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

Single Technology Appraisal

Axicabtagene ciloleucel for treating diffuse large B-cell lymphoma and primary mediastinal B-cell lymphoma after 2 or more systemic therapies [ID1115] Committee Papers

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

SINGLE TECHNOLOGY APPRAISAL

Axicabtagene ciloleucel for treating diffuse large B-cell lymphoma and primary mediastinal B-cell lymphoma after 2 or more systemic therapies [ID1115]

Contents:

1. Pre-Meeting Briefing

2. Company submission from Kite a Gilead company

3. Clarification letters

• NICE request to the company for clarification on their submission

• Company response to NICE’s request for clarification

4. Patient group, professional group and NHS organisation submission from:

• Bloodwise – endorsed by patient expert Phil Reynolds

• Royal College of Pathologists and British Society of Haematologists (joint submission) – endorsed by clinical expert Andrew McMillian

• NHS England

5. Expert statements from:

• Professor John Radford – clinical expert, nominated by Kite a Gilead company

• Claire Foreman – commissioning expert, nominated by NHS England

6. Evidence Review Group report prepared by CRD and CHE Technology Assessment Group, University of York

7. Evidence Review Group report – factual accuracy check

8. Evidence Review Group addendum

9. Technical engagement document prepared by NICE technical team and committee chair

10. Technical engagement responses from:

• Kite, a Gilead company

• Bloodwise

• Lymphoma Action

• NHS England

• Dr Andrew McMillan – clinical expert, nominated by Royal College of Pathologists

11. Post technical engagement clarification

• NICE request to the company for clarification

• Company response – prepared by Kite, a Gilead company

12. Company submission addendum from Kite, a Gilead company  Commercial offer

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

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

1

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

2

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

3

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

4

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

5

Confidential

==> picture [458 x 344] intentionally omitted <==

References:

Office for National Statistics. Cancer Registration Statistics, England, 2015. Office of National Statistics. Accessed November 2017.

High grade NHL. Cancer research UK. Accessed November 2017. Low grade NHL. Cancer research UK. Accessed November 2017.

Diffuse B-cell lymphoma. Lymphoma association. Accessed November 2017.

Survival for high grade lymphomas. Cancer Research UK. Accessed November 2017. Outcomes for R/R disease - Crump et al 2017

National Institute for Health and Care Excellence Pre-meeting briefing

6

Confidential

==> picture [458 x 344] intentionally omitted <==

References:

Office for National Statistics. Cancer Registration Statistics, England, 2015. Office of National Statistics. Accessed November 2017.

High grade NHL. Cancer research UK. Accessed November 2017. Low grade NHL. Cancer research UK. Accessed November 2017.

Diffuse B-cell lymphoma. Lymphoma association. Accessed November 2017.

Survival for high grade lymphomas. Cancer Research UK. Accessed November 2017. Outcomes for R/R disease - Crump et al 2017

National Institute for Health and Care Excellence Pre-meeting briefing

7

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission Table 2

National Institute for Health and Care Excellence Pre-meeting briefing

8

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

CAR T-cell therapy:

Engineered Autologous Cell Therapy is a process by which a patient’s own T-cells are collected and genetically altered to recognise and target antigens expressed on the cell surface of specific malignancies. (Source: Kochenderfer JN, Dudley ME, Kassim SH, et al.). Chemotherapy-refractory diffuse large B-cell lymphoma and indolent B-cell malignancies can be effectively treated with autologous T cells expressing an anti-CD19 chimeric antigen receptor. J Clin Oncol . 2015; 33(6):540-9.)

National Institute for Health and Care Excellence Pre-meeting briefing

9

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

10

==> picture [458 x 344] intentionally omitted <==

Source: Adapted from Figure 3 in company submission

National Institute for Health and Care Excellence Pre-meeting briefing

11

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Submission from NHS England

National Institute for Health and Care Excellence Pre-meeting briefing

12

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

13

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

14

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

15

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company’s submission Table 1

National Institute for Health and Care Excellence Pre-meeting briefing

16

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

17

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company’s submission Table 1

National Institute for Health and Care Excellence Pre-meeting briefing

18

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: ERG report page 13-14

National Institute for Health and Care Excellence Pre-meeting briefing

19

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Clinical and commissioning responses to technical engagement

National Institute for Health and Care Excellence Pre-meeting briefing

20

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

21

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

22

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission Table 5

mITT (n=108) = modified intention to treat ITT (n=119) = intention to treat The full analysis population included all enrolled patients (N = 111). The modified intention-to-treat (mITT) population included all patients treated with at least 1.0 x 10[6] anti-CD19 CAR T-cells/kg (N = 101) and was the analysis population used for all efficacy analyses in ZUMA-1 Phase 2. The safety analysis population included all patients treated with any dose of axi-cel (N = 101).

National Institute for Health and Care Excellence Pre-meeting briefing

23

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission Page 53-54

National Institute for Health and Care Excellence Pre-meeting briefing

24

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: ERG report pages: 38-46

National Institute for Health and Care Excellence Pre-meeting briefing

25

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company appendix Table 9

National Institute for Health and Care Excellence Pre-meeting briefing

26

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company appendix Table 9

National Institute for Health and Care Excellence Pre-meeting briefing

27

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

28

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Figures 5 & 6 company submission

National Institute for Health and Care Excellence Pre-meeting briefing

29

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Table 12 company submission

National Institute for Health and Care Excellence Pre-meeting briefing

30

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

31

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission pages 63-67 and ERG report page 54 Table 9

National Institute for Health and Care Excellence Pre-meeting briefing

32

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission pages 63-67 Tables 15 & 18

National Institute for Health and Care Excellence Pre-meeting briefing

33

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission page 66 Figure 12

National Institute for Health and Care Excellence Pre-meeting briefing

34

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: ERG report pages 50-56

National Institute for Health and Care Excellence Pre-meeting briefing

35

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company response to clarification pages 5-8 Figure 3

National Institute for Health and Care Excellence Pre-meeting briefing

36

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company response to clarification Tables 3 & 4

National Institute for Health and Care Excellence Pre-meeting briefing

37

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Figure 1 response to clarification after technical engagement

National Institute for Health and Care Excellence Pre-meeting briefing

38

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

39

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission Table 19

National Institute for Health and Care Excellence Pre-meeting briefing

40

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: ERG report page 57

National Institute for Health and Care Excellence Pre-meeting briefing

41

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

42

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Figure 14 company submission

National Institute for Health and Care Excellence Pre-meeting briefing

43

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission Figure 18

National Institute for Health and Care Excellence Pre-meeting briefing

44

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission Figure 20

National Institute for Health and Care Excellence Pre-meeting briefing

45

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission Figure 23

National Institute for Health and Care Excellence Pre-meeting briefing

46

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Provided after clarification. Source: clarification response Figure 28 corrected graph

National Institute for Health and Care Excellence Pre-meeting briefing

47

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Response to clarification Figure 11

National Institute for Health and Care Excellence Pre-meeting briefing

48

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission Figure 31

National Institute for Health and Care Excellence Pre-meeting briefing

49

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission Pages 91-119

Time on treatment was not explicitly reported in the SCHOLAR-1 study and is not relevant to the ZUMA-1 trial, as axi-cel is given as a one-off infusion. For the modelled BSC regimens, the number of treatment cycles and the days per cycle for each component of each regimen are taken from the South East London Cancer Network and the Thames Valley Strategic Clinical Networks (CS page 116)

National Institute for Health and Care Excellence Pre-meeting briefing

50

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: ERG report pages 72-81

National Institute for Health and Care Excellence Pre-meeting briefing

51

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: ERG report pages 82-86

National Institute for Health and Care Excellence Pre-meeting briefing

52

==> picture [452 x 339] intentionally omitted <==

Source: ERG report Figure 18

53

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

54

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

55

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: ERG report Figure 17

National Institute for Health and Care Excellence Pre-meeting briefing

56

==> picture [452 x 339] intentionally omitted <==

Source: ERG report Table 14, company submission Table 43

57

==> picture [452 x 339] intentionally omitted <==

Source: ERG report Table 15, company submission Table 43 Only adverse events that had an incidence equal or greater than 10% were included in the economic model.

58

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: ERG report pages 91-102

National Institute for Health and Care Excellence Pre-meeting briefing

59

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission pages 128-132

National Institute for Health and Care Excellence Pre-meeting briefing

60

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company’s response to clarification Table 19

National Institute for Health and Care Excellence Pre-meeting briefing

61

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

62

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission pages 128-132

National Institute for Health and Care Excellence Pre-meeting briefing

63

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission Tables 51-54

Professional and social services - PSSRU Unit Costs of Health and Social Care & National Audit Office.

Healthcare professionals - NHS national schedule of reference costs & PSSRU Unit Costs of Health and Social Care.

Treatment follow up - NHS national schedule of reference costs.

Hospitalisation - NHS national schedule of reference costs, PSSRU Unit Costs of Health and Social Care, ZUMA-1 & Hospital Episode Statistics.

National Institute for Health and Care Excellence Pre-meeting briefing

64

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: ERG report pages 91-102

National Institute for Health and Care Excellence Pre-meeting briefing

65

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission table 59 & ERG report page 126

National Institute for Health and Care Excellence Pre-meeting briefing

66

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission Table 60

National Institute for Health and Care Excellence Pre-meeting briefing

67

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company response to clarification Questions B.10 & B.11

After company’s response to technical engagement clarification company acknowledges ICU stay should be 4 days rather than 1 (Table 6) 0.2% change from base case ICER.

National Institute for Health and Care Excellence Pre-meeting briefing

68

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission Figure 37

National Institute for Health and Care Excellence Pre-meeting briefing

69

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

See company submission Table 61 page 153 and response to technical engagement clarification questions page 4 for further details

National Institute for Health and Care Excellence Pre-meeting briefing

70

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company’s response to technical engagement clarification Table 2 & Company submission Table 61

NB. Scenario analysis provided by the company includes the costs of only 1 day of ICU care ~ After company’s response to technical engagement clarification company acknowledges this should be 4 days (Table 6) 0.2% change from base case ICER.

National Institute for Health and Care Excellence Pre-meeting briefing

71

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission Figure 35 & 36 ERG report Table 26, page 110

National Institute for Health and Care Excellence Pre-meeting briefing

72

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: company’s response to clarification, ERG report Table 27

National Institute for Health and Care Excellence Pre-meeting briefing

73

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Adapted from company’s response to clarification Figures 1 & 2

National Institute for Health and Care Excellence Pre-meeting briefing

74

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission table 59 & ERG report page 126

National Institute for Health and Care Excellence Pre-meeting briefing

75

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: ERG report Table 38, page 128

National Institute for Health and Care Excellence Pre-meeting briefing

76

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: ERG report Tables 33-34

National Institute for Health and Care Excellence Pre-meeting briefing

77

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: ERG report Tables 35-37

National Institute for Health and Care Excellence Pre-meeting briefing

78

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: ERG report Table 37 – All changes made to costs, QALYs do not change.

National Institute for Health and Care Excellence Pre-meeting briefing

79

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

Source: Company submission table 59 & ERG report page 126

National Institute for Health and Care Excellence Pre-meeting briefing

80

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

81

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

82

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

83

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

84

CONFIDENTIAL

==> picture [452 x 339] intentionally omitted <==

National Institute for Health and Care Excellence Pre-meeting briefing

85

NATIONAL INSTITUTE FOR HEALTH AND CARE EXCELLENCE

Single technology appraisal

Axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115]

Document B Company evidence submission

15 February 2018

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 1 of 164

1 of 164

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 template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 2 of 164

2 of 164

Contents

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 3 of 164

3 of 164

Tables

Table 1: The decision problem ................................................................................... 7 Table 2: Technology being appraised ...................................................................... 12 Table 3: Summary of outcomes for R/R aggressive B-cell NHL patients treated with current SoC .............................................................................................................. 18 Table 4: ESMO recommended treatment strategies for R/R DLBCL[18] ..................... 20 Table 5: Clinical effectiveness evidence ................................................................... 28 Table 6: Summary of the trial methodology for ZUMA-1 .......................................... 30 Table 7: ZUMA-1 outcomes presented in the submission ........................................ 33 Table 8: Baseline characteristics of patients in ZUMA-1, Phase 2 ........................... 34 Table 9: Summary of statistical analyses ................................................................. 37 Table 10: EQ-5D-3L utility scores from the ZUMA-1 safety management cohort ..... 47 Table 11: Baseline characteristics in the ZUMA-1 and SCHOLAR-1 studies ........... 54 Table 12: ORR and CR in ZUMA-1 and SCHOLAR-1 .............................................. 57 Table 13: OS in ZUMA-1 and SCHOLAR-1 .............................................................. 59 Table 14: OS by response in ZUMA-1 and SCHOLAR-1 ......................................... 61 Table 15: Standardised comparison of ORR and CR in ZUMA-1 and SCHOLAR-1 (including standardisation for ECOG status) ............................................................ 62 Table 16: Standardised comparisons of survival in ZUMA-1 and SCHOLAR-1 (including standardisation for ECOG status) ............................................................ 63 Table 17: Standardised comparison of ORR and CR in ZUMA-1 and SCHOLAR-1 (including standardisation for subsequent ASCT) .................................................... 63 Table 18: Standardised comparisons of survival in ZUMA-1 and SCHOLAR-1 (including standardisation for subsequent ASCT) .................................................... 64 Table 19: Summary of key safety events from ZUMA-1 Phase 2 (mITT population) 68 Table 20: Most frequent Grade ≥3 treatment emergent adverse events occurring in ≥10% of patients and SAEs occurring in ≥2 patients in ZUMA-1 Phase 2 (mITT population) ............................................................................................................... 68 Table 21: Change in incidence of key adverse events between the interim analysis and the primary analysis of ZUMA-1 Phase 2 .......................................................... 70 Table 22: End-of-life criteria ..................................................................................... 83 Table 23: Summary list of published cost-effectiveness studies .............................. 86 Table 24: SCHOLAR-1 data source scenarios ......................................................... 93 Table 25: Data sources of clinical parameters used in the model ............................ 99 Table 26: Overall survival for axi-cel: PSM with single parametric curves coefficients ............................................................................................................................... 100 Table 27: Overall survival for axi-cel: PSM with single parametric curves goodness of fit statistics .............................................................................................................. 101 Table 28: Overall survival for axi-cel: mixture-cure model coefficients ................... 103 Table 29: Overall survival for axi-cel: mixture cure model goodness of fit statistics 104 Table 30: Overall survival for BSC: curve fit coefficients ........................................ 107 Table 31: Overall survival for BSC: goodness of fit statistics ................................. 108 Table 32: Overall survival for BSC: mixture-cure model coefficients ...................... 110 Table 33: Overall survival for BSC: mixture cure model goodness of fit statistics .. 111 Table 34: Progression-free survival for axi-cel: curve fit coefficients ...................... 112 Table 35: Progression-free survival for axi-cel: PSM with single parametric curves goodness of fit statistics ......................................................................................... 113 Table 36: Cycles per course and days per cycle .................................................... 116

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 4 of 164

4 of 164

Table 37: Optimal combinations of vial sizes for conditioning chemotherapy by BSA ............................................................................................................................... 117 Table 38: Optimal combinations of vial sizes for BSC by BSA ............................... 117 Table 39: Incidence of Grade 3+ axi-cel-related AEs occurring in ≥10% subjects (N=101) .................................................................................................................. 119 Table 40: Incidence of Grade 3+ conditioning chemotherapy-related AEs occurring in ≥10% subjects (N=101) .......................................................................................... 119 Table 41: Characteristics and results of included utility studies ............................. 123 Table 42: Adverse event disutilities ........................................................................ 125 Table 43: Summary of utility values for cost-effectiveness analysis ....................... 126 Table 44: Unit costs of leukapheresis ..................................................................... 128 Table 45: Unit costs of conditioning chemotherapy ................................................ 128 Table 46: Malignant lymphoma non-elective long-stay HRGs ................................ 129 Table 47: Malignant lymphoma elective inpatient HRGs ........................................ 130 Table 48: Malignant lymphoma elective inpatient excess bed day HRGs .............. 131 Table 49: Unit costs of chemotherapies that are not based on BSA ...................... 132 Table 50: Unit costs of chemotherapies based on BSA ......................................... 132 Table 51: Resource use and costs associated with professional and social services ............................................................................................................................... 134 Table 52: Resource use and costs associated with healthcare professionals ........ 135 Table 53: Resource use and costs associated with treatment follow-up ................ 136 Table 54: Resource use and costs associated with hospitalisation ........................ 137 Table 55: Proportion of patients who underwent post-refractory SCT, using different SCHOLAR-1 data (scenario analyses) ................................................................... 140 Table 56: Allogeneic SCT HRGs ............................................................................ 140 Table 57: Costs of allogeneic SCT follow-up .......................................................... 141 Table 58: Summary of variables applied in the economic model base case .......... 142 Table 59: Key model assumptions ......................................................................... 143 Table 60: Base-case results without patient access scheme ................................. 146 Table 61: Scenario analysis results ........................................................................ 152 Table 62: Validation of the de novo cost-effectiveness analysis ............................ 155

Figures

Figure 1: Axi-cel anti-CD19 CAR construct and mode of action ................................. 9 Figure 2: Process of manufacturing and administering axi-cel ................................. 11 Figure 3: Clinical pathway of care for patients with R/R aggressive NHL and proposed placement for axi-cel ................................................................................ 22 Figure 4: Duration of response in the ZUMA-1 updated analysis (n=108, median follow-up 15.4 months) ............................................................................................. 40 Figure 5: Progression-free survival in the ZUMA-1 updated analysis (n=108, median follow-up 15.4 months) ............................................................................................. 42 Figure 6: Overall survival in the ZUMA-1 updated analysis (n=108, median follow-up 15.4 months) ............................................................................................................ 44 Figure 7: CAR T-cell expansion in ZUMA-1 (mITT) .................................................. 45 Figure 8: Peak Number of Anti-CD19 CAR T-cells in Blood (/μL) by Best Response (mITT) ...................................................................................................................... 46

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 5 of 164

5 of 164

Figure 9: AUC for Number of Anti-CD19 CAR T-cells in Blood (/μL) by Best Response (mITT) ..................................................................................................... 46 Figure 10: Duration of response among the 19 DLBCL/PMBCL/TFL patients in the NCI study ................................................................................................................. 50 Figure 11: Comparison of OS between ZUMA-1 (mITT, Phase 2 primary analysis) and SCHOLAR-1 ...................................................................................................... 60 Figure 12: Overall Survival ZUMA-1 versus SCHOLAR-1 (SCHOLAR-1 Survival-RR Analysis set, ZUMA-1 mITT Analysis Set) ................................................................ 65 Figure 13: PRISMA flow diagram for cost-effectiveness studies .............................. 85 Figure 14: Model structure ....................................................................................... 88 Figure 15: OS and PFS in ZUMA-1 combined Phase 1 and 2 (August 2017 data cut) ................................................................................................................................. 89 Figure 16: OS and PFS in ZUMA-1 combined Phase 1 and 2: cumulative hazard plots .......................................................................................................................... 90 Figure 17: Overall survival in BSC: comparison of SCHOLAR-1 datasets ............... 94 Figure 18: Overall survival for axi-cel: KM with single parametric curves ............... 100 Figure 19: Overall survival for axi-cel: PSM with single parametric curves logcumulative hazard plot ........................................................................................... 101 Figure 20: Overall survival for axi-cel: KM with mixture cure model parametric curves ............................................................................................................................... 104 Figure 21: Overall survival for axi-cel: mixture-cure model log-cumulative hazard plot ............................................................................................................................... 105 Figure 22: Overall survival for axi-cel: mixture-cure method .................................. 106 Figure 23: Overall survival for BSC: PSM with single parametric curves logcumulative hazard plot ........................................................................................... 107 Figure 24: Overall survival for BSC: PSM with single parametric curves logcumulative hazard plot ........................................................................................... 108 Figure 25: Overall survival for BSC: PSM with single parametric curves selected distribution .............................................................................................................. 109 Figure 26: Overall survival for BSC: KM with mixture cure model parametric curves ............................................................................................................................... 110 Figure 27: Overall survival for BSC: mixture-cure model log-cumulative hazard plot ............................................................................................................................... 111 Figure 28: Progression-free survival for axi-cel: KM with single parametric curves 112 Figure 29: Progression-free survival for axi-cel: PSM with single parametric curves log-cumulative hazard plot ..................................................................................... 113 Figure 30: Progression-free survival for axi-cel: PSM with single parametric curves selected distribution ................................................................................................ 114 Figure 31: Progression-free survival for BSC, constructed from BSC overall survival ............................................................................................................................... 115 Figure 32: PRISMA flow diagram for utility studies ................................................ 121 Figure 33: Lifetime Markov trace for axi-cel ........................................................... 147 Figure 34: Lifetime Markov trace for BSC .............................................................. 147 Figure 35: PSA scatter plot at a £50,000 threshold ................................................ 148 Figure 36: Cost-effectiveness acceptability curve .................................................. 149 Figure 37: One-way sensitivity analysis: Tornado diagram .................................... 150

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 6 of 164

6 of 164

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

Decision problem, description of the technology and clinical care pathway

B.1.1. Decision problem

The 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.

