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

Single Technology Appraisal

Olaparib for maintenance treatment of BRCAmutated ovarian, fallopian tube and peritoneal cancer after response to first-line platinum-based chemotherapy [ID1124]

Committee Papers

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

NATIONAL INSTITUTE FOR HEALTH AND CARE EXCELLENCE

SINGLE TECHNOLOGY APPRAISAL

Olaparib for maintenance treatment of BRCA-mutated ovarian, fallopian tube and peritoneal cancer after response to first-line platinum-based chemotherapy [ID1124]

Contents:

The following documents are made available to consultees and commentators:

The final scope and final stakeholder list are available on the NICE website.

1. Company submission from AstraZeneca

2. Company response to NICE’s request for clarification

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

• a. Ovacome b. Ovarian Cancer Action

4. Expert personal perspectives from:

• a. Prof. Charley Gourley, chair of medical oncology and honorary consultant – clinical expert nominated by AstraZeneca

• b. Prof. Jonathan Ledermann, professor of medical oncology – clinical expert nominated by the British Gynaecological Cancer Society

• c. Rebecca Rennison, director of public affairs and services – patient expert nominated by Target Ovarian Cancer

• d. Florence Wilks – patient expert nominated by Ovarian Cancer Action e. Prof. Peter Clark – NHS England Cancer Drugs Fund clinical lead

5. Evidence Review Group report prepared by School of Health and Related Research (ScHARR)

6. Evidence Review Group – factual accuracy check

7. Technical engagement response from AstraZeneca

8. Technical engagement responses from experts: a. Prof. Jonathan Ledermann, professor of medical oncology – clinical expert nominated by the British Gynaecological Cancer Society

9. Technical engagement response from consultees and commentators: a. Royal College of Physicians joint with ACP-NCRI-RCR

10. Evidence Review Group critique of company response to technical engagement prepared by School of Health and Related Research

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

11. Final Technical Report

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

NATIONAL INSTITUTE FOR HEALTH AND CARE EXCELLENCE

Technology appraisal

Olaparib for maintenance treatment of newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer, after response to first-line platinum-based chemotherapy [ID1124]

Company evidence submission Document B

November 2018

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Contents

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table of Figures

Figure 1 Current and proposed position use of olaparib in the treatment pathway for patients with

BRCA-mutated advanced ovarian cancer .................................................................................... 16 Figure 2 SOLO1 study schema[63,64] ....................................................................................................... 22 Figure 3 SOLO1 patient disposition ...................................................................................................... 26 Figure 4 Kaplan-Meier plot of PFS (investigator-assessed) ................................................................. 32 Figure 5 Proportion of patients who remained progression-free........................................................... 32 Figure 6 Kaplan-Meier plot of PFS by BICR assessment ..................................................................... 34 Figure 7 Kaplan-Meier plot of PFS2 ...................................................................................................... 36 Figure 8 Mean change in FACT-O TOI score from baseline ................................................................ 37 Figure 9: Mean EQ-5D-5L weighted health state index score .............................................................. 38 Figure 10 Forest plot of progression-free survival by subgroup ........................................................... 39 Figure 11 Kaplan-Meier plot of TTD ...................................................................................................... 41 Figure 12 Most common AEs reported in SOLO1[67] .............................................................................. 43 Figure 13 Kaplan-Meier plot of PFS, showing recommended time for treatment discontinuation ........ 47 Figure 14 Model schematic ................................................................................................................... 63 Figure 15 Illustration of the partitioned survival calculation .................................................................. 65 Figure 16 SOLO1 PFS Kaplan-Meier curve .......................................................................................... 74 Figure 17 Cumulative hazards plot of PFS ........................................................................................... 75 Figure 18 Schoenfeld residuals of PFS ................................................................................................ 75 Figure 19 Visual representation of fitted parametric models to entire data set .................................... 77 Figure 20 Visual representation of fitted parametric models to PFS from month 24 onwards ............. 78 Figure 21 SOLO1 OS Kaplan-Meier plot .............................................................................................. 84 Figure 22 Cumulative hazards plot of OS ............................................................................................. 85 Figure 23 Schoenfeld residuals of OS .................................................................................................. 85 Figure 24 Fit of independent models to the Kaplan-Meier for OS in SOLO1 ....................................... 87 Figure 25 Fit of independent models to the post-24-month Kaplan-Meier period for OS in SOLO1 .... 88 Figure 26 Best fitting function for PFS2 ................................................................................................ 96 Figure 27 Illustration of model approach combining PFS and OS with modelling of long-term survival status and gains in median PFS2 to OS (overlaid with Kaplan-Meier data) ................................ 98 Figure 28 Kaplan-Meier plot for time to first subsequent PARP inhibitor therapy SOLO1 study ........ 112 Figure 29 Kaplan-Meier plot for TTD germline BRCA sub-group of the Study 19 .............................. 113 Figure 30 Schematic of calculation of the proportion of patients on subsequent PARP inhibitor treatment in each model cycle .................................................................................................... 114 Figure 31 Cost-effectiveness plane for olaparib versus routine surveillance ..................................... 123 Figure 32 Cost-effectiveness acceptability curve for olaparib versus routine surveillance ................ 123 Figure 33 Tornado diagram................................................................................................................. 124

Table of Tables

Table 1 Decision problem for NICE appraisal [ID1124] .......................................................................... 6 Table 2 Technology being appraised ...................................................................................................... 7 Table 3 Summary of FIGO staging classification for ovarian, fallopian tube, and primary peritoneal cancer ........................................................................................................................................... 10 Table 4 Chemotherapy regimens for first-line treatment of advanced ovarian cancer ......................... 12 Table 5 Common chemotherapy regimens for platinum-sensitive recurrent ovarian cancer ............... 14 Table 6 Common chemotherapy regimens for platinum-resistant recurrent ovarian cancer ................ 14 Table 7 Eligibility criteria for the identification of studies reporting relevant clinical evidence .............. 20 Table 8 Clinical effectiveness evidence ................................................................................................ 21 Table 9 SOLO1 inclusion and exclusion criteria ................................................................................... 23 Table 10 SOLO1 patient baseline characteristics ................................................................................. 27 Table 11 Quality assessment results for SOLO1 .................................................................................. 29 Table 12 Primary analysis of PFS (Investigator-assessed) .................................................................. 31 Table 13 Summary of PFS sensitivity analyses .................................................................................... 33

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 14 Summary of secondary efficacy analyses ............................................................................. 35 Table 15 Best overall response for patients with baseline evidence of disease................................... 36 Table 16 Duration of treatment exposure ............................................................................................. 40 Table 17 Summary of adverse events .................................................................................................. 42 Table 18 Summary list of published cost-effectiveness studies in patients with BRCA-mutated ovarian cancer ........................................................................................................................................... 56 Table 19 Summary of the de novo economic analysis ......................................................................... 59 Table 20 Features of the economic analysis and comparisons with previous appraisals in the relapse/recurrent advanced ovarian cancer setting ..................................................................... 69 Table 21 Summary of separate AIC and BIC goodness of fit data for PFS .......................................... 76 Table 22 Fitted parameters for the log-normal distribution fitted to PFS from month 24 onwards ....... 80 Table 23 Prediction of Kaplan-Meier data and long-term extrapolation of PFS with olaparib using the Kaplan-Meier and parametric model (“piecewise”), and fully parametric model methods (“entire data”) ............................................................................................................................................ 81 Table 24 Prediction of Kaplan-Meier data and long-term extrapolation of PFS with placebo using the Kaplan-Meier and parametric model (“piecewise”), and fully parametric model methods (“entire data”) ............................................................................................................................................ 82 Table 25 Summary of separate AIC and BIC goodness of fit data for OS ........................................... 86 Table 26 Literature estimates of OS following first-line platinum chemotherapy .................................. 90 Table 27 Prediction of Kaplan-Meier data and long-term extrapolation of OS with olaparib using the Kaplan-Meier and parametric model (“piecewise”), and fully parametric model methods (“entire data”) ............................................................................................................................................ 91 Table 28 Prediction of Kaplan-Meier data and long-term extrapolation of OS with placebo using the Kaplan-Meier and parametric model (“piecewise”), and fully parametric model methods (“entire data”) ............................................................................................................................................ 92 Table 29 Relationship between median TFST, TSST, PFS2 and OS in the gBRCA cohort of S19 ..... 94 Table 30 Predicted relationship between median TFST, TSST, PFS2 and OS in SOLO1 .................. 94 Table 31 Prediction of Kaplan-Meier data and long-term extrapolation of OS with placebo using the Kaplan-Meier and parametric model (“piecewise”) with application of the treatment effect from PFS2 ............................................................................................................................................. 97 Table 32 Summary of AEs included in the economic model ................................................................ 99 Table 33 Utility values associated with specific disease stages/states .............................................. 101 Table 34 Disutility values associated with AEs, and assumed duration of events ............................. 104 Table 35 Summary of utility values for cost-effectiveness analysis .................................................... 104 Table 36 Summary of drug related costs ............................................................................................ 108 Table 37 Drug acquisition costs – subsequent therapies received by patients in the SOLO1 study 109 Table 38 Chemotherapy recommended dose and duration of treatment ........................................... 110 Table 39 Subsequent IV drug administration costs ............................................................................ 110 Table 40 Unit costs and monthly frequency of resource use associated with the PF and PD states for BSC ............................................................................................................................................ 116 Table 41 Unit costs and monthly frequency of resource use associated with the PF and PD states for olaparib ....................................................................................................................................... 116 Table 42 Resource costs (per week) associated with the monitoring and management of patients treated with olaparib or routine surveillance ............................................................................... 116 Table 43 Unit costs for AEs in the model ............................................................................................ 117 Table 44 Overall summary of assumptions in the model .................................................................... 119 Table 45 Base-case results (1.5% discounting rate for costs and effects) ......................................... 121 Table 46 Average results based on the probabilistic sensitivity analysis (10,000 iterations) ............. 122 Table 47 Results of scenario analyses conducted ............................................................................. 125

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

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

• Ovarian cancer is relatively rare, aggressive and typically diagnosed at an advanced stage. Women with ovarian cancer in England face poor prognosis, with a 5-year survival rate of 30.6%, compared to the European mean of 37.6%.

• First-line treatment for women with newly diagnosed advanced ovarian cancer is curative in intent and aims to achieve complete remission. The mainstay of treatment involves cytoreductive surgery and platinum-based doublet chemotherapy. Patients with BRCA-mutated ovarian cancer receive the same first-line treatment as those with non-BRCA-mutated disease.

• Despite a good initial response to first-line treatment, the majority of patients with advanced ovarian cancer relapse within a 3-year period. Recurrent ovarian cancer is currently considered incurable, so there is high unmet medical need for effective and well-tolerated treatment options that delay or prevent the time to first relapse.

• The current standard of care for patients with advanced ovarian cancer after surgery and first-line platinum-based chemotherapy treatment is routine surveillance. No active maintenance treatment options are licensed or currently in use within the NHS.

• Olaparib is the first and only personalised medicine for patients with newly diagnosed BRCA-mutated advanced ovarian cancer, who are in response (complete or partial) to platinum-based chemotherapy.

• The magnitude of benefit observed with olaparib versus routine surveillance in the pivotal Phase III SOLO1 trial is unprecedented in this disease setting, with a 70% reduction in the risk of progression or death versus placebo, and a minimum estimated 3-year improvement in median PFS.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

B.1.1 Decision problem

This appraisal covers the full marketing authorisation for olaparib (LYNPARZA™) as a maintenance treatment for patients with newly diagnosed BRCA-mutated advanced ovarian cancer who are in response (complete or partial) after first-line platinum-based chemotherapy. The decision problem to be addressed is presented in Table 1.

Table 1 Decision problem for NICE appraisal [ID1124]

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Abbreviations: BRCA, breast cancer susceptibility gene; HRQoL, health-related quality of life; OS, overall survival; NHS, National Health Service; PFS, progression-free survival; PFS2, time from randomization to second progression or death; QALY, quality-adjusted life-year; TFST, time to first subsequent therapy; TSST, time to second subsequent therapy.

B.1.2 Description of the technology being appraised

Table 2 presents a summary of the key product attributes of olaparib. The draft Summary of Product Characteristics (SmPC) is included in Appendix C, and the European Public Assessment Report will be provided to NICE once available (anticipated xxxxxxxx).

Table 2 Technology being appraised

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Abbreviations: BRCA, breast cancer susceptibility gene; EMA, European Medicines Agency; PARP, poly-ADPribose polymerase; SmPC, Summary of Product Characteristics.

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

Disease overview

‘Ovarian cancer’ is a non-specific term used to describe cancers that originate in the ovary, fallopian tube and primary peritoneum. It is relatively rare, aggressive and typically diagnosed at an advanced stage, as symptoms tend to be vague and nonspecific (eg abdominal pain, fatigue and bloating) and there are currently no effective screening tests.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Recent studies have shown that survival outcomes for ovarian cancer in the UK are among the worst in Europe.[3-6] This is attributed to delayed diagnosis, low awareness of symptoms, variability in surgical outcomes and restricted access to innovative medicines.[5,6] The 5-year survival rate for ovarian cancer in England is 30.6%, compared to the European mean of 37.6%.[5]

This appraisal proposes olaparib as a maintenance treatment option for patients with newly diagnosed advanced BRCA-mutated ovarian cancer, who are in response (complete or partial) to first-line platinum-based chemotherapy, based on unprecedented benefit demonstrated in the Phase III randomised controlled trial, SOLO1.

Diagnosis and staging

National Institute for Health and Care Excellence (NICE) and the British Gynaecological Cancer Society (BGCS) guidelines recommend that initial investigations for suspected ovarian cancer should be performed if a woman (particularly if aged ≥50 years) reports having any of the following symptoms persistently/frequently:[7,8]

• Abdominal distention (bloating)

• Feeling full and/or loss of appetite

• Pelvic or abdominal pain

• Increased urinary urgency and/or frequency

Other symptoms of ovarian cancer may include irregular periods, lower abdominal and back pain, constipation, nausea, anorexia, dyspepsia, and extreme fatigue.

Initial investigations for women who present with ovarian cancer symptoms in the primary care setting should include clinical examination, ultrasound and measurement of serum cancer antigen 125 (CA-125) levels. If ovarian cancer is suspected, patients should be referred to secondary care for additional tests, including a computed tomography (CT) scan, to confirm the presence and extent of spread of disease.[8,9]

Ovarian cancer is surgically staged according to the International Federation of

Gynaecology and Obstetrics (FIGO) classification (Table 3).[10] Patients with Stage III Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved Page 9 of 144

or IV advanced ovarian cancer face poor prognosis,[11] with 5-year relative survival rates of 18.6 and 3.5%, respectively.[12] The English National Cancer Registration and Analysis Service reported in 2016 that 57.9% of women diagnosed with ovarian cancer in England had Stage III (locally advanced) or Stage IV (metastatic) disease at diagnosis.[13]

Table 3 Summary of FIGO staging classification for ovarian, fallopian tube, and primary peritoneal cancer

Source: Adapted from Prat et al, 2014[10] Abbreviation: FIGO, International Federation of Gynecology and Obstetrics.

