base case as the more conservative and most representative analysis set for all patients potentially to be treated in clinical practice, as compared to the IAS set.

Comparator data taken from SCHOLAR-5 are believed to offer a high degree of external validity and generalisability, with the blended comparator representative of the lack of standard of care in current practice and treatments received across the primary analyses cohort generally reflected in the UK cohort. Although at face value the inclusion of PI3Ki treatments (12%; Figure 12) suggests a misalignment with clinical practice in England, where such treatment is not routinely reimbursed and is anticipated to come from existing (but no longer) compassionate access to idelalisib in England around the time of receiving its European licence.

Axi-cel is already reimbursed for the treatment of adult patients with r/r DLBCL and primary mediastinal large B-cell lymphoma (PMBCL) after two or more lines of systemic therapy and NHS England service provisions for CAR T-cell therapies are well established, with no or very minimal impact on further site qualification, patient referral or management expected with the introduction of axi-cel in FL at fourth-line setting. For the treatment of adult patients with r/r FL on the 4L+ treatment pathway, axi-cel is expected to fit into the current service provisions and there are no additional or different infrastructure or personnel needs.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 72 of 155

B.2.13.4. Axi-cel as an end-of-life therapy

With increasing relapses and lines of treatment, patients with FL face the stark reality of a progressively shortened lifespan.[11-14] The emotional impact of this devastating reality must be heavily felt by patients who reach ≥ 4 lines of treatment, where there is no established standard of care and patient prognosis is extremely poor.[1, 28] When approved, axi-cel is expected to create a step change for the treatment of adult patients with r/r FL after receiving 4L+ care. Furthermore, at this late-stage of the treatment pathway, Gilead believes axi-cel would be adopted by clinicians as an end-of-life therapy in NHS England.

Based on SCHOLAR-5 data that provide the most up to date and comparable estimates for current care in the axi-cel target population, and clinical expert elicitation, the current life expectancy of these patients is short, at approximately 3 years.[21] While Gilead acknowledges that this is slightly longer than the NICE criteria of a short life expectancy, which is “normally less than 24 months”[48] , Gilead believes that in real terms, patients with r/r FL receiving ≥ 4 lines of treatment have a strikingly short life expectancy, which is approximately 3 years. Furthermore, axi-cel is expected to extend life expectancy by at least the requisite 3 months, based on first quartile estimates and economic model extrapolations. Table 22 summarises the data for axi-cel with regards to the NICE criteria for life-extending treatment at the end of life.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 73 of 155

Table 22: End-of-life criteria

==> picture [29 x 13] intentionally omitted <==

Cost effectiveness

B.3.1. Published cost-effectiveness studies

A systematic review of the literature was conducted to identify economic evaluations/cost-effectiveness analyses of relevance to the decision problem. Searches were initially conducted in May 2020, with an update performed in May 2021 to align with NICE requirements. Full search strategies, inclusion and exclusion criteria and the Preferred Reporting Items for Systemic Reviews and Meta-Analyses flow diagram are provided in Appendix G. To identify relevant literature, the following electronic databases were searched (in addition to conference proceedings and health technology assessment databases):

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 74 of 155

• MEDLINE[®] In-Process

• Embase[®] and MEDLINE

• EconLit

• NHS Economic Evaluation Database

Details of the published cost-effectiveness studies identified in the systemic literature review (SLR) as relevant to this submission are provided in Appendix G.

As shown in Table 23, three prior NICE single technology appraisals in r/r FL published within the last 10 years were identified as relevant to this appraisal. These were TA604 (idelalisib), TA627 (lenalidomide with rituximab) and TA629 (obinutuzumab with bendamustine [TA472 CDF review]).[33, 49-51] Throughout this submission, insights were drawn from these appraisals in r/r FL.

In addition to the prior NICE appraisals in r/r FL, due to the unique mechanism of action and innovative nature of CAR T-cell therapy, insights were drawn from the mock appraisal of regenerative therapies and cell therapy products (such as CAR T- cell therapies) published by Hettle et al. (2017)[52] , as well as the three completed NICE single technology appraisals for CAR T-cell therapies in advanced, previously treated lymphoma indications. These appraisals include TA559 (axi-cel in DLBCL and PMBCL), TA567 (tisagenlecleucel-T in DLBCL) and TA677 (KTE-X19 in r/r mantle cell lymphoma [MCL]).[45, 53, 54]

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 75 of 155

Table 23: Summary list of published NICE appraisals in r/r FL

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 76 of 155

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 77 of 155

B.3.2. Economic analysis

B.3.2.1. Patient population

The patient population considered in the analysis is '''''''''''''''' ''''''''' '''''' '''''' '''''''''''' ''''''''''''' ''''' '''''''''''' '''''''''' ''''' '''''''''''''''''''''' '''''''''''''''' , in line with the anticipated marketing authorisation for axi-cel. As discussed in Section B.1.1, the wording in the final NICE appraisal scope is ‘adults with r/r non-Hodgkin lymphoma’. As such, this submission focusses on a subtype of iNHL, those with r/r FL.

As described in Section B.2.3, ZUMA-5 evaluated the safety and efficacy of axi-cel for the treatment of patients with r/r iNHL. Specifically, ZUMA-5 included patients with r/r FL (Grades 1–3a) or MZL (nodal or extranidal) who received two or more prior lines of therapy, including an anti-CD20 monoclonal antibody combined with an alkylating agent.

The economic analysis, which evaluates axi-cel in a 4L+ position, therefore considers a subgroup of the FL population eligible for the ZUMA-5 study; a group in which there is a high unmet need for safe and efficacious treatment.

Two clinical experts were interviewed on 10 September 2021 to ensure the economic modelling approach in this submission was consistent with clinical expectations in NHS England practice. As discussed in Section B.2.3.1, these consultants agreed that the baseline characteristics of patients enrolled in ZUMA-5 were generally representative of patients who would be considered for axi-cel within its anticipated marketing authorisation in clinical practice. However, it was noted that patients enrolled in ZUMA-5 could potentially be considered younger and fitter than the typical 4L+ FL patient in the UK.[1] Although this was the case, it was also noted that the proportion of 4L+ FL patients enrolled in ZUMA-5 who were classified as POD24 was higher than that expected in clinical practice[1] , suggesting that the patients within the trial may be higher risk than expected in clinical practice (which is not uncommon in the clinical trial setting).

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 78 of 155

B.3.2.2. Model structure

A de novo cost-effectiveness model was developed in Microsoft Excel[®] . The proportion of patients in each health state at a given time point was determined using a partitioned survival analysis modelling approach.

Partitioned survival models in oncology are typically comprised of three health states (pre-progression, progressed disease and death), with the proportion of patients in each health state over time derived directly from the independently modelled OS and PFS projections. This structure reflects the natural history of disease (progressive) and separates the pre- and post-progression states, which in turn helps to capture potential differences in costs and HRQL.

As shown in Table 23, all prior NICE appraisals in r/r FL published within the last 10 years used a progression-based three-state partitioned survival approach. In TA472 (as reported in TA627), the Evidence Review Group considered the manufacturers state transition model unreliable and consequently a partitioned survival analysis was submitted.[51] In TA604, multiple comparisons were submitted with alternative structures; however, the Committee preferred the comparison using a partitioned survival analysis approach.[33] In TA627, a partitioned survival analysis approach was selected.[51] In these appraisals, the alive health states were further divided into onand off-treatment periods; however, as axi-cel is administered as a one-time infusion in the first model cycle, health states were not explicitly divided by treatment status in this analysis.

Furthermore, the study published by Hettle et al (2017) and the completed single technology appraisals for CAR T-cell therapies in advanced, previously treated lymphoma indications, used a progression-based three-state partitioned survival model.[45, 52-54] Notably, in TA567, an initial decision tree was used to account for patients who received leukapheresis but did not go on to have CAR T-cell therapy.[53] In this analysis, consistent with TA559 and TA677, the costs for patients who underwent leukapheresis but did not go on to receive axi-cel infusion in ZUMA-5 were accounted for using cost multipliers (described further in Section B.3.5.1.1).[45, 54]

Figure 16 presents the model structure diagram. All patients begin in the preprogressed health state. In each model cycle, patients may remain progression-free,

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 79 of 155

their disease may progress, or they may enter the death state. Once a patient has experienced disease progression, they may remain in the current state or they may enter the death state. Death is an absorbing health state. Figure 17 graphically represents how OS and PFS extrapolations are used to determine the proportion of patients in each health state at time ‘T’ in a partitioned survival structure.

In TA559 and TA567, patients with DLBCL who remained pre-progression for 2 years were captured as ‘long-term survivors’.[45, 53] It was assumed that long-term survivors, who had a heightened risk of death compared with the age-equivalent general population, did not incur further resource use and experienced improved HRQL. Similarly, in TA677, consistent long-term survivorship assumptions were followed for patients with MCL who remained alive and free of progression at 5 years following CAR T-cell therapy.[54] Here, a similar approach to capturing long-term survivors is undertaken, using the later time point of 5 years in line with TA677.

Figure 16: Model structure diagram

==> picture [455 x 119] intentionally omitted <==

==> picture [455 x 118] intentionally omitted <==

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 80 of 155

Figure 17: Partitioned survival analysis; health state occupancy example

==> picture [484 x 97] intentionally omitted <==

==> picture [484 x 97] intentionally omitted <==

==> picture [484 x 97] intentionally omitted <==

Key : OS, overall survival; PD, progressed disease; PF, progression free; PFS, progression-free survival; T, chosen time.

Note: superscript[T] denotes ‘at time T’.

B.3.2.2.1. Model settings

In line with the NICE reference case, the analysis perspective is that of the NHS and the Personal Social Services (PSS) in England for costs and direct health effects for individual patients for outcomes.[48]

As stated in the NICE reference case[48] , the time horizon for estimating clinical and cost effectiveness should be sufficiently long to reflect all important differences in costs or outcomes between the technologies being compared; a lifetime horizon is therefore used in the base case. In the analysis, a time horizon of 40 years constitutes a lifetime, based on the starting age of 4L+ patients in ZUMA-5 ('''''' years) and the modelled OS extrapolations (discussed in Section B.3.3.3).