Table 1: The decision problem

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115]

© Kite/Gilead (2018). All rights reserved

7 of 164

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115]

© Kite/Gilead (2018). All rights reserved 8 of 164

B.1.2. Description of the technology being appraised

Axicabtagene ciloleucel (YESCARTA® described as axi-cel in the submission) is the first in a breakthrough class of chimeric antigen receptor (CAR) T-cell therapies, which consists of autologous human T-cells that have been engineered to express a novel cell surface receptor fragment antibody that will identify and lock onto CD19 bearing cells. The T-cell receptor complex is comprised of a single-chain variable region fragment (scFv) with specificity for CD19, that is linked to an intracellular signalling part comprised of signalling domains from CD28 and CD3ζ molecules arranged in tandem.[4] Further details of the structure of the anti-CD19 CAR construct, the innovative mechanism of action and method of administration of axi-cel are described in Figure 1 and Table 2.

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

==> picture [450 x 330] intentionally omitted <==

Key: axi-cel, axicabtagene ciloleucel; CAR, chimeric antigen receptors; scFv, single-chain variable region fragment. Source: ZUMA-1 CSR[5]

Axi-cel is currently being developed for the treatment of patients with

relapsed/refractory (R/R) diffuse large B-cell lymphoma (DLBCL), transformed

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 9 of 164

9 of 164

follicular lymphoma (TFL) and primary mediastinal large B-cell lymphoma (PMBCL), where malignant B-cells express the CD19 antigen on their surface. Although these malignancies are somewhat clinically and pathologically distinct from one another, they are treated similarly in clinical practice – with a chemotherapy regimen containing rituximab in both front line and salvage. The current prognosis for patients with relapsed/refractory disease is extremely poor where the patient is ineligible to ASCT and therefore is left with no curative options. There is a significant unmet need in this group. A recent large, pooled, database analysis reported an objective response rate (ORR) to the next line of treatment of 26% (complete response [CR] 7%) and a median overall survival (OS) of 6.3 months.[6] In the pivotal ZUMA-1 trial, axi-cel produced a previously unseen response rate of 82% (CR = 58%) with median OS not reached in this patient population, suggesting a long term freedom from disease and cure in a significant proportion of patients.[7]

Since they are derived from the patient’s own T-cells, axi-cel is a highly innovative approach that provides a complete personalised immunotherapy, that targets and eliminates CD19-expressing B-cells.[5] Axi-cel is given as a single infusion, single treatment and the timescale from collection of the patient’s white blood cells by leukapheresis, through transportation to the manufacturing facility, product manufacture, delivery to the clinical centre and conditioning chemotherapy, culminating in administration of axi-cel to the patient, is 21–24 days (Figure 2). For further details of the manufacturing process of axi-cel, see Appendix M.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 10 of 164

10 of 164

Figure 2: Process of manufacturing and administering axi-cel

==> picture [569 x 162] intentionally omitted <==

Source: Axi-cel SmPC[8] and ZUMA-1 CSR[5]

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 11 of 164

Engineered Autologous Cell Therapy is a process by which a patient’s own T-cells are collected and genetically altered to recognise and target antigens expressed on the cell surface of specific malignancies.[9] The ability to genetically engineer human T-cells to express CARs may overcome some of the main limitations of the endogenous immune system and of other cancer immunotherapies, providing a new approach to treat cancer. CARs are synthetic immunoreceptors whose extracellular domain is typically an antibody-derived scFv that recognises a tumour cell surface protein. Engineering T-cells with a CAR involves using a replication-incompetent retroviral vector containing the CAR construct to transduce T-cells. This approach has already been demonstrated in a range of studies and has opened possibilities for the treatment of patients with a wide variety of cancer types including B-cell malignancies expressing the CD19 antigen.[5]

The safety profile of axi-cel is well described, with established protocols to manage adverse events (AEs) to ensure an acceptable risk-benefit ratio for the target patient population, whose therapy options are otherwise limited to palliative. Increasing familiarity with the side effect profile of CAR-T cells in general and axi-cel in particular has meant the incidence and severity of adverse events is decreasing over time.[5]

The summary of product characteristics (SmPC) for axi-cel is provided in Appendix D.

Table 2: Technology being appraised

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 12 of 164

12 of 164

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 13 of 164

13 of 164

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 14 of 164

14 of 164

Key: ALL, acute lymphoblastic leukaemia; axi-cel, axicabtagene ciloleucel; ASCT, autologous stem cell transplant; CAR, chimeric antigen receptors; CHMP, Committee for Human Medicinal Products; CLL, chronic lymphocytic leukaemia; DLBCL, diffuse large B-cell lymphoma; EMA, European Medicines Agency; FL, follicular lymphoma; IL, interleukin; mAb, monoclonal antibody; MCL, mantle cell lymphoma; PMBCL, primary mediastinal large B-cell lymphoma; TFL, transformed follicular lymphoma.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 15 of 164

15 of 164

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

Summary

• Aggressive subtypes of B-cell NHL include DLBCL, PMBCL, and TFL. Although they have distinct clinical, pathological and molecular characteristics, the approach to management is generally similar, which consists of combination therapy with rituximab and chemotherapy.

• In the UK around 70% of newly diagnosed cases of DLBCL are cured with standard of care frontline therapy[14] , approximately 10% of patients have primary refractory disease and 20% of patients who do respond to front-line treatment will relapse.[14] Failure to achieve a good response to salvage chemotherapy indicates a poor prognosis.[15]

• Outcomes for R/R patients are generally poor, with only a small minority of patients achieving long-term survival with salvage therapies with a median overall survival of 6.3 months.[6] For relapsed or refractory (R/R) patients who are ineligible for ASCT, outcomes are even worse, with median overall survival between 3.3 and 3.9 months.[16, 17]

• Treatment options for R/R DLBCL, PPMBCL and TFL patients are limited:

  • First line R/R patients who are eligible for ASCT receive salvage chemotherapy (e.g. RICE, R-DHAP, R-GDP) followed by ASCT if chemo-responsive.

  • First line R/R patients who are ineligible for ASCT receive platinum- and/or gemcitabinebased regimens or are considered for clinical trials. Very few of these patients are long-term survivors.[18]

  • Second line or later R/R patients have extremely limited treatment options, including clinical trials with novel agents or palliative care, with allogeneic transplantation also an option for those small numbers patients who are eligible but only if they achieve good partial remission with further salvage therapy.[18]

• There are significantly high levels of unmet need among target patients for axi-cel, who have primary refractory disease or are non-responsive to salvage therapy (and therefore ineligible for ASCT).

Disease overview

Non-Hodgkin lymphoma (NHL) comprises a heterogeneous group of cancers originating primarily in B-cells (and, to a lesser extent, in T-cells and natural killer cells). The prognosis depends on the histologic type, stage, along with other factors including the patient’s age and comorbidities, tumour genetics, the amount of lactate dehydrogenase (LDH) in the blood.[19] Aggressive subtypes of B-cell NHL include DLBCL, PMBCL, and TFL. All of these express CD-19 antigen on the cell surface.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 16 of 164

16 of 164

DLBCL

DLBCL is the most common subtype of B-cell NHL, accounting for approximately 30% of newly diagnosed cases of NHL.[20] With an annual incidence of around 5.9 per 100,000 in adults, DLBCL is the most common lymphoma subtype, accounting for around 40% of the total based on the HMRN Network. Around 3,400 people are diagnosed with DLBCL each year in England and Wales.[1, 2, 21] Newly diagnosed DLBCL patients in the UK have a median age of 61 (ranging from 19 to 88 in a large UK-based RCT).[14]

PMBCL

PMBCL arises from thymic (medullary) B-cells and has distinct clinical, pathological, and molecular characteristics from other subtypes. PMBCL represents approximately 5% of patients diagnosed with NHL[22] , and each year, approximately 380 people are diagnosed with PMBCL in England/Wales.[1, 2, 21] PMBCL is typically identified in younger patients (median age 35 years) and, unlike DLBCL, has a female predominance.[23-25]

PMBCL typically presents as a large, fast-growing mass, with invasion usually limited to the anterior-upper mediastinum.[23]

TFL

Follicular lymphoma (FL) is the second most common form of NHL in Western countries, accounting for approximately 20% of patients diagnosed with NHL.[26] FL can occur at any age, but the average age at diagnosis is around 65.[27] Some patients with FL will transform to a high grade DLBCL (known as TFL), which is much more aggressive and is associated with worse outcomes than FL.[28] Histological transformation to TFL occurs at an annual rate of approximately 3%. Thus, TFL accounts for approximately 1% of all NHL cases.[29] Each year, approximately 662 people are diagnosed with TFL in the England and Wales.[1, 2, 21]

Outcomes for relapsed or refractory patients

Outcomes for R/R patients treated with standard of care (SoC) are poor, with low levels of response and limited survival. Table 3 presents a summary of reported outcomes from the evidence base for R/R patients.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 17 of 164

17 of 164

Table 3: Summary of outcomes for R/R aggressive B-cell NHL patients treated

with current SoC

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 18 of 164

18 of 164

Burden of disease – Quality of life

Limited data are available on the health-related quality of life (HRQL) burden for patients with R/R DLBCL, PMBCL and TFL ineligible for ASCT. One small, singlearm, open-label, Phase 1/2 study in 30 patients aged ≥60 years with R/R DLBCL who had received one or two prior chemotherapy regimens, but were ineligible for ASCT, compared baseline Functional Assessment of Cancer Therapy – General (FACT-G) scores at baseline against general population and cancer-specific norms.[42] It was reported that these patients had lower FACT-G scores than both the general population and cancer-specific norms (75.5 vs 80.1 and 80.9, respectively), with the comparison to the cancer-specific population reaching a pre-defined minimally important difference (MID) threshold of 5-points (which is in line with commonly accepted MID thresholds[43] ). Within the FACT-G individual components, baseline scores for emotional wellbeing, functional wellbeing and physical wellbeing were significantly worse (lower) when compared to the general population and cancer population norms. However, the social/family wellbeing score was significantly better. All of the comparisons for the individual component scores reached a pre-defined MID threshold of 2-points (which is in line with commonly accepted MID thresholds[43] ).

Given this evidence, there is likely to be a high HRQL burden for these patients.

Clinical pathway of care

Clinical guidelines

Clinical guidelines specific to R/R DBLCL, PMBCL and TFL are extremely limited.

The British Society for Haematology (BSH) guidelines state that the most commonly used conditioning regimen for ASCT is BCNU, etoposide, cytarabine and melphalan (BEAM).[3]

The NICE clinical pathway recommends the use of rituximab, gemcitabine, dexamethasone and cisplatin (R-GDP) as a salvage therapy or as a consolidation therapy with ASCT or allogeneic SCT.[44] Pixantrone monotherapy is recommended for patients with R/R aggressive NHL if they have been previously treated with rituximab and they are receiving third- or fourth-line treatment.[44, 45] However, in a

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 19 of 164

19 of 164

recent clinical ad-board, clinicians confirmed that very few patients are actually treated with pixantrone monotherapy in NHS England.[1, 2] Furthermore, BSH Guidelines do not recommend pixantrone as a treatment option for patients with R/R DLBCL.[3]

The European Society for Medical Oncology (ESMO) guidelines for treating R/R DLBCL[18] are presented in Table 4; the relevant comparators to consider for axi-cel would be those for patients who are not eligible for transplant.[18]

Table 4: ESMO recommended treatment strategies for R/R DLBCL[18]

Clinical pathway

Figure 3 presents the clinical pathway of care for patients with DLBCL, PMBCL and TFL and highlights the patients for whom axi-cel therapy would be considered. A similar figure showing the proportions of patients who progress through each section of the pathway and the proportions that clinicians expect they would treat with CAR T-cell therapy are presented in Section 3.1 of the budget impact analysis submission. The eligible population for axi-cel is estimated to consist of 972 patients in 2018.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 20 of 164

20 of 164

There are four groups of patients who would be considered eligible for axi-cel therapy:

• Patients who were refractory after first-line therapy (primary refractory)

• Patients who relapsed after first-line therapy, but were ineligible for ASCT following second-line therapy for reasons of age and comorbidities (as very small number of patients)

• Patients who relapsed after first-line therapy, and would be eligible for ASCT at second-line but who do not respond to salvage therapy

• Patients who relapsed after first-line therapy, were eligible and treated with chemotherapy and ASCT and subsequently relapse (a small number of these patients who are young may progress to allogeneic SCT)

Ineligibility for ASCT is based on a number of factors[1, 2] , including:

• Age >70 years or ≥65 with comorbidities

• Inadequate response to salvage therapy or early relapse (within 12 months) after first ASCT.

• Relapse after second or later line of therapy

• Failure to mobilise stem cells for ASCT

• Presence of significant comorbidities or unresolved toxicities

Figure 3 presents the current treatment options for patients with R/R DLBCL,

PMBCL and TFL. As confirmed by clinicians in a recent clinical ad-board, PMBCL and TFL are generally treated using regimens similar to those used for DLBCL.[1, 2]

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 21 of 164

21 of 164

Figure 3: Clinical pathway of care for patients with R/R aggressive NHL and proposed placement for axi-cel

==> picture [634 x 346] intentionally omitted <==

Key: ASCT, autologous stem cell transplantation; BEAM, BCNU, etoposide, cytarabine and melphalan; HDCT, high-dose chemotherapy; R-CHOP, rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone; R-DHAP, rituximab, cisplatin, cytarabine, dexamethasone; R-GDP, rituximab, gemcitabine, dexamethasone and cisplatin; R-ICE, rituximab, ifosfamide, carboplatin, etoposide; SoC, standard of care.

Notes: a, For second-line salvage therapy, patients may be treated with an option that they did not use for first-line salvage, i.e. one of R-DHAP, R-GDP or R- ICE; b, a small proportion of patients who relapse after ASCT may be eligible to receive allogenic SCT, and would be considered for conditioning therapy, followed by allogenic SCT if they are able to achieve a response.

Source: BSH guidelines for DLBCL treatment[3] , NICE clinical pathway for DLBCL[44] ; EMSO guidelines for treating R/R DLBCL[18]

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115]

© Kite/Gilead (2018). All rights reserved

22 of 164

The current standard of care (SoC) for first-line treatment of aggressive B-cell NHL is R-chemotherapy.[3, 18, 44, 46]

Patients who are refractory to first line therapy (primary refractory) will usually receive salvage therapy with the objective of ASCT. Outcomes in this group are extremely poor.[15] Clinicians agreed that as long as the disease was not progressing too rapidly to allow CAR T manufacture then this patient group could be eligible for axi-cel therapy (Figure 3: Clinical pathway of care for patients with R/R aggressive NHL and proposed placement for axi-cel).[1, 2] Clinicians also stated that given the poor response of these patients to subsequent lines of therapy, and their limited expected survival, early CAR T treatment should be considered instead of follow-on chemotherapy, and care should be taken to ensure that any chemotherapy used is not impactful to future CAR T treatment (such as min-BEAM).[2]

For patients who relapse after first-line therapy and are deemed ineligible for ASCT, the only current treatment options are either platinum- and/gemcitabine-based regimens, or to be entered into a clinical trial (Figure 3).[18] There may be a small group of patients who are not eligible for ASCT whose comorbidities preclude ASCT but not axi-cel therapy.

Apart from the groups described in the two paragraphs above, other patients that can be considered for axi-cel are:

• Chemo-responsive patients who proceed to ASCT but then relapse

• Those who do not respond to chemotherapy, and therefore cannot proceed to ASCT

Limitations with current treatment and unmet need

For newly diagnosed DLBCL, PMBCL and TFL patients, rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone (R-CHOP) is SoC.[25, 47] A large prospective database study demonstrated that 71% of patients with newly diagnosed DLBCL remained event-free 2 years after diagnosis, when treated with SoC R-CHOP.[48] Furthermore, the OS in these patients was equivalent to that of the age- and sex-matched general population.[48] Similarly, a large UK randomised controlled study (RCT) confirmed the potential for positive outcomes in first-line

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 23 of 164

23 of 164

treatment with R-CHOP, with 75% of patients achieving 2-year progression free survival (PFS) and 2-year OS rates of up to 83%.[14] Therefore, for these patients, remaining event-free (in ongoing CR) for 2-years is consistent with long term disease free survival or cure.

In DLBCL, PMBCL and TFL around 10% of patients will not respond to treatment (primary refractory).[14] Outcomes in patients with refractory disease are poor with a median OS of less than 6 months. For primary refractory patients, published ORRs to second-line chemotherapy range from 0 to 23% (Table 3)[30-35] , and primary refractory disease was found to be a significant risk factor for failing response to second-line therapy.[36] Furthermore, most of these patients are not eligible for transplant due to their chemotherapy-resistant disease. Published ORR for patients refractory to second- or third-line therapy were 18% and 14%, respectively (Table 3).[32, 37] Similarly, a recent pooled database analysis (SCHOLAR-1) demonstrated that only 2.9% of primary refractory patients and 10% of patients refractory to second-line or later therapy achieved CR (ORR 20.2% and 26.1%, respectively) with SoC (Table 3).[6] Furthermore, primary refectory and refractory to second-line or later therapy patients had median OS of 7.1 months and 6.1 months, respectively (Table 3).[6] In an analysis of SCHOLAR-1 versus ZUMA-1, standardised for subsequent ASCT, which is likely to be more aligned to the relevant patient population for axi-cel, median OS was only 3.9 months (see Section B.2.9).

For DLBCL, PMBCL and TFL who respond to first-line SoC, around 20% of all patients will relapse.[14] Results of the PARMA trial[30] demonstrated superior outcomes for second-line chemotherapy plus autologous stem cell transplant (ASCT) compared with second-line chemotherapy alone in patients with relapsed DLBCL, leading to the adoption of second-line chemotherapy plus ASCT as the SoC for the relapsed population.[49] However, although ASCT is a treatment option for patients with relapsed disease, studies in relapsed B-cell NHL indicate that only half of patients who respond to second-line therapy and are then able to proceed to ASCT.[30, 36, 39, 50] Of those half will subsequently relapse and not be cured. So only 25% have a potential for cure.[15]

Relapse after ASCT, has a poor prognosis in patients with DLBCL. In those relapsing in less than 12 months it is particularly poor. Similarly, in the Collaborative

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 24 of 164

24 of 164

Trial in Relapsed Aggressive Lymphoma (CORAL), 4-year event-free survival (EFS) for patients who had early relapse after ASCT was significantly lower than that of patients who relapsed more than 12 months after ASCT (48% vs 56%; p<0.05; Table 3).[39] In a recent analysis of the CORAL data, the median OS was significantly shorter among patients who relapsed <6 months after ASCT (5.7 months) compared with those relapsing � 12 months after ASCT (12.6 months, P=0.0221; Table 3). The SCHOLAR-1 database study demonstrated that 14.7% of patients with early relapse after ASCT achieved CR and median OS of 6.2 months with SoC (Table 3).[6] In an analysis of data from the PARMA trial, the ORR to subsequent therapy was 40% for those with an early relapse and 69% for those with relapse more than 12 months after ASCT (p<0.0001), and 8-year OS rates were 13% and 29% (p<0.0001) for the two subsets (Table 3).[38] Similarly, in the Similar results are found in other studies.[40] Patients with R/R DLBCL ineligible for transplant have a poor prognosis with a median OS ranging from 4 to 13 months for patients with transplant-ineligible R/R DLBCL.[51] In the CORAL trial, 129 of 193 patients who received third-line therapy, but did not undergo subsequent transplant, had worse survival than patients who underwent ASCT or allogeneic SCT (34/193 patients): median survival was 3.3 months vs 11.1 months, respectively, and 2-year OS was 9.3% vs 33.9%, respectively.[17] Finally, a large Canadian database study demonstrated an OS of 3.9 months in R/R disease patients who are ineligible for ASCT.[16] In the recent SCHOLAR-1 database study, patients who were ineligible for ASCT had OS of 5.1 months. For SCHOLAR-1 patients matched to ZUMA-1 trial patients for comparison, median OS was even lower at 3.9 months.