BRCA mutations

Approximately 20–25% of ovarian cancers are associated with a BRCA

mutation .[14-18] Patients with BRCA-mutated ovarian cancer tend to develop disease at a younger age than those with non-BRCA-mutated ovarian cancer, are more likely to respond to treatment with platinum agents and PARP inhibitors, but have a higher risk of developing visceral metastases.[14,19,20] Similar clinical outcomes are observed in patients with BRCA-mutated ovarian cancer, regardless of whether the mutation is germline (inherited) or somatic (acquired) in origin.[21-26]

Testing to determine BRCA mutation status is already considered routine practice for UK patients with ovarian cancer as it provides important information about prognosis, the likelihood of response after platinum-based chemotherapy and/or PARP inhibitors, and risk of developing future breast or ovarian cancers.[8,27-29] This also enables family members to be tested and, if found to carry the BRCA mutation, to make decisions about reducing their risk of developing BRCA-related cancers, including undergoing preventative surgery.[30]

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Treatment pathway

First-line treatment for women with newly diagnosed advanced ovarian cancer is curative in intent and aims to achieve complete remission.[31,32] The mainstay of treatment involves cytoreductive surgery and platinum-based doublet chemotherapy. Patients with BRCA-mutated ovarian cancer receive the same first-line treatment as those with non-BRCA-mutated disease.

Surgery for advanced ovarian cancer is intensive and aims to achieve complete resection with no residual visible disease, as this is associated with a significantly improved progression-free survival (PFS) and overall survival (OS).[28,33-35] A maximal surgical effort is required, including intestinal resection, peritoneal stripping, diaphragmatic resection, removal of bulky para-aortic lymph nodes and splenectomy. Surgery is quickly followed by chemotherapy to reduce the risk of disease recurrence. The standard first-line regimen is carboplatin in combination with paclitaxel, both administered intravenously every 3 weeks, for six cycles (Table 4).[8,28,36] The combination of cisplatin and paclitaxel is equally effective but is more toxic and less convenient to administer. Docetaxel or pegylated liposomal doxorubicin hydrochloride (PLDH) may be given as alternatives in patients who cannot tolerate paclitaxel.[8,28,36] Bevacizumab in combination with carboplatin and paclitaxel is not recommended by NICE for first-line treatment of advanced ovarian cancer, but funding is available through the Cancer Drugs Fund for use at less than the recommended dose, in a small subgroup of patients with sub-optimally debulked Stage III or Stage IV ovarian cancer, provided strict criteria are met.[37,38] The majority of patients with newly diagnosed BRCA-mutated advanced ovarian cancer who would be considered for olaparib maintenance treatment would not be eligible under the CDF criteria for bevacizumab.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 4 Chemotherapy regimens for first-line treatment of advanced ovarian cancer

Source: Webber et al 2017,[39-41] Table 1[42] aConsider for patients at risk of poor tolerance to conventional schedules due to performance status or comorbidities;[b] For patients with suboptimally debulked stage III or stage IV disease. Abbreviations: AUC, area under the concentration–time curve; IP, intraperitoneal; IV, intravenous

After response to first-line platinum-based chemotherapy, the current standard

of care for patients with advanced ovarian cancer is routine surveillance. No active maintenance treatment options are licensed or currently in use within the NHS.

Current clinical practice guidelines currently recommend that patients with advanced ovarian cancer attend regular follow-up visits after completion of first-line treatment every 3 months for the first 2 years, then every 6 months for up to 5 years (or until progression occurs).[8,28,43] The basic format of a follow-up consultation involves

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved Page 12 of 144

symptom review, clinical examination (with or without measurement of CA-125 levels), and assessment of psychosocial and supportive care needs. Radiologic imaging is not routinely performed unless indicated by clinical signs or symptoms of disease.

Data from the University of Edinburgh Ovarian Cancer Database demonstrate that if a patient remains in remission with no evidence of disease for 5 years after diagnosis, they have a low risk of further recurrence after this timepoint (Appendix M). Preventing or delaying recurrence enables women with ovarian cancer to avoid disease-related symptoms, the need for further chemotherapy and associated toxicities, thereby improving physical and emotional wellbeing, ability to carry out activities of daily living, family duties, and ability to work.[44]

Despite a good initial response to first-line platinum-based chemotherapy, the - majority of patients with advanced ovarian cancer relapse within a 3 year period.[28,45] Recurrent ovarian cancer is currently considered incurable, so there is high unmet medical need for effective and well-tolerated treatment options that prevent or delay relapse.

At present, there are few effective treatment options for patients with recurrent ovarian cancer. Treatment goals focus on preserving quality of life and extending time to progression and time free from chemotherapy, rather than on cure.[47-50]

In general, patients who remain relapse-free for 6 months or longer after completion of a first-line platinum-based regimen are considered to have platinum-sensitive disease and a higher likelihood of responding to re-treatment with platinum agents. Those who relapse within 6 months of first-line platinum are considered to have platinum-resistant disease, with life-expectancy of fewer than 12 months.[28,45]

The likelihood and duration of response to chemotherapy markedly diminishes with each subsequent line and there is a high risk of developing cumulative toxicities (eg hypersensitivity, neurotoxicity, alopecia and ototoxicity).[48,51] Unfortunately, some patients are unable to benefit even from a second-line platinum chemotherapy, having become platinum resistant after first-line treatment.

Common chemotherapy regimens for platinum-sensitive and platinum-resistant recurrent ovarian cancer are presented in Table 5 and Table 6.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 5 Common chemotherapy regimens for platinum-sensitive recurrent ovarian cancer

Source: Webber et al 2017,[52-54] Table 2 Abbreviations: AUC, area under the concentration–time curve; PLD, pegylated liposomal doxorubicin

Table 6 Common chemotherapy regimens for platinum-resistant recurrent ovarian cancer

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Source: Webber et al 2017,[55-58] Table 3 Abbreviations: PLD, pegylated liposomal doxorubicin

Maintenance treatment with a poly(ADP-ribose) polymerase (PARP) inhibitor may be considered for patients with platinum-sensitive relapsed ovarian cancer who meet specific criteria (Figure 1). At present:

• Niraparib may be considered as a maintenance treatment option for patients with germline BRCA-mutated platinum-sensitive relapsed high-grade serous epithelial ovarian, fallopian tube or primary peritoneal cancer, who are in response to second-line platinum-based chemotherapy, if the CDF criteria for use are met (TA528)[38,59]

• Olaparib capsules are recommended as a maintenance treatment option for patients with germline or somatic BRCA-mutated platinum-sensitive relapsed ovarian, fallopian tube or peritoneal cancer, who are in response to third- or laterline platinum-based chemotherapy (TA381)[60]

• NICE is currently evaluating olaparib tablets as a maintenance treatment option for patients with platinum-sensitive relapsed ovarian cancer who are in response to second- or later-line platinum-based chemotherapy [ID1296].[61] xxxxxxxx for this

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Proposed use of olaparib

Olaparib is proposed as a maintenance treatment option for patients with newly diagnosed BRCA-mutated ovarian cancer, who are in response (complete or partial) after first-line platinum-based chemotherapy, in line with the anticipated EMA licence. The current and proposed treatment pathways for patients with BRCAmutated advanced ovarian cancer within the NHS in England are illustrated in Figure 1. It is expected that patients will only receive one course of treatment with a PARP inhibitor within the clinical management pathway for advanced ovarian cancer, as there is currently no trial data to support re-treatment with a PARP inhibitor.

Figure 1 Current and proposed position use of olaparib in the treatment pathway for patients with BRCA-mutated advanced ovarian cancer

==> picture [432 x 199] intentionally omitted <==

a This appraisal proposes olaparib tablets as a maintenance treatment option for patients with newly diagnosed BRCA-mutated ovarian cancer who are in response to first-line platinum-based chemotherapy [ID1124].

b Maintenance treatment with niraparib should only be considered as an option for treating patients with BRCA-mutated platinum-sensitive relapsed ovarian cancer who are PARP-inhibitor–naïve and in response (complete or partial) to second-line platinum-based chemotherapy, and the Cancer Drugs Fund criteria for use are met [TA528].15

c The capsule formulation of olaparib is currently recommended as an option for treating patients with BRCA-mutated platinum-sensitive relapsed ovarian cancer who are PARP-inhibitor–naïve and in response to third- or later-line platinum-based chemotherapy [TA381]17

d NICE is currently evaluating the tablet formulation of olaparib as a maintenance treatment option for patients with platinum-sensitive relapsed ovarian cancer who are PARP-inhibitor–naïve and in response to second- or later-line platinum-based chemotherapy [ID1296].18 xxxxxxxxxxxxxxxxxxxxxxxxxx.

B.1.4 Equality considerations

No equality issues related to the use of olaparib have been identified or are foreseen.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

B.2 Clinical effectiveness

• SOLO1 was a high-quality, double-blinded, international, Phase III randomised placebo-controlled trial of olaparib maintenance treatment in patients with newly diagnosed BRCA-mutated advanced ovarian cancer, who were in complete or partial response to first-line platinum-based chemotherapy (N=391). In total, 22 of 391 patients (5.6%) were included from six UK sites.

• The trial population was relatively young (median age 53 years), and most patients the majority were in complete clinical remission with no evidence of disease at study entry (81.8%). Per protocol, most patients in both arms of the trial received study treatment for 2 years or until disease progression; 10% of patients in the olaparib arm and 2.3% in the placebo arm continued treatment beyond 2 years at the investigator’s discretion, as they were considered to have residual disease that was stable (i.e. not progressing).

• SOLO1 met its primary endpoint, demonstrating a 70% reduction in the risk of disease progression or death and a minimum estimated 3-year improvement in median progression-free survival (PFS) with olaparib versus placebo in the proposed patient group (HR 0.30; P <0.0001). More than twice as many olaparib-treated patients were progression-free at 3 years after randomisation compared to placebo (60.4 versus 26.9%). The magnitude of PFS benefit observed in SOLO1 is unprecedented in patients with newly diagnosed advanced ovarian cancer, and far exceeds that observed in - previous first line chemotherapy trials.

• Secondary endpoint analyses demonstrate that the benefits of olaparib treatment persist beyond progression, and significantly extends time free from chemotherapy treatment and associated toxicities:

  • Time to second progression or death (PFS2): HR 0.50, P =0.0002

  • Time to first subsequent therapy or death (TFST): HR 0.30, P <0.0001

  • Time to second subsequent therapy or death (TSST): HR 0.45, P <0.0001

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

• Among the subset of patients who had evaluable disease at baseline, more than twice as many patients achieved complete response with olaparib versus placebo, increasing the likelihood of long-term remission (xxxxxxxx xxxxx xxxxx).

• OS data are currently immature (21.0% data maturity). The observed data shows that patients continue to experience a benefit from olaparib after stopping treatment around the 2-year timepoint. There is also no evidence that olaparib impacts the ability of patients to receive and respond to subsequent therapy. It is likely that the PFS benefit observed with olaparib in SOLO1 will translate to a meaningful improvement in OS, but further follow-up is needed to determine the magnitude of long-term benefit. Final OS analyses will be event-driven and are planned to be conducted at 60% data maturity ( xxxxxxxxxxxxx).

• Importantly, the clinical benefits of olaparib observed in the SOLO1 trial were not associated with detriment to health-related quality of life (HRQoL), as assessed using the disease-specific FACT-O questionnaire or EQ-5D-5L.

• Olaparib was generally well tolerated with adverse events (AEs) that tended to be mild or moderate in severity, and manageable without dose reduction or treatment discontinuation. The safety profile of olaparib observed in SOLO1 was consistent with that observed in previous studies.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

B.2.1 Identification and selection of relevant studies

Search strategy

A two-part systematic literature review was designed to identify published studies that report the use of health technologies in adult patients with ovarian cancer who have a BRCA mutation and who have previously received first-line platinum-based chemotherapy. The clinical systematic review (discussed here) identified studies that reported clinical evidence, whereas the non-clinical systematic review identified studies that reported economic, health state utility values and cost-of-illness evidence (discussed in Section B.3.1). The searches were conducted using the search strings presented in Appendix D, which include a mixture of free text and Medical Subject Headings (MeSH) terms.

The literature searches were conducted on 13 June 2018 using the MEDLINE and MEDLINE In-Process, Embase and the Cochrane Central Trials Register electronic databases from 1974 to 2018. Supplementary searches included searching relevant appraisal data (manufacturer submissions and evidence review/assessment group reports) from the previous NICE health technology assessments in ovarian cancer, and reviewing abstracts from the following congresses for up to 3 years prior to the search date:

• American Society of Clinical Oncology (ASCO) annual meetings (2016–2018)

• European Society for Medical Oncology (ESMO) congresses (2016–2017)[a]

• Society of Gynecologic Oncology (SGO) annual meetings on Women’s Cancer (2016–2018)

• International Society of Pharmacoeconomics and Outcomes Research (ISPOR) US (2016–2018) and European (2016–2017)[b] congresses

• Health Technology Assessment International (HTAi) annual meetings (2016– 2017)[c]

a ESMO 2018 congress had not occurred at the time of searching

b ISPOR Europe 2018 congress had not occurred at the time of searching

c HTAi 2018 had not occurred at the time of searching

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Study selection

The abstracts of the publications identified were screened by two independent reviewers to determine whether they met the predefined populations, interventions, comparators, outcomes, and study design (PICOS) eligibility criteria (Table 7), in accordance with the 2009 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Any disputes in eligibility were discussed and resolved. When there was no resolution, disputes were reconciled by a third reviewer. The full texts of all publications identified in the initial screening were obtained and assessed against the eligibility criteria.

Table 7 Eligibility criteria for the identification of studies reporting relevant clinical evidence

Abbreviations: BRCA, breast cancer susceptibility gene; HRQoL, health-related quality of life; OS, overall survival; PFS, progression-free survival; RCT, randomised controlled trial.

Identified trials

In total, the literature search described above identified two clinical studies that

reported results for a targeted maintenance treatment in patients with BRCA-mutated

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

ovarian cancer after first-line platinum-based chemotherapy: the SOLO1 trial of olaparib (NCT01844986) and the AGO-OVAR 16 trial of pazopanib (NCT00866697). Pazopanib is not licensed for use in the proposed population and is outside the scope of the current appraisal. The submission therefore presents clinical evidence reported for SOLO1.

B.2.2 List of relevant clinical effectiveness evidence

SOLO1 was the only identified trial to provide clinical evidence on the efficacy and safety of olaparib as a first-line maintenance treatment in the patient group proposed for this appraisal. The clinical evaluation presented in this submission is based on this trial.