The model incorporates a cycle length of 28 days, deemed sufficient to capture relevant changes in health. A half-cycle correction is applied to costs and outcomes, except for costs that are known or assumed to occur at the start of the model or at the start of each cycle. Costs applied at the start of the model or cycle include axi-cel

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 81 of 155

treatment-related costs, comparator costs (acquisition and administration) and AE costs (except for ongoing intravenous immunoglobulin [IVIG] therapy).

Costs and health outcomes are discounted each cycle at an annual rate of 3.5%, as per the NICE reference case.[48] Given the unique mechanism of CAR T-cell therapies and promising implications of sustained disease clearance and potential long-term survivorship, non-reference case discount rates of 1.5% are explored in the scenario analysis.

B.3.2.3. Intervention technology and comparators

B.3.2.3.1. Intervention

The intervention considered in the analysis is axi-cel. Axi-cel is incorporated into the economic evaluation according to its anticipated marketing authorisation and in line with the decision problem described in Section B.1.1.

As described in Section B.1.2, axi-cel is an autologous anti-CD19 CAR T-cell therapy that recognises and eliminates all CD19 expressing target cells, including B-cell malignancies and normal B-cells. Axi-cel was the first in a breakthrough class of CAR T-cell therapies. The therapy consists of genetically engineered T-cells which recognise CD19-expressing cancer cells, leading to an immune response which results in the direct killing of the target cancer calls.

Axi-cel was initially indicated by the European Medicines Agency for the treatment of adult patients with r/r DLBCL and PMBCL after two or more lines of systemic therapy.[43] In England, axi-cel is recommended for use via the CDF as an option for r/r DLBCL or PMBCL.[45] As discussed in Table 2, application for a marketing

authorisation extension of axi-cel to '''''''''''''' ''''''''''''''''''' '''''''''' '''''''''''''''''''''' ''''' '''''''''''''''''''''' ''''''''

'''''''''''''''''' '''''''''''''''''''''''''' ''''''''' '''''''''' ''''''''''''' ''''' ''''''''''' ''''''''''' ''''' '''''''''''''''''''''' ''''''''''''''''''''' has been

submitted.

Axi-cel is a one-time immunotherapy of 2 x 10[6 ] CAR T-cells/kg of body weight used autologously and administered intravenously. Patients must be treated with a lymphodepleting conditioning chemotherapy consisting of cyclophosphamide 500 mg/m[2] intravenously and 30 mg/m[2] fludarabine intravenously for 3 days before infusion of axi-cel is administered.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 82 of 155

B.3.2.3.2. Comparator

As discussed in Section B.1.3, despite the clinical pathway for FL being well established at early treatment lines, there is no established standard of care for adult patients with 4L+ r/r FL. At this stage in the pathway, treatment decisions are made on a case-by-case basis considering factors such as patient fitness, treatment goals, response and durability of response to prior therapy.

Given that there is no true standard of care for patients following treatment with lenalidomide plus rituximab (recommended by NICE in TA627) or obinutuzumab with bendamustine (recommended by NICE in TA629)[50, 51] , the comparator considered in the economic model is ‘current 4L+ care’, which consists of a basket of treatments.

As described in Table 1, rituximab monotherapy and best supportive care were included in the final scope issued by NICE. However, rituximab monotherapy is only recommended as an option for the treatment of r/r FL when all alternative treatment options have been exhausted (that is, if there is resistance to or intolerance of chemotherapy). As with best supportive care, if rituximab monotherapy was being considered for use in patients with 4L+ r/r FL, it would be reserved for patients not fit enough to receive intensive active treatment, thereby constituting a cohort of patients widely considered not suitable or appropriate for consideration of CAR T-cell therapy. As such, rituximab monotherapy and best supportive care are not considered within the blend of treatments comprising current 4L+ care.

Of the other comparators listed in the final scope, it is expected that obinutuzumab with bendamustine (used for patients who are rituximab-relapsed) and lenalidomide with rituximab would typically be used earlier in the treatment pathway than the 4L+ setting; a sentiment endorsed by the UK Clinical Experts interviewed in September 2021.[1] However, these treatments have been considered as part of the blend which comprises current 4L+ care given the lack of alternative or established treatment approaches in 4L+ care and, thereby, the necessity to potentially resort to their use in these later lines of therapy.

The source of comparative efficacy data for the current 4L+ care arm of the model is described in Section B.3.3.1.2. The basket cost of treatment in the current 4L+ care arm of the model is calculated as a weighted average; derived from the distribution

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 83 of 155

of treatment regimens that comprise the blend (described in Section B.3.5.1.2). To assess the impact on comparator costs and consequently cost-effectiveness results, the treatments comprising the blend and corresponding distributions are varied in scenario analysis.

B.3.3. Clinical parameters and variables

The clinical parameters used to inform the axi-cel and current 4L+ care arms in the economic model, and their respective sources, are summarised in Table 24 and discussed in more detail throughout this section and, in the case of AE rates, Section B.3.4.3.

As described in Section B.2.3, the primary endpoint of the ZUMA-5 trial was the ORR (defined as the incidence of patients achieving CR or PR as determined by central assessment). In line with prior appraisals of CAR T-cell therapies in advanced lymphoma indications, and in line with the model structure (Section B.3.2.2), ORR is not directly reflected in the cost-effectiveness analysis, but is indirectly captured (with respect to long-term survivorship for patients who respond well to CAR T-cell therapy).

Table 24: Data sources of clinical parameters used in the model

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 84 of 155

B.3.3.1. Efficacy overview

B.3.3.1.1. Axi-cel

Axi-cel PFS and OS estimates are based on patient-level data collected in the Phase 2 ZUMA-5 study. Specifically, all survival analyses for axi-cel were conducted using the mITT population in the cohort of 4L+ FL patients of ZUMA-5 (N = ''''''; all subjects with 4L+ FL treated with axi-cel at a target dose of 2 x 10[6] CAR T-cells/kg).

Latest available (September 2020 database lock) axi-cel PFS and OS Kaplan–Meier data are presented in Section B.2.9.1 (Figure 14and Figure 15, respectively). Despite a considerable follow-up period, data for axi-cel are still relatively immature, with median survival not yet having been reached for any endpoint. An additional ZUMA-5 data cut became available at the time of submission of this dossier, with an additional 6 months of follow-up (i.e. 24 months). It was not possible to include this data prior to submission; however, it is acknowledged that this data may provide a more informative data source for clinical inputs.

As discussed in Section B.2.13, although immaturity of the ZUMA-5 endpoints represents a positive signal for the longer-term benefits of axi-cel treatment for patients with r/r FL, there is uncertainty around the magnitude of this benefit in clinical practice. In other lymphomas, axi-cel survival curves are starting to plateau, representing the possibility of healthy long-term survival for a proportion of patients.[46, 47] Although this is yet to be shown in an indolent lymphoma setting, it is expected that this long-term survival benefit will translate to the r/r FL setting, based on the unique mechanism of action and transformative nature of CAR T-cell therapy.

This expectation of healthy long-term survival for a proportion of patients treated with CAR T-cell therapy was confirmed in interviews with clinical experts from the UK, whereby it was noted that it is reasonable to assume that a proportion of patients with r/r FL who are treated with axi-cel may have mortality hazards that behave more in line with the general population after a given time point.[1] However, as described above, the follow-up duration of the currently available clinical data is not sufficient to fully substantiate this effect in this more indolent lymphoma. As described in NICE Decision Support Unit (DSU) Technical Support Document (TSD) 21, when extrapolating clinical trial data to estimate lifetime outcomes, standard parametric

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 85 of 155

models are limited with respect to the type of hazards they can represent; more flexible models may be required where mortality hazards are expected to have complex shapes in the longer term.[57]

As described in NICE TSD 21, flexible parametric models using restricted cubic splines can enable hazard and survival functions with complex shapes to be accurately modelled. However, these models will generally provide extremely good fits within the range of the observed data, given a sufficient number of knots have been used, but this does not mean that their extrapolations will be reliable.[57]

In the prior NICE appraisals of CAR T-cell therapies in advanced lymphoma indications, mixture cure models were used to better reflect long-term survival expectations for patients following CAR-T infusion.[45, 53, 54] However, as reported in NICE TA567, robust estimation of mixture cure models require data from studies with sufficient follow-up times that exceed the anticipated time point of cure.

In this case, flexible parametric models or formal mixture cure modelling were not considered plausible due to the immaturity of the ZUMA-5 data at the latest available database lock (September 2020) at the time of submission. Nevertheless, standard parametric models alone cannot capture the expected benefit of axi-cel in the longer term.

To reflect the clinical opinion on long-term outcomes, we take an approach whereby OS and PFS are informed using standard parametric survival extrapolations until a specified time point. After which, for a proportion of patients who are considered long-term survivors, mortality is informed by the age-matched general population adjusted with an SMR. Those who are not captured as long-term survivors continue to follow the hazard of the parametric model. Notably this approach was requested by the Evidence Review Group in prior NICE appraisal for a CAR T-cell therapy in MCL.[54] This approach is also similar to that undertaken in prior NICE appraisals of treatments for acute lymphoblastic leukaemia, whereby it was assumed that patients who survived 4 to 5 years were cured (as reported in TA554).

In line with clinical opinion elicited during the interview to validate the economic modelling approach in this appraisal, it is assumed that 25% of patients treated with axi-cel are captured as long-term survivors in the base case survival estimates. This

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 86 of 155

value, which is used to estimate respective ‘cure fractions’ for the PFS and OS based on the proportion of patients progression-free/alive at the selected time point (discussed below), is varied in sensitivity and scenario analysis.

As discussed in Section B.3.2.2, in line with the prior NICE appraisal for a CAR T-cell therapy in MCL and in line with clinical validation interviews, a time point of 5 years was selected to explicitly capture long-term survivors. Alternative long-term survivorship timepoints of 2, 7 and 10 years are tested in scenario analysis. The 2- year timepoint is explored in line with TA559 and TA567 (patients with DLBCL), while 7 and 10 years are explored due to the more indolent disease evaluated in this appraisal.

It is not assumed that the mortality rate of long-term survivors would return to that of the age-matched general population, a heightened risk of death for long-term survivors is therefore captured using an SMR. In line with the prior NICE appraisal in DLBCL, in the absence of FL-specific data identified in the literature, an SMR of 1.09 was adopted. Alternative SMRs for long-term survivors of 1 (equivalent to general population) and 1.20 (based on feedback from clinical experts in the UK) are tested in scenario analysis.