Therefore, there is an obvious unmet need in these patients:

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 25 of 164

25 of 164

• Patients who were refractory after first-line therapy (primary refractory)

• Patients who relapsed after first-line therapy, but were ineligible for ASCT following second-line therapy for reasons of age and comorbidities

• Patients who relapsed after first-line therapy, and would be eligible for ASCT at second-line but who do not respond to salvage therapy

• Patients who relapsed after first-line therapy, were eligible and treated with chemotherapy and ASCT and subsequently relapse

B.1.4. Equality considerations

There is the potential for treatment in NHL to raise some issues of equality. The odds ratios for developing DLBCL for men compared with women is 1.7.[52] Treatment effects also seem to be influenced by gender, with outcomes improving by a greater amount for women with the introduction of rituximab, i.e. moving from first-line CHOP to R-CHOP as the SoC (from 68% to 84% and from 64% to 77%, respectively), and a higher median OS for women (90.6 months compared to 55 months; hazard ratio [HR]: 1.2, p=0.0681).[53] Therefore, women generally have better outcomes than men, with median PFS of 90.6 months compared to 55 months (HR: 1.2; p=0.02).[54] This difference was most significant in patients over 60 years of age.[54] Therefore, as more men than women are likely to develop NHL and their treatment outcomes are expected to be worse, there are likely to be a greater number of men progressing to the R/R setting, where their prognosis will be even worse. In addition to this, older patients are likely to be considered ineligible for stem cell transplant (age ≥70 is an ineligibility criteria) due to the burden of salvage chemotherapy and would also be less likely to be able to receive more aggressive chemotherapy options, either to achieve a response, or to achieve response in order to receive a stem cell transplant. Conditioning chemotherapy for CAR-T is less burdensome than salvage therapy, making axi-cel an acceptable alternative for this group of patients. The subgroup analyses from ZUMA-1 (see Appendix E) also showed consistent results for patients by both age and gender, which would help to reduce some of the equality associated with SoC for patients with DLBCL, PMBCL and TFL.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 26 of 164

26 of 164

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

Clinical effectiveness

Axi-cel provides a potentially curative therapy options for patients with otherwise limited treatment options and high unmet need (median follow-up 15.4 months; N=108)

• The OS rate at 18-months was 52% and PFS rate at 15-months was 41%, with few events occurring towards the end of the KM curves. The curves are stable at this point suggesting patients may be cured

Axi-cel demonstrated significant clinical benefits for these patients (median follow-up 15.4 months; N=108)

• Response was durable, with 42% remaining in response at the data-cut off (median follow-up 15.4 months), including 40% with CR

• Median OS has not yet been reached (95% CI: 12.0, not reached), with OS rates of 78%, 59% and 52% at 6-, 12- and 18-months, respectively

  • This suggests that, if current survival trends continue, then median OS could be greater than 18-months

• Overall median PFS was 5.8 months (95% CI: 3.3, not reached) but in patients achieving a CR median PFS has not been reached suggesting many of these patients will have long term disease free survival and potential cure

Axi-cel demonstrated significant improvements compared to SoC

• A Cox model of survival indicated a xxxxxxxxxxxxxx of risk of death with axi-cel (xxxxxxxx)

• Odds ratios for ORR and CR were xxxxxxxxxxxxx, demonstrating significant improvements for axi-cel (xxxxxxxx)

AEs associated with CAR T are manageable with existing safety protocols and the majority

of events resolved within a month of axi-cel infusion (median follow-up 8.7 months; N=101)

• xxx of patients experienced CRS, but only xxx experienced ≥ grade 3. There were xxxxxxxxxxx but all Grade ≤3 CRS events xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx xxxxxxxxx

• xxx of patients experienced neurological events, however the majority were mild. Xxx xxxxxx xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx xxxxxxxxxxxxxxxxxxxxx

• Grade 4 neutropenia, thrombocytopenia, and anaemia occurred in xxxxxxxxxxxx of patients, respectively. xxxxxxxxxxxxxxxxxxxxxxxxx

Increased clinician experience led to reductions in AE rates over the duration of the trial

• Between the interim and primary analyses Grade ≥ 3 AEs xxxxxxxxxxxxxxxxxxxxxxxx, CRS xxxxxxxxxxxxxxxxxxxxxxxxxx, neurological events xxxxxxxxxxxxxxxxxxxxxxx and there were xxxxxxxxxx

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 27 of 164

27 of 164

B.2.1. Identification and selection of relevant studies

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

B.2.2. List of relevant clinical effectiveness evidence

Evidence in support of axi-cel comes from the following sources:

• ZUMA-1 study

  • ZUMA-1, combined Phase 1 and 2 data (median follow-up 15.4 months)[7]

  • ZUMA-1, Phase 2 data (median follow-up 8.7 months)[5, 7]

  • ZUMA-1, Phase 1 data[55]

• NCI-09-C-0082: NCI proof-of-concept study[56]

• National Cancer Institute (NCI) preliminary dose-finding study[9, 57]

The pivotal, regulatory evidence to support axi-cel is the ongoing, single-arm, Phase 1/2 study, ZUMA-1, and this study is the focus of this submission. A summary of the ZUMA-1 study is presented in Table 5.

Table 5: Clinical effectiveness evidence

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 28 of 164

28 of 164

NCI-09-C-0082 (a National Cancer Institute [NCI] proof-of-concept study)[56] and an NCI preliminary dose-finding study[9, 57] were not used to populate the economic model but are included in section B.2.6. The results of these studies provide longerterm evidence to support axi-cel. NCI-09-C-0082 and the NCI preliminary dosefinding study were not included in the economic model because evidence was used from the pivotal, regulatory ZUMA-1 study, including combined data from the 108 Phase 1 and 2 patients.

A clinical SLR was performed, in line with NICE guidance, in order to identify evidence relevant to this submission. Full details on the methods of the SLR are presented in Appendix D. Due to the large amounts of heterogeneity between the comparator studies identified, and the limited evidence available in a comparable population to the ZUMA-1 trial and the anticipated axi-cel licence (the majority of the comparator studies included patients with R/R disease after first-line treatment, compared to the heavily pre-treated patient population in ZUMA-1), it is extremely difficult to make any valid comparisons between these studies and ZUMA-1. Therefore, it was considered more appropriate to use studies for which patient-level data were available to inform a historical comparator study; SCHOLAR-1. There were still differences between SCHOLAR-1 and ZUMA-1 (including a more severe, more heavily pre-treated population in ZUMA-1 and higher use of ASCT in SCHOLAR-1, which may bias outcomes against axi-cel), but the availability of

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 29 of 164

29 of 164

patient-level data meant that comparisons could be performed that would attempt to account for some of these differences in the patient populations and allow for a more appropriate comparison. Details on the studies identified in the clinical SLR,

including methods, baseline characteristics and outcomes, all compared to ZUMA-1, are presented in Appendix D. The results of the comparison of ZUMA-1 to SCHOLAR-1, including further discussion on the limitations of this analysis, is presented in Section B.2.9.

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

ZUMA-1

A summary of the methodology used in the Phase 1/2 clinical trial, ZUMA-1, is presented in Table 6.

Table 6: Summary of the trial methodology for ZUMA-1

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 30 of 164

30 of 164

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 31 of 164

31 of 164

Table 7 presents the outcomes of the ZUMA-1 trial that are presented in the

submission by data-cut and whether they were used to inform the economic model.

Due to the differences in analysis timepoints, and the patients included in these

analyses (i.e. Phase 2 alone or Phase 1 and 2 combined), it is important to be clear Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 32 of 164

32 of 164

which data are being used. The focus of the submission is on the updated analysis, when all 108 Phase 1 and 2 patients had been followed up for at least 12 months (median follow-up 15.4 months), where these data are available, as this provides longer-term evidence in support of axi-cel.

Table 7: ZUMA-1 outcomes presented in the submission

Baseline characteristics

Baseline characteristics for patients in ZUMA-1 Phase 2 are summarised in Table 8. The median age of patients in the ZUMA-1 trial was 58, with 24% of patients aged ≥65, which is similar to the median age of 61 for these patients in clinical practice in the UK.[14] UK clinical experts agreed that the patients treated in ZUMA-1 were likely to be reflective of the UK patients who would be considered for treatment with CAR T

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 33 of 164

33 of 164

therapy and they believed that the ZUMA-1 trial population overall was generally reflective of patients who would be seen in clinical practice.[1, 2]

It is also important to note that ZUMA-1 patients were a heavily pre-treated patient population (69% had received ≥3 prior treatments and 40% of patients received ≥4 prior treatment), which suggests that patients have received all standard available therapies, none of which had been effective, and are likely to have limited, palliative options remaining to them.

Table 8: Baseline characteristics of patients in ZUMA-1, Phase 2

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 34 of 164

34 of 164

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

For Phase 2 of the ZUMA-1 study, Cohort 1 (DLBCL patients) and Cohort 2 (PMBCL and TFL patients) were designed to differentiate between a treatment that has a true response rate of 20% or less and a treatment with a true response rate of 40% or more. The hypothesis is that the ORR for patients treated with axi-cel in Cohorts 1 and 2 is significantly greater than 20%. The pre-specified 20% control response rate was based on a review of published outcome data for patients with refractory

DLBCL, defined as those who either never responded (i.e. progressive disease [PD] or stable disease [SD] as best response to the last line of therapy) or relapsed within 12 months after ASCT.

The full analysis population included all enrolled patients (N = 111). The modified intention-to-treat (mITT) population included all patients treated with at least 1.0 x 10[6] anti-CD19 CAR T-cells/kg (N = 101) and was the analysis population used for all efficacy analyses in ZUMA-1 Phase 2. The safety analysis population included all patients treated with any dose of axi-cel (N = 101).

Two pre-specified interim analyses (IA1 and IA2) and one primary analysis were planned. IA1 was a futility analysis conducted when 20 patients in the mITT set of Cohort 1 had the opportunity to be assessed for response at the 3-month disease assessment. IA2 was to be conducted when 50 patients in the mITT set of Cohort 1 had the opportunity to be assessed for response at the 3-month disease

assessment. The nominal alpha level used for the assessment of efficacy in IA2 was 0.017. The primary analysis of Cohorts 1 and 2 combined was to be conducted when 72 patients in the mITT set of Cohort 1 and 20 patients in the mITT set of Cohort 2 had had the opportunity to be assessed for response at the 6-month disease

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 35 of 164

35 of 164

assessment. Both the IA1 and IA2 analyses have been completed. Cohort 1 met the primary endpoint at IA2. Therefore, in the primary analysis, which is the focus of this dossier, the inferential testing is presented only for Cohorts 1 and 2 combined. Nine additional patients had been enrolled at the time when the inferential analysis was met; all were treated with axi-cel and are included in the mITT assessment of efficacy.

The pre-specified primary analysis of the primary endpoint compared ORR for the 92 patients in the mITT (inferential) analysis set to the prespecified rate of 20% using a 1-sided exact binomial test. The nominal 1-sided alpha used to test for efficacy in this combined set was 0.0075. This analysis used the investigator’s assessment of response according to International Working Group (IWG) 2007 criteria.[58] The ORR was also analysed using response based on a central review. Duration of response (DoR), PFS, and OS were analysed using the Kaplan–Meier (KM) method.

A summary of the statistical analyses used in the ZUMA-1 Phase 2 study are presented in Table 9.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 36 of 164

36 of 164

Table 9: Summary of statistical analyses

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115]

© Kite/Gilead (2018). All rights reserved

37 of 164

See Appendix D for the number of participants eligible to enter the study and the CONSORT flow diagram for ZUMA-1.

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

ZUMA-1 was considered to be a good quality study and was conducted according to Good Clinical Practices (GCP). See Appendix D for full details of the quality assessment for ZUMA-1.

B.2.6. Clinical effectiveness results of the relevant trials

The primary analysis of ZUMA-1 was conducted at the point when the pre-specified 92 patients from the Phase 2 portion of the study could be evaluated 6 months after axi-cel infusion, which resulted in a median follow-up was 8.7 months.[5]

To evaluate the durability of response with axi-cel, an updated analysis was performed when the 108 patients in the Phase 1 and 2 portions of ZUMA-1 had been followed for a minimum of 1 year, which resulted in a median follow-up of 15.4 months.[7]

The focus of this submission is the updated analysis (median follow-up 15.4 months), where available, as this provides more long-term data and is consistent with the data used to inform the economic model. Relevant pharmacokinetic and pharmacodynamic outcomes are presented using the primary analysis results (median follow-up 8.7 months), as these outcomes were not updated for the latest data-cut. The results of the Phase 2 primary analyses are presented in Appendix L and were generally consistent with the updated results.

ZUMA-1, updated analysis (N = 108; median follow-up 15.4 months)

Response and duration of response

In the updated analysis of ZUMA-1 Phase 1 and 2 patients (N = 108; median followup 15.4 months), the ORR was 82%. At the data cut-off (median follow-up 15.4 months), 42% remained in response, including 40% with a CR. Of the 7 patients in Phase 1 of the study, 3 had an ongoing CR at 24 months. Of the patients who did

not have a CR at the time of the first tumour assessment (1 month after the infusion

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 38 of 164

38 of 164

of axi-cel), 23 patients (11 of 35 with a PR and 12 of 25 with SD) subsequently had CR in the absence of additional therapies as late as 15 months after treatment.

ORRs response rates were consistent across key covariates, including the use of tocilizumab or glucocorticoids. Figure 4 presents the KM curve for the DoR. The median DoR was 11.1 months (95% CI, 3.9 to could not be estimated). The figure shows that a higher proportion of patients who achieved CR were able to remain in response long term. There is a long tail on the KM curve after 5 months, with few events occurring and no new events after 11-12 months. The majority of patients who only achieve PR lose this response very quickly (within the first couple of months), however, the KM curve is again flat after 2-months with few events and around 10% of patients able to maintain response. As ORR includes both CR and PR patients it is a combination of the two, but again it is important to note the long, flat tail on the KM curve. This suggests that axi-cel can be seen as a curative therapy option, especially for patients who can achieve CR.

Only '''''''''''' patients who had a response subsequently received an allogenic SCT, which emphasises that axi-cel should be considered as a definitive therapy aiming to provide patients with a cure and not a bridge to subsequent therapy. However, receipt of axi-cel does not preclude patients who may have been ineligible for SCT to start with from subsequently receiving SCT when responding to axi-cel therapy.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 39 of 164

39 of 164

Figure 4: Duration of response in the ZUMA-1 updated analysis (n=108, median follow-up 15.4 months)

==> picture [700 x 293] intentionally omitted <==

Key: CI, confidence interval; mo, months; NE, could not be estimated; NR, not reached. Source: Neelapu et al., 2017[7]

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 40 of 164

Progression-free survival

Figure 5 presents the KM curve for PFS. The median duration of PFS was 5.8 months (95% CI, 3.3, not reached), with PFS rates of 49% (95% CI, 39 to 58) at 6 months, 44% (95% CI, 34 to 53) at 12 months, and 41% (95% CI, 31 to 50) at 15 months. The KM curve for PFS also has a long tail from around 5 to 6 months, following an initial drop, again suggesting the potential for cure, with few patients experiencing progression after they have remained progression-free for this initial 6- month period.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 41 of 164

41 of 164

Figure 5: Progression-free survival in the ZUMA-1 updated analysis (n=108, median follow-up 15.4 months)

==> picture [700 x 258] intentionally omitted <==

Key: CI, confidence interval; mo, months; NE, could not be estimated. Source: Neelapu et al., 2017[7]

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 42 of 164

Overall survival

Figure 6 presents the KM curve for OS. The median OS was not yet reached (95% CI, 12.0 months, not reached), with OS rates of 78% (95% CI, 69 to 85) at 6 months, 59% (95% CI, 49 to 68) at 12 months, and 52% (95% CI, 41 to 62) at 18 months. A total of 56% of patients were still alive at the time of the data cut-off (median followup 15.4 months). Again, the KM curve has a long tail, with few events occurring after 10 months. This supports the narrative from the DoR and PFS results with patients who are able to achieve and maintain CR considered to be cured, with few events occurring in those patients.

There are two groups of patients seen with axi-cel therapy, those who respond to therapy and are able to maintain this response and survival long-term and can be considered cured, and those who do not respond and continue to progress. This explains the flattening of the KM curves, as those patients who are not able to achieve CR drop out, while those who are able to achieve CR maintain their response and can be considered cured.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 43 of 164

43 of 164

Figure 6: Overall survival in the ZUMA-1 updated analysis (n=108, median follow-up 15.4 months)

==> picture [700 x 264] intentionally omitted <==

Key: CI, confidence interval; mo, months; NE, could not be estimated; NR, not reached. Source: Neelapu et al., 2017[7]

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 44 of 164

ZUMA-1, primary analysis (median follow-up 8.7 months)

CAR T-cell levels

Following the administration of the CAR-T cells to the patient the cells rapidly multiple (“the expansion phase”), peaking in the circulation within 1-2 weeks after administration before declining slowly thereafter (see Figure 7). Expansion was significantly associated with response to axi-cel, with an area under the curve (AUC) within the first 28 days xxx times as high in patients who had a response compared to those who did not. The correlation between peak and AUC T-cell numbers and response appears in Figure 8 and Figure 9. Most patients at 180 days post infusion, and xxx patients with ongoing CRs at 24 months, still had detectable circulating CAR-T cells. Persistence of anti-CD19 CAR-Ts would ensure ongoing destruction of malignant CD19+ cells and ongoing CR and remission. However, persistence of CAR-T cells in the circulation may not be required for long term remission as in the longer term NCI follow up ongoing remission could be present in the absence of detectable CAR-T cells in the blood and recovery of non-malignant CD19+ B cells.[57] Presumably in this scenario the malignant clones are eradicated completely leading to long term remission.

Figure 7: CAR T-cell expansion in ZUMA-1 (mITT)

==> picture [452 x 219] intentionally omitted <==

Key: CAR, chimeric antigen receptor; mITT, modified intend-to-treat Notes: Serial blood samples were analyzed for CAR T-cell levels and serum biomarkers in all 101 patients who were treated with axi-cel. Figure shows CAR T-cell expansion and persistence with median values and interquartile ranges (Q1 and Q3). Source: Neelapu et al, 2017[7]

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 45 of 164

45 of 164

Figure 8: Peak Number of Anti-CD19 CAR T-cells in Blood (/μL) by Best Response (mITT)

==> picture [453 x 109] intentionally omitted <==

==> picture [453 x 109] intentionally omitted <==

Key: CAR, chimeric antigen receptor; CR, complete response; mITT, modified intend-to-treat; PR, partial response.

Notes: To apply the log scale on y-axis, zero values were adjusted by adding 0.001. Peak is defined as the maximum number of CAR T measured post infusion. Diamonds represent mean values; circles represent the outliers

Figure 9: AUC for Number of Anti-CD19 CAR T-cells in Blood (/μL) by Best

Response (mITT)

==> picture [453 x 111] intentionally omitted <==

==> picture [453 x 110] intentionally omitted <==

Key: CAR, chimeric antigen receptor; CR, complete response; mITT, modified intend-to-treat; PR, partial response.

Notes: To apply the log scale on y-axis, zero values were adjusted by adding 0.001. Area under curve (AUC) is defined as the area under curve in a plot of number of CAR T-cells against scheduled visit from Day 0 to Day 28. Diamonds represent mean values; circles represent the outliers. Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 46 of 164

46 of 164

ZUMA-1, safety management cohort

A safety management cohort of 34 patients was studied to examine the impact of pre-emptive safety management. This cohort also captured HRQL data for these patients using the EQ-5D-5L at screening, Week 4, Month 3 and Month 6 post axicel infusion, as well as results by response category and for progression-free and progressed patients. A crosswalk algorithm was used to convert EQ-5D-5L to EQ5D-3L (as preferred by NICE) and then a UK valuation algorithm was applied to convert EQ-5D-3L descriptive scores to the EQ-5D-3L index with UK populationbased health utility values. This is discussed as part of the economic analysis in Section B.3.4.

Health-related quality of life

Patients experienced xxx xxx xxx in utility scores from screening (xxx) to Week 4 (xxxx), most likely because of a disutility associated with the timing of the transient toxicities associated with CAR T therapy (see Section B.2.10). By Month 3 and Month 6, the patient utilities had xxx xxx xxx xxx xxx xxx xxx xxx xxx xxx xxx xxx xxx

xxx xxx xxx, respectively), showing that patient HRQL is improved by axi-cel therapy. This is particularly evident when the results are broken down by response category and by health state, with patients in response experiencing xxx xxx xxx xxx xxx xxx

xxx xxx xxx xxx xxx xxx xxx xxx xxx, respectively) than patients who have not responded to treatment (xxxxx and xxxxx, respectively) and patients with

progression-free disease experiencing xxx xxx xxx xxx (xxxx) than patients with progressed disease (xxxxx).

Table 10: EQ-5D-3L utility scores from the ZUMA-1 safety management cohort

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 47 of 164

47 of 164

Supporting evidence for axi-cel (including longer-term data)

In addition to the pivotal Phase 2 data from ZUMA-1, supporting evidence for axi-cel is available from the following three studies:

• ZUMA-1, Phase 1 data[55]

• NCI-09-C-0082: NCI proof-of-concept study[56]

• National Cancer Institute (NCI) preliminary dose-finding study[9, 57]

ZUMA-1, Phase 1[55]

ZUMA-1, Phase 1: Methodology

As described in Section B.2.3, ZUMA-1 was a Phase 1/2, multicentre, open-label, single-arm study. The primary objective of Phase 1 was to evaluate the safety of axicel regimens in seven patients with refractory DLBCL, PMBCL, and TFL. Methods, inclusion/exclusion criteria and treatment were the same as described for Phase 2 in Section B.2.3, and the primary endpoint was the incidence of dose limiting toxicities.