Table 8 Clinical effectiveness evidence

Abbreviations: FIGO, International Federation of Gynecology & Obstetrics; PR, partial response

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

B.2.3 Summary of trial methodology

Trial design

SOLO1 was a high-quality, international, Phase III, randomised, double-blind, placebo-controlled trial that assessed the efficacy and safety of olaparib versus placebo in patients with newly diagnosed advanced BRCA-mutated ovarian cancer who were in response (complete or partial) following first-line platinum-based chemotherapy (N=391).[62,63] The trial design is summarised in Figure 2, and described in further detail below.

Figure 2 SOLO1 study schema[63,64]

==> picture [452 x 157] intentionally omitted <==

Source: Clinical Study Report Olaparib D0818C00001[62] Abbreviations: BICR, blinded independent central review; ECOG, Eastern Cooperative Oncology Group; FACTO, Functional Assessment of Cancer Therapy – Ovarian Cancer; FIGO, International Federation of Gynecology and Obstetrics; HRQoL, health-related quality of life; PFS, progression-free survival; PFS2, time to second progression or death; RECIST, Response Evaluation Criteria in Solid Tumours; TOI, Trial Outcome Index

Eligibility criteria

Patients were eligible for inclusion in the SOLO1 trial if they had newly diagnosed BRCA-mutated advanced (FIGO stage III or IV) high-grade serous or high-grade endometrioid ovarian, fallopian tube or primary peritoneal cancer, and were in complete or partial response to first-line platinum-based chemotherapy, with no clinical evidence of disease progression on the post-treatment scan:

• Complete response was defined as no evidence of measurable or nonmeasurable disease on the end of chemotherapy scan and a normal CA-125, according to the Response Evaluation Criteria in Solid Tumours (RECIST)

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

• Partial response was defined as ≥30% reduction in RECIST measurable or nonmeasurable disease demonstrated from the start to finish of previous chemotherapy or no radiological evidence of disease on the end of chemotherapy scan with a CA-125, which had not decreased to within the normal range[63,64]

Patients with stage III disease must have had an upfront or interval attempt at optimal cytoreductive surgery, and those with stage IV disease must have had either a biopsy and/or upfront or interval cytoreductive surgery. BRCA mutation status may have been determined by either germline or tumour testing, provided that the test was conducted in an accredited laboratory. Further details of the SOLO1 eligibility criteria are presented in Table 9.

Table 9 SOLO1 inclusion and exclusion criteria

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Source: Data on file: D0818C00001 Clinical Study Report[63,64] Abbreviations: AML acute myeloid leukaemia; CA, cancer antigen; CYP, cytochrome P450; ECOG, Eastern Cooperative Oncology Group; FIGO, International Federation of Gynaecology and Obstetrics; MDS, myelodysplastic syndrome; PARP, poly(ADP-ribose) polymerase; PD, progressive disease; SD, stable disease.

Interventions

Eligible patients were randomly assigned in a 2:1 ratio to receive either olaparib tablets (300 mg twice daily) or matching placebo, using an Interactive Voice Response System (IVRS) / Interactive Web Response System (IWRS). Randomisation was performed within 8 weeks after their last dose of chemotherapy (last dose was the day of the last infusion) and stratified based on complete or partial response to first-line platinum chemotherapy (complete or partial). The first patient was randomised into the study on 3 September 2013, and the last patient on 6 March 2015.

The majority of patients received study treatment for up to 2 years or until objective radiological disease progression . At the 2-year timepoint, patients with complete response (no radiological evidence of disease) were required to stop study treatment. Those with residual evidence of stable disease could continue to receive study treatment in a blinded manner at the investigator’s discretion.[63,64]

Following discontinuation of the trial intervention, further treatment was at the discretion of the investigator. Any further systemic anticancer treatment was collected until death, loss to follow-up or withdrawal of consent.[63,64]

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Primary and secondary endpoints

The primary endpoint in SOLO1 was investigator-assessed PFS, defined as the time from randomisation to objective disease progression on imaging according to modified RECIST 1.1 or death by any cause.[63,64]

Patients had tumour assessments at baseline and every 12 weeks for the first 3 years, and then every 24 weeks relative to the date of randomisation until objective disease progression. PFS was analysed using a log-rank test stratified by response to first-line platinum chemotherapy. To show a consistency of effect with the investigator assessment of PFS, a sensitivity analysis of PFS was also performed using blinded independent central review (BICR) of progression status. Other sensitivity analyses were also carried out to demonstrate the robustness of the result (Section B.2.4).[63,64]

Predefined secondary endpoints included PFS2, TFST, TSST, OS, best overall response, HRQoL and AEs.[63,64]

Locations

SOLO1 was conducted across 15 countries (Australia, Brazil, Canada, China, France, Israel, Italy, Japan, Netherlands, Poland, Russia, South Korea, Spain, UK, US). In total, 22 of 391 patients (5.6%) were included from six UK centres.

Trial population

Patient disposition

Between 3 September 2013 and 6 March 2015, 391 patients were randomised into the SOLO1 trial; 260 patients were assigned to the olaparib arm and 131 to the placebo arm. All patients received their allocated treatment except one patient who received no treatment in the placebo arm due to withdrawal (Figure 3).[64]

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Figure 3 SOLO1 patient disposition

==> picture [432 x 307] intentionally omitted <==

Source: Data on file: D0818C00001 Clinical Study Report. Figure 3 Abbreviations: AE, adverse event.

At the data cut-off (DCO) for the primary analysis (17 May 2018), the median duration of follow-up across both treatment arms was 41 months.[64] The majority of patients in both arms of the trial had discontinued study treatment, but were still alive and continued to be followed for long-term survival. In the olaparib arm, the most common reason for treatment discontinuation was completion of the 2-year treatment as per protocol (47.3 vs 26.9% in the placebo arm), 26 patients (10.0%) had received maintenance treatment beyond 2 years at the investigator’s discretion, and 13 patients (5%) were still receiving treatment with olaparib at the DCO.[64] In the placebo arm, the most common reason for treatment discontinuation was objective disease progression (60.0% versus 19.6% in the olaparib arm), three patients (2.3%) had received maintenance treatment beyond 2 years, and one patient (<1%) was still receiving placebo at the DCO.[64]

Baseline characteristics

Patients randomised to the treatment groups were well-matched for baseline characteristics (Table 10).[64] The trial population was relatively young, with median

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

age of 53.0 years in both the olaparib and placebo arms, as expected for patients with BRCA-mutated ovarian cancer.[14] The majority of patients (81.8%) were in complete clinical response at study entry, with no evidence of residual disease, Eastern Cooperative Oncology Group (ECOG) performance status of 0, and a CA125 level within the normal range.[64]

Table 10 SOLO1 patient baseline characteristics

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved Page 27 of 144

Source: Data on file: D0818C00001 Clinical Study Report. Tables 11, 12 and 15[63,64] aother includes fallopian tube, peritoneum and omentum cancer (n=1), ovary and peritoneum (n=1) and tuboovary (n=1);[b] The seven randomised patients who did not have any debulking surgery all had Stage IV disease and were not required to have surgery as per protocol.

Abbreviations: ECOG, Eastern Cooperative Oncology Group; FIGO, International Federation of Gynaecology and Obstetrics; ULN, upper limit of normal; VUS, variants of unknown significance.

B.2.4 Statistical analyses

SOLO1 efficacy and safety analyses were performed in accordance with a comprehensive Statistical Analysis Plan, which is summarised in Section 5.7 of the Clinical Study Report.[64]

The primary endpoint was PFS, defined as the time from randomisation until the date of objective radiological disease progression according to modified RECIST 1.1 or death (by any cause in the absence of progression).[64] It was determined that, 206 PFS events in the study would provide the trial 90% power to show statistically significant PFS at the two-sided 5% level if the assumed true treatment effect were hazard ratio (HR) 0.62 (translating to an 8-month benefit in median PFS over 13 months on placebo).[64] PFS was planned to be analysed when approximately 196 events had occurred (50% data maturity) or after the last patient randomised had the opportunity to be on the study for at least 36 months, whichever came first.[64]

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Sensitivity analyses of PFS were also performed to test for sources of bias including:

• Evaluation time bias

• Attrition bias

• Ascertainment bias (BICR)

• Electronic case report form stratification variables

• Possible informative censoring

All efficacy and HRQoL endpoints were analysed using the full analysis set (FAS), which included all randomised patients on an intention-to-treat basis (ie based on treatment assigned at randomisation, regardless of whether treatment was received).

Summaries of safety and tolerability assessments were based on the safety analysis set (SAS), which included all patients who received at least one dose of randomised study medication.

B.2.5 Quality assessment

SOLO1 was performed in accordance with the principles of the Declaration of Helsinki, Good Clinical Practice guidelines, and the AstraZeneca policy of bioethics, under the auspices of an independent data and safety monitoring committee.[63] A complete quality assessment in accordance with the NICE-recommended checklist for assessment of bias in randomised controlled trials is presented in Table 5. The risk of bias in SOLO1 is confirmed as being low.

Table 11 Quality assessment results for SOLO1

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 29 of 144

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

B.2.6 Clinical effectiveness results

Primary endpoint: PFS

The magnitude of PFS benefit observed with olaparib in SOLO1 far exceeds - that observed in previous first line chemotherapy trials conducted in patients with newly diagnosed BRCA-mutated advanced ovarian cancer. Olaparib reduced the risk of progression or death by 70% versus placebo (HR, 0.30; P<0.00001), and extended median PFS by an estimated minimum of three years.[63,64]

Table 12 Primary analysis of PFS (Investigator-assessed)

Source: Data on file: D0818C00001 Clinical Study Report. Table 18[63,64] Abbreviations: CI, confidence interval; HR, hazard ratio; PFS, progression-free survival

The primary analysis of investigator-assessed PFS was conducted after 198 of the 391 patients enrolled in SOLO1 had progressed or died (50.6% data maturity), with a median follow-up duration of 41 months (17 May 2018 DCO). A smaller proportion of patients in the olaparib arm had progressed or died, compared to the placebo arm (39.2 vs 73.3%).[63,64] Median PFS in the placebo arm was 13.8 months , consistent with previously published advanced ovarian cancer trials (Appendix N). Median PFS in the olaparib arm had not been reached and was estimated to be at least 3 years longer than observed with placebo.

The Kaplan-Meier plot for PFS shows that the clinical benefits of olaparib occurred early, with increasing separation of the curves for olaparib versus placebo from the time of first assessment, 12 weeks after randomisation (Figure 4). There was no evidence of a change in the shape of the Kaplan-Meier plot after the 2-year timepoint when the majority of patients discontinued treatment as per protocol. This indicates that patients had a consistent and sustained benefit beyond treatment completion.[64]

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved Page 31 of 144

Landmark analyses show that a significantly greater proportions of patients in the olaparib arm remained progression-free with long-term follow-up, compared to placebo (Figure 5).[64]

Figure 4 Kaplan-Meier plot of PFS (investigator-assessed)

==> picture [452 x 233] intentionally omitted <==

Source: Data on file: D0818C00001 Clinical Study Report. Figure 6[63,64] Abbreviations: CI, confidence interval; PFS, progression-free survival

Figure 5 Proportion of patients who remained progression-free

==> picture [472 x 163] intentionally omitted <==

*Based on Kaplan-Meier estimates. DCO: May 2018; Median FU: olaparib, 40.7 months placebo, 41.2 months.[63,64] Abbreviations: DCO, data cut-off; FU, follow-up

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Pre-planned sensitivity analyses of PFS were highly consistent with the investigatorassessed PFS results, with hazard ratios ranging from 0.25 to 0.33 (Table 13). The BICR-assessed PFS result was consistent with the primary PFS analysis, with a median not reached in the olaparib arm versus 14.1 months in the placebo arm (HR 0.28; 95% CI 0.20, 0.39; P <0.001; 38.4% maturity). In sensitivity analyses conducted to evaluate for the risk of attrition bias and informative censoring bias, median PFS with olaparib was approximately 3 years longer than that observed with placebo.

Table 13 Summary of PFS sensitivity analyses

Abbreviations: BICR, blinded independent central review; CI, confidence interval; eCRF, electronic case report form; HR, hazard ratio; NC, not calculated; NR, not reached; PFS, progression-free survival[63,64]

There was a significant improvement in BICR PFS in the olaparib arm compared with the placebo arm (HR 0.28; 95% CI 0.20, 0.39; P <0.0001; Figure 6). The median PFS (months) was not reached for patients in the olaparib arm vs 14.1 months for patients in the placebo arm.[64]

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Figure 6 Kaplan-Meier plot of PFS by BICR assessment

==> picture [452 x 236] intentionally omitted <==

Source: Data on file: D0818C00001 Clinical Study Report. Figure 7[62] Abbreviations: BICR, blinded independent central review; CI, confidence interval; PFS, progression-free survival

Secondary endpoints: PFS2, TFST, TSST, OS, and best overall response

Secondary time-to-event endpoints analyses are summarised in Table 14, and show that:

• Olaparib significantly reduced the risk of PFS2, demonstrating that first-line use of olaparib does not diminish the benefit conferred by subsequent therapy (HR 0.50; 95% CI 0.35, 0.72; P =0.0002; Figure 7). This analysis is confounded by bias in favour of placebo, due to an imbalance in the proportion of patients who received subsequent treatment with a PARP inhibitor between the treatment arms (xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx).[63,64]

• Olaparib significantly extended time free from chemotherapy and potential associated toxicities, with significant improvement in both TFST and TSST

  • Consistent with results of the primary PFS analysis, olaparib reduced the risk of receiving first subsequent therapy or death by 70% versus placebo (HR 0.30; 95% CI 0.22, 0.40). There was an unprecedented 36.7-month difference in median TFST between treatment arms (median 51.8 months versus 15.1 months)[63,64]

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

• TSST results were similarly consistent with PFS2 analyses, with a 55% reduction in the risk of receiving first subsequent therapy or death with olaparib versus placebo (HR 0.45; 95% CI 0.32, 0.63; P <0.0001).[63,64]

• OS data are immature, as the majority of patients are still alive and participating in the study (21.0% maturity). At the time of analysis (17 May 2018 DCO), olaparib demonstrated a small numerical OS benefit, and median OS had not been reached in either treatment arm (HR 0.95; 95% CI 0.60, 1.53; P =0.8903). It is planned that final OS analyses will be conducted at approximately 60% maturity (xxxxxxxxxxxxxxxxxxxxxxxx).

Table 14 Summary of secondary efficacy analyses

Source: Data on file: D0818C00001 Clinical Study Report. Tables 22, 23, 25, 26[63,64] Abbreviations: CI, confidence interval; HR, hazard ratio; NR, not reached; OS, overall survival; PFS2, time from randomisation to second progression; TFST, time to first subsequent therapy; TSST, time to second subsequent therapy

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Figure 7 Kaplan-Meier plot of PFS2

==> picture [452 x 227] intentionally omitted <==

Abbreviations: CI, confidence interval; PFS2, time to second objective disease progression Source: Data on file: D0818C00001 Clinical Study Report. Figure 9[64]

Among the subset of patients who had evaluable disease (target or non-target

lesions) at study entry (N=90), there was a higher objective response rate observed with olaparib compared to placebo (xxxxxxxxxxxxxx; Table 15). In those patients with objective response, the median duration of response was xxxxxxxxxxxx for olaparib versus xxxxxxxxxx for placebo. More than twice as many patients achieved complete response with olaparib maintenance treatment, compared to placebo, increasing the likelihood of long-term remission (xxxxxxxxxxxxxxxxxxxxxx).