It is acknowledged that this approach is yet unproven and suffers from limitations inherent to the use of data with a non-optimal follow-up duration. Further study of CAR T-cell therapy in FL over time will provide more accurate and robust estimates; highlighting that axi-cel is likely to be a candidate for the CDF, whereby additional real-world and ZUMA-5 data collection would reduce some uncertainty in the longterm survival extrapolations. For the purpose of this submission, to reflect this uncertainty, scenario analyses are conducted around the proportion of patients who were captured as long-term survivors, the time point selected and the SMR compared with general population.

B.3.3.1.2. Current 4L+ care

As discussed in Section B.3.2.3.2, as there is no established clinical management in 4L+ r/r FL, the comparator arm in the cost-effectiveness model (current 4L+ care) comprises a blend of multiple therapies. Furthermore, as ZUMA-5 is a single-arm

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 87 of 155

trial, comparative efficacy data for current 4L+ care were sourced from an alternative data source.

As discussed in Section B.2.9, three published studies were identified as potentially relevant, however all were considered highly limited in their suitability as a source of comparative efficacy data and therefore no matching comparison for treatment-effect modification was conducted for these studies.

B.3.3.1.2.1. SCHOLAR-5

As described in Section B.2.9.1, SCHOLAR-5 is a retrospective, observational, multicentre and patient-level database with two primary objectives: characterising the natural history of iNHL and building an external comparator group to provide comparative data with the ZUMA-5 trial. Details of the SCHOLAR-5 study (including the methods, natural history, treatment patterns and comparative analysis) are described in Section B.2.9.1.

For the comparative analysis, ''''' patients from SCHOLAR-5 with 4L+ FL were available for analysis against the ZUMA-5 mITT population. Importantly for this appraisal, SCHOLAR-5 patient-level data were readily available to be analysed as a comparative data source to ZUMA-5, enabling the estimation of a relative treatment effect.

The SCHOLAR-5 cohort was used as an external control for the ZUMA-5 clinical trial, after alignment to the ZUMA-5 population using propensity score methods (SMR weighting) to adjust for known confounders. This enabled comparisons to be drawn by balancing patient characteristics between both data sources. Full details of the propensity scoring methods and outcomes are provided in the SCHOLAR-5 technical report provided in the reference pack.[21] Clinical validation was sought regarding the patient characteristics that impact prognosis and thus should be balanced between the two data sources (Table 14).

B.3.3.1.3. Comparative efficacy

Summary Kaplan–Meier plots are presented for PFS and OS for axi-cel (ZUMA-5) and current 4L+ care (propensity score weighted SCHOLAR-5) in Section B.2.9.1.3.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 88 of 155

Data from SCHOLAR-5 are relatively mature in comparison to ZUMA-5, due both to a longer available follow-up period and inferior outcomes leading to higher numbers of events taking place earlier in the follow-up. Notably, the PFS data from SCHOLAR-5 reach the x-axis (Figure 14). Note that as the DELTA study did not collect progression data for subsequent lines of therapy, which was used in the SCHOLAR-5 analysis, the number at risk post-weighting for the PFS endpoint is lower (n = '''''') than that for OS (n = ''''''').

B.3.3.1.4. Extrapolation

As the latest PFS and OS Kaplan–Meier data are incomplete (that is, there were patients still alive or progression-free and alive at database lock), extrapolation was required to estimate lifetime PFS and OS.

A range of standard parametric survival models were fitted to both the axi-cel and current 4L+ care PFS and OS data, in line with NICE DSU TSD 14.[58] In addition to the standard models specified in TSD 14, the gamma model was also included (which is equivalent to the generalised gamma model with a shape parameter of 1).[58] The parametric models explored were:

• Exponential

• Gamma

• Generalised gamma

• Gompertz

• Log-logistic

• Log-normal

• Weibull

As per NICE DSU TSD 14, it is generally considered unnecessary to rely on the proportional hazards assumption when patient-level data are available.[58] Furthermore, given the unique mechanism of action for axi-cel compared with current 4L+ care, it is considered unreasonable to assume proportional hazards between treatments. Curves were therefore fitted separately to each treatment arm.

As per NICE TSD 14[58] , goodness-of-fit and plausibility of extrapolation were assessed based on:

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 89 of 155

• Visual inspection of fitted curves

• Comparisons of Akaike information criterion (AIC) and Bayesian information criterion (BIC) statistics

• The plausibility of long-term extrapolation based on clinical expert opinion and expected survival from other data sources

As discussed in Section B.3.3.1, due to the immaturity of the ZUMA-5 data at the latest available data cut, formal mixture cure models were not considered a reasonable approach and therefore not included in the economic analysis. However, standard parametric curves were combined with SMR-adjusted population mortality using a 5-year landmark approach to capture a proportion of long-term survivors.

The following sections detail the approaches to modelling PFS and OS for the axi-cel and current 4L+ care treatment arms.

B.3.3.2. Progression-free survival

A summary of the base case approach to modelling PFS is provided in Table 25.

Table 25: Summary of base case approach used to model PFS, by treatment

arm

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 90 of 155

B.3.3.2.1. Axi-cel

Fitted models are graphically represented alongside ZUMA-5 PFS Kaplan–Meier data in Figure 18. Corresponding smoothed hazard plots are presented in Appendix O. AIC and BIC statistics and landmark estimates are presented in Table 26.

Figure 18: Axi-cel progression-free survival: standard parametric curves

==> picture [484 x 104] intentionally omitted <==

==> picture [484 x 103] intentionally omitted <==

==> picture [484 x 103] intentionally omitted <==

Key: KM, Kaplan–Meier. Note: standard parametric curves presented have not been corrected for background mortality.

Table 26: Axi-cel, PFS: standard parametric curve AIC and BIC statistics and landmark survival estimates

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 91 of 155

==> picture [452 x 190] intentionally omitted <==

----- Start of picture text -----
'''''''''''''''''''''''''''' ''''''''''' ''''''''''' ''''''''''''' ''''''''''''''' ''''''''''''''' '''''''''''''' '''''''''''''''' ''''''''''''''''
''''''''''''''''''' ''''''''''''' ''''''''''''' ''''''''''''' '''''''''''' '''''''''''''' '''''''''''''''' ''''''''''''''' '''''''''''''''
''''''''''''''''''''''''''' '''''''''''''' ''''''''''''' '''''''''''' ''''''''''''''' '''''''''''''''' '''''''''''''' '''''''''''''' ''''''''''''''
'''''''''''''''''
'''''''''''''''''''''''' ''''''''''''' '''''''''''''' '''''''''''''''' '''''''''''' ''''''''''''''' ''''''''''''''' ''''''''''''''' '''''''''''''''
''''''''''''''''''''''''''' '''''''''''''' '''''''''''''' ''''''''''''''' ''''''''''''''' '''''''''''''''' '''''''''''''''' '''''''''''''''' ''''''''''''''''
'''''''''''''''''''''''''''' '''''''''''''' ''''''''''''''' ''''''''''''''' '''''''''''' '''''''''''''''' ''''''''''''''' '''''''''''''''' ''''''''''''''''
''''''''''''''''''' '''''''''''''' ''''''''''''' ''''''''''''' ''''''''''''' ''''''''''''''' '''''''''''''''' '''''''''''''' ''''''''''''''
Key: AIC, Akaike information criterion; BIC, Bayesian information criterion; PFS, progression-free
survival
Notes: mean and median values are provided in units of months. Best statistically fitting model in
bold. Selected model in italics. Projected PFS values here are not accounting for background
mortality correction.
----- End of picture text -----

Based on visual inspection, all curves provide a similar fit to the Kaplan–Meier data. Similarly, the goodness-of-fit statistics are within 5 AIC across all models, and 5 BIC across all models apart from the generalised gamma, suggesting a very similar statistical fit to the observed portion of the data between parametric models.

The exponential, gamma and Weibull curves, which have the most conservative long-term extrapolations, predict similar outcomes, with ''''''''''''''''''' of patients alive and free of progression at 5 years following treatment with axi-cel. In clinical validation interviews, it was suggested that given the uncertainty of the long-term effects of axicel and immaturity of the data, the Weibull curve should form the model base case.

As previously discussed, in the interview with clinical experts it was noted that it is reasonable to assume that a proportion of patients with r/r FL who are treated with CAR T-cell therapy (25%) may have mortality hazards that behave more in line with the general population after 5 years.

The selected parametric curve (Weibull) was used to model PFS in the axi-cel arm for all patients up to Year 5. Beyond this point, the SMR-adjusted general population mortality hazard was applied to 25% of patients treated with axi-cel (which equates to approximately ''''''''''''''''''''''''''''''''' of the ''''''''''' of patients in PFS at Year 5). The remaining patients were assumed to follow the mortality hazard of the selected Weibull curve.

The resulting selected axi-cel PFS curve is presented in Figure 19.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 92 of 155

Figure 19: Axi-cel PFS: selected curve (25% long-term survivorship)

==> picture [484 x 108] intentionally omitted <==

==> picture [484 x 109] intentionally omitted <==

==> picture [484 x 108] intentionally omitted <==

Key: KM, Kaplan–Meier; PFS, progression-free survival.

In scenario analysis, the impact of capturing all patients who are alive and free of progression at 5 years as long-term survivors is tested (that is, all of the estimated ''''''''''' of patients in PFS are assumed to be long-term survivors and follow the SMRadjusted general population mortality hazard).

B.3.3.2.2. Current 4L+ care

Fitted models are graphically represented alongside SCHOLAR-5 PFS Kaplan– Meier data in Figure 20. Corresponding smoothed hazard plots provided in Appendix O. AIC and BIC statistics and landmark estimates are presented in Table 27.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 93 of 155

Figure 20: Current 4L+ care PFS: standard parametric curves

==> picture [484 x 101] intentionally omitted <==

==> picture [484 x 102] intentionally omitted <==

==> picture [484 x 101] intentionally omitted <==

Key: 4L+, fourth line or later; KM, Kaplan–Meier; PFS, progression-free survival. Note: standard parametric curves presented have not been corrected for background mortality.