ZUMA-1, Phase 1: Patient characteristics

Seven patients in the ZUMA-1 Phase 1 study were treated with axi-cel. Patients had a median age of 59 (range: 29 to 69), with 3 patients (43%) ≥65 years. 71% of patients were male and all 7 patients had DLBCL. Four patients had received prior ASCT, and six patients had received 3 or more prior therapies.

ZUMA-1, Phase 1: Efficacy results

Five of seven (71%) patients achieved an objective response within 1 month of axicel infusion, with four of seven (57%) achieving a CR. Three patients (43%) were in ongoing CR at 12 months post-infusion. All three patients with ongoing CR had previously relapsed within 5.8 months of ASCT.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 48 of 164

48 of 164

NCI proof-of-concept study[56]

NCI study: Methodology

The NCI study (NCI-09-C-0082) was a single-arm, open-label study. Twenty-two patients received a single dose of axi-cel 2 days after a low-dose chemotherapy conditioning regimen of cyclophosphamide plus fludarabine.

The primary study objective was to determine the safety and feasibility of the administration of the axi-cel cryopreserved anti-CD19 CAR T-cells following a nonmyeloablative chemotherapy regimen in participants with B-cell lymphoma. The secondary objective was to determine the in vivo survival of the axi-cel anti-CD19 CAR T-cells, and to determine if the treatment regimen could cause regression of B- cell malignancies.

NCI study: Patient characteristics

In the NCI study, 22 participants with advanced NHL received axi-cel preceded by low-dose chemotherapy. Nineteen patients had DLBCL/PMBCL/TFL (13 patients were “DLBCL, not otherwise specified”; 2 patients were “PMBCL”; 3 patients were “DLBCL, transformed from FL”; 1 patient was “DLBCL transformed from CLL”), two patients had follicular lymphoma (FL), and one patient had mantle cell lymphoma (MCL).

Eleven of 19 patients with DLBCL/PMBCL/TFL had chemotherapy-refractory lymphoma. Five other patients with DLBCL/PMBCL/TFL had lymphoma that had relapsed 10 months or less after ASCT as their last treatment prior to protocol enrolment. Eleven patients with DLBCL/PMBCL/TFL were high risk by second-line, according to the age-adjusted International Prognostic Index (sAAIPI). The median number of unique lymphoma therapies received before protocol enrolment was four (range one to seven).

NCI study: Efficacy results

The duration of response among the 19 patients with DLBCL/PMBCL/TFL is presented in Figure 10. Nine (47%) patients achieved CR, and four (21%) patients had PR, giving an ORR of 68%. In the nine (47%) patients who achieved CR, this was ongoing 7+ to 24+ months after axi-cel infusion (green bars). Of the remainder,

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 49 of 164

49 of 164

4 (21%) achieved PR of 1–14 months’ duration (blue bars), 2 (11%) had stable disease of 2 and 3 months’ duration (red bars), and 4 (21%) had progressive disease (PD).

Figure 10: Duration of response among the 19 DLBCL/PMBCL/TFL patients in

the NCI study

==> picture [325 x 307] intentionally omitted <==

Key: +, CR is ongoing; CLL, chronic lymphocytic leukaemia; CR, complete response; DLBCL, diffuse large B-cell lymphoma; FL, follicular lymphoma; PD, progressive disease; PMBCL, primary mediastinal B-cell lymphoma; PR, partial response; SD, stable disease; TFL, transformed follicular lymphoma.

Notes: Green bars, ongoing CR; blue bars, PR; red bars, SD. Source: Adapted from data presented in the NCI study[56]

Preliminary dose-finding study[9, 57]

Dose-finding study: Methodology and patient characteristics

A dose-finding study conducted at the NCI included seven refractory DLBCL patients

plus two refractory PMBCL patients. All patients received a single dose of axi-cel 2 days after a low-dose chemotherapy conditioning regimen of cyclophosphamide plus

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 50 of 164

50 of 164

fludarabine. It is worth noting that this conditioning chemotherapy dose was lower than that currently recommended in the draft SmPC.

Dose-finding study: Efficacy results

Four (44%) of these patients (two DLBCL, two PMBCL) had achieved complete remission during the 9-month period following axi-cel infusion.

Moreover, long-term follow-up of seven of the original nine DLBCL/PMBCL patients treated in the dose-finding study demonstrated that four (44%) patients (two DLBCL, two PMBCL) were in ongoing complete remission between 38+ and 56+ months after infusion of axi-cel. Out of these four patients with ongoing CR, recovery of normal B-cells was observed in three patients.

B.2.7. Subgroup analysis

In the ZUMA-1 study, pre-specified subgroup analyses were performed using key baseline and prognostic factors for ORR and the PFS rate at 6 months using the primary analysis data (N=101, median follow-up: 8.7 months) (the updated 12-month analysis was conducted to focus on OS and DoR and did not include an update of subgroup analyses), based on the following factors:

• ECOG PS (0 vs 1)

• Age (<65 vs ≥65 years)

• Disease type (DLBCL vs PMBCL vs TFL)

• Refractory to first-line therapy, i.e. primary refractory (yes vs no)

• Refractory to ≥2 lines of therapy (yes vs no)

• Number of prior chemotherapies (1 vs 2–3 vs ≥4)

• History of bone marrow involvement (yes vs no)

• Tumour burden (≤median vs >median)

• Sex (male vs female)

• Race (White vs Asian vs other)

• CD19 at baseline (positive vs negative)

• Refractory subgroup (refractory to ≥second-line therapy vs relapse post ASCT)

• Disease stage (I–II vs III–IV)

• IPI risk (0–2 vs 3–4)

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 51 of 164

51 of 164

• CD19 H-score (≤150 vs >150)

• CD4/CD8 ratio (>1 vs ≤1)

• Steroid use (yes vs no)

• Tocilizumab use (yes vs no)

However, the study was not designed to distinguish between these patient groups, and therefore, all tests are descriptive. A summary of the results of the subgroups analyses are presented in Appendix E. Both the ORR and PFS rate at 6 months were consistent across subgroups, with similar efficacy across all NHL prognostic factors. Of particular interest to NICE in the decision problem meeting was the potential difference between the different disease groups (DLBCL, PMBCL and TFL) and the difference between the primary refractory and relapsed populations. Both ORR and PFS rates at 6 months were consistent across all of these subgroups of patients, and all showed significant results for patients treated with axi-cel.

B.2.8. Meta-analysis

The main evidence for axi-cel came from one single-arm Phase 1/2 study, supported by data from two small single-arm, dose-finding/proof-of-concept studies, and a patient level historical control study. The patient level historical control study SCHOLAR-1 is used in place of a literature review/meta-analysis, and given the heterogeneity between the patient populations for relevant comparator treatments (where the majority of patients have received only one prior line of therapy), the availability of patient-level data to account for differences between patient characteristics and key prognostic factors is considered to be more rigorous and allows a more appropriate comparison.

B.2.9. Indirect and mixed treatment comparisons

Full details on the methods of the SLR are presented in Appendix D. Due to the large amounts of heterogeneity between the studies identified in the SLR and the ZUMA-1 study, which included much more heavily pre-treated patients compared to the majority of the SLR studies which were mostly patients after first-line treatment, direct comparison between these studies was not considered appropriate. Instead, the SCHOLAR-1 study was conducted using data from four sources for which patient-level data were available: MD Anderson Cancer Centre (MDACC) database;

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 52 of 164

52 of 164

Mayo Clinic and University of Iowa (MC/IA) Specialised Program of Research Excellence (SPORE) database; the National Cancer Institute of Canada (NCIC) Cancer Trials Group (CTG) randomised Phase 3 study LY.12; and the French Lymphoma Academic Research Organisation (LYSARC) randomised phase 3 Collaborative Trial in Relapsed Aggressive Lymphoma (CORAL) study. This would allow patients to be included that more closely matched the patient population of ZUMA-1 and would allow for adjustment to be made to account for any differences between patients and therefore allow for a more appropriate comparison.

Full details on the methodology of the SCHOLAR-1 study and the comparison to ZUMA-1 are presented in Appendix D. The analysis uses the updated, 12-month data from ZUMA-1 for the patients from the Phase 2 part of the study (N = 101).

ZUMA-1 and SCHOLAR-1 demographic and baseline/disease characteristics Demographic and disease characteristics among subjects in the SCHOLAR-1 evaluable set are presented in Table 11. Most patients had ECOG performance status ≤ 1 and Stage III–IV disease; ZUMA-1 did not include any ECOG ≥2 patients, but it did have a higher proportion of patients with Stage III-IV disease. Approximately one-third of evaluable patients in SCHOLAR-1 had high-intermediate to high-risk IPI risk classification, compared to almost half in ZUMA-1. ZUMA-1 also contained a higher proportion of patients with PMBCL and TFL. Overall, around 20% of patients had relapsed ≤ 12 months following ASCT. More patients in SCHOLAR-1 had a history of being primary refractory, but slightly more patients in ZUMA-1 had a history of being refractory to two consecutive lines of therapy. It is also important to note that around '''''''''' of patients in SCHOLAR-1 went on to receive ASCT after determination of refractory status (compared to only '''''''''''' in the ZUMA-1 trial). This needs to be taken into account when considering the SCHOLAR-1 results, as this could lead to an overestimation of survival outcomes for these patients. In addition, patients in the ZUMA-1 trial were more aggressively pre-treated compared to SCHOLAR-1 patients (40% of patients receiving 4 or more lines of therapy in ZUMA1 compared to only 0.2% in SCHOLAR-1).

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 53 of 164

53 of 164

Table 11: Baseline characteristics in the ZUMA-1 and SCHOLAR-1 studies

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115]

© Kite/Gilead (2018). All rights reserved

54 of 164

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115]

© Kite/Gilead (2018). All rights reserved

55 of 164

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115]

© Kite/Gilead (2018). All rights reserved

56 of 164

ZUMA-1 and SCHOLAR-1 results

ORR and CR

Table 12 presents a summary of the ORR, CR and PR rates in SCHOLAR-1 and ZUMA-1. There were 523 patients who were evaluable in the response analysis set. The pooled estimates from SCHOLAR-1, using a random-effects model, resulted in an ORR of 26%, CR of 7% and PR of 17.5%. These were significantly lower than the results from ZUMA-1 (82%, 54% and 28%, respectively), indicating that treatment with axi-cel results in significantly greater proportions of patients being able to achieve a response.

Table 12: ORR and CR in ZUMA-1 and SCHOLAR-1

Key: CI, confidence interval; CR, complete response; NA, not applicable; ORR, overall response rate. Notes: a, Treatment with axi-ce in ZUMA-1, first therapy after refractory determination in SCHOLAR1.

Source:

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 57 of 164

57 of 164

Survival

A summary of survival in the 603 survival evaluable subjects is provided in Table 13. The median OS in SCHOLAR-1 was 6.3 months, with 6-month, 1-, and 2-year OS rates of 53%, 28% and 20%. This is significantly lower than the 6-month (80%) and 1-year (55%) OS rates in ZUMA-1 (n=101), indicating that axi-cel treatment results in significantly greater survival for patients. This difference in survival for axi-cel patients is clearly demonstrated in the KM curves comparing ZUMA-1 and SCHOLAR 1 (Figure 11).

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 58 of 164

58 of 164

Table 13: OS in ZUMA-1 and SCHOLAR-1

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115]

© Kite/Gilead (2018). All rights reserved

59 of 164

Figure 11: Comparison of OS between ZUMA-1 (mITT, Phase 2 primary analysis) and SCHOLAR-1

==> picture [700 x 383] intentionally omitted <==

Key: CI, confidence interval; mITT, modified intent-to-treat; mo, months; OS, overall survival.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115]

© Kite/Gilead (2018). All rights reserved

60 of 164

Survival by response

A summary of OS by response, with survival derived from time of response for responding patients in SCHOLAR-1, is presented in Table 14. This analysis was conducted in the subset of SCHOLAR-1 patients in the survival RR analysis set who have non-missing response and response dates (n=372 subjects). This subset is used because survival time is landmarked from the time of response and hence a response date is required. As indicated, responding subjects experience longer survival than non-responding subjects, with subjects who attain CR experiencing 14.9 months median overall survival, and 1- and 2-year survival rates of 56% and 40%, respectively. This supports the message from the Maurer study, that there are a proportion of patients who achieve CR who are able to maintain this response long-term and be considered cured. The results from the ZUMA-1 study have higher survival rates at 6- and 12-months for CR patients, but the main benefit is from the increasing proportion of patients that are able to achieve CR with axi-cel therapy.

Table 14: OS by response in ZUMA-1 and SCHOLAR-1

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 61 of 164

61 of 164

Standardised comparison of ZUMA-1 and SCHOLAR-1

Due to the significant heterogeneity discussed previously between ZUMA-1 and SCHOLAR-1 standardised comparisons of response and survival between the two studies were conducted and summarised below.

Standardised comparisons of response (by ECOG status)

Standardising based on refractory subgroup and ECOG provides standardised estimates of response and complete response in SCHOLAR-1 of ''''''''' '''''''' '''''''''''', respectively. The standardised comparisons indicate approximately ''''''''''''''''''' ''''''''''' '''''''''''''''''' '''''''''''''''''''''''' in the incidence of response and CR between ZUMA-1 and SCHOLAR-1, with odds ratios for response and complete response of ''''''''''' ''''''''' '''''''''''. This demonstrates that patients treated with axi-cel are statistically significantly more likely to achieve response or CR than patients treated with SoC.

Table 15: Standardised comparison of ORR and CR in ZUMA-1 and SCHOLAR-

1 (including standardisation for ECOG status)

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

----- Start of picture text -----
ZUMA-1 SCHOLAR- Standardised [a] Standardised [a] Odds ratio
mITT 1 Response difference ratio (95% CI) (p-value [b] )
(N=101) (N=508) (95% CI)
ORR [c] ''''''' '''''' '''''''' '''''''''' '''''''''' ''''''''''' ''''''''''''' '''''''''''''
'''''''''''' ''''''''''''''''''''''
CR ''''''' ''''''' '''''' ''''''''' ''''''' ''''''''''' ''''''''''' ''''''''''
'''''''''''''' '''''''''''''''''''
Key: CI, confidence interval; CR, complete response; ECOG, Eastern Cooperative Oncology
Group; mITT, modified intent-to-treat; ORR, overall response rate.
Notes: a, standardised according to pre-specified stratification factors; b, CMH test stratified based
on pre-specified stratification factors; c, for the purpose of comparisons, the response rate in
ZUMA-1 is landmarked from the time of SCT for subjects in the SCT strata and therefore is
different than that reported for the primary endpoint of ZUMA-1.
Source: ZUMA-1 versus SCHOLAR-1 data on file
----- End of picture text -----

Standardised comparisons of survival (by ECOG status)

Standardising based on refractory subgroup and ECOG provides standardised median OS for the SCHOLAR-1 study of 5.8 months, with 3-, 6-, and 12- month

survival rates of ''''''''''''' '''''''''''' '''''''''''''', respectively. The standardised difference in

median survival time was not estimable because the ZUMA-1 median survival is not

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 62 of 164

62 of 164

yet reached. The standardized ratios of 3-, 6-, and 12-month survival rates are '''''''''''' ''''''''''''' ''''''''' ''''''''''', respectively, and the hazard ratio from the Cox model stratified by the covariates used was ''''''''''. This indicates that there is a statistically significantly lower risk of death for patients treated with axi-cel.

Table 16: Standardised comparisons of survival in ZUMA-1 and SCHOLAR-1

(including standardisation for ECOG status)

Standardised comparisons of response (by subsequent ASCT)

Standardising based on refractory subgroup and subsequent ASCT provides standardised estimates of response and complete response in SCHOLAR-1 of '''''''''''''' '''''''' ''''''''', respectively. The standardised comparisons indicate approximately ''''''''''''''''''' ''''''''' ''''''''''''''''' ''''''''''''''''''''''' in the incidence of response and CR between ZUMA-1 and SCHOLAR-1, with odds ratios for response and complete response of '''''''''''' ''''''''' ''''''''''''''''. This demonstrates that patients treated with axi-cel are statistically significantly more likely to achieve response or CR than patients treated with SoC.

Table 17: Standardised comparison of ORR and CR in ZUMA-1 and SCHOLAR-

1 (including standardisation for subsequent ASCT)

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 63 of 164

63 of 164

Notes: a, standardised according to pre-specified stratification factors; b, CMH test stratified based on pre-specified stratification factors; c, for the purpose of comparisons, the response rate in ZUMA-1 is landmarked from the time of SCT for subjects in the SCT strata and therefore is different than that reported for the primary endpoint of ZUMA-1. Source: ZUMA-1 versus SCHOLAR-1 data on file

Standardised comparisons of survival (by subsequent ASCT)

Standardising based on refractory subgroup and ECOG provides standardised median OS for the SCHOLAR-1 study of '''''''' '''''''''''''''''', with 3-, 6-, and 12- month survival rates of ''''''''''''' ''''''''''' '''''''''''''', respectively. The standardised difference in

median survival time was not estimable because the ZUMA-1 median survival is not yet reached. The standardized ratios of 3-, 6-, and 12-month survival rates are ''''''''''''' '''''''''''' ''''''' '''''''''', respectively, and the hazard ratio from the Cox model stratified by the covariates used was '''''''''. This indicates that there is a statistically significantly lower risk of death for patients treated with axi-cel. Figure 12 presents the KM curve for SCHOLAR-1 versus ZUMA-1 from this analysis and shows a significant improvement in survival for axi-cel patients.

Table 18: Standardised comparisons of survival in ZUMA-1 and SCHOLAR-1

(including standardisation for subsequent ASCT)

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 64 of 164

64 of 164

Figure 12: Overall Survival ZUMA-1 versus SCHOLAR-1 (SCHOLAR-1 Survival-RR Analysis set, ZUMA-1 mITT Analysis

Set)

==> picture [659 x 90] intentionally omitted <==

==> picture [659 x 90] intentionally omitted <==

==> picture [659 x 90] intentionally omitted <==

==> picture [659 x 90] intentionally omitted <==

Source: ZUMA-1 versus SCHOLAR-1 data on file

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115]

© Kite/Gilead (2018). All rights reserved

65 of 164

Uncertainties in the indirect and mixed treatment comparisons

The comparison between SCHOLAR-1 and ZUMA-1 makes the best use of the evidence that is available, as by using the patient-level data from SCHOLAR-1, the inclusion criteria could be matched to ZUMA-1. However, there remain a number of key limitations that should be considered when interpreting this evidence, which may have biased results against ZUMA-1. Although the inclusion criteria for the studies were matched, there remains a large amount of heterogeneity between the study populations. It is important to note that the population of ZUMA-1 are a heavily pretreated population; '''''''''''' of patients had received four or more prior lines of therapy, compared to only '''''''''''' in SCHOLAR-1. Heavily pre-treated patients are likely to have already exhausted all other treatments, have few options remaining to them, and would be expected to have poor outcomes. In addition, '''''''''''' of patients went on to receive subsequent ASCT, compared to '''''''''''' of patients in ZUMA-1. Therefore, the outcomes in SCHOLAR-1 may not be fully reflective of the SoC outcomes that would be expected in a comparable ZUMA-1 population. In an analysis standardised for patients receiving subsequent ASCT, the outcomes in SCHOLAR-1 were poorer and this resulted in a more favourable comparison between axi-cel and SoC.

It is also important to consider that ZUMA-1 is an open-label single-arm study, which is being compared to a historical cohort consisting of data from two RCTs and two observational, database studies. Such analyses can always be criticised, however the naïve comparison of SCHOLAR-1 compared to ZUMA-1 still shows the significant benefit of axi-cel treatment to these patients, and the analyses attempting to adjust for the imbalances between the patient populations indicates that these benefits are likely to be even greater in reality.

B.2.10. Adverse reactions

The two most commonly encountered toxicities with CAR T-cell therapies are cytokine release syndrome (CRS) and neurotoxicity. CRS is characterised by high fever, hypotension, hypoxia, and/or multiorgan toxicity; neurotoxicity manifests as a toxic encephalopathic state with symptoms of confusion and delirium, and occasionally seizures and cerebral oedema. These toxicities are manageable in most patients, although some require monitoring and treatment in the intensive-care setting, and fatalities can occur, as emphasised by the clinical trial experiences

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 66 of 164

66 of 164

reported to date. Accurate assessment and prompt management of toxicities can mitigate the adverse outcomes. The overall goal of management is to maximise the benefit from potentially curative cellular therapy while minimising the risk of life-threatening complications, particularly CRS and neurotoxicity.