Table 15 Best overall response for patients with baseline evidence of disease

==> picture [471 x 181] intentionally omitted <==

----- Start of picture text -----
xxxxxxxx Xxxxxxxx Xxxxxxxx
xxxx xxxx
xxxxxxxx
• xxxxxxxx xxxxxxxx xxxxxxxx
• xxxxxxxx xxxxxxxx xxxxxxxx
• xxxxxxxx xxxxxxxx xxxxxxxx
xxxxxxxx
• xxxxxxxx xxxxxxxx xxxxxxxx
• xxxxxxxx xxxxxxxx xxxxxxxx
• xxxxxxxx xxxxxx xxxxxxxx
• xxxxxxxx xxxxxx xxxxxxxx
----- End of picture text -----

Source: Data on file: D0818C00001 Clinical Study Report. Table 30 Abbreviations: SD, stable disease

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Patient-reported outcomes: FACT-O TOI and EQ-5D-5L

In SOLO1, patient-reported HRQoL was assessed using the Functional Assessment of Cancer Therapy – Ovarian (FACT-O) questionnaire. The main endpoint for HRQoL analysis was the FACT-O Trial Outcome Index (TOI), which is based on assessment of physical and functional wellbeing and ovarian cancer specific symptoms.[65] TOI scores range from 0 to 100, with higher scores indicating better HRQoL. A change of at least 10 points in TOI score was considered as a clinically relevant or a minimally important difference.[64]

Baseline scores for the TOI were relatively high, with mean scores of 73.6 and 75.0 for the olaparib and placebo arms respectively.[64] This is as expected in the first-line maintenance setting, as the majority of patients have minimal disease burden after response to platinum-based chemotherapy.

Figure 8 shows that no clinically meaningful changes in TOI score were observed over time in either treatment arm, and that olaparib maintenance treatment was not associated with any detriment to HRQoL.[64]

Figure 8 Mean change in FACT-O TOI score from baseline

==> picture [313 x 267] intentionally omitted <==

Source: Data on file: D0818C00001 Clinical Study Report. Figure 15[63,64] Abbreviation: FACT-O, Functional Assessment of Cancer Therapy – Ovarian; TOI, Trial Outcome Index

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

The impact of treatment and disease state on health state utility was assessed using the EQ-5D-5L index, a standardised measure of health status developed by the EuroQol Group in order to provide a simple, generic measure of health for clinical and economic appraisal.[66] There was no worsening or deterioration in mean EQ-5D5L index score over time for patients in the olaparib arm compared with patients in the placebo arm (Figure 9).[64]

Figure 9: Mean EQ-5D-5L weighted health state index score

==> picture [452 x 227] intentionally omitted <==

Source: Data on file: D0818C00001 Clinical Study Report. Figure 16[64] Abbreviations: bd, twice daily; EQ-5D-5L, EuroQol five dimensions, five levels

B.2.7 Subgroup analysis

The superiority of olaparib over placebo was maintained across all predefined subgroup analyses (Figure 10).[64] The only observed interaction was based on the stratification factor of whether patients had complete or partial response at study entry. Patients with complete response at study entry had a HR of 0.35 (95% CI 0.26, 0.49; median PFS olaparib not reached vs placebo 16.6 months). Patients with partial response had a HR of 0.19 (95% CI 0.11, 0.34; median PFS olaparib 28.6 months vs placebo 5.6 months).[64]

Full details of the methods and results of SOLO1 subgroup analyses are presented in Appendix E.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 38 of 144

Figure 10 Forest plot of progression-free survival by subgroup

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

Abbreviations: bd, twice daily; CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; ULN, upper limit of normal Source: Data on file: D0818C00001 Clinical Study Report. Figure 8[63,64]

B.2.8 Meta-analysis

Not applicable as SOLO1 was the only identified trial of olaparib to provide clinical effectiveness evidence relevant to this appraisal.

B.2.9 Indirect and mixed treatment comparisons

SOLO1 directly compared olaparib versus routine surveillance (placebo), the intervention and comparator of interest for this appraisal. For this reason, indirect and mixed treatment comparisons were not deemed necessary or appropriate to support the clinical effectiveness of olaparib in the proposed treatment setting.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

B.2.10 Safety and tolerability

Treatment exposure

At the time of analysis (17 May 2019 DCO), almost all patients in the SOLO1 trial had completed study treatment as per protocol (95.0% in the olaparib arm and 99.2% in the placebo arm). Only 26 patients in the olaparib arm (10.0%) and 3 patients in the placebo arm (2.3%) received treatment beyond the 2-year treatment period recommended for the majority of patients (Table 15). Median time to

treatment discontinuation or death (TTD) was xxxx months for olaparib, compared to xxxx months for placebo xxxxxxxxxxxxxxxxxxxxxxxxx; Figure 11).[64] The mean daily dose of study treatment administered was xxxxxxxx for olaparib and xxxxxxxx for placebo.[64]

Table 16 Duration of treatment exposure

Source: Data on file: D0818C00001 Clinical Study Report. Table 8, Table 31 and 11.2.1.2 Tables and figures[62]

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Figure 11 Kaplan-Meier plot of TTD

==> picture [452 x 227] intentionally omitted <==

Source: Data on file: D0818C00001 Clinical Study Report. Figure 11[62]

Adverse events

The safety and tolerability observed in SOLO1 was consistent with that

observed in previous studies . The majority of AEs in both treatment arms were mild to moderate in severity, intermittent in nature, and manageable using either standard supportive treatment or olaparib dose modification. The most common AEs reported in the olaparib treatment arm were nausea, fatigue, vomiting, anaemia and diarrhoea (

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 17 and Figure 12).[64]

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 17 Summary of adverse events

Note: Shown are data on adverse events that occurred in at least 15% of the patients in either trial group (except where noted) during the trial intervention or up to 30 days after discontinuation of the intervention.[a] Includes patients with anaemia, decreased haemoglobin level, decreased haematocrit, decreased red blood cell count, erythropenia, macrocytic anaemia, normochromic anaemia, normochromic normocytic anaemia and normocytic anaemia;[b] Includes patients with neutropenia, febrile neutropenia, neutropenic sepsis, neutropenic infection, decreased neutrophil count, idiopathic neutropenia, granulocytopenia, decreased granulocyte count, and agranulocytosis;[c] Thrombocytopenia occurred in less than 15% of the patients in each trial group, but the data are provided to complete the profile of hematologic toxic effects. The data include patients with thrombocytopenia, decreased platelet production, decreased platelet count, or decreased plateletcrit. Source: Data on file: D0818C00001 Clinical Study Report. Table 40 and Table 48.[64] Abbreviations: AE, adverse event; NA, not available.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Figure 12 Most common AEs reported in SOLO1[67]

==> picture [462 x 215] intentionally omitted <==

*Grouped term Abbreviation: AE, adverse event

AEs of grade ≥3 were reported in 39.2% of patients receiving olaparib and 18.5% of patients receiving placebo (

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 17). Consistent with the known safety profile of olaparib, the only AEs of grade ≥3 reported in more than 3% of patients were anaemia (21.5% for olaparib versus 1.5% for placebo), neutropenia (8.5% versus 4.6%) and diarrhoea (3.1% versus 0%).[64]

Serious AEs were reported in 20.8% of patients in the olaparib arm and 12.3% of patients in the placebo arm. The most commonly reported serious AE in the olaparib arm of the SOLO1 trial was anaemia (6.5% versus 0% for placebo).[63,64]

No AEs that occurred during the trial intervention or up to 30 days after discontinuation of the intervention resulted in death.[64]

Adverse events of special interest

During the long-term collection of safety data three cases of acute myeloid leukaemia (AML) (1.2%) and no cases of myelodysplastic syndrome (MDS) were identified in patients in the olaparib arm.[64] All three cases of MDS/AML occurred beyond treatment discontinuation and 30-day follow-up, and resulted in death. Because the deaths due to MDS/AML occurred >30 days after treatment

discontinuation they were not classified as treatment-emergent AEs with an outcome of death. No cases of MDS/AML were observed in the patients receiving placebo.

New primary malignancies occurred in five patients (1.9%) in the olaparib group and three patients (2.3%) in the placebo group. Pneumonitis/ interstitial lung disease occurred in five patients (1.9%) in the olaparib group and no patients in the placebo group.[64]

B.2.11 Ongoing studies

In addition to the ongoing SOLO1 trial, AstraZeneca is undertaking a comprehensive clinical trial programme to investigate the efficacy and safety of olaparib across multiple indications, including:

• Newly diagnosed ovarian cancer maintenance plus bevacizumab (PAOLA-1, NCT02477644)

• BRCA-mutated platinum-sensitive relapsed ovarian cancer (SOLO2, NCT01874353)

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

• Re-treatment of patients with platinum-sensitive relapsed ovarian cancer, after previous progression on a PARP inhibitor (OReO, NCT03106987)

• BRCA-mutated human epidermal growth factor receptor 2 (HER2)-negative metastatic breast cancer (OlympiAD, NCT02000622)

• BRCA-mutated high-risk HER2-negative breast cancer (OlympiA, NCT02032823)

• BRCA-mutated pancreatic cancer (POLO, NCT02184195)

• Metastatic castration-resistant prostate cancer (PROfound, NCT02987543)

B.2.12 Innovation

There have been few advances in treatment options for advanced ovarian cancer for more than 20 years since carboplatin and paclitaxel became the standard chemotherapy regimen against which newer treatment strategies are now evaluated.[68] Trials that have investigated using triplet platinum-based chemotherapy or sequential doublets, dose-dense regimens, anti-angiogenic agents and other targeted maintenance treatments have shown modest benefit with no more than a 30% reduction in the risk of progression or death compared to standard platinumbased chemotherapy (3-weekly carboplatin and paclitaxel), and no change in median PFS of more than 6 months (see Appendix N).

Olaparib is the first and only personalised medicine for patients with newly diagnosed BRCA-mutated advanced ovarian cancer. The magnitude of benefit observed with olaparib versus placebo in the SOLO1 trial is substantial and practicechanging, with a 70% reduction in the risk of disease progression or death in the proposed patient group versus placebo, and a minimum estimated 3-year improvement in median PFS. This is by far the greatest PFS benefit that has been observed to date in trials of first-line treatments for advanced ovarian cancer and may be one of the largest improvements in PFS ever observed in solid tumours. Importantly, the survival benefits observed with olaparib were achieved with an acceptable safety profile and no detrimental impact to patients’ HRQoL.[62]

Although no other PARP inhibitors are currently available for women with newly diagnosed advanced ovarian cancer, it is relevant to note that the safety profile of olaparib may be superior to that observed with other PARP inhibitors (e.g. niraparib

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

or rucaparib), due to improved target selectivity, as less off-target binding and bone marrow sequestration.[69]

B.2.13 Interpretation of clinical evidence

As described previously, no active treatment options are currently available for patients with newly diagnosed BRCA-mutated advanced ovarian cancer, after response to first-line platinum-based chemotherapy. While a small proportion of patients may have long-term remission or even cure, the majority of patients relapse and require retreatment within the first 3 years. There is, therefore, high unmet need for effective and well-tolerated treatment options that prevent or delay recurrence.

SOLO1 was a high-quality, Phase III international, randomised controlled trial of olaparib maintenance treatment versus the current standard of care (ie routine surveillance/placebo) in a large sample of patients with newly diagnosed BRCAmutated advanced ovarian cancer (N=391). The trial population was relatively young (median age 53.0 years) and had a good response to initial surgery and first-line platinum-based chemotherapy. The majority of patients (81.8%) were in complete clinical remission at baseline, with no evidence of disease, ECOG performance status of 0, and normal CA-125 levels. In total, 22 of 391 patients (5.6%) were included from six UK sites.

Clinical effectiveness

The magnitude of PFS benefit observed with olaparib in SOLO1 far exceeds - that observed in previous first line chemotherapy trials conducted in patients with newly diagnosed advanced ovarian cancer (Section B.2.13 and Appendix N). Olaparib reduces the risk of disease progression or death in by 70% versus placebo, and extends median progression-free survival (PFS) by a minimum estimate of 3 years (HR 0.30; P<0.0001).[63,64]

After 41 months of follow-up, the majority of patients (73.3%) in the placebo arm of the SOLO1 trial had progressed or died, confirming the high unmet need for improved treatment options for women with BRCA-mutated advanced ovarian cancer. Median PFS was 13.8 months , consistent with previously published

advanced ovarian cancer studies. The literature suggests that median PFS ranges

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

from 10.6 to 20.7 months when assessed prior to initiation of first-line treatment, and from 12.3 to 13.2 months when assessed after response to first-line platinum-based chemotherapy (Appendix N). In contrast, fewer than half (39.2%) of patients in the olaparib arm had progressed or died within the same follow-up duration. Median PFS in the olaparib arm of the SOLO1 trial had not been reached, but was estimated to be at least 3 years longer than observed in the placebo arm based on PFS sensitivity analyses conducted to assess for the risk of attrition bias (median 49.9 months for olaparib versus 13.8 months for placebo; 36.1-month difference) and informative censoring bias (median 46.9 months versus 11.8 months; 35.1-month difference). It is further validated by analyses of median TFST (median 51.8 months versus 15.1 months; 36.7-month difference).[63,64]

Kaplan-Meier analyses of PFS showed that the clinical benefits of olaparib occurred early, with increasing separation of the curves for olaparib versus placebo from the time of first assessment (12 weeks after randomisation; Figure 13). More than twice as many olaparib-treated patients were progression-free at 3 years after randomisation compared with placebo-treated patients (60.4 vs 26.9%).[64] These data are of clear clinical significance, as disease progression is commonly associated with development or worsening of ovarian cancer-related symptoms, the need for further cytotoxic chemotherapy, deterioration of physical and emotional well-being, and decreased ability to carry out activities of daily living, family duties, and/or work.[44]

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Figure 13 Kaplan-Meier plot of PFS, showing recommended time for treatment discontinuation

==> picture [432 x 239] intentionally omitted <==

Source: Data on file: D0818C00001 Clinical Study Report. Figure 6[63,64] Abbreviations: CI, confidence interval; PFS, progression-free survival

Highly consistent PFS results were observed across all pre-planned sensitivity and subgroup analyses, supporting the robustness of the primary endpoint analysis. Importantly, in the subgroup of patients who had partial response to first-line platinum-based chemotherapy, olaparib reduced the risk of progression or death by 81% versus placebo (N=71, HR 0.19, 95% CI 0.11, 0.34; median PFS 28.6 months versus 5.6 months).[64] More than twice as many patients who entered the trial with evaluable disease achieved complete response with olaparib compared to placebo (xxxxxxxxxxxx) These data further emphasise the important role for olaparib in maintaining local control and preventing or delaying recurrence in patients with newly diagnosed BRCA-mutated advanced ovarian cancer versus the current standard of care (routine surveillance).

