CENTER FOR DRUG EVALUATION AND RESEARCH
`
`APPROVAL PACKAGE FOR:
`
`APPLICATION NUMBER
`
`21-462
`
`Statistical Review(s)
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1120-0001
`
`

`
`STATISTICAL REVIEW AND EVALUATION - ~~
`
`DEPARTMENT OF HEALTH AND HUMAN SERVICES
`PUBLIC HEALTH SERVICE
`FOOD AND DRUG ADMINISTRATION
`CENTER FOR DRUG EVALUATION AND RESEARCH
`
`MEDICAL DIVISION:
`BIOMETRICS DIVISION:
`
`Oncology Drug Products (HFD-150)
`Division of Biometrics I (HFD-710)
`
`NDA NUMBER:
`
`NDA 21-462
`
`DRUG NAME:
`
`ALIMTA® (pemetrexed, 1~Y231514) 500 mg Vials
`
`INDICATION:
`
`Treatment of Malignant Pleural Mesothelioma
`
`SPONSOR:
`
`Eli Lilly and Company
`
`DOCUMENTS REVIEWED:
`1. Cover letter and documents (CDER REC’D Dates: 24-OCT-2002, 22-
`NOV-2002 and 26-NOV-2002) including SAS data base
`2. Cover letter (CDER REC’D Dates: 6-DEC-2002) including the pdf file for
`review’s aids, SAS data sets, and SAS programs for the efficacy analyses
`STATISTICAL KEY WORDS: Log-rank test, proportional hazard model,
`Kaplan-Meier estimate, hazard ratio, multiple comparison, Bonferroni adjustment
`
`STATISTICAL REVIEWER:
`
`Yong-Cheng Wang, Ph.D. (HFD-710)
`
`ACTING STATISTICAL TEAM LEADER: Ning Li, Ph.D. (HFD-710)
`
`DBI DEPUTY DIRECTOR:
`
`Kooros Mahjoob, Ph.D. (HFD-710)
`
`CLINICAL REVIEWER:
`
`Robert M. White Jr., M.D. (HFD-150)
`
`CLINICAL TEAM LEADER:
`
`John Johnson, M.D. (HFD- 150)
`
`PROJECT MANAGER:
`
`Patricia Garvey (HFD- 150)
`
`Distribution: NDA 21-462
`HFD- 150/Garvey
`HFD- 150/White
`HFD- 150/Johnson
`HFD-710/Wang
`HFD-710/Li
`HFD-710/Mahjoob
`HFD-710/Chi
`HFD-700/Anello
`
`File and Date: C:kAAAkNDALMimta (21-462)kREVIEWkAlimta.doc
`
`12-9-2003
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1120-0002
`
`

`
`STATISTICAL REVIEW AND EVALUATION
`
`Table of Contents
`
`EXECUTIVE SUMMARY OF STATISTICAL FINDINGS ................... 1
`
`I.I RECOMMENDATIONS AND CONCLUSIONS .....................................................
`1.2 BRIEF OVERVIEW OF CLINICAL STUDIES ..................................................... 1
`1.3 STATISTICAL ISSUES AND FINDINGS ............................................................. 2
`
`2
`
`INTRODUCTION ......................................................................................... 4
`
`2.1 OVERVIEW ................................................................................................... 4
`2.1.1 Background ......................................................................................... 4
`2.1.2 Major Statistical Issues ....................................................................... 5
`2.2 DATA SOURCES ............................................................................................ 5
`
`STATISTICAL EVALUATION ..................................................................
`
`3.1
`EVALUATION OF EFFICACY ................ " .......................................................... 6
`1.1 Study JMCH ........................................................................................ 6
`3.
`3.1. I. 1 Introduction ............................ . ........................................................ 6
`3.1.1.2 Statistical Issues .............................................................................. 6
`3.1.1.3 Study Objectives ............................................................................. 7
`3.1.1.4 Efficacy Endpoints .......................................................................... 7
`3.1.1.5 Sample Size Considerations ............................................................ 8
`3.1.1.6 Stratification .................................................................................... 9
`3.1.1.7 Interim Analysis .......................................................... : ................... 9
`3.1.1.8 Efficacy Analysis Methods ............................................................. 9
`3.1.1.9 Sponsor’s Results and Reviewer’s Findings/Comments .............. 10
`3.1.1.9.1 Baseline Characteristics ........................................................ 10
`3.1.1.9.2 Primary Efficacy Analyses ................................................... 12
`3.1.1.9.3 Secondary Efficacy Analyses ............................................... 14
`3.1.1.10 Sponsor’s Conclusions and Reviewer’s Conclusions/Comments ..... 17
`EVALUATION OF SAFETY ........................................................................... 18
`
`3.2
`
`FINDINGS IN SPECIAL/SUBGROUP POPULATIONS ...................... 18
`
`4.1 GENDER ..................................................................................................... 19
`4.2 R~CE .......................................................................................................... 20
`4.3 AGE .......................................................................... . ................................ 22
`4.4 OTHER SPECIAL]SUBGROUP POPULATIONS ................................................ 23
`
`SUMMARY AND CONCLUSIONS ......................... ~ ............................... 23
`5.] STATISTICAL ISSUES AND COLLECTIVE EVIDENCE ..................................... 23
`5.2 CONCLUSIONS AND RECOMMENDATIONS ................................................... 24
`
`6 APPENDICES ............................................................................................. 25
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1120-0003
`
`