Furthermore, clinical evidence from ZUMA-1 has revealed that with increasing clinician experience in the use of axi-cel (and in monitoring and treating these AEs), there is a demonstrable decrease in the incidence of these events, and this experience is likely to be of great benefit to the wider clinical community.[5]

Due to the limited clinical experience with axi-cel within the UK, the manufacturer is working with NHS England in order to identify a small number of centres with relevant expertise in managing the side effect profile associated with cell based therapy.

CAR-T-cell-therapy-associated TOXicity (CARTOX) Working Group has been formed in order to develop a consistent approach to the monitoring, grading, and management of toxicities.[59, 60] This group includes representatives from multiple institutions and multiple medical disciplines, including haematological oncology, solid-tumour oncology, stem-cell transplantation, neurology, critical care, immunology, and pharmaceutical sciences. The CARTOX Working Group has now collectively developed recommendations and a practical guide for monitoring, grading, and management of CRS and neurotoxicity in adult patients.[59, 60]

As the majority of AEs associated with CAR T therapy occur soon after infusion, and this is what was observed in the ZUMA-1 study, only detailed safety data are available from the primary analysis (median follow-up 8.7 months). Additional AEs that occurred after this point, for the updated analysis (median follow-up 15.4 months) are presented to support this case.

Summary of safety data from ZUMA-1, primary analysis

A summary of the safety events from the primary analysis (at a median follow-up of 8.7 months) of the Phase 2 part of ZUMA-1 is provided in Table 19. Across the combined 101 treated participants, 96 (95%) experienced a Grade 3 or higher AE. The most commonly occurring AEs are presented in Table 20.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 67 of 164

67 of 164

Table 19: Summary of key safety events from ZUMA-1 Phase 2 (mITT

population)

Table 20: Most frequent Grade ≥3 treatment emergent adverse events

occurring in ≥10% of patients and SAEs occurring in ≥2 patients in ZUMA-1 Phase 2 (mITT population)

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 68 of 164

68 of 164

Deaths

Among the 10 patients who underwent leukapheresis, but were not treated with axicel, eight died: six patients died due to PD; one patient died from myelodysplastic syndrome (MDS) that occurred 1 year after starting other off-study therapy, and one patient died from tumour lysis syndrome (TLS) that was considered related to conditioning chemotherapy. Further details regarding these patients are presented below.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 69 of 164

69 of 164

Among the 101 patients treated with axi-cel, 30 died. Two deaths occurred within 30 days after the axi-cel infusion. One of these two deaths was an AE (pulmonary embolism) deemed unrelated to conditioning chemotherapy or axi-cel; the other event was a death due to PD. Twenty-five of the 30 patients died from disease progression. Two patients died after PD and initiation of other cancer therapy. Three patients had fatal AEs, which are discussed in more detail below.

Changes in AE rates between the interim and primary analysis of ZUMA-1

Throughout the ZUMA-1 study clinician experience in the recognition and management of toxicities increased remarkably as observed from the marked reduction in the incidence of severe CRS and neurotoxicity in the latter part of the study. Compared to patients recruited from the study start until the primary analysis,

a there was a ''''''' ''''''''''''''''''''''''''''''''''''''' ''''''''''''''''''''''' '''''' ''''''''''''''' '''' ''''' '''''''''''''' ''''''''''' ''''''''''''' '''''''''''

''''' ''''''''''' '''''''''''' and '''' ''''''' ''''''''''''''''''''''''''''''''''''''' '''''''''''''''''''''' '''''' ''''''''''''''' '''''' '''''''''''''''''''''''' '''''''''''''

''''''''''''' '''''''''''' ''''' '''''''''' '''''''''''' in patients recruited after the interim analysis. Importantly, there were '''''''' ''''''''''''''''''''''''''' '''''''''''''' ''''''' ''''''''' ''''''' '''''''''''''''''''' ''''''''''''''''''''''' '''''''''''''''''''''''' '''''''''

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

'''''''''' '''''''''''' ''''''''''' ''''''''''' ''''''''' '''''''''''''''''' '''''''''''''''''''' (Table 21).

Table 21: Change in incidence of key adverse events between the interim analysis and the primary analysis of ZUMA-1 Phase 2

Cytokine release syndrome (CRS)

CRS, the most-common toxicity of cellular immunotherapy, is triggered by the

activation of T-cells on engagement of their T-cell receptors (TCRs) or CARs with

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 70 of 164

70 of 164

cognate antigens expressed by tumour cells. The activated T cells release cytokines and chemokines as do bystander immune cells. CRS is a clinical constellation of symptoms including fever, nausea, fatigue, myalgias, malaise, hypotension, hypoxia, coagulopathy and capillary leak, and/or multiorgan toxicity. Besides the constitutional symptoms, some severe cases can experience significant hemodynamic instability and or other organ toxicity. Mild-to moderate CRS usually is self-limiting and can be managed with close observation and supportive care. Severe CRS must require intensive medical management with tocilizumab alone or with steroids. Patients at high risk of severe CRS include those with bulky disease, comorbidities, and those who develop early onset CRS within 3 days of cell infusion.[59, 60]

In ZUMA-1 study, overall, ''''''''''''''''''''''' '''''''' '''''''''''''''''''''''''''' '''''''''''''''''' ''''''''''''''''''''''' '''''''' '''''''''''''''''''''''''''' ''''''''''''''''''''''''''''''''''''''''' '''''''' '''''''''''''''''''''''''''' ''''''''''''''''''''''''''''''''''''''''' '''''''' '''''''''''''''''''''''''''' ''''''''''''''''''''''''''''''''''''''''' '''''''' '''''''''''''''''''''''''''' ''''''''''''''''''''''''''''''''''''''''' '''''''' '''''''''''''''''''''''''''' ''''''''''''''''''''''''''''''''''''''''' '''''''' '''''''''''''''''''''''''''' ''''''''''''''''''

The most common CRS symptom of any grade ''''''''''''''''''''''' '''''''' '''''''''''''''''''''''''''' ''''''''''''''''''

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

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

All of the Grade <3 CRS events resolved within a median of 8 days and all CRS events did resolve, with the exception of the two Grade 5 events (hemophagocytic lymphohistiocytosis and anoxic brain injury following cardiac arrest), both of which followed ongoing CRS events that started within the first week after the cell infusion.

Neurological events

Neurotoxicity is another prominent toxicity of CAR-T cell therapy with published reports of 20–64%, including grade ≥3 in 13–52%.[59] It typically manifests as a toxic encephalopathy; the most common symptoms include encephalopathy, headache, delirium, anxiety, tremor, aphasia; other manifestations of neurotoxicity such as decreased level of consciousness, confusion, seizures and cerebral edema have also been observed in clinical trials of CAR-T cells. In general, the mild clinical signs are self-limited and resolve within days; more severe symptoms may require

supportive care alone or with dexamethasone and can be complete resolved within 4

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 71 of 164

71 of 164

weeks. However, some deaths caused by this unexpected toxicity have been documented.[59]

The manifestation of neurotoxicity can be biphasic; the first phase occurs concurrently with high fever and other CRS symptoms, typically within the first 5 days after cellular immunotherapy, and the second phase occurs after the fever and other CRS symptoms subside, often beyond 5 days after cell infusion. The pathophysiological mechanism underlying CRES remains to be determined. Two potential explanations can be postulated – either through passive diffusion of cytokines into the brain or by trafficking of T cells into the CNS. Neurotoxicity is primarily managed with supportive care.[59, 60]

In ZUMA-1 trial, overall, ''''''' ''''''''''''''' '''''''''''''''''''''' ''''''''''''''''''''''''''''''' ''''''''''''''''''''''''''''''' '''''''''''''''''' ''''''' ''''''''' '''''''''''''''''' '''''''''''' ''''' '''''''''''''' '''''''''''' ''''' ''''''''''''' ''''''''''''''' ''''''''''' ''''''''''''''' '''''''''''' '''''''''''''''''' ''''' ''''''''''''''. The most common neurologic event of any grade '''''''''' '''''''''''''''''''''''''''''''''''''''' '''''''''''''''''' ''''''''''''''''''''' '''''''' '''''''''''''''''''''''''''' '''''''''''' '''''''''''''''''' '''''''''''''''''' ''''''''''''''''''' ''''''''''''''''''' ''''''''''''''' '''''''''''''''''''''''''' ''''''''''''' '''''''''''''''''''' '''''''''''''' '''''''''''''''''''' ''''''''''''''''''''''''' ''''''''''''' '''''''''' '''''''''''''' ''''''''''''''' '''''''''''''''''''''' '''''''''''''''

==> picture [447 x 118] intentionally omitted <==

----- Start of picture text -----
'''''''''' '''''''''''''''''' ''''' ''''''''''''''''''''''''' ''''''''''''' '''''''''''' '''''''''''''' ''''''''' '''''' '''''''''''''' '''''''''''''''''''' ''''''''''''' ''''
'''''''''''''''''' '''''' '''''' '''''''''''' ''''''''' ''''''''' ''''''''''''''''''''''''''' ''''' ''''''''' ''''''''''''' ''''' ''''''''''''''' '''' ''''''''''''''''''
'''''''''''''''''''''''''' '''''''''''''''''' '''' ''''' ''''''''''''''' '''''''''''''''''''''''''''''''' '''''''''''''''' '''''''''''' '''''''''''''''''''''''' '''''' '''''' '''''''''''''
''''' '''''''''''''''''''''''' '''''''''' '''''''''''' ''''''''''''''''''''''' ''''''''''''' '''' '''''' '''''''''''''' ''''''''''''''''''''''''' '''''''''''''''' ''''''''''''
'''''''''''''''''''''''''''''''''''''''' '''''''''''''''' ''''''''''''''''''''''''' ''''''''''''''' '''''''''''' '''''''''''''''''''' '''''''''''''' '''''''''' ''''''''''''''''''''''''''''''
'''''''''''''
----- End of picture text -----

As discussed above, as investigator’s experience with axi-cel therapy and with

monitoring and managing neurological events increased over the course of the

==> picture [449 x 56] intentionally omitted <==

----- Start of picture text -----
ZUMA-1 study, ''''''''''''' '''''' '''''''''''''''' ''''' '''''' ''''''''''''''''''' ''''''''''''''''''''''''''''''''''' '''''''''''''''''' ''''''''''''''''''''''''''
'''''''''''' ''''''''''' '''''''''''''''' '''''''' ''''''' '''''''''''''''''''''' ''''' ''''''''' '''''''''''''''' ''''''''''''''''''''' ''''' '''''''''' ''''''''''''''''' '''''''' '''''''
'''''''''''''''''''' ''''''''''''''''''' ''''''''''''''''''' '''''''' '''''''''''''''''' ''''''''''''''''''' '''''''' '''''''' '''''''''''''''' '''''''''''''''''' (Table 21).
----- End of picture text -----

Cerebral oedema

No cases of cerebral oedema were reported.

Cytopenias

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 72 of 164

72 of 164

Cytopenias were consistent with the known toxicities of the conditioning regimen of cyclophosphamide and fludarabine. Grade 4 neutropenia, thrombocytopenia, and anaemia occurred in ''''''''''''''' ''''''''''' ''''''''''' ''''''''' of patients, respectively. Prolonged (duration >30 days) Grade 4 neutropenia or thrombocytopenia occurred in ''''' '''''' of patients each. It is important to note that '''''' '''''''''''''''''''''''' ''''''''''''''' ''' ''''''''''''''''''''''''

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

==> picture [450 x 139] intentionally omitted <==

----- Start of picture text -----
'''''''''''''' '''''''''''' '''''' ''''''''''''' '''' '''''''''''''''''''''''' ''''''''''''''''' ''''''''''''''''' '''''''''''''''''''''''''' ''''''''''''''' ''''' ''''''''''''''''''''''''''
''''''''' ''''''''''''''''' ''''''''' '''''''''''' ''''''''''''' '''''''' '''''''''''''''''' '''''' '''''''''''''''''' ''''''''' ''''''''''''''''' '''''''' ''''''''''''''''''''''
''''''''''''''''''''''''' ''''''''''''''' ''''''' '''''''''''''''' '''''' ''''''''' '''''''''''''''''''''''''''''' ''''''''' '''''''''''''''''' ''''''''''''''''''''''''' ''''''''''''''''''
'''' ''''''''''''''''''''''''''''' '''''''''' ''''''''''''''''' ''''''''' '''''''''''' '''''''''''' ''''''''' '''''''''''''''''' ''''' '''''''''''''''''' '''''''''''' ''''''''''''''''
'''''''''''''''''''' ''''''''''' ''''' ''''''''''' ''''''''''''''''' '''''''''' '''''''''' ''' ''''''''''''''''' ''' ''''''''''''''''''''''''' '''''''''' ''''''''''''''''' ''''''''''
'''''''''''' ''''''''''''' '''''''''' '''''''' '''''''''''''''''''' ''''''' ''''''''''''''' ''' ''''''''''''''''''''''''' '''''''''''''''''''''' '''''' '''''''''''''' ''''' '''''
'''''''''''''''
----- End of picture text -----

The proposed prescribing information recommends monitoring of blood counts during treatment with axi-cel. In the ZUMA-1 trial, patient blood counts were monitored at 2 weeks, 4 weeks, 2 months, 3 months and then every 3 months up to 24 months, following axi-cel infusion. In clinical practice, patient blood counts would be monitored as per local standard practice, as is currently done for SoC chemotherapy treatments.

Infections

==> picture [448 x 226] intentionally omitted <==

----- Start of picture text -----
'''''''''''''''''''''''''''' '''''''''''''''''''' '''''''''''''' ''''''''''''''''''''''''''''' '''''''''''''''''''''' '''''''''' ''''''''''''''''''''''''' '''''''''''' ''''' ''''''''
'''''''''''''''' '''''''''''''''''''''''''' '''' '''''''''''''''''''''' '''''''''''' '''''''''''''' ''''''''''' ''''''''''''' ''''''''''''''''''''''' '''''''''' '''''''''''' '''''''''''
'''''''' ''''''''' ''''''''''''' '''''''''''''''''''''''''''' '''''''''''''''''''''''' ''''''''''''' ''''''''''''''''''' ''''' ''''''''''' '''' '''''''''''''''''' ''''''''''''
''''''''''''''''' '''''''''' '''''''''''''''''''' '''''''''''''' ''''''''''' '''''''''''''''''''' '''''''''''''' ''''''''''''''''''''''''''''' '''''''''''''''''' '''''''''''''''''''''''
''''''''''''' '''''''''''''''' '''''''''''''' '''''''''''''' ''''''''' ''''''''''''''''''''''' ''''''''''''''.
''''''''''''''''''''''''' ''''''''''''''''' '''''''''''' '''''''''' ''''''''''''' '''' '''''''''''''''''' '''''''''' '''''''' ''''''''''''''' '''''''''''''' ''''''''' '''''''''''''''' ''''
''''''''''''''''' ''''''''' ''''''''''' ''''''''''''''''''''''' '''''''''''''''''''' ''''' '''''''''' ''''' '''''''''''''''''''' '''''''''''' '''''''''''''' '''' ''''''''''
'''''''''''''''''''''' ''''''''''''.
'''''''''''' '''''''''''''''' ''''''''''' ''''''''''''''''' ''''' '''''''''' ''''' '''''' '''''''''''''''''''''''' '''''''''''''''''' ''''' '''''''''''''''''''', defined as
prior to conditioning chemotherapy and axi-cel infusion. '''''''''''' '''''''''''' '''''''''' '''''' '''''''' ''''''''
----- End of picture text -----

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved

73 of 164

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

The proposed prescribing information recommends that axi-cel should not be administered to patients with active systemic infections and that prophylactic antimicrobial agents should be administered according to standard institutional guidelines (and this is how patients were managed within the ZUMA-1 trial).

Grade 5 adverse events/Deaths

Overall, ''''''''''''''' ''''''''''''' '''''''''''''''''' ''''''''''''''''''''''''''''''''' '''' ''''''''''''' '''' ''''''''' '''''''''''' '''''''''' ''''''' ''''''''''''''''''''''''''''''''''' '''''''''''''''' '''''''''''''''''''''''''''''''''''''''' ''''''''''''''''''''''''''''''''''''''''''' '''''''' '''''''''''''''' ''''''''''''''' '''' '''''''' '''''''''''''''' ''''' '''''''''''''' '''''''''' '''''''''' ''''''''''''''''''''''''' '''''' ''''''''''''''' '''''''''''''''''''''''''' '''''''''''''''''''''''''''' ''''''' '''''''''' ''''''''''''''''''''''''''''''''' '''''''''''''''''' ''''' ''''''''' '''''''''''' '''''''''' ''''''''''''''''''' ''''''''''''' '''''''' '''''''''''''''''' '''''''''''''''''''.

Patients unable to receive axi-cel

In total, 111 patients were enrolled in the Phase 2 part of the ZUMA-1 study and all 111 patients were leukapheresed. Eight patients (7%) were unable to progress to conditioning chemotherapy; two patients (2%) died due to progressive disease, four patients (4%) had AEs, and two patients (2%) had non-measurable disease.

Therefore, 103 patients progressed to conditioning chemotherapy, and of these patients, 101 went on to receive axi-cel; one patient died due to tumour lysis syndrome, which was considered to be related to conditioning chemotherapy, and one other patient experienced an AE that prevented them from receiving axi-cel.

Summary of safety data from ZUMA-1, updated analysis (N = 108; median follow-up 15.4 months)

After the data cut-off for the primary analysis until the updated analysis, ten patients had serious AEs (including nine infections in eight patients). There were no new events associated with the CRS or neurological events related to axi-cel treatment. Forty-four patients (44%) died from causes that included disease progression (in 37 patients), AEs (in three patients, including two with the above-mentioned axi-cel– related events associated with CRS and one with pulmonary embolism that was not related to axi-cel), and other causes after disease progression and subsequent

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 74 of 164

74 of 164

therapies that were not related to axi-cel (in four patients). One death that was not associated with axi-cel was previously reported in Phase 1 of ZUMA-1.[55] There were no new deaths from AEs after the primary analysis. No cases of replicationcompetent retrovirus or axi-cel treatment-related secondary cancers were reported.

No other studies reported additional AEs.

B.2.11. Ongoing studies

Follow-up in the ZUMA-1, Phase 2 study is ongoing.[7]

No other studies investigating axi-cel in patients with R/R DLBCL, PMBCL or TFL are due to provide additional evidence within the next 12 months.

B.2.12. Innovation

Axi-cel is the first in a breakthrough class of CAR T-cell therapies and is an innovative approach that provides complete personalised immunotherapy. Axi-cel is given as a single infusion and single treatment rather than the recurrent cycles of traditional chemotherapy and their associated toxicity. It offers a significant benefit in the potential treatment landscape for R/R DLBCL, PMBCL and TFL patients, who are ineligible to transplant and associated with a median life expectancy of 3.3 to 6.3 months.[6, 16, 17, 41] Due to elimination of malignant B-cell in the treatment group , axicel has been shown to have a durable complete response of 40% and median life expectancy has not yet been reached at 15.4 months. The American Society of clinical Oncology (ASCO) named CAR-T cell therapy the advance of the year.[61]

As described in Section B.1.3, current treatment options for patients with R/R DLBCL, PMBCL and TFL ineligible for ASCT are extremely limited and generally consist of palliative care, with poor outcomes and expected survival as low as 3.3 months.[6] As described in Section B.2.6 and Section B.2.9, patients treated with axicel achieve much higher rates of overall response compared to current SoC (82% compared to 26%), with a durable complete response (40% of patients remained in complete response at a median follow-up of 15.4 months).[5] Survival data in the pivotal ZUMA-1 study so far suggests that axi-cel will significantly improve survival outcomes for these patients, for whom few other treatment options remain. Current literature suggests that patients with DLBCL who achieve event-free survival at 24 Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 75 of 164

75 of 164

months following first-line immunochemotherapy treatment have a subsequent OS equivalent to that of the age- and sex-matched general population, which means these patients can be considered to be cured.[48] Long-term follow-up data of patients from the NCI study[57] demonstrated that following axi-cel treatment, patients who achieve CR can experience a durable, long-term response with associated increases in survival; 44% were in ongoing CR between 38+ and 56+ months after axi-cel infusion, and three out of these four patients had recovery of normal B-cells. These data suggest that axi-cel has the potential to increase the OS of R/R DLBCL, PMBCL and TFL patients towards that of the age- and sex-matched general population.

Risks of therapy are known and include CRS and neurologic events that have an early onset, generally within 2 weeks after the therapy, and are mostly reversible (>95%). Alongside the understanding that tocilizumab use does not affect efficacy, strategies have been further developed to identify these AEs earlier and manage them more aggressively, so that the rates of severe CRS and neurologic events have decreased overtime.