Table 27: Current 4L+ care: PFS: standard parametric curve AIC and BIC

statistics and landmark survival estimates

==> picture [452 x 255] intentionally omitted <==

----- Start of picture text -----
Proportion pre-progression at each
landmark value
Mean Median
Model AIC BIC
PFS PFS
6 5
1 year 2 years
months years
'''''''''''''''''''''''''''' ''''''''''' ''''''''''' ''''''''''' ''''''''''' ''''''''''''''' '''''''''''''''' ''''''''''' '''''''''''''
'''''''''''''''''''' ''''''''''''' '''''''''''''' '''''''''' '''''''''' '''''''''''''' '''''''''''''' ''''''''''''' '''''''''''
''''''''''''''''''''''''''''
' ''''''''''''''''''' '''''''''''''' '''''''''''''' '''''''''' '''''''''' '''''''''''''' '''''''''''''' '''''''''''' ''''''''''''
'''''''''''''''''''''''''' ''''''''''''' ''''''''''''' '''''''''' '''''''''' '''''''''''''' ''''''''''''''''' '''''''''''' ''''''''''''
''''''''''''''''''''''''' '''''''''''' ''''''''''''' ''''''''''''' '''''''''''' ''''''''''''''' '''''''''''''' '''''''''''''' '''''''''''''
'''''''''''''''''''''''''''' '''''''''''' '''''''''''' '''''''''''' '''''''''''' '''''''''''''' '''''''''''''' ''''''''''' '''''''''''''
''''''''''''''''''' '''''''''''' ''''''''''''' '''''''''''' ''''''''''' '''''''''''''''' ''''''''''''''''' ''''''''''' ''''''''''''''
Key: 4L+, fourth line or later; AIC, Akaike information criterion; BIC, Bayesian information criterion;
PFS, progression-free survival
Notes: Mean and median values are provided in units of months. Best statistically fitting model in
bold. Selected model in italics. Projected PFS values here are not accounting for background
mortality correction.
----- End of picture text -----

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 94 of 155

Based on the goodness-of-fit statistics and visual interpretation of trends, the exponential model provides the best fit to the Kaplan–Meier data. Goodness-of-fit statistics are within 5 points across models, except for BIC for the generalised gamma and log-logistic models. Given the maturity of the SCHOLAR-5 PFS data, the choice of survival extrapolation does not have a large impact on the long-term PFS estimate, with median PFS ranging from '''''''' ''''' ''''''' ''''''''''''''''', and ''''''''''''' ''''' '''''''''''' of patients remaining alive and free of progression at 5 years across all parametric models.

In the validation interview with clinical experts, it was noted that all current 4L+ care standard parametric PFS curves were clinically plausible.[1] The exponential curve was therefore selected as the model base case based on the goodness-of-fit.

Figure 21 presents the selected axi-cel and current 4L+ PFS curves used in the model base case.

Figure 21: Axi-cel versus current 4L+ care, PFS: selected

==> picture [484 x 109] intentionally omitted <==

==> picture [484 x 108] intentionally omitted <==

==> picture [484 x 109] intentionally omitted <==

Key: 4L+, fourth line or later; KM, Kaplan–Meier; PFS, progression-free survival.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 95 of 155

B.3.3.3. Overall survival

A summary of the base case approach to modelling OS is provided in Table 28.

Table 28: Summary of base case approach used to model OS, by treatment

arm

B.3.3.3.1. Axi-cel

Fitted models are graphically represented alongside ZUMA-5 OS Kaplan–Meier data in Figure 22. Corresponding smoothed hazard plots are presented in Appendix O. AIC and BIC statistics and landmark estimates are presented in Table 29.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 96 of 155

Figure 22: Axi-cel overall survival: standard parametric curves

==> picture [484 x 101] intentionally omitted <==

==> picture [484 x 102] intentionally omitted <==

==> picture [484 x 101] intentionally omitted <==

Key: KM, Kaplan–Meier. Note: standard parametric curves presented have not yet been corrected for background mortality.

Table 29: Axi-cel, OS: standard parametric curve AIC and BIC statistics and

landmark survival estimates

==> picture [452 x 255] intentionally omitted <==

----- Start of picture text -----
Proportion pre-progression at each
landmark value
Mean Median
Model AIC BIC
OS OS
6 1 5
2 years
months year years
'''''''''''''''''''''''''''' ''''''''''' '''''''''' '''''''''''''''' '''''''''''''''' '''''''''''''' ''''''''''''''' '''''''''''''''' ''''''''''''''''
''''''''''''''''' '''''''''''' ''''''''''''''' ''''''''''''''' ''''''''''''''''' ''''''''''''''' ''''''''''''''' '''''''''''''' ''''''''''''''
''''''''''''''''''''''''''''
'' '''''''''''''''''' ''''''''''''' '''''''''''' ''''''''''''''' ''''''''''''''' ''''''''''''''' '''''''''''''' '''''''''''''' '''''''''''''''''
'''''''''''''''''''''''' '''''''''''''' '''''''''''''' '''''''''''' '''''''''''''' '''''''''''''' '''''''''''''' ''''''''''''''''' ''''''''''''''''
''''''''''''''''''''''''' '''''''''''''' '''''''''''' ''''''''''''''' ''''''''''''''''' '''''''''''''''' ''''''''''''''' '''''''''''''''' ''''''''''''''''
''''''''''''''''''''''''' '''''''''''' ''''''''''''''' '''''''''''''''''' '''''''''''''''''' '''''''''''''''' ''''''''''''''' '''''''''''''''' '''''''''''''''
'''''''''''''''' ''''''''''''' '''''''''''' ''''''''''''''' ''''''''''''''' '''''''''''''' '''''''''''''' ''''''''''''''''' '''''''''''''''
Key: AIC, Akaike information criterion; BIC, Bayesian information criterion; PFS, progression-free
survival
Notes: Mean and median values are provided in units of months. Best statistically fitting model in
bold. Selected model in italics. Projected OS values here are not accounting for background
mortality correction.
----- End of picture text -----

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 97 of 155

Based on the goodness-of-fit statistics, the exponential model provides the best fit to the Kaplan–Meier data, although it should be noted the observed OS data at database lock are immature. Furthermore, the AIC and BIC statistics are within 5 points across all models, except for BIC for the generalised gamma and log-normal models. As with PFS, the exponential, gamma and Weibull OS models produce similar longer-term extrapolations, with '''''''''''''' ''''' ''''''''''''''''' of patients alive at 5-years following treatment with axi-cel.

It was suggested in the validation interview with clinical experts that the generalised gamma and Gompertz curves may be ruled out based on the clinical plausibility of the long-term extrapolations. Similarly, the log-normal model can be excluded based on the plausibility of the long-term extrapolation, with ''''''''''' of patients modelled to be alive at 40 years prior to any background mortality adjustment.

Consistent with PFS, the Weibull curve was selected in the base case. The selected parametric curve was used to model OS in the axi-cel arm for all patients up to Year 5. Beyond this point, the SMR-adjusted general population mortality hazard was applied to 25% of patients treated with axi-cel (which equates to approximately ''''''''''''''''''''''' of the ''''''''''' of patients estimated to be alive at Year 5). The remaining patients were assumed to follow the mortality hazard of the selected Weibull curve.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 98 of 155

Figure 23: Axi-cel overall survival: selected curve (25% long-term survivorship)

==> picture [484 x 108] intentionally omitted <==

==> picture [484 x 109] intentionally omitted <==

==> picture [484 x 108] intentionally omitted <==

Key: KM, Kaplan–Meier; OS, overall survival.

As previously discussed, the impact of assuming all patients who are alive and free of progression at 5-years are long-term survivors is tested in scenario analysis (that is, approximately ''''''''' of patients alive at 5 years are assumed to be long-term survivors).

B.3.3.3.2. Current 4L+ care

Fitted models are graphically represented alongside SCHOLAR-5 OS Kaplan–Meier data in Figure 24. Corresponding smoothed hazard plots provided in Appendix O. AIC and BIC statistics and landmark estimates are presented in Table 30.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 99 of 155

Figure 24: Current 4L+ care overall survival: standard parametric curves

==> picture [484 x 102] intentionally omitted <==

==> picture [484 x 101] intentionally omitted <==

==> picture [484 x 101] intentionally omitted <==

Key: KM, Kaplan–Meier. Note: standard parametric curves presented have not been corrected for background mortality.

Table 30: Current 4L+ care: OS: standard parametric curve AIC and BIC statistics and landmark survival estimates

==> picture [452 x 255] intentionally omitted <==

----- Start of picture text -----
Proportion pre-progression at each
landmark value
Mean Media
Model AIC BIC
OS n OS
6 5
1 year 2 years
months years
'''''''''''''''''''''''''' ''''''''''''' ''''''''''''' '''''''''''''' ''''''''''''' ''''''''''''''' '''''''''''''' '''''''''''''''' ''''''''''''''
'''''''''''''''''''' ''''''''''''' ''''''''''''' ''''''''''''' ''''''''''''' ''''''''''''''' '''''''''''''''' '''''''''''''' '''''''''''''''
''''''''''''''''''''''''''''
'' '''''''''''''''''' ''''''''''' ''''''''''' '''''''''''''''' ''''''''''''' ''''''''''''''' '''''''''''''''' ''''''''''''''''' ''''''''''''''''
''''''''''''''''''''''' '''''''''''''' ''''''''''''' '''''''''''''''' ''''''''''''' ''''''''''''''''' '''''''''''''''' ''''''''''''''' '''''''''''''''
'''''''''''''''''''''''''''' '''''''''''''' ''''''''''''' '''''''''''''''''' '''''''''''''' '''''''''''''''' '''''''''''''''' ''''''''''''''' ''''''''''''''
''''''''''''''''''''''''''' ''''''''''''' ''''''''''''' ''''''''''''''' ''''''''''''' '''''''''''''''' '''''''''''''''' '''''''''''''''' ''''''''''''''''
'''''''''''''''''' '''''''''''''' '''''''''''''' '''''''''''' ''''''''''''' ''''''''''''''' '''''''''''''''' ''''''''''''''' '''''''''''''''
Key: AIC, Akaike information criterion; BIC, Bayesian information criterion; PFS, progression-free
survival.
Notes: mean and median values are provided in units of months. Best statistically fitting model in
bold. Selected model in italics. Projected PFS values here are not accounting for background
mortality correction.
----- End of picture text -----

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 100 of 155

Based on the goodness-of-fit statistics and visual interpretation of trends, the generalised gamma model provides the best fit to the Kaplan–Meier data; however both the generalised gamma and Gompertz models may be excluded from contention based on the plausibility of the long-term extrapolation. The generalised gamma and Gompertz models predict '''''''''' ''''''''' '''''''''' of current 4L+ care patients are alive at 40 years (aged '''''''''), respectively (prior to any background mortality adjustment).