In interpreting SOLO1 PFS analyses, it is important to note that the vast majority of patients in both arms of the trial discontinued study treatment at, or before, the 2- year timepoint, as per protocol (90.0% for olaparib and 97.7% for placebo). A much smaller proportion of patients discontinued treatment prior to the 2-year timepoint due to objective disease progression with olaparib versus placebo (19.6% versus

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 49 of 144

60.0%, respectively). There was no evidence of change in the shape of the KaplanMeier plot after the 2-year timepoint when the majority of patients discontinued treatment as per protocol, indicating consistent and sustained benefit beyond treatment completion (Figure 13).[64]

Intermediate clinical endpoints of PFS2, TFST and TSST provide information about the long-term outcomes and reflect real-life treatment decisions and patient experience. At the time of analysis (17 May 2018 DCO), almost half as many patients in the olaparib arm had started a first subsequent therapy or died compared to placebo (38.1 vs 71.8%). Crossover was not permitted within the SOLO1 trial design; however, xxx of olaparib-treated patients and xxxxx of placebo-treated patients received post-progression treatment with a PARP inhibitor at the investigator’s discretion.

Despite the imbalance in subsequent PARP inhibitor use, a statistically significant and clinically meaningful improvement in PFS2 was observed with olaparib versus placebo (HR 0.50; 95% CI 0.35, 0.72; P =0.0002).[63,64] This demonstrates that firstline use of olaparib does not diminish ability of patients to receive and respond to subsequent treatment, should the disease progress.

TFST and TSST analyses were highly consistent with the PFS and PFS2 analyses, demonstrating that olaparib significantly extends time free from chemotherapy treatment and associated toxicities which negatively impact on patient HRQoL . Olaparib reduced the risk of receiving first subsequent therapy or death by 70% versus placebo and extended median TFST by 36.7 months (HR 0.30; 95% CI 0.22, 0.40).[63,64] There was a 55% reduction in the risk of receiving first subsequent therapy or death with olaparib versus placebo, despite the imbalances in subsequent PARP inhibitor use described above (HR 0.45; 95% CI 0.32, 0.63; P <0.0001).[63,64] Importantly, duration of median TFST in the olaparib arm of SOLO1 (51.8 months) was substantially longer than median TSST in the placebo arm (40.7 months). This suggests that on average, patients in the placebo arm had received two lines of subsequent chemotherapy by the time that patients in the olaparib arm were initiating their first subsequent chemotherapy regimen.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Together these data indicate three compelling reasons why use of olaparib in the first-line setting is more beneficial to patients, compared with use of PARP inhibitors in the later-lines :

1. First-line treatment offers the only chance of long-term remission or cure for women with advanced ovarian cancer. The estimated minimum 3-year improvement in median PFS observed with olaparib in SOLO1 far exceeds that observed in trials of PARP inhibitors in the recurrent setting, which ranged from from 6.9 months to 15.5 months (see Appendix O).[21,25,70,71]

2. The majority of patients with advanced BRCA-mutated ovarian cancer respond well to first-line platinum-based chemotherapy and are thus likely to benefit from olaparib maintenance treatment. In contrast, the rates of response to platinum agents decrease sharply with each subsequent recurrence.

3. For the majority of patients who receive olaparib in the first-line setting, the maximum duration of treatment duration is 2 years. Only 10% of patients in SOLO1 continued to receive olaparib beyond this time. In contrast, patients who receive PARP inhibitors for platinum-sensitive relapsed ovarian cancer are treated until progression. In Study 19, 11% of patients who received olaparib capsules for platinum-sensitive relapsed ovarian cancer remained on treatment, without progression, for more than 6 years.

SOLO1 OS data are currently immature, as the majority of patients are still alive and participating in the study (21.0% data maturity). The interim analysis showed a small numerical OS benefit with olaparib versus placebo (HR 0.95; 95% CI 0.60, 1.53; P =0.8903), but is biased in favour of placebo due to an imbalance in post-

progression use of PARP inhibitors outside of the study (xxxxxxxxxxxxx

xxxxxxxxxxxxx). This form of bias is widely recognised in ovarian cancer clinical trials, particularly in the adjuvant setting.[72-75] The observed data shows that patients continue to experience a benefit from olaparib after stopping treatment around the 2- year timepoint. There is also no evidence that olaparib impacts the ability of patients to receive and respond to subsequent treatment. It is likely that the minimum

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

estimated 3-year improvement in median PFS observed with olaparib in SOLO1 will translate to an improvement in OS, but the magnitude of long-term benefit is uncertain. Final OS analyses will be event-driven and are planned to be conducted at 60% data maturity (xxxxxxxxxxxxxxx).

Patient-reported outcomes: FACT-O TOI and EQ-5D-5L

FACT-O and EQ-5D-5L analyses demonstrate that the clinical benefits of olaparib observed in the SOLO1 trial were not associated with any detriment in HRQoL or health status versus placebo, despite the longer duration of therapy. This is important as patients with newly diagnosed advanced BRCA-mutated ovarian cancer do not currently receive any active treatment after response to first-line platinumbased chemotherapy. There was no deterioration in HRQoL after the 2-year timepoint, when the majority of patients discontinued treatment.

Safety and tolerability

Olaparib was generally well tolerated in patients with newly diagnosed BRCAmutated advanced ovarian cancer, with AEs that tended to be mild or moderate in severity, and manageable without dose reduction or treatment discontinuation.[63,64] The most commonly reported AEs in the olaparib group of SOLO1 were nausea, fatigue/asthenia, vomiting and anaemia, consistent with the AE profile observed in previous studies conducted in the platinum-sensitive relapsed setting (Study 19, SOLO2).[70,71]

Collectively, these data suggest that the safety and tolerability profile of olaparib is suitable for use as a maintenance treatment option in patients with newly diagnosed BRCA-mutated advanced ovarian cancer. It should be noted that olaparib has been approved for use in the platinum-sensitive relapsed setting since 2015, so medical oncologists who specialise in the treatment of ovarian cancer will already be familiar with recommendations for managing AEs.

Strengths and limitations of the clinical evidence

Strengths

• SOLO1 was a robust, high-quality, double-blinded randomised placebo-controlled

trial that directly compared the intervention and comparator of interest for this

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

appraisal in a large sample of patients with newly diagnosed BRCA-mutated advanced ovarian cancer, who were in response (complete or partial) to first-line platinum-based chemotherapy (N=391). The quality assessment presented in [Section B.2.5] confirmed the risk of bias within this study to be low.

• At the time of analysis (17 May 2018 DCO) all patients in SOLO1 had been followed for a minimum duration of 36 months. The median duration of follow-up was 41 months, providing confidence in the robustness of efficacy and safety results.

• The primary endpoint, PFS, is the Gynecological Cancer Intergroup (GCIG) preferred endpoint for ovarian cancer clinical trials conducted in this disease setting.[76] The magnitude of PFS benefit observed in SOLO1 is unprecedented in newly diagnosed advanced ovarian cancer and far exceeds that observed in previous first-line chemotherapy trials (Appendix N). Highly consistent results were observed across the primary analysis of PFS and all pre-defined sensitivity and subgroup analyses.[63,64]

• The secondary endpoints of PFS2, TFST, TSST and best overall response were consistent with the primary PFS analyses, demonstrating clinically meaningful and statistically significant benefits for olaparib versus placebo. These endpoints are directly relevant to clinical practice and supported by robust analyses.

• The study also included the assessment of patient-reported HRQoL, symptoms and health status as measured using the FACT-O TOI and EQ-5D-5L, demonstrating no detriment.

Limitations

• At the time of analysis (17 May 2018 DCO), the majority of patients in both arms of the SOLO1 trial were still alive and participating in the study. Median PFS, PFS2 and TSST in the olaparib arm had not been reached, and OS data in both trial arms were immature (overall 21.0% data maturity). Whilst there is high confidence in the robustness of clinical effectiveness data presented in this submission, AstraZeneca recognise that there is a degree of uncertainty around the magnitude of clinical benefit that will be realised with further long-term followup. Further analyses of time-to-event endpoints will be event-driven and are anticipated xxxxxxxx.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

B.3 Cost-effectiveness

Summary of the cost-effectiveness analysis

• A three-state cohort-based partitioned survival model was developed to evaluate the cost-effectiveness of olaparib versus routine surveillance in patients with newly diagnosed advanced BRCA1/2-mutated high grade epithelial ovarian, fallopian tube or primary peritoneal cancer who are in response (complete or partial) to first-line platinum-based chemotherapy.

• The model structure comprises three health states of progression-free (PF), progressed disease (PD) and death, and is populated with clinical data (timeto-event outcomes, EQ-5D health state utilities, and adverse events) from the SOLO1 study and clinical literature

• PFS and OS were modelled using a piecewise modelling approach based on observed Kaplan-Meier data up to year 2 (end of olaparib treatment for 90% of patients in SOLO1), and survival functions fitted to data from year 2 onwards. PFS and OS were modelled up to a lifetime horizon of 50 years • Patients that remain PF for at least 7-years after response to first-line platinum chemotherapy were considered as long-term survivors of ovarian cancer and are no longer at risk of relapse. All events after year 7 landmark were modelled as deaths unrelated to ovarian cancer. Alternative landmarks were applied in sensitivity analysis. • The base case predicted that olaparib provided xxxxxx additional QALYs, with an incremental cost of xxxxxxxx. The cost per QALY gained versus routine surveillance was £11,830. In the probabilistic analysis, the corresponding cost per QALY gained was £11,941, and olaparib has a 99% probability of being cost-effective at a willingness to pay threshold of £30,000.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

B.3.1 Published cost-effectiveness studies

A systematic literature review of studies reporting the economic evaluation, health state utility (HSU) and cost-of-illness of patients with newly diagnosed BRCA mutated advanced (FIGO Stage III–IV) ovarian cancer following first-line platinumbased chemotherapy was conducted on 25 May 2018. Full details of the cost effectiveness systematic literature review are presented in Appendix G.

In total, 26 studies met the inclusion criteria for the systematic literature review. Of these, 4 reported cost-of-illness data, 2 reported HSU values (HSUV) and 15 reported the economic evaluation of treatments for ovarian cancer. The cost of illness and HSUV studies are summarised in sections B.3.4 and B.3.5 of the submission, respectively. A summary of the economic evaluation studies is provided below.

All 15 economic evaluation studies reported the cost-effectiveness of maintenance PARPi treatment in patients with a BRCA mutation and platinum-sensitive ovarian cancer that had a complete or partial response to therapy after at least two lines of platinum chemotherapy. The studies used data from Study 19 (NCT00753545), ARIEL3 (NCT01968213), and ENGOT-OV16/NOVA (NCT01847274). Only one study reported the cost-effectiveness of treatment from a UK perspective.[60] None of the identified studies reported on the cost-effectiveness of maintenance therapy in the first-line setting.

The search of published evaluations was supplemented by hand-searching of manufacturer submission and evidence review/ assessment group reports from previous NICE Health Technology Appraisals (HTA). Two published HTAs relating to the treatment of patients with ovarian cancer and a BRCA mutation were identified: NICE TA381 and NICE TA528.[59] Further detail on the evidence review group’s assessment of TA381 was published in a secondary reference by Tappenden et al. (2017).[77]

Both TA381 and TA528 were conducted in patients eligible for maintenance PARPi therapy after two prior lines of platinum-based chemotherapy. Of note, in all

appraisals, the evidence reviews groups (ScHARR and BMJ-TAG) stated preference

for the use of partitioned survival modelling in predicting the lifetime costs and health Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved Page 55 of 144

effects of treatment. Further, in TA528, the committee had considered the modelling of OS gain based on an assumed ratio of PFS gain (values ranging from 1:1 to 2:1 for OS to PFS gain) when recommending on the use of niraparib maintenance therapy, due to uncertainty in the OS data available at the time of its assessment.

At the time of writing (November 2018), NICE was also undertaking an appraisal of olaparib tablets for maintenance treatment of patients with platinum-sensitive relapsed ovarian cancer [ID1296].[61] As in previous appraisals, the evidence review group and committee expressed a preference for the use of partitioned survival modelling for estimating cost-effectiveness of maintenance treatment. A summary of the included studies and HTAs with results in British Pounds (GBP) is presented in Table 18 below.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 18 Summary list of published cost-effectiveness studies in patients with BRCA-mutated ovarian cancer

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved Page 57 of 144

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 58 of 144

Abbreviations: BRCA, breast cancer susceptibility gene; ERG, Evidence Review Group; ICER, incremental cost-effectiveness ratio; NR, not reached; PD progressed disease; PF, progression free; QALYs, quality-adjusted life year.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 59 of 144

B.3.2 Economic analysis

As the systematic literature review did not identify an existing economic evaluation of maintenance therapy in newly diagnosed patients with ovarian cancer, a de novo decision analytic model was constructed in Microsoft[®] Excel to estimate the incremental cost-effectiveness of olaparib maintenance therapy versus routine surveillance in this setting. Key characteristics of the de novo analysis are shown in Table 19. Further detail is provided in subsequent sections.

Table 19 Summary of the de novo economic analysis

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Patient population

In line with the NICE scope, the de novo economic analysis evaluates the costeffectiveness of olaparib tablets versus routine surveillance in the maintenance treatment of patients with newly diagnosed advanced BRCA1 - and BRCA2 -mutated high grade epithelial ovarian, fallopian tube or primary peritoneal cancer who are in response (complete or partial) to first-line platinum-based chemotherapy. This population is consistent with the FAS of the SOLO1 study, and the primary source of clinical data in the economic analysis. The baseline characteristics of the SOLO1 population is summarised in Table 10 of the submission. The majority of patients randomised to treatment in SOLO1 had:

• No residual disease, having had cytoreductive surgery (>97%) and 82% having a complete response to their platinum chemotherapy.

• A good performance status

• And CA-125 levels within the normal range

The SOLO1 population is considered representative of patients eligible for maintenance treatment after first-line platinum chemotherapy in clinical practice in England.

Intervention technology and comparators

The intervention is the tablet formulation of olaparib at the recommended daily dose of 300 mg (two 150 mg tablets) taken twice daily. This dosage is aligned to the anticipated European Marketing Authorisation for olaparib in this indication.

As in SOLO1 and the draft SmPC,[82] olaparib tablets are to be administered up to disease progression or unacceptable toxicities for a maximum of 2 years in patients

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

with no residual disease. For patients with residual disease, the draft SmPC includes the option of continuing treatment beyond 2 years, as permitted within SOLO1 (see Section B.2.3 for further detail). Treatment beyond 2 years is captured within the sensitivity analysis. As olaparib is provided in a convenient tablet formulation, no additional healthcare support (eg inpatient visits) is required beyond the dispensing and monitoring of therapy. These costs are captured in the economic analysis.