Other oncology treatments, such as current chemotherapies and even newer immunotherapies, involve long-term, multiple and regular clinical visits. The potential for patients receiving axi-cel to achieve long-term, durable response and potentially be cured avoids multiple future hospital visits for additional treatments and diseaserelated monitoring. Therefore, axi-cel treatment visits will have less impact on patients HRQL, and given that the treatment is intended for patients who would otherwise be considered as being at end-of-life (see Section B.2.13), anything that reduces their time in hospital or their number of clinic visits can be considered to be a massive benefit to the patients.[1, 2] These quality of life benefits are unlikely to have been captured in the HRQL/QALY data used in the economic analysis.

In clinical practice, axi-cel will be given as a single-time administration. This means that patient adherence is effectively 100%. Patient compliance is also expected to be high, with 92% of patients identified as suitable to receive axi-cel in the ZUMA-1 study going on to receive treatment.[5] In the Juliet study[62] for the closest alternative CAR T therapy, tisagenlecleucel, this was only 60%. This highlights the consistency and effectiveness of the manufacturing process for axi-cel. As there is the potential

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 76 of 164

76 of 164

for patients receiving axi-cel to achieve long-term, durable response and potentially be cured, there would be reduced ongoing direct costs expected for these patients after axi-cel administration and AEs have been satisfactorily managed; there will likely be a monitoring period of 1–2 years after axi-cel administration[5] , at which point, according to published data, a patient could be considered cured.[1, 2, 48]

Overall, axi-cel has demonstrated a positive benefit-risk profile and offers a new and effective treatment option for patients with relapsed or refractory aggressive B-cell NHL. The intended patient population is well defined and has high unmet medical need, with no curative options, no standard therapy, and short expected survival. The unprecedented and consistent treatment effect observed with axi-cel is a complete step change for these patients and supports recommendation for the use of axi-cel for patients with relapsed/refractory aggressive B-cell NHL who are ineligible for ASCT.

B.2.13. Interpretation of clinical effectiveness and safety evidence

Summary and discussion of the available evidence to support axi-cel

From the ZUMA-1 study results, axi-cel is a highly effective treatment option for patients with R/R DLBCL, PMBCL and TFL who are ineligible to receive ASCT.[7] In the updated analysis (median follow-up 15.4 months), 82% of patients achieved a response, with a durable complete response (40% of patients remained in complete response). Median OS was not reached at 12 months (95% CI: 12.0, not reached) and the OS at 18 months was 52%. This is a significant improvement on the 3.3–6.6 months expected OS with current SoC.[6, 16, 17, 41] For patients achieving CR, 83.1% were still alive at 12 months, highlighting the benefit of axi-cel treatment for these patients. Median PFS in ZUMA-1 was 5.8 months (95% CI: 3.3, not reached) with 41% of patients still progression-free at 15 months. The KM curves for OS and PFS also have long, flat tails, with few events occurring (i.e. few incidents of progression or deaths) towards the end of the curves. This indicates the potential for long term disease free survival and potential cure.

In this disease there is potential for long term survival in those that achieve CR.[14, 48] The evidence from ZUMA-1 indicates that the benefit of axi-cel will be in increasing

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 77 of 164

77 of 164

the number of patients who are able to reach and remain in CR and are therefore able to experience these long-term survival benefits.

The ZUMA-1 population was a heavily pre-treated patient population, with 69% of patients having received 3 or more prior therapies and therefore the outcomes may underestimate the efficacy in the licensed indication where patients have been less heavily treated.

Axi-cel patients treated in ZUMA-1 are compared to a patient-level pooled analysis of SoC in patients with refractory DLBCL, PMBCL and TFL in SCHOLAR-1, where the ORR was 26%, 7% of patients achieved CR, and the median OS was 6.3 months.[6] In a standardised comparison of ZUMA-1 and SCHOLAR-1 (accounting for patients with ECOG 2-4 in SCHOLAR-1), the median OS in SCHOLAR-1 was only ''''''' ''''''''''''''''

''''''''''' '''' '''''''''''''''''''' ''''''''' '''''' ''''''''''' '''''''''''''''''''''''''''', suggesting a ''''''''' '''''''''''''''''''''' in the risk of death for axi-cel patients. In a standardised comparison of ZUMA-1 and SCHOLAR1 (accounting for patients receiving subsequent ASCT), the median OS in SCHOLAR-1 was only '''''''' ''''''''''''''''''''' '''''''''' ''' ''''''''''''''''''' ''''''' ''''' ''''''''''''' ''''''''''''''''''''''''''', suggesting a ''''''''''' '''''''''''''''''''''''''' in the risk of death for axi-cel patients.

The safety profile of CAR T therapy is well described and includes CRS, neurological events, infections and cytopenias. In the ZUMA-1 study 95% of patients experiencing a grade ≥3 AEs (the most important were ''''''''''''''''''''''''''''' ''''''''''''''''' '''''''''''''''''''

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

'''''''''''''''''''''''''''''''' '''''''''''''''). Established treatment protocols are now in place for the management of CRS, neurological events and cytopenias, and as a result, in the second part of the trial the majority of these events were manageable and reversible within the first month after infusion (see Section B.2.10). Management of these AEs is already available as part of standard practice within NHS England.[1, 2] Axi-cel is also likely to be administered in specialist centres in the UK, which will result in even greater increases in experience in the long-term. This increased experience and shared knowledge is likely to be of additional benefit for clinicians throughout the UK,[1, 2] allowing them to better manage the toxicities associated with CAR T therapies and potentially lead to reductions and improvements in management of these AEs, both when CAR T therapies are introduced with further reductions over time.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 78 of 164

78 of 164

Strengths and limitations of the evidence base

There are a number of strengths of the evidence to support axi-cel. ZUMA-1 was a relatively large clinical trial (given the orphan nature of the population of interest) in a population directly relevant to the decision problem, including reasonable numbers of patients of all relevant subtypes (DLBCL, PMBCL and TFL) and with compelling results in a condition with significant unmet medical needs, no curative options, no standard therapy and short expected survival. The results from ZUMA-1 were consistent across all patient subgroups and supports axi-cel as a viable treatment option for the entire population likely to be specified in the marketing authorisation (as per the SmPC). Clinical opinion confirmed that the population treated in ZUMA-1 was reflective of the population anticipated to be treated with CAR T therapy in NHS England.[1, 2] The outcomes used in the trial are consistent with those that would be captured as part of standard practice in NHS England, and clinical opinion confirmed that the results seen in ZUMA-1 would be expected to be the same for patients treated in the UK.[1, 2]

Due to the curative nature of the treatment and the high proportion of patients in the ZUMA-1 trial still alive and in ongoing CR at a median follow-up of 15.4 months, median OS has yet to be reached. This creates some uncertainty in terms of fully assessing the benefits of axi-cel, with the potential to underestimate the efficacy of treatment. The fact that 52% of patients with limited other treatment options, poor expected survival and high unmet need are still alive at 18 months (compared to expected survival of 3.3–6.6 months with SoC) is a significant benefit. The NCI studies[9, 56, 57] provide support to the long-term duration of response observed in ZUMA-1, with 47% in ongoing CR between 7+ and 24+ months and 44% in ongoing CR between 38+ and 56+ months, respectively. UK clinicians agreed that they would expect the long-term DoR data seen in ZUMA-1 to translate to substantial improvements in survival for these patients.[1, 2]

ZUMA-1 was conducted as a single-arm study, which makes comparison to SoC difficult. However, as patients with aggressive B-cell NHL that is refractory or relapsed after two prior therapies have no other curative options and no specified SoC, a single-arm study was considered the most appropriate. Simon et al.[63]

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 79 of 164

79 of 164

suggest the following criteria for judging whether a single-arm study can support traditional approval, which the data for axi-cel support:

• The drug mechanism of action is supported by strong scientific rationale and/or preclinical data

  • The mechanism of action of axi-cel is via T-cell mediated killing of CD19+ target cells. Pharmacokinetic studies in ZUMA-1 and the NCI study demonstrate levels of anti-CD19 CAR T-cells are associated with objective response. Immune checkpoint inhibitors[64, 65] and blinatumomab are FDA approved therapies that activate or re-direct T-cells to treat cancer. Data from NCI 09-C-0082, the longest ongoing study of anti-CD19 CAR T-cells in B-cell malignancies have demonstrated the potential for anti-CD19 CAR T-cells to induce responses in patients with advanced B-cell malignancies.[9, 56, 57, 66]

• The drug is intended for a well-defined patient population

  • The indicated population to be treated, represents a homogeneous population with respect to outcome, with no curable options or SoC. In clinical practice, these patients are managed with salvage therapies, with limited response and poor survival (ORR, CR and OS in SCHOLAR-1 of 26%, 7%, and 6.3 months, respectively [primary analysis] and 20%, 6%, and 3.9 months, respectively [standardised analysis]).

• The drug produces substantial, durable tumour responses that clearly exceed those offered by any existing available therapies

  • The results presented in Section B.2.6 and B.2.9 and summarised at the beginning of Section B.2.13 show that axi-cel demonstrates substantial and durable tumour response that clearly outweigh outcomes currently experienced by patients treated with SoC.

• The benefits outweigh the risks

  • The benefits of axi-cel outweigh the risks. Axi-cel is administered as a single dose following conditioning chemotherapy, and the majority of AEs, which can be severe, occur within 30 days of infusion, are well defined, generally reversible, and manageable, with no apparent long-term consequences other than B-cell aplasia. The rates of severe CRS and neurologic events decreased

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 80 of 164

80 of 164

over the course of the trial. (See Section B.2.10 and the summary at the beginning of B.2.13).

Limited HRQL data are available for patients with R/R DLBCL, PMBCL and TFL. The analysis performed on the safety management cohort of ZUMA-1 contains the first set of health utility values that will be published for this population. However, this analysis only contains small patient numbers, and therefore, the results are difficult to interpret. Discussion with UK clinical experts confirmed the assumption that CR implies the disease no longer adversely impacts patient’s HRQL;[1] therefore, the substantial proportion of patients achieving a durable CR when treated with axi-cel are also likely to benefit from associated significant improvements in HRQL. Clinical experts also agreed that after 2 years in CR a patient at this stage of disease could be considered cured,[1] which supports the other available evidence in this area.[48] Therefore, it can be assumed that patients who remain in CR for 2 years and beyond would return to a similar level of HRQL as patients in the general population.

There are a number of difficulties in comparing axi-cel to SoC. Given the lack of availability of studies in a comparable population to ZUMA-1 and the large amounts of heterogeneity between ZUMA-1 and the studies identified in the clinical SLR, the approach was taken to use studies for which patient level data were available, to match the inclusion criteria to ZUMA-1 (SCHOLAR-1). However, although this comparison matched on inclusion criteria and is the most appropriate comparison to consider, there remains a large amount of heterogeneity between the study populations, which may have biased the results against ZUMA-1. The population of ZUMA-1 are a heavily pre-treated population; '''''''''''' '''''' '''''''''''''''''''' '''''''''' '''''''''''''''''''' '''''''''''

'''''' ''''''''''''' ''''''''''' '''''''''''' '''''' '''''''''''''''''' ''''''''''''''''''''''''' '''''' '''''''''' '''''''''''' '''' '''''''''''''''''''''''''''''''. Heavily pre-treated patients are likely to have already exhausted all other treatments, have few options remaining to them, and would be expected to have much poorer outcomes. Therefore, the outcomes in SCHOLAR-1 may not be fully reflective of the SoC outcomes that would be expected in a comparable ZUMA-1 population. Also, ''''''''''' of patients in SCHOLAR-1 went on to receive subsequent ASCT, compared to ''''''' patients in ZUMA-1. This is likely to have improved outcomes for SCHOLAR-1 patients. In an analysis standardised for patients receiving subsequent ASCT, the outcomes in SCHOLAR-1 were poorer and this resulted in a more favourable

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 81 of 164

81 of 164

comparison between axi-cel and SoC (see Section B.2.9). However, the naïve comparison of SCHOLAR-1 compared to ZUMA-1 still shows the significant benefit to these patients of axi-cel treatment, and the analyses attempting to adjust for the imbalances between the patient populations indicates that these benefits are likely to be even greater in reality.

Generalisability

The ZUMA-1 trial was conducted in 24 centres; 23 across the US and one in Israel. As this represents a large geographical area, and manufacturing of axi-cel was so successful (92% of patients identified to receive axi-cel went on to receive treatment) and was delivered to patients within such a short time frame (median: 17.0 days; range: 14–51 days) no issues are anticipated in providing axi-cel to patients in the UK. However, as there were no UK centres, no UK patients were enrolled in the study. The median age of patients in the ZUMA-1 trial was 58, which is similar to the median age for these patients in clinical practice, which is 61 for patients with DLBCL and TFL.[14] PMBCL patients are generally younger, with a median age of 35,[23-25] and this could have pulled down the median age in ZUMA-1. UK clinical experts agreed that the patients treated in ZUMA-1 were likely to be reflective of the UK patients who would be considered for treatment with CAR T therapy and they believed that the ZUMA-1 trial population overall was generally reflective of patients who would be seen in clinical practice.[1] Therefore, they also assumed that the treatment outcomes seen for patients in ZUMA-1 would be what they would expect for patients treated in NHS England.[1]

Axi-cel as an end-of-life therapy

Table 22 presents the evidence to support axi-cel as an end-of-life therapy, in line with the NICE criteria. In summary, with current SoC, median OS for R/R DLBCL, PMBCL and TFL patients is as low as 3.3 months and is therefore within the 24 months normally specified by NICE. Median OS was not reached in the ZUMA-1 study, but with a lower 95% confidence interval of 12.0 months and an 18-month OS rate of 52%, it seems that axi-cel is likely to offer an extension to life of greater than the 3 months specified by NICE. Therefore, axi-cel should be considered as an endof-life therapy, in that it is a treatment option for patients who would otherwise be considered as being at end-of-life; however, it should be seen as a definitive,

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 82 of 164

82 of 164

curative therapy for those patients who are able to achieve CR and maintain this response beyond 2 years. The number of patients eligible to receive axi-cel every year is limited (approximately 970 patients in 2018).

Table 22: End-of-life criteria

Key: ASCT, autologous stem cell transplantation; axi-cel, axicabtagene ciloleucel; DLBCL, diffuse large B-cell lymphoma; NHS, National Health Service; OS, overall survival; PMBCL, primary mediastinal B-cell lymphoma; R/R, relapsed or refractory; SoC, standard of care; TFL, transformed follicular lymphoma.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 83 of 164

83 of 164

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

Cost effectiveness

B.3.1. Published cost-effectiveness studies

• In appendix G, describe and compare the methods and results of any published cost-effectiveness analyses available for the technology and/or the comparator technologies (relevant to the technology appraisal).

• See section 3.1 of the user guide for full details of the information required in appendix G.

A systematic review of the published literature was conducted to identify all relevant economic evaluations/modelling studies for the treatment of adult patients with R/R diffuse large B-cell lymphoma (DLBCL).

The search was conducted on 27 September 2017 using the following electronic databases:

• MEDLINE and Embase (using Embase.com)

• MEDLINE In-Process (using Pubmed.com)

• EconLit (using Ebsco.com)

• The Cochrane Library (using wiley.com), including the following:

  •  National Health Service Economic Evaluation Database

  • Health Technology Assessment Database

Additionally, conference proceedings from the last 2 years (2016–2017) and data

available on HTA websites were searched to identify recently completed or ongoing studies of interest.

A total of 931 potentially relevant papers or abstracts were identified for this review. Fourteen duplicate records were excluded. After preliminary screening of abstracts, 864 records were excluded, and 53 records were included for secondary screening. After secondary screening of full text articles, 51 studies were excluded. One study was identified from HTA or conference searches. Due to the publication of multiple articles for the same study, 2 studies were extracted from 3 publications.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 84 of 164

84 of 164

Figure 13 presents the PRISMA flow diagram of studies identified for the costeffectiveness review.

Figure 13: PRISMA flow diagram for cost-effectiveness studies

==> picture [439 x 136] intentionally omitted <==

==> picture [439 x 136] intentionally omitted <==

==> picture [439 x 136] intentionally omitted <==

Key: PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; HTA, health technology assessments.

A summary of the published cost-effectiveness studies is presented in Table 23.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 85 of 164

85 of 164

Table 23: Summary list of published cost-effectiveness studies

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115]

© Kite/Gilead (2018). All rights reserved

86 of 164

Full details of the systematic review methods and results are provided in Appendix G.

B.3.2. Economic analysis

A de novo model was built to assess the cost-effectiveness of axi-cel in the treatment of adult patients with R/R DLBCL, primary mediastinal B-cell lymphoma (PMBCL) and transformed follicular lymphoma (TFL) who are ineligible for autologous stem cell transplant (ASCT).

Identified previous cost-effectiveness models did not assess CAR T therapies and may not be appropriate to model axi-cel, which has a very different mechanism of action and superior efficacy compared to current standard of care (see Section B.2). Therefore, a de novo model was developed.

More recently, NICE commissioned the Centre for Reviews and Dissemination/Centre for Health Economics, University of York (the York team) to explore the suitability of current NICE technology assessment guidelines for the assessment and appraisal of regenerative medicines and cell therapy products (e.g. CAR T therapy). As a result, the York team published a report detailing the findings, which also included two examples of de novo cost-effectiveness models for CAR T therapy for the treatment of acute lymphocytic leukaemia (ALL) as a “bridge” to stem cell transplant (SCT) or with “curative intent”.[68] This report and the de novo models developed will be referred to as the York study in this document. Although the costeffectiveness models were based on hypothetical data, these models, especially the model for the CAR T therapy with “curative intent”, are highly relevant to this analysis. Therefore, the de novo model developed for this analysis has drawn heavily on the York study.

Patient population

As outlined in Section B.1.1, in accordance with the anticipated European licence,

axi-cel is indicated for ''''''' ''''''''''''''''''''''' ''''' '''''''''''' ''''''''''''''''''' ''''''''''' '''''''''''' ''''''''''''' ''''''''''''''''''''''''''

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

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

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 87 of 164

The chosen population is in line with what was discussed in the decision problem and the scope, and reflects the population of ZUMA-1, from which efficacy and safety data for axi-cel are derived.

Model structure

A partitioned survival model with three health states (pre-progression, postprogression and death) was selected as the model structure. The three-state partitioned survival model is widely used in oncology modelling, including NICE submissions[69] , and is also the method used by the York study for the modelling of CAR T therapy with “curative intent”.[68] A state transition approach is not used for this analysis because the lack of PFS data in the SCHOLAR-1 study (which provides efficacy data for the comparator arm), and hence it is not possible to directly estimate pre-progression survival or post-progression survival, which are required for the state transition model for the comparator arm. Furthermore, the state transition approach assumes that OS is dependent on PFS (i.e. post progression survival is explicitly used to model the transition probabilities from progressed to death), which is not in line with the expected long-term survivors in the axi-cel arm.

All patients enter the model in the pre-progression health state, having progressed on previous treatment(s) for either DLBCL, PMBCL or TFL. Patients remain in the pre-progression health state until they experience disease progression or die. Once patients enter the post-progression state, where they remain until death. The model structure is presented below (Figure 14).

Figure 14: Model structure

==> picture [343 x 97] intentionally omitted <==

==> picture [343 x 97] intentionally omitted <==

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 88 of 164

88 of 164

In a partitioned survival model, OS and PFS are modelled independently and the proportions of patients who are progression-free, progressed and dead over time are derived directly from the OS and PFS curves which are calculated as “PFS”, “OS – PFS”, and “1-OS”, respectively.[69]

The OS and PFS KM data for axi-cel based on the latest ZUMA-1 combined Phase 1 and 2 data cut (August 2017) are presented in Figure 15.[7] The corresponding hazard plots are presented in Figure 16. Despite the relatively short follow-up period and small number of patients at risk, it seems a plateau began to emerge in the PFS data from around 6 months, and a similar trend for OS also seems to be evident after around 16 months. The flat tails of OS and PFS are suggestive of a proportion of patients experiencing long-term remission and survival.

Figure 15: OS and PFS in ZUMA-1 combined Phase 1 and 2 (August 2017 data cut)

==> picture [452 x 182] intentionally omitted <==

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 89 of 164

89 of 164

==> picture [452 x 179] intentionally omitted <==

Key: PFS, progression-free survival; OS, overall survival. Note: OS and PFS data are from the 12-month data cut.

Figure 16: OS and PFS in ZUMA-1 combined Phase 1 and 2: cumulative hazard

plots

==> picture [397 x 244] intentionally omitted <==

Key: PFS, progression-free survival; OS, overall survival.

Modelling OS for axi-cel

The modelling of OS for axi-cel is based on the patient level data collected in the latest combined Phase 1 and 2 ZUMA-1 data cut (August 2017).[7] All survival analyses for axi-cel were conducted using the modified intent-to-treat (mITT)

population from combined Phase 1 and 2 of ZUMA-1 (N = 108; i.e. those patients

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 90 of 164

90 of 164

who received at least 1 x 10[6] anti-CD19 CAR T-cells/kg body weight). This population will be referred to as Phase 1/2 ZUMA-1 in subsequent sections.