During the clinical validation interview, the gamma curve was selected as the preferred OS extrapolation based on the plausibility of the extrapolation, with a median OS of ''''''' years and ''''''''''' of patients alive at 5 years.[1] The gamma curve is therefore used to inform the base case OS extrapolation in the current 4L+ care arm.

Figure 25 presents the selected axi-cel and current 4L+ curves used in the model base case.

Figure 25: Axi-cel versus current 4L+ care, OS: selected

==> picture [484 x 109] intentionally omitted <==

==> picture [484 x 108] intentionally omitted <==

==> picture [484 x 108] intentionally omitted <==

Key : KM, Kaplan–Meier; OS, overall survival.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 101 of 155

B.3.3.4. Summary of selected survival extrapolations

Figure 26 presents a summary of the selected survival extrapolations for PFS and OS, capturing 25% of patients treated with axi-cel as long-term survivors at 5 years.

Figure 26: Axi-cel versus current 4L+ care, OS and PFS: selected

==> picture [484 x 110] intentionally omitted <==

==> picture [484 x 110] intentionally omitted <==

==> picture [484 x 110] intentionally omitted <==

Key : KM, Kaplan–Meier; PFS, progression-free survival; OS, overall survival.

B.3.4. Measurement and valuation of health effects

B.3.4.1. Health-related quality of life data from clinical trials

The NICE reference case stipulates that health effects should be expressed in QALYs.[48] To generate QALYs, life years are weighted by a utility value, representative of HRQL. Patient utility is treated as an index, where an index value of 0 represents death and a value of 1 represents full health. As no HRQL data were collected in ZUMA-5 or SCHOLAR-5, utility values identified in the literature were used to inform the analysis (discussed in Section B.3.4.3).

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 102 of 155

B.3.4.2. Mapping

As there were no HRQL data collected in ZUMA-5, de novo mapping analyses were not conducted for this submission.

B.3.4.3. Health-related quality of life studies

In line with the search for economic evaluations and cost and resource use studies, a systematic review of HRQL evidence in patients with r/r FL was conducted. Original searches were conducted in May 2020, with an updated search run in May 2021. The study identification process, search strategies and a description of the included utility studies is presented in Appendix H. In total, 25 studies identified in the HRQL and utility SLR were included for data extraction. No additional studies of relevance were identified in the updated searches. Of the 25 included studies, 16 reported HRQL elicited via a generic patient-reported outcome instrument; of these 16 studies, 15 included a version of the EQ-5D. The NICE reference case specifies that the EQ-5D is the preferred measure of HRQL in adults.

The SLR identified six studies that reported health state utility scores in a r/r FL population, many of which refer to the same set and source of utility values (Wild et al. [2006]/Pettengell et al. [2007]).[22, 59] As was the case in TA627[51] , Wild et al. was not included within the SLR itself as only the abstract is available and utility values are not directly reported in the abstract; however, information on this study has been gathered from many economic evaluations for FL patients.

It is understood that Wild et al. and Pettengell et al. report components of the same study, in which 222 patients (aged 18 years and over with histologically confirmed FL and an Eastern Cooperative Oncology Group performance status of 0–2) were recruited from eight sites across the UK and completed several patient-reported outcome measures. Patients were analysed according to five disease states: 'active disease-newly diagnosed', 'active disease-relapsed', 'partial response', 'complete response' and 'disease free'. Pettengell et al. assessed patient HRQL using the Functional Assessment of Cancer Therapy – Lymphoma (FACT-Lym) instrument and administered the Hospital Anxiety and Depression Scale, to measure psychological morbidity, and the Work Productivity and Activity Impairment Scale to assess the influence of the disease upon activity and productivity. Pettengell et al. identified a clear relationship between disease status and HRQL in their study.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 103 of 155

Patients with active relapsed disease reported worse HRQL outcomes across FACTLym domains, in comparison to those in remission, partial responders to therapy, and those with newly diagnosed disease. This result was robust to the authors’ statistical analyses (ordinary least squares linear regression of the FACT-Lym Total Outcome Score upon scores from each contributory domain). As also reported in TA627, from the Wild et al. study, utility scores were obtained using the EQ-5D questionnaire and by grouping the disease state categories:

• Progression-free utility of 0.805 (standard error: 0.018)

• Progressed (off-treatment) utility of 0.7363 (assuming combined health states of active disease – newly diagnosed/relapsed)

• Progressed (on-treatment) utility value of 0.62 (assuming single health state active disease – relapsed)

In TA604, utility values from Pettengell et al./Wild et al. were used to inform the costeffectiveness analysis.[33] In TA627, utility values reported in Pettengell et al./Wild et al. were used in the economic model in scenario analysis. However in the TA627 company base case, utility values were derived for R[2] and R-chemo using a mixed effects model fit to HRQL data collected in the AUGMENT study. In the TA627 Final Appraisal Determination, it was reported that patients in AUGMENT had HRQL values that were higher than the general population for the same age group, in all health states. It was noted in the Final Appraisal Determination that clinical experts said someone with FL would not have higher quality of life than a member of the general population in any health state, and at best their quality of life would be equal to a member of the general population at the same age.[51] The TA627 Committee’s preferred approach was to cap progression-free utility at the general population and use relative decrements in other health states. In TA629 (TA472 CDF resubmission), utility estimates from the GADOLIN study were used.[54]

Table 31 presents a summary of the utility values identified in the literature for patients with r/r FL. The HRQL data used in the base case analysis and scenario analysis is discussed in Section B.3.4.5.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 104 of 155

Table 31: Summary of utility values identified in the literature

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 105 of 155

B.3.4.4. Adverse reactions

As reported in NICE TA677, since the approved access of CAR T-cell therapies through the CDF in NHS England, clinicians have become increasingly comfortable with toxicity management for CAR T-cell therapies.[54] However, it is acknowledged that there still may be short-term impactful AEs for many patients following treatment with axi-cel. Therefore, a comprehensive approach to capturing these in the model for the axi-cel arm has been taken. For the current 4L+ care arm, a more simplistic approach has been taken.

B.3.4.4.1. Adverse event frequencies

The model attempts to capture the impact of experiencing AEs on both costs and HRQL. The model considers Grade ≥ 3 treatment-related AEs occurring in ZUMA-5 patients. For AEs of clinical importance for CAR T-cell therapies (CRS requiring tocilizumab treatment, and hypogammaglobulinaemia associated with B-cell aplasia), AEs of all grades were included in the model, in line with previous CAR T- cell therapy appraisals.

Specifically, the following AEs were modelled for axi-cel:

• Grade ≥ 3 axi-cel-related AEs occurring in 5% or more of subjects in ZUMA-5

• Grade ≥ 3 treatment-emergent CRS occurring in ZUMA-5 ('''''''''''''' of patients) and any grade CRS requiring treatment with tocilizumab (''''''''''''''''' of patients)

• The proportion of patients who received immunoglobulin treatment ('''''''''''''''''' of patients)

Table 32 presents the incidence of Grade ≥ 3 treatment-related AEs occurring in 5% or more of patients treated with axi-cel, informed by ZUMA-5. For current 4L+ care, AE frequencies were sourced from clinical trial data (reported in NICE TA627) for individual treatments comprising the basket[54] , before being weighted by the current 4L+ care treatment distributions (described further in Section B.3.5.1.2). Conservatively, Grade ≥ 3 AEs for the treatments comprising current 4L+ care were only included in the cost-effectiveness analysis if they also occurred in 5% or more of ZUMA-5 patients. Current 4L+ care AEs included in the analysis are presented in Table 33.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 106 of 155

Table 32: Grade ≥ 3 adverse event data, axi-cel (ZUMA-5)

Table 33: Adverse event data, current 4L+ care (reported in TA627)

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 107 of 155

B.3.4.4.2. Adverse event utility decrements

Consistent with the approach used by Hettle et al., and the base case approach used in TA559 and TA677, it is assumed that those experiencing Grade ≥ 3 CRS have a quality of life of zero (i.e. the utility decrement is set to be the negative of the utility value in the progression-free health state).[45, 52, 54] The CRS utility decrement, which is applied as a one-off in the first model cycle with other AE utility decrements, was calculated using the median time to CRS resolution observed in ZUMA-5 (6 days).

Also in line with the methods used by Hettle et al. and in TA677, a disutility for hypogammaglobulinaemia was not applied as it is not thought to result in a reduction of HRQL.[45, 52]

For other Grade ≥ 3 AEs occurring in more than 5% of patients, a utility decrement of 0.15 was applied. The utility decrement is consistent with the approach taken in prior NICE appraisals in advanced lymphoma indications (TA677 and TA567)[53, 54] , and was originally derived from a cost-effectiveness analysis by Guadagnolo et al. (2006) in patients after primary treatment for Hodgkin lymphoma.[61]

AEs related to the axi-cel arm are expected to occur in the short term after initial treatment; therefore, a one-off QALY decrement is applied in the first model cycle. As a simplifying modelling assumption, AE disutility values in the current 4L+ arm are also applied as a one-off QALY decrement in the first model cycle, despite AEs occurring over the current 4L+ care treatment duration in practice. Table 34 presents the AE durations, which are combined with the AE frequencies and disutility values to calculate the one-off QALY decrement.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 108 of 155

Table 34: Adverse event durations

B.3.4.5. Health-related quality-of-life data used in the cost-effectiveness analysis

In the absence of HRQL data from ZUMA-5, committee-preferred assumptions from the NICE appraisal of lenalidomide with rituximab for treated FL (TA627) were taken into account when forming the base case analysis.[60] As discussed in Section B.3.4.3, in TA627, general population utility values were used in the progression-free state, with relative decrements from AUGMENT used in the progressed state.