The comparator is ‘routine surveillance’, comprising patient observation, follow-up, and general supportive or symptomatic care.

Time horizon

In line with the NICE reference case, a lifetime horizon (50 years) from the date of starting maintenance treatment was used in the base case. This covers the period over which all important differences in costs or outcomes between olaparib tablets and routine surveillance would be observed, including those relating to the subset of patients expected to achieve long-term survival after first-line platinum chemotherapy (see Section B.2.1).

Discounting

The discount rate used in the base case for both costs and outcomes is 1.5% per annum. Section 6.2.19 of the 2013 NICE methods guide[79] recommends that if it is likely that based on the evidence presented, long term health benefits are likely to be achieved, a discount rate of 1.5% should be considered by the committee. The evidence presented herein demonstrates that patients in this setting are highly likely to have long term health benefits (ie >30 years). This assumption is also in line with recommendations made in the green book for discount rates to be applied to health and life values.[78]

A discount rate of 3.5% is tested in sensitivity analyses.

Perspective

The model adopts a NHS/PSS perspective as recommended by the NICE reference case.[79] This includes resource use and costs associated with disease management, treatment acquisition, adverse events and end-of-life care.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved Page 63 of 144

Model structure

A three-state cohort-based partitioned survival (or ‘area-under the curve’) model was developed to assess the cost-effectiveness of olaparib versus routine surveillance. This modelling approach is consistent with the preferred approaches of review groups and committees in previous NICE appraisals of maintenance treatment in ovarian cancer (TA310, TA508), and is consistent with the approaches adopted in the majority of economic evaluations submitted to the NICE for the HTA of treatments for advanced cancer.[83,84] An illustration of the model state structure is provided in Figure 14 and the calculation method is shown in Figure 15.

Figure 14 Model schematic

==> picture [331 x 206] intentionally omitted <==

Note: Health state transitions are not explicitly modelled in the partitioned survival analysis. The direction of transition in the model is provided as an illustration.

The health states are defined as:

• Progression-free after response to first-line chemotherapy (PF)

• Radiologically confirmed progressed disease (PD)

• Death, from any cause

The three states are mutually exclusive and fully exhaustive, meaning that patients

must occupy one of the states at any given time. The PF and PD status of the cohort is modelled on the primary PFS endpoint of SOLO1 as assessed by study

investigators. PFS was assessed according to the modified RECIST criteria version

1.1, which defines progression as:

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved Page 64 of 144

• Appearance of new lesions in patients with clinical complete response at entry

• At least a 20% increase or absolute 5mm increase in the sum of diameters of target lesions taking as reference the smallest sum on study in patients with partial response at entry

The death state captures deaths from cancer and non-cancer related causes. In current clinical practice and without the use of olaparib maintenance therapy, approximately 10–20% of patients with stage III–IV epithelial ovarian cancer will be classified as long-term survivors having remained PF beyond 5–10 years since diagnosis.[85] This ‘exceptional’ responder group is expected to achieve long-term remission and experience mortality risks approaching that of the general population, matched by age and gender.[46]

To reflect long-term survival in the model, the survival rate for PFS after a chosen landmark time was set equal to all-cause mortality rates for persons with a BRCA mutation that have no evidence of cancer. The landmark time at which a patient is a long-term survivor of ovarian cancer varies across the literature and includes survival of >5–10 years after initial diagnosis. In the base case, a landmark of 7 years was selected based on:

• Expert advice that patients free of progression 5 years after completing 2 years of olaparib therapy are expected to be ‘exceptional’ responders and considered for discharge to primary care

• Evidence suggesting that relapse after 5 years of disease-free survival is rare in ovarian cancer[86]

• Data from the Edinburgh Ovarian Cancer Database suggest that the rate of relapse following diagnosis of ovarian cancer reduces to zero at approximately 7-8 years (Appendix M for further information)

Alternative landmarks of 5, and 10 years were considered in sensitivity analysis.

The use of olaparib maintenance therapy in this setting is expected to increase the proportion of patients with long-term survival due to the substantial increase in PFS, as observed in SOLO1. The potential for increased long-term survival in patients treated with olaparib is supported by data in the later line ovarian cancer setting from Study 19 showing that a higher proportion of patients achieve 'exceptional’ response

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved Page 65 of 144

to olaparib and survive beyond 5 years (35.5% versus 20.86% for placebo) in the 2[nd] or later line recurrent ovarian cancer setting. Further detail on the modelling of OS is provided in later sections.

The health states of PF and PD represent clinically relevant landmarks in the treatment of patients with advanced ovarian cancer after surgery and response to first-line platinum-based chemotherapy, by reflecting a state of disease remission (PF after response to first-line chemotherapy) and the return of disease (PD) with its associated morbidity and mortality burden to the patient. The onset of progression in a maintenance setting also marks the transition from a state of “inactive” disease to a state of progressive disease, requiring a shift in the follow-up and the management of patients alongside the administration of further treatment with its associated costs to the NHS. Furthermore, the selected health states are consistent with the clinical endpoints assessed in SOLO1 including the primary endpoint of radiological PFS, and the key secondary endpoint of OS.

A graphical illustration of the partitioned survival method used in this submission is provided in Figure 15, below.

Figure 15 Illustration of the partitioned survival calculation

==> picture [439 x 243] intentionally omitted <==

Abbreviations: OS, overall survival; PF, progression-free.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

As outlined in the Decision Support Unit (DSU) review of partitioned survival analysis (TSD19), the partitioned survival method uses PFS and OS curves to directly estimate the proportion of patients occupying each state over time. The proportion occupying the PF state are estimated directly from the cumulative survival probabilities for PFS, while the proportion occupying the PD state are estimated from the cumulative survival of OS minus the cumulative survival of PFS. The term “partitioned survival” refers to the use of PFS to partition the area under the OS curve to those alive and PF (PFS) and those alive and in PD (OS minus PFS), as described previously. The numbers occupying the death state are estimated from one minus the OS curve. State occupancy is evaluated at monthly intervals equivalent to 30.44 days (365.25/12). The partitioned survival approach makes direct use of parametric survival curves fitted to the key primary and secondary time-toevent endpoints of SOLO1: PFS and OS. As noted in TSD19, partitioned survival modelling is well understood, intuitive, easy to communicate and has been accepted by NICE in previous ovarian cancer appraisals.

Alongside PFS and OS, the model independently simulates the time on treatment with olaparib using Kaplan-Meier data on the time from randomisation to discontinuation of study drug in SOLO1. This ensures that modelled drug costs for olaparib reflect drug usage in SOLO1, including the time on treatment for those that discontinue therapy early (eg prior to progression and before completing the full 2 years of treatment) due to unacceptable toxicity.

The drug costs of PARP therapy (olaparib capsules or niraparib) administered after progression in SOLO1 in a second or later line maintenance setting were included in the analysis to reflect their use in SOLO1, and expected use in clinical practice in England and Wales given NICE guidance recommending niraparib maintenance after 2 courses of platinum chemotherapy (TA582) and olaparib capsules after 3 or more courses of platinum chemotherapy (TA381). In SOLO1, xxxxxxxxxxxxx) of placebo patients received a subsequent PARP versus xxxxxxxxxxx of olaparib patients.

To enable the application of discounting of subsequent PARP costs as per the NICE reference case, subsequent PARP use was modelled using data on the time to first

subsequent PARP therapy in SOLO1 and data on the time on treatment of olaparib Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved Page 67 of 144

capsules from the germline BRCA sub-population of the phase II trial, Study 19. The time to first PARP use in SOLO1 were used to estimate the proportion of patients’ starting therapy by model cycle, while time on treatment data from S19 were used to estimate the proportions on therapy relative to the start of PARP treatment. When combined, these data estimate the proportion of patients receiving subsequent PARP treatment by cycle in the model.

While the use of subsequent PARP is expected to differ across treatment arms in the model, evidence from S19 and SOLO1 suggest that the overall use of platinum and non-platinum chemotherapy is likely to be similar and therefore have a limited impact on the incremental results of the analysis.[63,87] To simplify the analysis, the drug and administration costs of subsequent platinum and non-platinum therapy were therefore included as one-off costs on progression. These costs were hence not fully adjusted for the effects of discounting, as considered for subsequent PARP therapy. However, as these costs are similar across the arms, albeit expected to occur on average later in the olaparib arm, this is not expected to impact the results.

Other costs captured in the analysis include AEs and the costs of routine follow-up and disease and treatment monitoring. AE costs were captured as a one-off cost at the start of the model, and included grade 3 or above anaemia, neutropenia and diarrhoea. Follow-up and monitoring costs were modelled on the PF and PD status of the cohort. The rate of resource consumption for patients occupying the PF state were assumed to vary over time to reflect the changing pattern of follow-up while on maintenance treatment (<2 years), up to discharge (7 years) and beyond this time point. Further detail is provided in Section B.3.5.

Consistent with the NICE reference case, the health benefits of treatment were measured in terms of quality adjusted life years (QALYs) using EQ-5D-based HSUVs evaluated using UK general population preference weights. EQ-5D-5L data routinely collected in SOLO1 were mapped to EQ-5D-3L HSUV using the Van Hout et al crosswalk algorithm,[88] as recommended by NICE. As with previous NICE appraisals in ovarian cancer, HSUV were assigned to the states of PF and PD. Given the use of a lifetime horizon (ie 50 years), the HSUV assigned to both PF and PD were adjusted for the gradual decline in health status expected with age, using

the regression analysis of general population EQ-5D-3L HSUV from Ara et al.[80] The Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved Page 68 of 144

effects of AEs on HSUV were modelled as a one-off QALY loss applied at the start of the model.

In developing the model, various alternatives to partitioned survival modelling (including Markov and semi-Markov state transition modelling) were considered but not judged applicable to this appraisal for the following reasons:

• Time in state methods (as used in TA528) do not allow for the discounting of costs and outcomes over time and are therefore not in line with the NICE reference case. They also do not consider state occupancy over time and potentially over simplify the treatment pathway

• Markov modelling requires estimates of transition probabilities between the states of PF, PD and death. For transitions that occur post-randomisation, e.gg progression to death (or post-progression survival), the events rates observed in SOLO1 are likely subject to bias from informative censoring due to the much later progression in the olaparib arm (e.g. fewer post-progression events may be observed for olaparib than placebo due to a shorter observation period arising from the delayed progression for olaparib) and from selection bias due to responders having not progressed at the time of analysis. Further detail on these issues is provided in NICE TSD19.

A comparison of methods selected for this appraisal and the approaches adopted in previous ovarian cancer appraisals is provided in Table 20. The approaches used in this submission closely match the preferred methods of the committees and review groups in previous ovarian cancer appraisals.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 20 Features of the economic analysis and comparisons with previous appraisals in the relapse/recurrent advanced ovarian cancer setting

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 70 of 144

Abbreviations: BNF, British National Formulary; CMU, Commercial Medicines Unit; EQ-5D, EuroQol 5-dimension Questionnaire; EQ-5D-3L, 3-level EuroQol 5-dimension Questionnaire; NA, not applicable; NR, not reported; OS, overall survival; QALY; quality-adjusted life year; TA, technology appraisal.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved Page 71 of 144

B.3.3 Clinical parameters and variables

All clinical data used in the analysis were obtained from the SOLO1 study and based on data from the FAS population, analysed at the primary DCO of 17 May 2018. PFS was modelled based on the primary endpoint of modified RECIST v1.1 as assessed by the study investigator, while OS was modelled on the secondary endpoint of time from randomisation to death from any cause.

Survival curves for PFS were extrapolated up to a landmark of 7 years, after which point an adjusted all-cause mortality rate was assumed (see previous sub-section on model structure for justification B.2.4). The adjustment for long-term survival was not applied directly to OS to avoid assigning all-cause mortality rates to deaths from the PD state. OS data were modelled up to the point where the cumulative survival probabilities for OS were predicted to be equal to or less than the cumulative survival of PFS, at which point, the OS curve followed the trajectory of PFS. This reflects the longer-term trend of survivors being those with an “exceptional” response having not progressed and is a logical constraint in the model to avoid negative numbers occupying the PD state (eg if OS < PFS and PD = OS – PFS, then PD <0).

Further detail on the modelling of PFS and OS is available in the following sections. The general method of survival modelling is detailed below and applies to both PFS and OS.

An illustration of the approach to modelling PFS and OS is provided in Figure 27.

General method of survival analysis

The process of survival model fitting followed the approaches recommended by the Decision Support Unit (Latimer 2011),[83] and approaches accepted in previous appraisals in cancer.

This approach included:

• An assessment of log-cumulative hazard and suitable residual plots to assess whether proportional hazards (or odds or accelerated failure time) can be assumed

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

• If plots were not parallel then independent functions were fitted to each arm, and if plots showed non-straight lines, consideration was given to other flexible modelling techniques

• Standard parametric models, including Exponential, Weibull, Log-normal, Loglogistic, Gompertz, and Generalised Gamma, were fitted to the entire data set. Covariates for patient characteristics were not included in the parametric analysis because baseline characteristics were balanced across treatment arms in the SOLO1 study population.

In support of the methods recommended by the DSU, we further considered the use of “piecewise” modelling methods similar to those accepted in other NICE appraisals in adjuvant and advanced cancers (TA428, TA531, TA519).[89-91] These methods involve the fitting of survival functions to different regions of the survival curve in order to improve on model fit or provide more plausible long-term extrapolations.[92] In the case of SOLO1, the use of a “piecewise modelling” method is justified on the basis that ;

• The use of a single survival curve fitted to the entre data set may not yield plausible estimates of long-term survival given the presence of “exceptional” responders in both the routine surveillance and olaparib arms of the model. The use of models fitted to the later portion of the curve may better capture the longterm survival trend expected in this population by excluding survival data from those with early progression (eg PFS <2 years)

• In SOLO1, olaparib maintenance treatment was limited to 2 years in patients that had a complete response at entry (81.8% of patients). As noted previously, there was no evidence of change in the shape of the Kaplan-Meier plot after the 2-year timepoint indicating consistent and sustained benefit beyond treatment completion. To explore this further, and to resolve any uncertainty over the continued and sustained benefit of olaparib beyond this time point, we explored the use of survival curves to the post-24-month period.

To align with the design of SOLO1, survival curves were fitted to the post-2-year period of study follow-up for both PFS and OS and compared alongside the models fitted to the entire data set. This time point is before the median follow-up for PFS of

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved Page 73 of 144

SOLO-1 (approximately 41 months) thereby retaining sufficient data to support longterm extrapolations.