As stated above, the partitioned survival approach is used whereby OS is fitted independently (from PFS). The steps and processes suggested in TSD 14 were followed for the choice of appropriate modelling and extrapolation methods.[70] Specifically, the visual fit, statistical fit and clinical plausibility are all considered when assessing the plausibility of different approaches/models for OS.

Apart from the standard single parametric curves suggested in TSD 14, the more complex method of a mixture cure model[71, 72] was also applied to model OS for axicel. Regarding the statistical model’s name, “cure” relates to the assumption that a proportion of patients will experience long-term remission and thus will have a mortality rate equivalent to the age and sex matched population. The mixture cure model has been used as the base case method because: (i) there is a strong biomedical rationale for believing a proportion of those patients treated with axi-cel will have an excellent long-term prognosis (with a risk of mortality similar to the general population, and (ii) the standard parametric curves did not provide a plausible estimate of the OS for axi-cel arm (see Section B.3.3). A mixture cure model is based on clinical rationale and has the potential to more accurately model the OS for axi-cel, especially the tail of the curve, due to the likelihood for a significant proportion of patients being long-term survivors in the axi-cel arm.

A detailed description of the methodology for mixture cure modelling is presented in Appendix L. A similar novel approach was also tested in the York study, where spline models were used to provide more options for the modelling of OS for CAR T therapy.[68] Notably, the spline model in the York study resulted in the most plausible extrapolation of OS (based on hypothetical OS data that have a plateau in the tail) and provided the best statistical fit based on Akaike/Bayesian information criteria (AIC/BIC). Spline models are not used in this analysis as the extrapolations based on spline models rely most strongly on data observed towards the end of the curve due to the fitting of knots, which allow the curve to account for changes in hazard. In the case where data are sparse towards the end of the KM, then the suitability of extrapolations would require careful consideration as these would be informed mostly by the end of the curve. The spline models also do not have a strong clinical

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 91 of 164

91 of 164

rationale compared to the mixture cure model in this case. The details of the mixture cure model are described in Section B.3.3.

The mixture-cure models fitted to ZUMA-1 data have been published.[73] The updated analysis presented here used a more recent ZUMA-1 data cut-off (August 2017 data cut-off) and explored more parametric options for mixture-cure models. Furthermore, the latest UK general population mortality was used in the model to estimate mortality for “cured” patients.

Modelling OS for the comparator

The modelling of OS for the comparator is based on the patient level data that was collected in the SCHOLAR-1 study.

Subjects in SCHOLAR-1 may be refractory to therapy at multiple times throughout the treatment course. Therefore, refractory subgroup was classified in 2 ways. The first was based on the refractory status at the first time in the treatment course the subject was determined to be refractory (“first refractory categorisation”). The second was based on the refractory status at the last time in the treatment course the subject was determined to be refractory (“last refractory categorisation”). The “first refractory categorisation” maximises the subject cases included in the SCHOLAR-1 analysis. The latter is consistent with how analyses of the ZUMA-1 study were conducted and therefore more appropriate to be used for comparisons of SCHOLAR-1 with the ZUMA-1 study. Based on “last refractory categorisation”, 593 SCHOLAR-1 evaluable patients were used in this analysis, among which 562 patients were evaluable for survival.

For the model base case, the SCHOLAR-1 data were adjusted by removing patients with an ECOG score of 2–4. This adjustment was performed because only patients with ECOG 0–1 were recruited in ZUMA-1 trial based on the trial protocol.

Additional to the preferred base case, three further options of SCHOALR-1 data were presented as scenario analyses. The base case and scenarios of SCHOLAR-1 data used in the model are summarised in Table 24.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 92 of 164

92 of 164

Table 24: SCHOLAR-1 data source scenarios

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 93 of 164

93 of 164

comparability between ZUMA-1 and SCHOLAR-1

The KMs for the different SCHOLAR-1 data sources are presented in Figure 17. The figure shows that removing ECOG 2–4 patients or propensity score adjustment appear to have a minimal impact on KM survival of the unadjusted overall SCHOLAR-1 patients. The only adjustment that resulted in significantly different (worse in this case) KM survival is the adjustment with both ECOG 2–4 and postrefractory SCT patients removed. This is likely due to the removal of patients with long-term survival by excluding patients with successful SCT after the refractory treatment. Because a relatively high proportion of patients in SCHOLAR-1 are given SCT compared to ZUMA-1 (''''''''''' ''''''''' ''''''''' respectively), this highlights the bias in the SCHOLAR-1 data due to the improved survival benefit of SCT.

Figure 17: Overall survival in BSC: comparison of SCHOLAR-1 datasets

==> picture [437 x 133] intentionally omitted <==

==> picture [437 x 134] intentionally omitted <==

For the base case, an adjustment was made by removing ECOG 2–4 patients, and

no formal statistical adjustment (e.g. propensity score analysis) was performed. Propensity score analysis was only performed on the overall SCHOLAR-1 patients (i.e. not removing ECOG 2–4) and only matched to the ZUMA-1 Phase 2 patients

(n=101). However, Figure 17 shows that the propensity score adjustment makes Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 94 of 164

94 of 164

very little difference to the corresponding unadjusted data. Therefore, it is expected that the propensity score analysis would only make a small difference when applied to a subset of SCHOLAR-1 patients (removing ECOG 2–4) and when matched to combined Phase 1 & 2 ZUMA-1 (n=108). Therefore, the most appropriate base case is considered the crude adjustment with ECOG 2–4 patients removed.

The patient level data (unadjusted or adjusted) for the base case and scenarios are used to fit parametric survival curves to model and extrapolate OS for the comparator arm. The use of patient level data to fit various types of standard and more advanced parametric curves for the BSC arm is one advantage of this analysis compared to other single arm studies where only aggregate data are available for the comparator arm(s). The steps and processes suggested in TSD 14 for the choice of appropriate modelling and extrapolation methods were followed.[70] Specifically, the visual fit, statistical fit and clinical plausibility are all considered when assessing the plausibility of different approaches/models for OS.

Apart from the standard single parametric curves suggested in TSD 14, the more complex method of a mixture cure model[71, 72] was also applied to model OS for BSC, which is consistent to how the axi-cel arm is modelled.

Modelling PFS for axi-cel

The modelling of PFS for axi-cel is based on the patient level data collected in the latest combined Phase 1 & 2 ZUMA-1 data cut (August 2017, n=108), using the partitioned survival approach. In general, PFS has much less impact on the estimated LYs, QALYs and ICERs compared to the OS. There is also more certainty regarding the modelling of PFS compared to OS due to the relative maturity of PFS data (see Figure 15).

To apply the partitioned survival approach, PFS is fitted independently, and the steps and processes suggested in TSD 14 were followed for the choice of appropriate modelling and extrapolation methods.[70] The standard single parametric curves (exponential, Weibull, Gompertz, log-normal, log-logistic, generalised gamma) were fitted, and the curve used in the base case was chosen based on statistical fit, visual inspection and clinical plausibility.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 95 of 164

95 of 164

Modelling PFS for the comparator

Progression status was not collected in SCHOLAR-1; therefore, the following 3 options were included in the model for the modelling of PFS for the comparator:

• Apply a time-dependent ratio to the comparator OS to derive PFS, where the timedependent ratio is derived directly from the modelled OS and PFS for axi-cel (base case)

• Assuming PFS=0 for the comparator (scenario analysis)

• Assuming PFS is the same as OS for the comparator (scenario analysis)

Modelling utility and cost and resource use

Health-related quality of life (HRQL) data were collected in a safety management cohort of ZUMA-1 (87 EQ-5D-5L observations from 34 patients), as detailed in B.3.4. This was used as the base case source for model utilities. Values for patients receiving second- and subsequent-line treatment for renal cell carcinoma (RCC)[75] were used in the scenario analyses. These were 0.76 for the pre-progression health state and 0.68 for the post-progression health state, and were used in the base case in the pixantrone company submission.[45]

To model cost and resource use, costs that were considered in the York report were used.[68] For drug and administration costs, the following sources were used: eMit, NHS reference costs, MIMS, South East London Cancer Network and Thames Valley Strategic Clinical Networks. For model resource use, NICE TA306, the Personal Social Services Research Unit (PSSRU), NHS reference costs and the National Audit Office End of Life Care report were used.

General model settings

The cost-effectiveness analysis assumes a lifetime time horizon (44 years). This approach is considered to be appropriate, given that axi-cel is associated with reduced mortality and expected long-term survivors and NICE guidance states that the time horizon should be long enough to reflect all important differences in costs or outcomes between the technologies being compared.

The model uses a cycle length of 1 month. This is anticipated to capture all the

relevant changes in the modelled cohort, considering the median OS in the BSC arm Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 96 of 164

96 of 164

is expected to be approximately 6 months. Half-cycle correction is implemented using the life table method.[76]

Costs and efficacy were discounted at 3.5% in the model base case, based on NICE method guidance[77] and the York study.[68] Due to the potential for axi-cel to provide long-term survival (the model estimates a mean undiscounted OS of >10 years) and HRQL benefits, and given that the total acquisition cost of axi-cel is incurred within the first model cycle, an alternative discount rate of 1.5% was used in a scenario analysis. This scenario analysis is especially relevant if the NICE committee decides that axi-cel qualifies for the use of a 1.5% discount rate based on the NICE method guide (section 6.2.19).[77]

Intervention technology and comparators

Intervention

The intervention, axi-cel, is implemented in the model as per the expected marketing authorisation, which is expected in June 2018, and is reflective of the decision problem described in B.1.1. Axi-cel has been granted Priority Medicines (PRIME) regulatory support for DLBCL in the European Union (EU) and has been designated an orphan product.

Axi-cel is a therapy in which a patient’s T-cells are engineered to express a chimeric antigen receptor (CAR), which recognises the antigen CD19 expressed on the surface of B-cell lymphomas and eliminates the target cells. The process of generating and administering the engineered T-cells is described in Section B.1.2.

Axi-cel is administered as a single intravenous infusion in the hospital setting. All patients receive lymphodepleting low-dose conditioning chemotherapy of 500 mg/m[2] cyclophosphamide and 30 mg/m[2] fludarabine during the 3 days prior to the infusion of anti-CD19 CAR T-cells.

Comparator

The comparator considered in this economic evaluation is BSC as a blended comparator including several therapy options, which are assumed to have the same efficacy and safety. This is in line with the NICE treatment pathway, which recommends that salvage therapy with multi-agent chemotherapy be offered to

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 97 of 164

97 of 164

individuals who are not eligible for ASCT. The use of BSC as a single comparator was also deemed appropriate based on interviews with UK clinicians.

In light of the recommendation from NICE (TA306) for the use of pixantrone monotherapy for treating multiply relapsed or refractory aggressive non-Hodgkin's B- cell lymphoma[45] , interviews with UK clinicians were carried out to establish the level of use of pixantrone in clinical practice. The UK clinicians interviewed stated that they did not use pixantrone due to disappointing clinical experience.[78] Therefore, following clinical validation, pixantrone was not included as a potential treatment in the BSC arm or as a separate comparator in the model.

During clinician interviews, several treatment regimens were identified, with no universal standard of care. The Oxford University Hospitals (OUH) NHS Foundation Trust derived a list of regimens used in UK clinical practice. These identified regimens are assumed to have equal efficacy to the regimens used in SCHOLAR-1 (which is used as the source of BSC efficacy) and were validated in the clinical adboard. The regimens include:

• Gemcitabine and methylprednisolone (GEM)

• Gemcitabine, methylprednisolone and cisplatin (GEM-P)

• Rituximab, gemcitabine, cyclophosphamide, vincristine and prednisolone (RGCVP)

• Rituximab, vinblastine and prednisolone (RVP)

B.3.3. Clinical parameters and variables

Clinical data from the Phase 1/2 trial, ZUMA-1, and a multi-cohort retrospective analysis of two randomised studies and two registries, SCHOLAR-1, were used to inform the model base case. Specifically, patient level data from ZUMA-1 were used to inform the efficacy of axi-cel, and patient level data from SCHOLAR-1 were used to derive efficacy for the BSC arm, due to the absence of a comparator arm in ZUMA-1. However, a limitation of SCHOLAR-1 is that it does not contain PFS data.

Data utilised from the trials include:

• OS

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 98 of 164

98 of 164

• PFS (ZUMA-1 only)

• Body surface area (for drug dosing)

• AE rates (ZUMA-1 only)

The data sources for each clinical parameter are detailed in Table 25.

Table 25: Data sources of clinical parameters used in the model

Efficacy data

For the BSC arm, survival data from SCHOLAR-1 are relatively mature (>85% dead at the end of follow-up [approximately 15 years after diagnosis]). In contrast, survival data derived from ZUMA-1 for axi-cel are immature (48% dead at the end of the 2- year follow-up), and therefore, the appropriate extrapolation of the data was vital to allow the long-term treatment effects to be estimated.

Overall survival – axi-cel arm

Standard single parametric curves for partitioned survival approach

A partitioned survival model (PSM) with standard single parametric curves has been widely used in oncology modelling and was employed in the York report. A variety of single parametric survival curves were fit and applied to the axi-cel OS data in the model using the coefficient data presented in Table 26. These are presented graphically in Figure 18.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 99 of 164

99 of 164

Table 26: Overall survival for axi-cel: PSM with single parametric curves coefficients

Figure 18: Overall survival for axi-cel: KM with single parametric curves

==> picture [430 x 129] intentionally omitted <==

==> picture [430 x 128] intentionally omitted <==

The goodness of fit statistics, in terms of AIC and BIC, are presented in Table 27 for each curve fit option.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 100 of 164

100 of 164

Table 27: Overall survival for axi-cel: PSM with single parametric curves goodness of fit statistics

Plots of modelled cumulative hazard over time and observed cumulative hazard over time are presented in Figure 19.

Figure 19: Overall survival for axi-cel: PSM with single parametric curves logcumulative hazard plot

==> picture [398 x 122] intentionally omitted <==

==> picture [398 x 123] intentionally omitted <==

Key: KM, Kaplan–Meier.

Based on AIC and BIC statistics, the loglogistic and exponential curve fits provide

the best statistical fit. However, none of the standard parametric curves provide a

plausible extrapolation of the long-term survival for axi-cel arm. Based on the

mechanism of action of axi-cel and expert opinion from clinicians, it is expected that

the tail seen towards the end of the observed KM for axi-cel (based on the Phase 1/2 Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 101 of 164

101 of 164

ZUMA-1 data) would continue and plateau as these are likely long-term survivors and would have the same mortality as the gender- and age-matched general population. Furthermore, by Year 10, the estimated OS based on all standard parametric curves are lower than the observed OS for the comparator arm based on the SCHOLAR-1 data; which is deemed clinical implausible given the significant survival benefit for axi-cel compared to SCHOLAR-1 data observed during the ZUMA-1 trial period. Therefore, standard single parametric curves were not deemed clinically plausible, and were therefore not considered further in this analysis.

Mixture cure models for partitioned survival approach

As described previously, a mixture-cure model was used as an alternative method to model OS for the axi-cel treatment arm to be used in the partitioned survival approach. The mixture cure model was estimated using the Phase 1/2 ZUMA-1 patient level data for which a logistic regression was used to model the probability that patients experienced long-term remission, and parametric models were used to estimate the survival for those without long-term remission. In the model, patients with long-term remission were assumed to have the age- and gender-matched background mortality, derived from population life tables.[76] Expert clinical opinion suggests that patients with long-term remission can be assumed to have the same mortality as age- and gender-matched general population. The model has the option to assume additional mortality risk (compared to the general population) for longterm remission patients over and above the general population mortality, and these were tested in the scenario analyses.

Given the expectation and clinical opinion that a proportion of patients experience long-term remission based on Phase 1/2 ZUMA-1 KM data (see Figure 15), and given the plateau shown in the relatively mature SCHOLAR-1 data for BSC patients (Figure 17), the mixture cure model is deemed a more appropriate method for accounting for these long-term survivors. The rationale and methodology behind mixture cure models is further discussed in Appendix L.

The logistic regression model estimated the “cure fractions” (proportion of patients in long-term remission), and three standard parametric models were fitted to estimate survival on the proportion of patients not experiencing long-term remission: Weibull,

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 102 of 164

102 of 164

gamma and lognormal. The parameter information for each of these is presented in Table 28.

It should be noted that the estimated cured fractions differ dependent on the parametric curve chosen to represent the patients not in long-term remission, as it is the combination of the estimated cure fraction, general population mortality for patients in long-term remission and the estimated survival (Weibull, gamma and lognormal) for patients not in long-term remission that jointly produces the overall OS for the axi-cel arm.

Table 28: Overall survival for axi-cel: mixture-cure model coefficients

Combining the estimated cure fraction, the general population mortality (for “cured” patients) and the fitted parametric curves for “not cured”, Figure 20 shows the overall estimated OS for each mixture-cure model compared to the observed OS KM.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 103 of 164

103 of 164

Figure 20: Overall survival for axi-cel: KM with mixture cure model parametric

curves

==> picture [431 x 128] intentionally omitted <==

==> picture [431 x 129] intentionally omitted <==

Key: MCM, mixture cure model.

Statistical goodness of fit of each distribution is presented in Table 29 using AIC and BIC statistics.

Table 29: Overall survival for axi-cel: mixture cure model goodness of fit

statistics

Figure 21 shows plots of modelled cumulative hazard over time and observed cumulative hazard over time.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 104 of 164

104 of 164

Figure 21: Overall survival for axi-cel: mixture-cure model log-cumulative

hazard plot

==> picture [398 x 122] intentionally omitted <==

==> picture [398 x 123] intentionally omitted <==

Key: KM, Kaplan–Meier; MCM, mixture cure model.

The comparison of modelled OS with observed OS and modelled cumulative hazard over time with observed cumulative hazard over time suggested that Weibull and gamma mixture-cure models provided plausible OS and hazard predictions compared to the observed KM.

The estimated cure fraction for the lognormal mixture-cure model is close to 1% (see Table 28), and consequently, there is not much difference between the lognormal mixture-cure model and the standard lognormal model. The rationale to assess standard parametric curves applies, and hence, the lognormal mixture-cure model is deemed clinically implausible and excluded from further analyses.

Considering statistical fit, visual inspection and the clinical rationale for a substantial proportion of the treated population having long-term remission, the Weibull mixturecure model was used in the base case analysis of OS for axi-cel and is presented below in Figure 22.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 105 of 164

105 of 164

Figure 22: Overall survival for axi-cel: mixture-cure method

==> picture [363 x 116] intentionally omitted <==

==> picture [363 x 117] intentionally omitted <==

Key: MCM, mixture cure model.

The gamma mixture-cure model is tested in a scenario analysis.

Overall survival – BSC arm

Standard Parametric survival curves for OS for BSC

Using the SCHOLAR-1 patient level data with ECOG 2–4 removed, standard single parametric survival curves were fitted for the modelling of OS for the comparator arm (see Table 30 and Figure 23 for coefficients estimated for each parametric curve and the fitted curves compared to adjusted SCHOLAR-1 OS).

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 106 of 164

106 of 164

Table 30: Overall survival for BSC: curve fit coefficients

Figure 23: Overall survival for BSC: PSM with single parametric curves logcumulative hazard plot

==> picture [424 x 131] intentionally omitted <==

==> picture [424 x 131] intentionally omitted <==

AIC and BIC statistics for each curve fit are presented in Table 31.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 107 of 164

107 of 164

Table 31: Overall survival for BSC: goodness of fit statistics

Plots of modelled cumulative hazard over time and observed cumulative hazard over time are presented in Figure 24.

Figure 24: Overall survival for BSC: PSM with single parametric curves logcumulative hazard plot

==> picture [399 x 122] intentionally omitted <==

==> picture [399 x 123] intentionally omitted <==

Key: K-M, Kaplan–Meier.

Based on AIC and BIC statistical fit, Gompertz is the best fit curve. This is also supported by visual inspection and clinical opinion. Therefore, the Gompertz curve was chosen as the most suitable single parametric curve as it has a good visual fit, reasonably good AIC/BIC statistics and better represents the long-term survivors as observed in the SCHOLAR-1 data. This is presented in Figure 25.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 108 of 164

108 of 164

Figure 25: Overall survival for BSC: PSM with single parametric curves

selected distribution

==> picture [424 x 131] intentionally omitted <==

==> picture [424 x 131] intentionally omitted <==

For each of the alternative SCHOLAR-1 data sources that are explored as part of the scenario analysis, the Gompertz distribution was also chosen as the best fit to the SCHOLAR-1 OS data based on visual inspection and AIC/BIC statistics.

Mixture cure models for OS for BSC

Consistent with the approach used to model axi-cel OS, a mixture cure model was estimated using the SCHOLAR-1 patient level data with ECOG 2–4 removed and followed the same methodology as for the axi-cel arm.