A consistent approach was taken here, with age- and sex-matched general population utility values calculated from Ara and Brazier, 2010.[62] The relative decrement for the progressed state (0.969) was calculated as the AUGMENTderived progression-free utility values for lenalidomide with rituximab (0.847) over the corresponding progressed (off treatment) utility (0.821). In scenario analysis, the impact of calculating the relative decrement for the progressed state using rituximab monotherapy data from AUGMENT (rather than lenalidomide with rituximab data) is explored. Furthermore, a scenario is presented which uses the absolute decrement of -0.026 taken from the AUGMENT mixed effects model to calculate utility in the progressed disease state.

Table 35 describes the utility values applied in the base case cost-effectiveness model and the sources they are taken from.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 109 of 155

It is acknowledged that, in the absence of HRQL data collected in the pivotal trial, utility estimates used in the base case analysis are uncertain. Although the assumptions made align with Committee preferences in a recent appraisal in r/r FL, it should be noted that TA627 considered patients at an earlier line of therapy. As such, several scenarios have been explored using alternative utility values identified in the literature:

• Wild et al.

  • Progression-free utility of 0.805 (0.018)

  • Progressed utility of 0.7363 (off-treatment value applied)

• AUGMENT (lenalidomide with rituximab data)

  • Progression-free utility of 0.847

  • Progressed utility of 0.821 (off-treatment value applied)

• AUGMENT (rituximab monotherapy data)

  • Progression-free utility of 0.840

  • Progressed utility of 0.813 (off-treatment value applied)

• GADOLIN

  • Progression-free utility of 0.822 (0.010; on treatment value applied)

  • Progressed utility of 0.758 (0.024)

When alternative values identified in the review of the literature are used to capture utility for patients in the pre-progression and progressed disease health states in scenario analysis, an adjustment is applied over time. This assumes utility declines with age in line with general population trends, as represented by Health Survey for England data modelled by Ara and Brazier.[62]

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 110 of 155

Table 35: Summary of utility values for cost-effectiveness analysis

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 111 of 155

B.3.5. Cost and healthcare resource use identification, measurement and valuation

A systematic review of the literature was conducted to identify relevant published cost and resource use data. Searches were conducted alongside those presented in Section B.3.1 for economic evaluations. As reported in Appendix I, and in line with the economic evaluation SLR reported in Appendix G, searches were initially conducted in May 2020, with an update performed in May 2021.

Arguably one of the most relevant sources of cost and resource use identification to this appraisal was the most recent NICE STA in previously treated FL (TA627).[51] Furthermore, due to the mechanism of action and innovative nature of CAR T-cell therapies in advanced lymphoma indications, TA559, TA567 and TA677 have also been identified as useful in informing costs and resource use data and assumptions in this economic evaluation.[45, 53, 54]

In line with the NICE reference case, the perspective on costs is that of the NHS and PSS in England.[48] Costs are derived from typical sources for economic evaluations conducted in a UK setting, including:

• The Monthly Index of Medical Specialities (MIMS) for branded treatment costs[63]

• The drugs and pharmaceutical electronic market information tool (eMIT) for generic treatment costs[64]

• National Schedule of NHS costs (or NHS reference costs) 2019/20 for service/healthcare activity costs[65]

• The PSS Research Unit (PSSRU) Unit Costs of Health and Social Care 2020 for staff costs and inflation indices[66]

• Published literature sources

B.3.5.1. Intervention and comparators’ costs and resource use

B.3.5.1.1. Axi-cel

As described in Section B.1.2, axi-cel is produced using patient T-cells, which are extracted via leukapheresis. During the manufacturing of a CAR T-cell therapy, some patients may require bridging therapy to remain in a stable condition and eligible for

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 112 of 155

CAR T-cell infusion. Prior to receiving axi-cel, which is given as a single infusion treatment, patients receive conditioning chemotherapy. In line with ZUMA-5, patients are hospitalised prior to axi-cel administration and are expected to be monitored in an inpatient setting for at least 7 days following infusion.

The following treatment-related costs are therefore considered within the axi-cel arm of the model:

• Leukapheresis

• Bridging therapy

• Conditioning chemotherapy

• Axi-cel drug acquisition costs

• Axi-cel infusion and monitoring hospitalisation costs

As discussed in Section B.2.3, '''''' patients with 4L+ r/r FL were enrolled in ZUMA-5 (at the commencement of leukapheresis). Of the ''''''' enrolled patients, '''''' were treated with conditioning chemotherapy and axi-cel.

While axi-cel is administered as a one-off infusion, '''''' of the ''''' treated patients in the ZUMA-5 4L+ mITT analysis set were retreated with axi-cel. Despite this observation from the trial, retreatment with axi-cel is not expected to occur in clinical practice in England and does not form part of the expected marketing authorisation, therefore costs of axi-cel retreatment are not considered relevant to the cost-effectiveness analysis.

As treatment with axi-cel is administered as a single infusion, treatment-related costs in the axi-cel arm of the model are applied as a one-off cost in the first model cycle, for simplicity.

B.3.5.1.1.1. Leukapheresis costs

In alignment with two of the previous NICE appraisals for CAR T-cell therapies in advanced, previously treated lymphoma indications (TA677 and TA559), the cost of leukapheresis was calculated as the weighted average of stem cell and bone marrow harvest.[45, 54] Table 36 presents unit costs sourced from the latest NHS reference costs (2019/2020).[65]

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 113 of 155

Table 36: Leukapheresis unit costs (NHS reference costs 2019/20)[65]

Although '''''' patients in ZUMA-5 underwent leukapheresis, only '''''' received axi-cel. To reflect costs more accurately, an uplifting factor of ''''''''''' '''''''''''''''' was applied to the initial weighted average cost of leukapheresis, resulting in a leukapheresis cost of ''''''''''''''''''''''''' [£1,953.38 x (''''''''''''''')] per patient.

B.3.5.1.1.2. Bridging therapy costs

In ZUMA-5, bridging therapy could be administered to patients between leukapheresis and the administration of conditioning chemotherapy, at the discretion of the treating investigator. Some patients may require bridging therapy to remain in a stable condition and eligible for CAR T-cell infusion. However, in the 4L+ mITT cohort of ZUMA-5, only '''''''''''' ''''''''''''''''''' ''''''''''''' received bridging therapy.

'''''''''' '''''''''''''''''' ''''''''''''''' ''''''''''' '''''''''''''''''''' ''''''''''''''''''''' ''''''''''''''''' '''''''''' ''''''''''''''''' ''''''''' '''''''''''''''''''''' '''''''''

''' '''''''''''''''''''' '''' '''' '''''''''''. Other bridging therapies included dexamethasone, etoposide, carboplatin, ifosfamide, and mitoxantrone. However, the dose unit and dose frequency were not known for all bridging treatments.

As a simplifying assumption, it was assumed that one dose of rituximab monotherapy is representative of bridging therapy, for the small number of patients who were treated. This is considered a conservative assumption given that rituximab, which is a branded therapy, has higher treatment costs than other generic chemotherapies. The cost of a single dose of rituximab (£1,251.65; see comparator costs in Section B.3.5.1.2 for details) was applied to the proportion of enrolled patients who received bridging therapy in ZUMA-5, resulting in a total bridging therapy cost of '''''''''''''''''' [£1,251.65 x (''''''''''')].

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 114 of 155

B.3.5.1.1.3. Conditioning chemotherapy costs

A lymphodepleting chemotherapy regimen of intravenous cyclophosphamide 500 mg/m[2] and intravenous fludarabine 30 mg/m[2] on the fifth, fourth, and third day before axi-cel infusion was considered. Table 37 presents conditioning chemotherapy unit costs, sourced from the latest eMIT.[64]

Table 37: Conditioning chemotherapy unit costs (eMIT)[64]

Drug wastage was considered for conditioning chemotherapy by assuming that patients receive only whole vials, and vial sharing is not permitted. A normal distribution was fitted to mean body surface area (BSA) data to calculate a distribution of the number of vials of cyclophosphamide and fludarabine needed for one administration. Mean BSA (1.99 m[2] ) was estimated using mean height and weight data from the mITT population of ZUMA-5 in the Du Bois formula.[67] BSA standard deviation was assumed to be 10% of the mean value. The resulting distribution was used to more accurately calculate the number of vials required for the average patient.

The resulting acquisition costs of conditioning chemotherapy were £17.50 and £39.51 for cyclophosphamide and fludarabine, respectively. In line with prior appraisals in CAR T-cell therapies[45, 53, 54] , it was assumed conditioning chemotherapy is administered in hospital over 3 days in an elective inpatient setting. A hospitalisation cost per day of £903.20 (see below; Table 38) was applied to the total conditioning chemotherapy cost.

The total cost of 3 days of conditioning chemotherapy, including drug acquisition and hospitalisation, was £2,880.65 per patient.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 115 of 155

B.3.5.1.1.4. Axi-cel acquisition costs

Axi-cel has a one-off cost of £280,451.00 (list price), which includes production and delivery (Kite, data on file). A patient access scheme (PAS) of ''''''''''' (simple discount) for axi-cel has been agreed with NHS England, the resulting one-off acquisition cost is ''''''''''''''''''''''''''''''.