The analysis was performed on all patients that were censored for PFS/OS or had a PFS/OS event after month 24. Event times were re-baselined to estimate the time from month 24 to progression or death (eg time from randomisation to progression or death minus 24 months). In the Excel model, the cumulative survival probabilities from this analysis were applied to the proportion with PFS or OS at month 24 to predict outcomes beyond this time. For consistency, the same time point was used for both olaparib and routine surveillance.

The two methods, “entire data set” and “piecewise”, were then assessed based on:

• Goodness of fit (AIC/BIC),

• Fit to Kaplan-Meier plot and landmark survival probabilities, and

• Clinical plausibility of model extrapolations and relevant UK data

Alternative approaches to estimating plausible OS projections were also performed, as outlined in the later sections.

Relevant and clinically plausible best fitting model was selected for the base case. Alternative plausible models were then considered in sensitivity analysis.

Progression-free survival up to the landmark for long-term survival

At the time of DCO there were 198 PFS events (50.6% maturity) with more events on the routine surveillance arm than the olaparib arm (73% routine surveillance vs 39% olaparib, respectively). After a median follow-up of approximately 41 months, the median was not reached for patients in the olaparib arm versus 13.8 months for patient in the placebo arm. The sample sizes for the analysis of PFS from

randomisation (“entire data set”) and PFS from month 24 (“piecewise”) were 131 and 41 for placebo, and 260 and 172 for olaparib, respectively. The Kaplan-Meier plot for PFS (randomisation to progression or death) is shown in Figure 16 below.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Figure 16 SOLO1 PFS Kaplan-Meier curve

==> picture [374 x 271] intentionally omitted <==

Source: Data on file: D0818C00001 Clinical Study Report. Figure 6[63,64] Abbreviations: bd, twice daily

The corresponding Kaplan-Meier plot for PFS from month 24 onwards (with fitted survival models) is shown in Figure 20 . The plot clearly demonstrates that progression rates remain lower for the olaparib arm versus placebo, indicating continued benefit of treatment despite the cessation of study drug in the majority of patients.

Inspection of the log cumulative hazards (Figure 17) and Schoenfeld residual plots (Figure 18) for PFS suggest that treatment effect is likely to vary over time. Following the DSU process, independent models were therefore fitted to each arm of the study including to the entire data and to the post-24-month period.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Figure 17 Cumulative hazards plot of PFS

==> picture [287 x 287] intentionally omitted <==

Figure 18 Schoenfeld residuals of PFS

==> picture [249 x 250] intentionally omitted <==

The Akaike information criterion (AIC) and Bayesian information criterion (BIC) statistics for PFS are presented below in Table 21.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 21 Summary of separate AIC and BIC goodness of fit data for PFS

Abbreviations: AIC, Akaike information criterion; BIC, Bayesian information criterion; BICR, blinded independent central review; PFS, progression-free survival.

According to AIC, the best fitting models to the entire data set was the loglogistic for olaparib and generalise gamma for routine surveillance, and the log-normal for both arms in the post 24-month period. In general, the models fitted to the entire data set produced reasonable visual predictions of the Kaplan-Meier plot for PFS for the olaparib arm but poorly estimated several regions of the Kaplan-Meier plot for placebo (Figure 19 below). With the post-24-month analysis, most models produced a reasonable and consistent prediction of the Kaplan-Meier data for olaparib and placebo; the key exception being exponential for olaparib.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Figure 19 Visual representation of fitted parametric models to entire data set

==> picture [558 x 280] intentionally omitted <==

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 78 of 144

Figure 20 Visual representation of fitted parametric models to PFS from month 24 onwards

==> picture [337 x 280] intentionally omitted <==

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 79 of 144

The fit of the parametric models to Kaplan-Meier estimates of PFS at landmarks of 3, 6, and every 6 months up to 4 years for olaparib and placebo are shown in Table 23 and Table 24, respectively. The predictions have been colour-coded to show the accuracy of the models in predicting landmark survival in SOLO1 (green is within 1.0% of Kaplan-Meier estimates; amber is within 1–3% of Kaplan-Meier estimates and red is >3.0% deviation from Kaplan-Meier estimates). For ease, the survival data used in the first 24 months of the “piecewise” method are excluded from the summary tables given that outcomes during this period were directly estimated from the Kaplan-Meier data and therefore exactly reproduces the estimates from the study.

In general, as shown in Table 23, most models including those fitted to the entire data or as Kaplan-Meier up to month 24 and parametric model thereafter (“piecewise” method), generated plausible estimates of landmark survival for the olaparib arm. Overall, the “piecewise” method yielded the fewest amber or red predictions (>1% deviation from landmark estimates), in part, due to the use of Kaplan-Meier data up to month 24 and through its improved prediction of survival in the post-24-month period. The model with the fewest deviations of >1.0% from the landmark survival in SOLO1 was the “piecewise” method with log-normal survival from month 24 onwards.

In contrast to olaparib, there was greater deviation in the fit of the different survival models and methods (“entire data” or “piecewise”) to the placebo arm of SOLO1, as shown in Table 24. All the models fitted to the entire data set (e.g. as per DSU guidance) had at least three red predictions (>3.0% of landmark survival) of the Kaplan-Meier estimates from SOLO1. Other than the survival estimate at 4 years, which is highly uncertain due to the small numbers at risk (n=1 in placebo arm), the “piecewise” method again yielded the fewest amber or red predictions. As with olaparib, the model with the fewest red or amber predictions for routine surveillance was the “piecewise” method with log-normal survival from month 24 onwards.

The survival estimates for routine surveillance were further compared to relapse-free survival data for patients with BRCA1 or BRCA2 mutations registered to the Edinburgh Ovarian Cancer Database. This database contains outcome data for patients diagnosed with ovarian cancer in the South East region of Scotland (N > 4000). Relapse data from a subset of patients with BRCA mutated high grade serous ovarian cancer (n=160) were assessed using Kaplan-Meier methods (Appendix M).

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved Page 80 of 144

Relapse-free survival at 3 ,4, 5 and 7 years were compared to estimates from the routine surveillance SOLO1 survival models. The method that yielded the closest prediction of the long-term RFS data xxxxxxxxxxxxxx from the UK database was the “piecewise” method with log-normal survival from month 24 onwards

xxxxxxxxxxxxxxxxxxxx

Based on statistical goodness of fit (AIC), fit to the Kaplan-Meier estimates, and prediction of landmark progression in a UK population, the preferred method for modelling PFS in the base case was the “piecewise” method with log-normal survival from month 24 onwards for both olaparib and routine surveillance. The fitted coefficients for the log-normal survival distributions are provided in Table 22.

Table 22 Fitted parameters for the log-normal distribution fitted to PFS from month 24 onwards

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

----- Start of picture text -----
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx
xxxxxxxx
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx
xxxxxxxx
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx
xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx
----- End of picture text -----

Clinically relevant alternatives to the base case method include the generalised gamma fitted to the entire data set, and the “Piecewise” method with log-logistic in the post-24-month period. These options were considered in sensitivity analysis.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 23 Prediction of Kaplan-Meier data and long-term extrapolation of PFS with olaparib using the Kaplan-Meier and parametric model (“piecewise”), and fully parametric model methods (“entire data”)

Green cells correspond to prediction of within 1.0% of Kaplan-Meier estimate, amber cells are prediction of within 1.0–3.0% of Kaplan-Meier estimate and red is greater than 3.0% difference to Kaplan-Meier estimate.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 82 of 144

Table 24 Prediction of Kaplan-Meier data and long-term extrapolation of PFS with placebo using the Kaplan-Meier and parametric model (“piecewise”), and fully parametric model methods (“entire data”)

==> picture [699 x 269] intentionally omitted <==

----- Start of picture text -----
20.40% 17.00% xxxxxxxxxxxxxx
20.40% 17.00%
20.40% 17.00% 20.40% 17.00% 20.40% 17.00% 20.40% 17.00% 20.40% 17.00% 20.40% 17.00% 20.40%
20.40% 17.00% 20.40% 17.00% 20.40% 17.00% 20.40% 17.00% 20.40% 17.00% 20.40% 17.00% 20.40% 17.00% 20.40%
20.40% 17.00% 20.40% 17.00% 20.40% 17.00% 20.40% 17.00% 20.40% 17.00% 20.40% 17.00% 20.40% 17.00% 20.40%
20.40% 20.40% xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx
20.40% 20.40% xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx
Xxxxxxx 20.40% 20.40% xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx
Xxxxxxxxx 20.40% 20.40% xxxxxxxxxxxxxxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx
xxxxxxx
20.40% 20.40% xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx
20.40%xxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx
20.40%
xxxxxxx
17.00% 20.40% xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx 19.40% 15.30% 9.60% 3.80% 3.70%
20.40% 20.40% xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx 18.50% 14.40% 8.60% 3.00% 3.00%
20.40% 20.40% xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx 18.00% 15.20% 11.30% 7.10% 6.90%
20.x
xxxxxxxxx 20.40% 20.40% xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx 19.10% 16.00% 11.50% 6.60% 6.40%
20.40% 20.40% xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx 22.40% 19.30% 15.00% 10.20% 10.00%
20.40%xxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx xxxxxxx
20.40% 22.80% 20.40% 17.00% 12.80% 12.60%
xxxxxxx
----- End of picture text -----

Green cells correspond to prediction of within 1.0% of Kaplan-Meier estimate, amber cells are prediction of within 1.0–3.0% of Kaplan-Meier estimate and red is greater than 3.0% difference to Kaplan-Meier estimate.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 83 of 144

Progression-free survival beyond the landmark for long-term survival

As outlined previously, the survival rate for PFS after the landmark of 7 years was modelled on the all-cause mortality rate for persons with a BRCA mutation without evidence of cancer. These patients are considered as “exceptional” responders whose disease is unlikely to return, with their mortality risk approaching that of the age and gender matched general population.

The mortality rate for persons with a BRCA mutation and no evidence of cancer was estimated from age- and gender-matched all-cause mortality data from the office for national statistics, adjusted for the potential excess mortality risk of having a BRCA1 or BRCA2 mutation. The excess mortality risk was modelled using a hazard ratio for mortality of 1.26 (0.00, 3.42), based on the excess mortality of female carriers of the BRCA1 or BRCA2 mutation, aged 51–60 years and who have an absence of

melanoma and cancer of the breast, ovary, and pancreas, as reported by Mai et al.[93] This hazard ratio was applied throughout the lifetime of the cohort, eg assuming proportional hazards, as a simplifying assumption. The impact of varying the hazard ratio for BRCA1/2 all-cause mortality on results was assessed in sensitivity analysis.

Overall survival

At the time of PFS analysis, the interim OS data were highly immature (82/391 death events, 21% maturity) and the median OS was not reached in either treatment arm. The HR suggested no OS detriment for patients in the olaparib arm (HR=0.95). The Kaplan-Meier plot is characterised by separation in the curves favouring olaparib after 1 year that is sustained to approximately 3 years. The Kaplan-Meier plot for OS (randomisation to death) is shown below.

Following the methods used for PFS, the analysis of OS included the fitting of survival models to the entire data set and to the post-24-month period. The sample sizes for the analysis of OS from randomisation (“entire data set”) and OS from month 24 (“piecewise”) were xxxx and xxx for placebo, and xxx and xxx for olaparib, respectively.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Figure 21 SOLO1 OS Kaplan-Meier plot

==> picture [240 x 227] intentionally omitted <==

Inspection of the log cumulative hazards (Figure 22) and Schoenfeld residual plots (Figure 23) for OS suggest that treatment effect is likely to vary over time for at least the first 24 months. Following the DSU process, independent models were therefore fitted to each arm of the study including to the entire data and to the post-24-month period.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Figure 22 Cumulative hazards plot of OS

==> picture [306 x 307] intentionally omitted <==

Figure 23 Schoenfeld residuals of OS

==> picture [286 x 257] intentionally omitted <==

The AIC and BIC statistics for OS are presented below.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 25 Summary of separate AIC and BIC goodness of fit data for OS

Abbreviations: AIC, Akaike information criterion; BIC, Bayesian information criterion; BICR, blinded independent central review; PFS, progression-free survival.

According to AIC, the best fitting models to the entire data set were the log-logistic for olaparib and generalised gamma for routine surveillance. The AIC statistics generally suggest that for the olaparib arm any of the log-normal, log-logistic or Weibull provide equally relevant statistical fits to the data. The models with the lowest AIC scores for the post-24-month period were exponential for olaparib and generalised gamma for routine surveillance. Again, the AIC statistics generally suggest that all the models fitted to the olaparib arm provided a statistically relevant fit to the data, whereas for routine surveillance, the generalised gamma performed best in terms of AIC.

The fit of the models to the Kaplan-Meier plots for OS are shown in Figure 24 for the entire data set and Figure 25 for the models fitted to the post-24-month period. Due

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

to the data maturity for OS at the time of the PFS analysis, there exists uncertainty surrounding the extrapolation of OS using both the entire data set and post-24month period, resulting in a wide range of potential future OS estimates. This is most notable for the placebo arm, which had a smaller sample size than the olaparib arm due to the 2:1 randomisation in SOLO1 and showed uncharacteristic flattening of the OS curve from approximately 3-years which is clinically implausible.

Figure 24 Fit of independent models to the Kaplan-Meier for OS in SOLO1

==> picture [452 x 227] intentionally omitted <==

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Figure 25 Fit of independent models to the post-24-month Kaplan-Meier period for OS in SOLO1

==> picture [359 x 309] intentionally omitted <==

The fit of the parametric models to Kaplan-Meier estimates of OS at landmarks of 3, 6, and every 6 months up to 4 years for olaparib and placebo are shown in Table 27 and Table 28, respectively. The predictions have been colour-coded to show the accuracy of the models in predicting landmark survival in SOLO1. As observed for PFS, the “piecewise” models generated the best overall fit to the OS landmark data for placebo, with the fewest amber or red predictions, while both sets of methods yielded plausible estimates of landmark survival for the olaparib arm (Table 27).

For olaparib, the parametric models fitted to the entire data set and using the “piecewise” method predict that the cumulative probability of OS will range from 64% to 73% at 5 years, decreasing to 42% to 63% at 7 years. For logical consistency and plausibility, the PFS and OS curve cannot cross. Therefore, as previously discussed, the extrapolated OS curve is used until it crosses the PFS curve. From this point on, the OS follows the trajectory of the PFS curve which is being driven by all-cause

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

mortality as these patients are “exceptional” responders who are assumed to have a mortality rate similar to all-cause mortality.

The predicted median OS ranges from xxxxxxxx (xxxxxx and xxxxxxxx, gompertz fitted to the entire data set) to xxxxxxxx (xxxxxxxx and xxxxxxxx, exponential fitted to the entire data set). For consistency with the PFS analysis, the “piecewise” method was used for modelling of OS, with the log-logistic model selected for the post-24month period based on goodness of fit and conservative and plausible median OS estimate for olaparib (xxxxxxxx, xxxxxxxx and xxxxxxxx). Alternative methods including the “piecewise” method with Gompertz (2nd best fitting) and “piecewise” method with log-logistic (3rd best fitting) were considered in sensitivity analysis.