The logistic regression model estimated the “cure fractions” and three parametric models were fitted to estimate survival on the proportion of patients not experiencing long-term remission: Weibull, gamma and lognormal. The parameter information for each of these is presented in Table 32.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 109 of 164

109 of 164

Table 32: Overall survival for BSC: mixture-cure model coefficients

Combining the estimated cure fraction, the general population mortality (for “cured” patients) and the fitted parametric curves for “not cured”, Figure 26 shows the fit of all the distributions to the observed OS KM.

Figure 26: Overall survival for BSC: KM with mixture cure model parametric

curves

==> picture [430 x 128] intentionally omitted <==

==> picture [430 x 129] intentionally omitted <==

Key: MCM, mixture cure model.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 110 of 164

110 of 164

The goodness of fit of the curve fits used for the “not cured” proportion are presented in Table 33 using AIC and BIC statistics.

Table 33: Overall survival for BSC: mixture cure model goodness of fit statistics

Figure 27 shows plots of modelled cumulative hazard over time and observed cumulative hazard over time.

Figure 27: Overall survival for BSC: mixture-cure model log-cumulative hazard plot

==> picture [399 x 122] intentionally omitted <==

==> picture [399 x 123] intentionally omitted <==

Key: K-M, Kaplan–Meier; MCM, mixture cure model.

The comparison of modelled OS and observed OS and modelled cumulative hazard

over time and observed cumulative hazard over time suggest that all parametric models fit the KM very well and are tested in scenario analyses. The mixture-cure models are not used in the base case as the standard (simpler) Gompertz parametric model has a good statistical and visual fit.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 111 of 164

111 of 164

Progression-free survival – axi-cel arm

As discussed previously, a partitioned survival approach was used to model PFS for the axi-cel arm. A variety of single parametric survival curves were fit and applied to the axi-cel PFS data in the model using the coefficient data presented in Table 34. The curve fits are presented graphically in Figure 28.

Table 34: Progression-free survival for axi-cel: curve fit coefficients

Figure 28: Progression-free survival for axi-cel: KM with single parametric

curves

==> picture [430 x 128] intentionally omitted <==

==> picture [430 x 128] intentionally omitted <==

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 112 of 164

112 of 164

The AIC and BIC goodness of fit statistics for each parametric curve fitted to axi-cel PFS are presented in Table 35.

Table 35: Progression-free survival for axi-cel: PSM with single parametric

curves goodness of fit statistics

Plots of modelled cumulative hazard over time and observed cumulative hazard over time are presented in Figure 29.

Figure 29: Progression-free survival for axi-cel: PSM with single parametric

curves log-cumulative hazard plot

==> picture [398 x 123] intentionally omitted <==

==> picture [398 x 123] intentionally omitted <==

Key: KM, Kaplan–Meier .

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 113 of 164

113 of 164

From the goodness of fit statistics, the Gompertz curve fit demonstrates the best statistical fit while also having a good visual fit and was used in the model base case. The log cumulative hazard plot further demonstrates that the Gompertz curve provides the most plausible fit to the observed data, particularly toward the end of the observed PFS.

Figure 30 presents the base case curve fit alongside the KM for axi-cel.

Figure 30: Progression-free survival for axi-cel: PSM with single parametric curves selected distribution

==> picture [417 x 123] intentionally omitted <==

==> picture [417 x 124] intentionally omitted <==

Alternative curve fits were explored as part of the scenario analysis.

Progression-free survival – BSC arm

Due to the lack of PFS data from SCHOLAR-1, PFS for BSC was estimated by assuming that the same ratio between PFS and OS at each time point in the axi-cel arm can be applied to the BSC arm. Applying this ratio to the OS data of the BSC arm allowed for a PFS curve to be constructed, as shown in Figure 31. It is

acknowledged that this is a limitation of the SCHOLAR-1 data but is thought to be the most appropriate method to overcome the lack of PFS data for the comparator.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 114 of 164

114 of 164

Figure 31: Progression-free survival for BSC, constructed from BSC overall

survival

==> picture [387 x 116] intentionally omitted <==

==> picture [387 x 117] intentionally omitted <==

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

The assumption that 100% of time spent alive in the BSC arm is spent in the progression-free state or that 100% of time spent alive in the BSC arm is spent in the progressed state were tested in scenario analyses.

Time on treatment and retreatment

Time on treatment was not explicitly reported in the SCHOLAR-1 study and is not relevant to the ZUMA-1 trial, as axi-cel is given as a one-off infusion. For the modelled BSC regimens, the number of treatment cycles and the days per cycle for each component of each regimen are taken from the South East London Cancer Network[79] and the Thames Valley Strategic Clinical Networks.[80] These are presented in Table 36.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 115 of 164

115 of 164

Table 36: Cycles per course and days per cycle

In the ZUMA-1 trial, although axi-cel was given as a one-off infusion, some patients were retreated in line with the trial protocol (10/108 subjects were retreated based on the August 2017 data cut; 9 patients from Phase 2 and 1 patient from Phase 1 trial). Although the expected market authorisation does not allow retreatment of axi-cel, for this analysis, these retreated patients were not removed or censored (at the time of re-treatment) for the PFS or OS endpoints. Based on the ZUMA-1 trial protocol, patients had to have progressed on axi-cel to be eligible for retreatment; therefore, censoring was not required for PFS. For OS, censoring at the time of re-treatment with axi-cel was not done to be consistent with subsequent treatments across all ZUMA-1 study patients, i.e. whether patients underwent retreatment or a new anticancer therapy following the initial axi-cel treatment. Based on best overall responses per investigator, among the 9 retreated patients from the Phase 2 trial, '''' '''''''''''''' ''''''''''' '''' '''''''''''''' patients had complete and partial response, respectively; '''' '''''''''''''' patient had stable disease and '''' '''''''''''''''' patients had progressed disease. It is therefore assumed that including the retreated patients in ZUMA-1 would have minimal impact on the OS for the axi-cel arm.

To account for additional costs of these axi-cel retreated patients in the model, additional costs for conditioning chemotherapy, cell infusion and monitoring were applied to the 9.3% retreated population (i.e. 10/108). As the quantity of axi-cel

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 116 of 164

116 of 164

initially manufactured is sufficient for the delivery of up to two treatments based on the ZUMA-1 trial protocol, no additional leukapheresis or axi-cel acquisition costs are applied to the re-treated patients.

Body surface area (BSA)

The dosing of the conditioning chemotherapy (500 mg/m[2] cyclophosphamide and 30 mg/m[2] fludarabine) is based on patient BSA, as are most of the BSC regimens. Therefore, estimation of the distribution around patient BSA for the patient population was required and was derived from the Phase 2 ZUMA-1 trial (n=101).

The optimal combinations of vial sizes (achieve lowest drug cost by assuming no vial sharing) were calculated for each range of BSA (see Table 37 and Table 38 for conditioning chemotherapy and BSC regimens, respectively). The proportions of patients belonging to each range of BSA were calculated using the ZUMA-1 patientlevel data.

Table 37: Optimal combinations of vial sizes for conditioning chemotherapy by BSA

Table 38: Optimal combinations of vial sizes for BSC by BSA

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 117 of 164

117 of 164

Adverse event rates

For axi-cel patients, AE rates were captured in the ZUMA-1 trial. The AE model inputs were based on the Phase 2 ZUMA-1 trial January 2017 data cut-off (n=101). Specifically, the following AEs were modelled:

• Grade 3 or higher axi-cel-related AEs occurring in ≥10% of subjects in ZUMA-1

• Grade 3 or higher conditioning chemotherapy-related AEs occurring in ≥10% of subjects in ZUMA-1

• Grade 3 or higher treatment-emergent cytokine release syndrome (CRS) occurring in ZUMA-1

No Grade 3 or higher leukapheresis-related AEs occurred in ≥10% of subjects in ZUMA-1. Grade 3 or higher leukapheresis-related AEs that occurred in <10% of patients include anaemia, decreased white blood cell count, decreased platelet count, decreased lymphocyte count, decreased neutrophil count and vomiting. Only one Grade 4 leukapheresis-related AE occurred (platelet count decreased).

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 118 of 164

118 of 164

The incidence of modelled Grade 3+ axi-cel related AEs by AE type are presented in Table 39, while the incidence of AEs due to conditional chemotherapy are presented in Table 40.

Table 39: Incidence of Grade 3+ axi-cel-related AEs occurring in ≥10% subjects

(N=101)

Table 40: Incidence of Grade 3+ conditioning chemotherapy-related AEs occurring in ≥10% subjects (N=101)

No AE data were collected in the SCHOLAR-1 trial. Therefore, as a conservative assumption, no AEs were modelled for the BSC arm. The rates of AE incidence affect both the quality of life and cost aspects of the model, which are discussed further in Sections B.3.4 and B.3.5, respectively.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 119 of 164

119 of 164

B.3.4. Measurement and valuation of health effects

Health-related quality-of-life data from clinical trials

HRQL data were collected in a safety management cohort of ZUMA-1. The key limitations of the EQ-5D data are that it is from a relatively small sample (87 observations from 34 patients), and EQ-5D-5L was used instead of EQ-5D-3L, which is preferred by NICE for technology appraisals.[81] As suggested by NICE, the crosswalk algorithm proposed by van Hout et al. (2012)[82] was used to convert EQ5D-5L to EQ-5D-3L, and the resulting data are used to derive utilities for progression-free and progressed patients in the model base case (see Table 43).

Alternative utilities identified from the literature were used in scenario analyses.

Health-related quality-of-life studies

In appendix H describe how systematic searches for relevant health-related quality-of-life data were done.

Identification of studies

A systematic review of the published literature was conducted to identify all relevant utility and HRQL studies for the treatment of adult patients with R/R DLBCL.

The search was conducted on 07 September 2017 using the following electronic databases:

• MEDLINE and Embase (using Embase.com)

• MEDLINE In-Process (using Pubmed.com)

• EconLit (using Ebsco.com)

• The Cochrane Library (using wiley.com), including the following:

  •  National Health Service Economic Evaluation Database

  • Health Technology Assessment Database

Additionally, conference proceedings from the last 2 years (2016–2017) and data

available on HTA websites were searched to identify recently completed or ongoing studies of interest.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 120 of 164

120 of 164

A total of 1,438 potentially relevant papers or abstracts were identified for this review. Studies were screened based on the information reported in their titles and/or abstracts. Of these, 5 were removed as duplicates, and 1,371 were excluded at the primary screening stage as they were not relevant to the research question.

Sixty-two articles were assessed in full for further evaluation. Of these, 61 were excluded, and 1 was included. Additionally, 3 studies were included from costeffectiveness review. Therefore, 4 citations were included for this review. Due to the publication of multiple articles for the same study, 3 studies were extracted from 4 publications.

Figure 32 presents the PRISMA flow diagram of studies identified for the utility review.

Figure 32: PRISMA flow diagram for utility studies

==> picture [451 x 101] intentionally omitted <==

==> picture [451 x 102] intentionally omitted <==

==> picture [451 x 101] intentionally omitted <==

==> picture [451 x 102] intentionally omitted <==

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 121 of 164

121 of 164

Key: PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; HTA, health technology assessments

Studies that met the inclusion criteria of the review

Of the three studies included, two were economic modelling studies from the US, and one was a UK HTA.

The study populations in all three studies consisted of DLBCL patients, although the line of therapy differed between studies. One study specifically considered relapsed DLBCL patients, one looked at DLBCL patients after first remission, and one only specified DLBCL.

Details of the data collected are presented in Table 41. A further description of the findings is provided in Appendix H.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 122 of 164

122 of 164

Table 41: Characteristics and results of included utility studies

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115]

© Kite/Gilead (2018). All rights reserved

123 of 164

Adverse reactions

AEs associated with axi-cel are detailed in Section B.3.3. These are expected to occur in the short term after the initial treatment of axi-cel; therefore, a one-off QALY decrement is applied in the first model cycle.

Utility decrements for anaemia, febrile neutropenia, neutropenia, platelet count decrease, pyrexia and thrombocytopenia were identified in the pixantrone submission to NICE.[45] As in the York study, it is conservatively assumed that those experiencing cytokine release syndrome (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).[68] Data on utility decrements for encephalopathy, hypophosphatemia, hypotension, leukopenia, decreased neutrophil count, and decreased white blood cell count were not identified. For each of these AEs, a disutility equal to the maximum of the identified non-CRS AE disutilities was assumed. This approach was also taken in the pixantrone submission to NICE.[68]

Total AE durations were calculated using patient-level data from ZUMA-1. Durations were calculated as the total number of days that each patient experiences a specific AE, even if that event was experienced more than once. This is consistent with the use of the proportion of patients experiencing each AE, rather than the rate of each event. AEs were ongoing in eight of 712 observations; given the small number of missing end dates, these observations were excluded. The impact of this approach is expected to be minimal.

AE disutilities and durations, and their respective data sources, are presented in Table 42.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 124 of 164

124 of 164

Table 42: Adverse event disutilities

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

In the model base case, health state utilities were based on EQ-5D data collected in a safety management cohort of ZUMA-1 (n=34, with 87 observations). Utilities derived from the NICE multiple technology appraisal of bevacizumab, sorafenib, sunitinib and temsirolimus in advanced/metastatic renal cell carcinoma[75] , which was also used in NICE TA306, were used in a scenario analysis.

The model also assumes that if patients have remained in the PFS state for 2 years, they are classed as being in long-term remission and are thus assumed to have equal utility values as the age and gender matched general population after this point.[85] This assumption is supported by the findings of Maurer et al. (2014), where the survival of DLBCL patients treated with immunochemotherapy was compared to

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 125 of 164

125 of 164

that of the general population.[48] It was found that, in the DLBCL patients who were disease-free at 24 months, there was no significant difference in subsequent survival compared with that for the general population.

In scenario analyses, a percentage decrement to the age and gender matched general population utility values are applied.

Table 43: Summary of utility values for cost-effectiveness analysis

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 126 of 164

126 of 164

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

In appendix I describe how relevant cost and healthcare resource data were identified.

Intervention and comparators’ costs and resource use

Axi-cel treatment costs

For axi-cel, the costs included in the model were the costs of leukapheresis, conditioning chemotherapy, acquisition cost of axi-cel, and cell infusion and monitoring. For simplicity, all costs associated with axi-cel are assumed to be incurred in the first model cycle, including those associated with retreatment.

Leukapheresis

In Phase 1/2 of ZUMA-1, 119 subjects underwent leukapheresis, 110 subjects received conditioning chemotherapy, and 108 subjects received axi-cel. Of the nine subjects not treated with either conditioning chemotherapy or axi-cel:

• Two subjects died due to disease progression prior to treatment

• Four subjects experienced AEs that precluded treatment

• Two subjects had non-measurable disease prior to treatment

• One subject discontinued due to the need for immediate therapy for disease

  • progression

The cost of leukapheresis was calculated as the weighted average of all healthcare resource groups (HRGs) for stem cell and bone marrow harvest in the NHS national

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 127 of 164

127 of 164

schedule of reference costs (Table 44).[86] This approach was also taken by the authors of the NICE regenerative medicines report.[68]

Table 44: Unit costs of leukapheresis

The weighted average cost of leukapheresis was calculated to be £1,284.77. An uplifting factor of 1.102 (119/108) was used to adjust the unit leukapheresis cost and used in the model, to account for patients who undergo leukapheresis but do not proceed to receive axi-cel. Therefore, the model cost of leukapheresis is £1,415.63.

Conditioning chemotherapy

Conditioning chemotherapy includes intravenous infusions of cyclophosphamide 500 mg/m[2] and fludarabine 30 mg/m[2] on the 5[th] , 4[th] and 3[rd] days prior to infusion of axicabtagene ciloleucel. Unit costs for cyclophosphamide and fludarabine were taken from the NHS drug and pharmaceutical electronic market information tool (eMit)[87] , and are presented in Table 45.

Table 45: Unit costs of conditioning chemotherapy

The costs of the conditioning chemotherapy were derived after calculating the optimal combination of the different vial sizes, considering the population BSA data based on patient level data from ZUMA-1 (Table 37). It was assumed that unused chemotherapy remaining in vials is wasted.

Conditioning chemotherapy is assumed to require a non-elective long-stay hospitalisation, in line with assumptions taken in the NICE regenerative medicines report.[68] The cost of a non-elective long-stay hospitalisation is calculated as the

weighted average of non-elective long-stay HRGs for malignant lymphoma, including Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 128 of 164

Hodgkin's and Non-Hodgkin's, in the NHS national schedule of reference costs[86] ; see Table 46.

Table 46: Malignant lymphoma non-elective long-stay HRGs

The weighted average cost of hospitalisation for conditioning chemotherapy was calculated to be £5,062.63.

A multiplier of 1.019 (110/108) was used to adjust both the conditioning chemotherapy cost and the hospitalisation cost for conditioning chemotherapy to account for the two patients in ZUMA-1 who were treated with conditional chemotherapy but not axi-cel. Furthermore, the additional costs of conditioning chemotherapy for the retreated patients were also accounted for, as discussed in Section B.3.3. Therefore, the model cost of conditional chemotherapy, including hospitalisation, is £5,856.77.

Drug acquisition

The acquisition cost of axi-cel is assumed to be a one-off cost of '''''''''''''''''''' including all shipping, engineering and generation of the CAR T-cells.

A multiplier is not applied to the acquisition cost of axi-cel to account for patients who were recruited in ZUMA-1 but did not receive the axi-cel treatment. It is assumed the cost of the drug will only be reimbursed if axi-cel is administered to the patient.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 129 of 164

129 of 164

Cell infusion and monitoring

The infusion of axi-cel and subsequent monitoring is assumed to incur the cost of an elective hospitalisation, in line with the assumption taken in the NICE regenerative medicines report.[68] However, the mean length of stay observed in the ZUMA-1 trial for axi-cel was 17.6 days, which is over a week longer than that reported for malignant lymphoma (including Hodgkin’s and non-Hodgkin’s) inpatient admissions in Hospital Episode Statistics[88] , which is 10.4 days. To cost this in the model, the weighted average of elective inpatient HRGs for malignant lymphoma, including Hodgkin's and Non-Hodgkin's, in the NHS national schedule of reference costs[86] was used, plus the weighted average cost of 7.2 elective inpatient excess bed day HRGs.

Table 47: Malignant lymphoma elective inpatient HRGs

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 130 of 164

130 of 164

Table 48: Malignant lymphoma elective inpatient excess bed day HRGs

The weighted average cost of elective inpatient HRGs was calculated to be £3,716.28. The weighted average cost of elective inpatient excess bed day HRGs was calculated to be £422.79. The total cost for cell infusion and monitoring was therefore calculated to be £6,760.37 (i.e. 3716.28 + [7.2 x 422.79]).

BSC treatment costs

The BSC arm is applied as a blended comparator (BSC), which is comprised of four different regimens, as detailed in Section B.1.1. The model applies costs for each regimen, multiplied by their distribution of use in the UK. In the base case, without an identified better source of evidence, the four regimens are assumed to be distributed equally, i.e. 25% of each regimen is prescribed.

Drug acquisition

The treatments included in each regimen are a mixture of both orally administered and intravenously administered drugs, based on BSA, as presented in Table 49 and Table 50, respectively. For the chemotherapies where dosing is dependent on BSA, the optimal combinations of the vial sizes required for each range of BSA were calculated.

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 131 of 164

131 of 164

Table 49: Unit costs of chemotherapies that are not based on BSA

Table 50: Unit costs of chemotherapies based on BSA

Treatment duration data are detailed in Table 36 and BSA data are presented in Table 38.

Drug administration

The administration of BSC is assumed to incur the cost of a non-elective hospitalisation, as described previously for the cell infusion and monitoring costs in axi-cel.

Health-state unit costs and resource use

Medical resource use required is dependent on progression status and is thus modelled by applying different costs for each health state. The model considers costs associated with the following:

• Professional and social services (Table 51)

• Health care professionals (Table 52)

• Treatment follow-up (Table 53)

Company evidence submission template for axicabtagene ciloleucel for treating relapsed or refractory diffuse large B-cell non-Hodgkin lymphoma [ID1115] © Kite/Gilead (2018). All rights reserved 132 of 164

132 of 164

• Hospital resource use (Table 54)

Resource use in each health state was estimated from a survey of three key opinion leaders, commissioned by the manufacturer of pixantrone to support the pixantrone submission to NICE.[45] It is acknowledged that the population addressed by pixantrone differs to that of axi-cel, and that the estimates are only based on the opinions of three clinicians. This uncertainty is dealt with in the scenario analyses, in which medical resource use costs are doubled and halved to assess the impact of changes in these costs.

The pixantrone submission to NICE considered three health states to which the medical resource use is applied:

• PFS on 3[rd] (or 4[th] ) line treatment

• PFS, discontinued 3[rd] or 4[th] line treatment

• PD