B.3.5.1.1.5. Axi-cel infusion and monitoring hospitalisation costs

Axi-cel is administered as a single infusion within 30 minutes. In line with prior NICE appraisals, CAR T-cell therapies are administered, and patients monitored, in an elective inpatient setting.[45, 54] Table 38 presents the cost of hospitalisation, which was calculated as the weighted average cost for malignant lymphoma, including Hodgkin lymphoma and non-Hodgkin lymphoma from NHS reference costs 2019/2020.[65]

Hospital Episode Statistics 2019/20 report the mean length of stay for malignant neoplasms of lymphoid, haematopoietic, and related tissues (Codes C81-C96) as 8.1 days.[68] As the mean duration of hospitalisation following axi-cel was longer in ZUMA-5 ('''''''''''' days) than that reported in the Hospital Episode Statistics, a per-day hospitalisation cost was calculated (£903.20) before being applied to the ZUMA-5 hospitalisation duration to avoid the potential underestimation of costs in the axi-cel arm.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 116 of 155

Table 38: Hospitalisation unit costs (NHS reference costs 2019/20)[65]

When using the average length of stay reported in ZUMA-5 ('''''''''' days) and the average calculated per-day hospitalisation cost of £903.20, the resulting axi-cel infusion and monitoring costs is '''''''''''''''''''''''''''

B.3.5.1.1.6. Summary of treatment-related axi-cel costs

Table 39 presents a summary of the treatment-related axi-cel costs, applied as a one-off in the first model cycle.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 117 of 155

Table 39: Axi-cel treatment costs per patient

B.3.5.1.2. Current 4L+ care

As discussed in Section B.3.2.3.2, current 4L+ care is modelled as a basket of therapies, as there is no established standard of care for patients with FL who have received three or more prior lines of therapy. The components of this basket were aligned with SCHOLAR-5 European treatment patterns (

Figure 12). The following assumptions were made regarding SCHOLAR-5 treatment patterns:

• Idelalisib is representative of the ‘PI3Ki based’ treatment category

• Rituximab plus bendamustine and obinutuzumab plus bendamustine are representative of the ‘CD20+Benda’ treatment category

• R-CHOP (cyclophosphamide, doxorubicin, vincristine and prednisolone) is representative of the ‘CD20+CHOP like’ treatment category

• Rituximab plus CVP (cyclophosphamide, vincristine and prednisolone) is representative of the ‘CD20+other chemo’ treatment category

• CVP is representative of the chemotherapy category

• R[2] is representative of the ‘R[2] and other immunomodulatory imide drug (IMiD) based’ category

However, since SCHOLAR-5 included patients outside of an English setting, some of the observed therapies are not reimbursed by NHS England for the treatment of FL. To better reflect the cost of the comparator mix in practice, all treatments that are not reimbursed by NHS England were removed from the basket. Similarly, SCHOLAR-5

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 118 of 155

included patients who received experimental treatments, which could not be costed and were therefore removed from the basket.

Finally, in validation interviews with NHS consultants, it was stated that CVP alone is unlikely to be used in 4L+ patients with r/r FL, and that rituximab plus bendamustine use is likely to be twice that of obinutuzumab plus bendamustine at 4L+.[1] These views were taken into consideration when costing current 4L+ care.

Table 40 presents the final re-weighted treatment distributions comprising current 4L+ care costs in the base case analysis.

Table 40: Distribution of current 4L+ care therapies

Table 41 and Table 42 present unit costs (list price) and dosing schedules for the set of treatments included in the re-weighted current 4L+ care blend, respectively. In the first instance, generic treatment costs were sourced from eMIT; branded treatment costs were then sourced from MIMS.[63, 64]

As described in Section B.3.5.1.1.3 for conditioning chemotherapy costs, wastage was considered for treatments administered intravenously by assuming that patients require whole vials and that vial sharing is not permitted. For treatments administered per m[2 ] of BSA, all available vial sizes were included in the model and a

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 119 of 155

distribution fitted to calculate the average number of vials required for one administration.

For the two included treatments that are administered orally, the most efficient pack size of tablets was included based on the dosing schedule. For lenalidomide, the 21 x 20 mg pack was included; and for prednisolone, the 28 x 20 mg pack was included. Wastage was conservatively not considered for oral treatments in the current 4L+ care arm of the model, by calculating the cost of the number of tablets required for a 28-day model cycle.

Table 41: Current 4L+ care unit costs (list price)

Table 42: Current 4L+ care – dosing schedules

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 120 of 155

Table 43 presents administration costs considered in the cost-effectiveness model. Administration costs for treatments administered intravenously were sourced from NHS reference costs 2019/20.[65] The administration complexity was specified for each treatment, before the cost of subsequent elements of a chemotherapy cycle were applied for the remaining administrations in a given cycle. Oral therapies are conservatively assumed to have zero administration or dispensing costs in the current 4L+ care arm.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 121 of 155

Table 43: Administration costs (NHS reference costs 2019/20)[65]

Table 44 summarises the per 28-day model cycle drug and administration costs in the current 4L+ care arm (including drug wastage for intravenous therapies).

Table 44: Drug and administration costs per 28-day model cycle

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 122 of 155

In the absence of time on treatment (ToT) data reported in SCHOLAR-5, median treatment durations in the current 4L+ care arm were estimated using the reported number of treatment cycles in the relevant SmPC (Table 45). To cost treatment regimens on a per-cycle basis, while appropriately discounting over time, exponential ToT curves were fitted to the estimated treatment durations.

Table 45: Treatment durations (summary of product characteristics)

Per-cycle drug and administration costs for each treatment in the current 4L+ care blend (Table 44) were applied to the respective estimated ToT curve. Finally, single weighted per-cycle current 4L+ care treatment and administration costs were calculated using the distribution of comparator treatments shown in Table 40.

B.3.5.1.3. Subsequent treatment costs

As discussed in Section B.3.2.3.2, despite the clinical pathway for FL being well established at early treatment lines, there is no established standard of care for adult

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 123 of 155

patients with r/r FL who are receiving 4L+ care. The distribution of subsequent therapies received in both the axi-cel and current 4L+ care arms of the model are therefore assumed equal to the re-weighted distribution of treatments in the comparator arm of the cost-effectiveness model (Table 40). Table 46 presents the one-off subsequent acquisition and administration costs applied in the model.

Table 46: Subsequent treatment costs

Subsequent treatment costs are applied once at the point of progression as a simplifying assumption. The proportion of patients receiving the cost of subsequent therapies is estimated as the number of patients who leave the pre-progression state in a given cycle, after accounting for the proportion of PFS events that are deaths as observed in ZUMA-5 (''''''''''').

As described in Section B.3.2.2.1, a time point of 5 years is used to determine when patients remaining progression-free are assumed to be long-term survivors and thus no longer experience the cost of subsequent systemic treatment in the axi-cel arm.

B.3.5.2. Health-state unit costs and resource use

Health-state-dependent resource use costs are assumed to be similar to those presented in previous FL NICE submissions and in the European Society for Medical Oncology (ESMO) guidelines.[51, 69] Resource use frequency in the pre-progression health state is further split by induction phase and maintenance phase (Table 47). As current 4L+ care comprises a blend of treatments, the weighted average length of the primary/induction phase was calculated (seven cycles).

In the validation interview with NHS Consultants, it was noted that in the first 6 months following CAR T-cell therapy, monitoring may be more frequent, with monthly blood tests. In the economic model, axi-cel resource use follows the pre-progression (induction) level for six cycles, and the maintenance level thereafter while patients remain progression free until Year 5. As previously discussed, after 5 years patients

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 124 of 155

who remain alive and free of progression are assumed to be long-term survivors and no longer experience resource use requirements.

Unit costs for haematologist visits and CT scans sourced from NHS reference costs 2019/20 are presented in Table 48.[65] Unit costs comprising diagnostic tests sourced from NICE TA627 are presented in Table 49 and uplifted to the latest cost year using PSSRU inflation indices.[51, 66]

Table 47: Resource use frequency

Table 48: Resource use unit costs (NHS reference costs 2019/20)

Table 49: Diagnostic test costs

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 125 of 155

B.3.5.3. Adverse event costs

As discussed in Section B.3.4.4.1, the model attempts to reflect the costs associated with the management of AEs. In line with the approach to utility decrements, as AEs related to the axi-cel arm are expected to occur in the short term after initial treatment, AE management costs are applied as a one-off in the first model cycle (except for hypogammaglobulinaemia). As a simplifying modelling assumption, AE management costs in the current 4L+ arm are also applied as a one-off, despite AEs occurring over the current 4L+ care treatment duration in practice.

B.3.5.3.1. Axi-cel adverse event costs

In line with NICE TA677, TA559 and Hettle et al., it is assumed the cost of managing AEs is captured by the initial infusion and monitoring hospitalisation costs (Section B.3.5.1.1.5) associated with the administration of CAR T-cell therapy, with the following exceptions:

• All patients experiencing an axi-cel-related Grade ≥ 3 AE (''''''''''''''''''' Table 32) are assumed to incur the cost of an additional bed day (£903.20; Table 38)

• Hypogammaglobulinaemia (B-cell aplasia) is managed with IVIG. Costs for IVIG are included for any patient in ZUMA-5 requiring this treatment (''''''''''')

• CRS is managed with tocilizumab. Costs for tocilizumab are included for any patient in ZUMA-5 requiring this treatment ('''''''''')

• Patients experiencing Grade 3/4 CRS are assumed to be managed in the intensive care unit ('''''''')

B.3.5.3.1.1. Hypogammaglobulinaemia

Treatment-emergent hypogammaglobulinaemia occurred in '''''' patients ('''''''''''' in the

4L+ mITT population of ZUMA-5), ''''''''''''''''''' ''''''''''' ''''' ''''''''''''' '''''''''''' '''''''''''' ''''''''''''' '''' '''''

''''''''''''''''. ''''''''''''''''''''''''''''''''''' '''''''''''''''''' '''''''''''''''' were treated with IVIG therapy.

Hypogammaglobulinaemia has not been applied as a one-off cost, as it requires ongoing treatment over a relatively long period of time. The cost for administration of a simple parenteral chemotherapy regimen in an outpatient setting was used for IVIG administration, in line with TA677, TA567 and TA559.[45, 53, 54] For the IVIG treatment costs, the immunoglobulin drug costs reported in MIMS were used; specifically, in line with TA677, it was assumed that Gammaplex[®] 5% solution for infusion would be

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 126 of 155

used in practice. Table 50 summarises the IVIG unit, measure, pack size and cost per pack.

Table 50: IVIG unit costs (MIMS)[63]

In line with the assumptions used in TA677 and TA559 (which were based on Hettle et al.) a dose of 0.5g/kg every 4 weeks was assumed.[45, 52, 54] Furthermore, IVIG was assumed to be administered to pre-progression patients for a duration of 12 months, consistent with the assumption used in TA677 and TA559.[45, 54] In TA677, it was reported that NHS consultants agreed that both the dosing regimen and assumed duration was sensible.[54] The consultants in TA677 added that there is awareness in clinical practice of the cost of IVIG therapy and that, as a result, wastage is likely to be minimised.[54] Despite this, as a conservative approach, wastage is accounted for when costing IVIG (Table 51).