For routine surveillance, the “piecewise” models predict that the cumulative probability of OS will range from xxx to xxx at 5-years, and from xxx to xxx at 10years, with median OS ranging from xxxxxxxx (xxxxxxxx and xxxxxxxx) to xxxxx xxxxx xx (xxxxx xx and xxxxx xx). The corresponding landmark probabilities of OS for the models fitted to the entire data set ranged from xxx to xxx at 5-years, and xxx to xxx at 10-years, with an associated median OS ranging from xxxxx xx (xxxxx xx and xxxxx xx) to xxxxx xx (xxxxx xx and xxxxx xx).

The survival estimates for the routine surveillance arm were compared to historical literature sources, and to advice received from two UK clinical experts. Relevant literature estimates are shown in Table 26, and were selected based on comparability to the SOLO1 population. UK clinical experts gave estimates of median OS for routine surveillance of between xxxxxxxx. When compared to these estimates, both methods (“entire data” and “piecewise”) significantly overpredict median OS in the routine surveillance population.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 26 Literature estimates of OS following first-line platinum chemotherapy

Note: these literature estimates correspond to survival from the date of starting platinum chemotherapy

In general, due to the pattern of OS for routine surveillance in the post-24-month period of SOLO1, the “entire data set” and “piecewise” models yielded implausible estimates of long-term OS for the routine surveillance arm, despite accurately predicting the OS curve within the study.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 27 Prediction of Kaplan-Meier data and long-term extrapolation of OS with olaparib using the Kaplan-Meier and parametric model (“piecewise”), and fully parametric model methods (“entire data”)

Green cells correspond to prediction of within 1.0% of Kaplan-Meier estimate, amber cells are prediction of within 1.0–3.0% of Kaplan-Meier estimate and red is greater than 3.0% difference to Kaplan-Meier estimate.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 92 of 144

Table 28 Prediction of Kaplan-Meier data and long-term extrapolation of OS with placebo using the Kaplan-Meier and parametric model (“piecewise”), and fully parametric model methods (“entire data”)

Green cells correspond to prediction of within 1.0% of Kaplan-Meier estimate, amber cells are prediction of within 1.0–3.0% of Kaplan-Meier estimate and red is greater than 3.0% difference to Kaplan-Meier estimate.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 93 of 144

Given that the olaparib models generated robust fits to the observed OS in SOLO1 and plausible long-term estimates of OS, an alternative method was sought to predict the OS for routine surveillance based on the estimates for olaparib. The DSU outlines that an alternative option to independently fitting models to each arm of a study is to use a single model fitted to both arms with a covariate for treatment effect. This approach assumes that difference in survival is driven by differences in the scale or rate parameters, and depending on choice of model, can yield proportional hazards, accelerated failure time or proportional odds model. These models assume a constant treatment effect on their respective scales, e.g. on the hazards, failure time or odds of survival.

The survival models fitted to the olaparib arm were used to predict OS for the routine surveillance arm with the use of a constant treatment effect to account for the expected longer-term difference in OS between placebo and olaparib. As OS data are immature in SOLO1, the estimates of treatment effect for olaparib versus placebo (and conversely, placebo versus olaparib) have not yet matured sufficiently to reliably inform the longer-term modelling of OS. Therefore, the incremental difference in survival between the two arms of SOLO1 was estimated from a surrogate endpoint.

The use of a surrogate endpoint to predict immature OS was recently accepted by the NICE appraisal committee for niraparib maintenance in platinum sensitive recurrent ovarian cancer (TA528), with gain in OS being estimated from an assumed ratio of gain in PFS ranging between 1:1 to 2:1. For SOLO1, it is acknowledged that the high rate of subsequent PARP use after progression (xxxx) on routine surveillance is likely to confound the post-progression survival period of the study. This is expected to weaken the relationship between PFS and OS when compared to the recurrent platinum sensitive setting. Therefore, to account for the effect of subsequent PARP-use on OS, we conservatively assume that the effect of treatment on is proportional to the effect observed on PFS2, which covers the period from randomisation to second progression or death. Unlike PFS, PFS2 captures the effects of subsequent PARP inhibitors given after first progression and is therefore considered a more appropriate surrogate of OS than PFS.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Surrogacy between PFS2 and OS is supported by data from S19, the most mature data set for maintenance PARP in ovarian cancer. While PFS2 was not measured in S19, long-term data on the TSST was collected as an EMA recommended surrogate of PFS2. In SOLO1, median PFS2 and median TSST were sufficiently similar to use surrogacy between TSST and OS in S19 as surrogacy for PFS2 to OS in SOLO1.

As in SOLO 1, subsequent PARP use confounded the effectiveness of olaparib in Study 19. Despite this, in Study 19 the median difference in TSST (xxxxxxxxxx), which accounted for subsequent PARP use, was subsequently transferred to the median difference in OS (xxxxxxxxxx), Table 29. This suggests that the effect of maintenance treatment with a PARP inhibitor observed on the endpoint of TSST or PFS2 is predictive of the longer-term effect of treatment on OS.

Table 29 Relationship between median TFST, TSST, PFS2 and OS in the gBRCA cohort of S19

Abbreviations: NA, not available; OS, overall survival; PFS2, time to second disease progression or death; TFST, time to first subsequent therapy; TSST, time to second subsequent therapy

Given that SOLO1 follows a similar pattern to Study 19 regarding effect and subsequent PARP use, it is therefore reasonable to assume that the median difference in TSST or PFS2 expected in SOLO1 may predict the median gain in OS. Median estimates of PFS2 for olaparib were not available, however, predictive modelling estimates it at xxxxxxxxx (Table 30, best fitting function of log-logistic).

With median estimates of PFS2 of xxxxxxxxxxx for routine surveillance, the predicted gain in median PFS2 for olaparib is approximately xxxxxxxx. It is therefore predicted that olaparib at first line will result in a 24-month gain in median OS.

Table 30 Predicted relationship between median TFST, TSST, PFS2 and OS in SOLO1

Abbreviations: NR, not reported; OS, overall survival; PFS2, time to second disease progression or death; TFST, time to first subsequent therapy; TSST, time to second subsequent therapy

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

The assumed xxxxxxxxx between PFS2 and OS was captured in the model by applying the relative effect of placebo versus olaparib from the PFS2 endpoint to the survival models used for OS (olaparib). The estimated treatment effect for PFS2 was derived from fitting a series of standard parametric models to the entire PFS2 data set in SOLO1, with treatment group as a covariate. The estimated covariate derived from each PFS2 model (e.g. Weibull) was then applied to the matching OS model (e.g. Weibull), to provide alternative estimates of OS for routine surveillance.

The resulting survival predictions for routine surveillance ranged from xxx to xxx at 10 years, with median OS ranging from xxv xxx xxx x (xx years and x months) to xxv xxx xxx x (xx years and x months). These predictions correspond with historical estimates (median OS of xv xxxxx and with clinical advice outlined previously. As shown in Table 31 and Figure 27, this method further accurately predicted landmark OS at month 30 in SOLO-1.

As per DSU guidance, the same model was applied to both arms of the analysis, e.g. Kaplan-Meier up to month 24 and log-logistic for extrapolation. For routine surveillance, Kaplan-Meier data was used up to month 24 and the treatment effect reported in Table 31 was applied to the post-24 month model.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 31 was applied to post-24-month log-logistic model for olaparib. In the base case, the predicted median gain in OS for olaparib versus routine surveillance is +24 months, which is consistent with the predicted gain in PFS2. Alternative methods including the “piecewise” method with log-normal and “piecewise” method with Weibull were considered in sensitivity analysis.

Figure 26 Best fitting function for PFS2

==> picture [342 x 321] intentionally omitted <==

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 31 Prediction of Kaplan-Meier data and long-term extrapolation of OS with placebo using the Kaplan-Meier and parametric model (“piecewise”) with application of the treatment effect from PFS2

Table note: Exponential and Gompertz are proportional hazards models with treatment effect expressed as the hazard ratio (HR). A HR>1 implies that placebo is less effective than olaparib, as expected. All other models are accelerated failure time methods (including Weibull), and treatment effect is expressed as an acceleration (AF) factor for placebo versus olaparib. An AF of less than 1.0 implies that the expected time to event for placebo is shorter than the corresponding time to death for olaparib. The predicted AFs of 0.71–0.72 imply that the expected time to death for placebo is 71–72% of the expected time to death for olaparib.

Green cells correspond to prediction of within 1.0% of Kaplan-Meier estimate, amber cells are prediction of within 1.0–3.0% of Kaplan-Meier estimate and red is greater than 3.0% difference to Kaplan-Meier estimate.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 98 of 144

Figure 27 Illustration of model approach combining PFS and OS with modelling of long-term survival status and gains in median PFS2 to OS (overlaid with Kaplan-Meier data)

==> picture [558 x 280] intentionally omitted <==

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 99 of 144

Adverse events

A detailed discussion on the AEs experienced by patients in the SOLO1 study is presented in Section B.2.10. The economic analysis only included AEs that were ≥ grade 3. The reason for this is these are the AEs that are likely to have an impact on the decision-making process as they may be associated with significant costs, and/or an impact on the HRQoL of patients. The AEs taken into consideration are presented in Table 32 below.

Table 32 Summary of AEs included in the economic model

Source: Data on file: D0818C00001 Clinical Study Report. Table 48; Abbreviation: AE, adverse event.

B.3.4 Measurement and valuation of health effects

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

EQ-5D-5L collected in SOLO1

In SOLO1, EQ-5D-5L assessments were planned at:

• Baseline (prior to randomization)

• Day 29

• Every 12 weeks (+/- 7 days) for 24 months or data cut-off for the primary

analysis

For patients who discontinued their allocated therapy, EQ-5D-5L assessments were

planned for the discontinuation visit and 30 days post last dose. For patients with

documented progression, EQ-5D-5L assessments were planned for every 12 weeks as part of scheduled follow up.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Mapping (EQ-5D-5L to EQ-5D-3L)

The SOLO1 trial collected health status data using EQ-5D-5L. The 3-level version (EQ-5D-3L) and the UK time trade-off value set are the reference case for HTA submissions, as defined by NICE. If EQ-5D-5L is collected, NICE recommend applying the mapping function developed by van Hout et al. to convert it to the EQ5D-3L for the reference-case analysis.[88,96] All completed EQ-5D-5L questionnaires that contained responses to all five health domains were mapped to EQ-5D-3L utilities using the crosswalk method by van Hout et al.[88]

Health-related quality-of-life studies

Published HSUVs were identified through a systematic literature review of studies reporting the HSU of patients with BRCA-mutated newly-diagnosed advanced ovarian cancer following response to platinum-based chemotherapy (see Appendix H). No studies were identified that reported HRQoL in the first-line maintenance therapy setting. Two published studies reported HSUV associated with maintenance therapy in the second-line setting in a population with ovarian cancer and a BRCA mutation.

Supplementary searches of relevant NICE HTAs (described previously in B.3.1): TA381,[60] TA528[59] and the ongoing appraisal of olaparib maintenance therapy in the second line setting ID1296,[97] identified additional EQ-5D data; however, no HSUVs were identified for patients with BRCA-mutated newly-diagnosed advanced ovarian cancer following response to platinum-based chemotherapy. A summary of the EQ5D-based HSUVs reported by these sources is provided in Table 33 .

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

Table 33 Utility values associated with specific disease stages/states

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 102 of 144

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 103 of 144

Abbreviations: HSUV, health-state utility value; OLS, ordinary least squares.

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124] © AstraZeneca (2018). All rights reserved Page 104 of 144

Adverse reactions

A one-off QALY adjustment for an AE is modelled based on its disutility (loss of utility) multiplied by its assumed duration. A summary of the AEs’ disutilities, durations and sources are presented in Table 34 .

Table 34 Disutility values associated with AEs, and assumed duration of events

Abbreviation: SE, standard error.

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

The base case analysis used EQ-5D-3L utility values derived from the SOLO1 study. This was considered the most robust and applicable source of utility data for this population, as it was directly collected in patients with BRCA-mutated newlydiagnosed advanced ovarian cancer following response to platinum-based chemotherapy, and no alternative values were identified in the systematic literature review.

There was no evidence of a meaningful difference in mean HSUV across treatment groups or by study visit; therefore, HSUV data were pooled across treatment groups to increase sample size in the analysis. The utility values used in the base case analysis are presented in Table 33 .

Table 35 Summary of utility values for cost-effectiveness analysis

The mapped EQ-5D-3L HSUVs from the SOLO1 study are consistent with the general population norms from Kind et al matched on age and gender (0.85 for a female aged 45-54 years; the mean age in SOLO1 is 53.5 years), and reflective of

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

the fact that at baseline, patients in SOLO-1 have no evidence of disease, and a good performance status.

HSUVs were adjusted over the lifetime time horizon by age-related decrements to reflect the aging of the cohort.

Mean HSUVs for the UK general population were estimated using the ordinary least squares (OLS) regression model published in Ara 2010[80] ( Equation 1 ) and used to apply age-related HSUV in previous advanced cancer appraisals (TA528, TA519).[59,60] This study explored the relationship between HSUVs, age, sex and history of CVD in Health Survey for England (HSE) data. In the 2003 and 2006 HSE surveys, a random sample of participants (individuals aged 16–98 years living in private households in England) completed the EQ-5D questionnaire (N=26,679) which were converted into preference-based HSUVs using time-trade off valuations from the UK general population (Dolan, 1996).[105]

Equation 1 OLS regression (Model 1) used to estimate the mean HSUVs for individuals in the general population

𝐸𝑄−5𝐷= 0.9508566 + 0.0212126 ∗𝑚𝑎𝑙𝑒−0.0002587 ∗𝑎𝑔𝑒−0.0000332 ∗𝑎𝑔𝑒[2]

In sensitivity analysis, decrements based on the weighted health state EQ-5D-3L index by age and sex for each 10-year age band presented in Kind 1999[81] were tested. For each age band, a monthly decrement was calculated and applied additively, per cycle (monthly [30.44 days]), in the economic model.

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

A systematic literature review was conducted to identify published resource use and cost data associated with the treatment and management of patients with newly diagnosed, advanced BRCA1/2-mutated high grade epithelial ovarian, fallopian tube or primary peritoneal cancer that has responded (completely or partially) to first-line platinum-based chemotherapy. See Appendix I for full details of how cost and resource use data were identified.

The costs in the economic model consisted of:

Company evidence submission for olaparib in patients with newly diagnosed BRCA-mutated advanced ovarian, fallopian tube or peritoneal cancer [ID1124]

© AstraZeneca (2018). All rights reserved

• Treatment related costs

  • Drug acquisition costs (including subsequent therapies)

  • Drug administration costs (including subsequent therapies)

• Disease monitoring and patient observation costs

• AEs costs

• End-of-life care costs

• BRCA testing costs (explored in a scenario analysis)