Table 51: Summary of IVIG costs applied in the model

B.3.5.3.1.2. CRS

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 127 of 155

As described in Section B.2.10.4.1, '''''''''''' of patients in the 4L+ mITT analysis set of ZUMA-5 experienced a CRS event. Most of these were Grade 1–2, and, in the full mITT analysis set of ZUMA-5, all CRS events resolved after a median duration of ''' days. The method for costing CRS was taken from previous CAR T-cell appraisals and Hettle et al.[45, 52, 54] It is assumed that patients experiencing a Grade 3/4 CRS event ('''''''' of patients) accrue the cost of an intensive care unit (ICU) hospitalisation. The cost of an ICU hospitalisation was calculated based on non-specific, general adult critical care costs from NHS reference costs 2019–2020.[65] A weighted average of the costs for supporting one and two organs was assumed based on feedback from clinicians during validation, equating to an daily ICU cost of £1,508.65. A duration of 4 days was assumed for the ICU stay, based on assumptions in the Final Appraisal Determination for TA559 and in the TA677 company submission.[45, 54] The final ICU cost for all patients with Grade 3/4 CRS is ''''''''''''''''''''.

Furthermore, '''''' patients ('''''''''''' of patients in ZUMA-5) were treated with a cytokine inhibitor drug – tocilizumab. The modelled cost of cytokine inhibitor drugs is £659.76, taken from NHS reference costs 2019-2020 (currency code PHCD00098 / High Cost Drugs).[65] It is assumed that this cost is the average cost per tocilizumab administration and covers both drug and administration costs. It is further assumed that only one administration of tocilizumab would be required. The total cost for CRS management including hospitalisation and tocilizumab treatment, applied upfront in the model is '''''''''''''''''''''.

B.3.5.3.2. Current 4L+ care adverse event costs

AE management costs in the current 4L+ care arm are presented in Table 52.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 128 of 155

Table 52: Adverse event management unit costs

B.3.5.4. Miscellaneous unit costs and resource use

A patient with end-stage cancer typically incurs costs at the end of life for palliative and hospice care. The publication by Round et al. (2015) is a standard source used for such costs in submissions to NICE.[70] Costs were taken from this publication and inflated to 2019/20 prices using PSSRU inflation indices.[66] As the publication does not specifically report an end-of-life care cost for patients with any form of haematological malignancy, the average cost for all cancer types reported was assumed. End-of-life care costs are applied as a one-off upon death (Table 53).

Table 53: End-of-life care costs (Round et al. [2015])[70]

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 129 of 155

B.3.6. Summary of base case analysis inputs and assumptions

B.3.6.1. Summary of base case analysis inputs

A summary of the variables included in the model, their base case values and distributions used to reflect uncertainty is provided in Appendix P.

Information on the uncertainty of each parameter, such as standard errors confidence intervals and sample sizes, was taken from the original source where available. Where uncertainty information was not reported, the standard error was assumed to be 10% of the mean value.

A normal distribution was used for costs, resource use frequencies, and durations as per the central limit theorem. A beta distribution was used for probabilities, proportions and utilities, acknowledging that such parameters can never be negative and cannot exceed 1. A log-normal distribution was used for the SMR applied to general population mortality, acknowledging that the SMR cannot be negative and is right-skewed. A multivariate normal distribution (using variance covariance matrices) was used to capture uncertainty in correlated parameters, such as survival parameters and utility regressions, while maintaining the correlation between parameters. In a number of the alternative utility sources used, no indication of covariance was reported in the source to account for correlation between preprogression and post-progression utility values. Therefore, to ensure that sampled values demonstrated face validity, a pragmatic approach was taken whereby the same random number was used to ensure that pre-progression utility was always greater than post-progression utility.

Inputs not associated with parameter uncertainty, such as the time horizon, discount rates and alternative modelling assumptions, were investigated in the scenario analysis only.

B.3.6.2. Assumptions

The approach to modelling has been designed to make the best use of the available data to inform the decision problem, in line with the NICE reference case and guidance on methods of appraisal. In the absence of key data, key assumptions have been necessary and have been made to minimise potential bias in the analysis

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 130 of 155

while tending towards conservative assumptions, where assumptions were necessary. Table 54 describes key modelling assumptions with justifications.

Table 54: Key model assumptions

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 131 of 155

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 132 of 155

B.3.7. Base case results

B.3.7.1. Base case incremental cost-effectiveness analysis results

Table 55 presents the base case incremental cost-effectiveness results for axi-cel (PAS price) versus current 4L+ care. Costs and QALYs are time-preference discounted at 3.5%, in line with the NICE reference case.[48] Markov traces for each arm are presented in Figure 27 and Figure 28.

Based on the outlook for current 4L+ care patients (in the absence of an established standard of care) and the hope and expectation of the transformative effect of CAR

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 133 of 155

T-cell therapy for patients with advanced lymphomas, axi-cel is expected to offer an incremental health effect of ''''''''''' undiscounted life years, or ''''''''''' discounted QALYs, with more time spent in the progression-free state for patients treated with axi-cel.

The estimated deterministic incremental cost-effectiveness model (ICER) of £48,272 for axi-cel (PAS price) versus current 4L+ care falls below the NICE decision-making threshold for treatments given end-of-life weighting.

Disaggregated results and model results compared with the clinical data are presented in Appendix J.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 134 of 155

Table 55: Base case results (with PAS)

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 135 of 155

Figure 27: Lifetime Markov trace for axi-cel

==> picture [421 x 103] intentionally omitted <==

==> picture [421 x 103] intentionally omitted <==

==> picture [421 x 103] intentionally omitted <==

Figure 28: Lifetime Markov trace current 4L+ care

==> picture [421 x 104] intentionally omitted <==

==> picture [421 x 103] intentionally omitted <==

==> picture [421 x 103] intentionally omitted <==

Key: 4L+, fourth line plus.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 136 of 155

B.3.8. Sensitivity analyses

B.3.8.1. Probabilistic sensitivity analysis

A probabilistic sensitivity analysis (PSA) was undertaken to explore the joint uncertainty of all model parameters based on their distributional information (see Appendix P) and their associated impact on cost-effectiveness results. To ensure convergence, all inputs were varied simultaneously over 2,000 iterations (rolling average incremental costs, life years [LYs] and QALYs were plotted on convergence graphs within the cost-effectiveness model and visually inspected).

All PSA iterations indicated that axi-cel provides an incremental QALY benefit versus current 4L+ care, at an increased total cost. When comparing average PSA results with deterministic results (Table 56), incremental costs are consistent and incremental QALYs slightly lower, leading to a slightly higher mean PSA ICER.

Figure 29 shows the scatter plot for 2,000 PSA iterations (with the axi-cel PAS). Due to the difference between the mean PSA and deterministic ICER, the analysis was re-run, specifically without varying the survival analysis parameters (Figure 30). Comparing Figure 29 and Figure 30 illustrates that the higher PSA ICER is partly due to the asymmetrical uncertainty distributions of interrelated survival analysis parameters resulting in non-normality in the sampled outcomes.

The cost-effectiveness acceptability curve in Figure 31 demonstrates that axi-cel (with PAS) is more likely to be a cost-effective treatment option when compared with current 4L+ care at a willingness-to-pay threshold of approximately £50,000 per QALY gained.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 137 of 155

Table 56: Mean results of PSA (2,000 runs) and comparison with deterministic results (with PAS)

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 138 of 155

Figure 29: Scatter plot of PSA with 2,000 iterations (with PAS)

==> picture [427 x 134] intentionally omitted <==

==> picture [427 x 135] intentionally omitted <==

Key: PAS, patient access scheme; PSA, probabilistic sensitivity analysis; QALY, quality-adjusted life year.

Figure 30: Scatter plot of PSA with 2,000 iterations (with PAS; excluding

survival analysis parameters)

==> picture [427 x 134] intentionally omitted <==

==> picture [427 x 135] intentionally omitted <==

Key: PAS, patient access scheme; PSA, probabilistic sensitivity analysis; QALY, quality-adjusted life year.

Note : For comparability, the scale of the axes is equivalent to that of the PSA including the variation of the survival analysis parameters.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 139 of 155

Figure 31: Cost-effectiveness acceptability curves for PSA with 2,000 iterations

(with PAS)

==> picture [455 x 96] intentionally omitted <==

==> picture [455 x 96] intentionally omitted <==

==> picture [455 x 96] intentionally omitted <==

Key: 4L+, fourth line or later; PAS, patient access scheme; PSA, probabilistic sensitivity analysis.

B.3.8.2. Deterministic sensitivity analysis

One-way sensitivity analysis (OWSA) was performed to evaluate the sensitivity of the model’s ICER to individual inputs, holding all else constant. Inputs with parametric uncertainty were set to the upper and lower limits of their 95% CIs, reported in Appendix P. Where CIs were not reported, upper and lower bounds were calculated from the mean, standard error and assumed distribution of each parameter. Figure 32 and Table 57 present the 10 parameters with the greatest impact on the ICER with descending sensitivity when their values were set to their upper and lower limits of the CIs. The results demonstrate that the model is relatively insensitive to reasonable variation in most parameters. The parameter with the greatest impact on the ICER is the proportion of patients receiving axi-cel that are long-term survivors and thus experience SMR-adjusted general population mortality from 5 years, as described in Section B.3.3.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 140 of 155

Figure 32: Tornado diagram showing OWSA results on ICER (with PAS)

==> picture [682 x 68] intentionally omitted <==

==> picture [682 x 69] intentionally omitted <==

==> picture [682 x 68] intentionally omitted <==

==> picture [682 x 68] intentionally omitted <==

Key: ICER, incremental cost-effectiveness ratio; OWSA, one-way sensitivity analysis; PAS, patient access scheme; R[2] , lenalidomide with rituximab.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 141 of 155

Table 57: OWSA results for the most influential model parameters on the ICER (with PAS)

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 142 of 155

B.3.8.3. Scenario analysis

Table 58 presents the scenario analyses conducted to assess structural and methodological uncertainty in the model. The scenarios that have the largest impact on the ICER relate to the discount rate, survival extrapolations and the long-term survivorship assumptions in the model.

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 143 of 155

Table 58: Scenario analysis results (with PAS)

Company evidence submission for axicabtagene ciloleucel for treating relapsed or refractory low-grade NHL [ID1685] ©Kite, a Gilead Company (2022). All rights reserved. 144 of 155