`
`1 Executive Summary of Statistical Findings
`
`1.1 Recommendations and Conclusions
`
`Based on the collective evidences and findings, in this statistical reviewer’s
`opinion the data and results of the Phase III Study H3E-MC-JMCH support the
`sponsor’s efficacy claim of ALIMTA® (pemetrexed, LY231514) 500 mg Vials
`with respect to the survival endpoint for the patients with Malignant Pleural
`Mesothelioma. The data and results of the study show that the primary endpoint,
`survival, is statistically significantly improved in new treatment arm as compared
`to control arm for the randomized and treated (RT) population (p-value=0.021).
`The secondary endpoints, time to progressive disease, time to treatment failure,
`and response rate, are also demonstrated statistically significant improvement in
`new treatment group compared to the control group. In the fully supplemented
`(FS) subgroup, efficacy results are similar to those findings in the RT population.
`The hazard ratios for both RT and FS populations showed the consistency of the
`magnitude of survival benefit.
`
`1.2. Brief Overview of Clinical Studies
`
`This application consists of report of results from the Study H3E-MC-JMCH
`(referred as Study JMCH here and after) in the patients with Malignant Pleural
`Mesothelioma (MPM).
`
`The registration Study JMCH was a multi-national, multi-center, single-blind, and
`parallel-arm Phase III trial with MPM patients randomized to LY231514 plus
`Cisplatin (LY/cis) and Cisplatin alone treatment arms. A total of 574 patients
`were entered into the study (that is, signed the Informed Consent Document); 456
`of these patients were randomized to a treatment arm; 448 of these patients were
`treated and constitute the randomized and treated (RT) population.
`
`LY/cis: Total: 226, Male: 184, Female: 42. Fully Supplemented (FS): 168,
`Partially Supplemented (PS) or Never Supplemented (NS): 58.
`
`Cisplatin alone: Total 222, Male: 181, Female: 41. Fully Supplemented: 163,
`Partially Supplemented or Never Supplemented: 59.
`
`LY/cis treatment: LY231514 was administrated at the dose of 500 mg/m2 as a 10-
`minute intravenous infusion, diluted in approximately 100 mL normal saline.
`Approximately 30 minutes after the administration of LY231514, Cisplatin was
`administered at the dose of 75 mg/m2 over 2 hours. Both drugs were administered
`on Day 1 of a 21-day period. This 21-day period defined one cycle of therapy.
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1120-0004
`
`

`
`Cisplatin alone treatment: approximately 100 mL normal saline was given as an
`intravenous infusion over approximately 10 minutes. Approximately 30 minutes
`after the administration of normal saline, Cisplatin was administered at 75 mg/m2
`over 2 hours of Day 1 of a 21-day period. This 21-day period defined one cycle
`of therapy.
`
`Both treatment arms: (1) Dexamethasone, 4mg (or an equivalent conicosteriod),
`was to be taken by all enrolled patients orally twice a day (BID) 1 day before, on
`the day of, ~d 1 day a~er each dose of LY231514, for prh-nary prophylaxis
`against rash. (2) Folic acid and vitamin B~2 for supplementation were a standard
`component of therapy for all patients participating in the study. Folic acid, 350
`I.tg to 1000 I-tg, was to be taken orally daily, beginning approximately 1 to 3
`weeks before the fzrst dose of therapy and continued daily for 1 to 3 weeks after
`the patient discontinued treatment. A vitamin B~2 injection, 1000 I.tg, was to be
`administered intramuscularly approximately 1 to 3 weeks before the first dose of
`therapy and should have been the fast dose of therapy and should have been
`repeated approximately every 9 weeks until the patient discontinued, study
`therapy. (3) Pre- and post-hydration for Cisplatin was administered according to
`institutional guidelines:
`
`The primary objective of Study JMCH was to compare survival in chemonaive
`patients with MPM when treated with LY231514 plus Cisplatin combination
`therapy to survival in the same patient population when treated with Cisplatin
`alone. The primary efficacy endpoint was the overall survival time.
`
`1.3 Statistical Issues and Findings
`
`Statistical Issues:
`
`* 456 patients were randomized to treatment arms out of which 8 of these
`patients were died from study disease before any dosing. The sponsor did not
`follow the statistical reviewer’s comments of IND 40061/SN298 that the
`primary survival analysis should be based on all patients as randomized. The
`sponsor did primary efficacy analysis based on the randomized and treated
`population which did not include those 8 patients.
`¯ The sponsor’ efficacy claim was based on the RT population and stated that in
`clinically, folic acid and vitamin B~2 would improve the clinical outcome
`regardless of the treatment arm. The results of the FS subgroup also support
`the efficacy claim.
`¯ There was a heterogeneous distribution for gender in the two treatment arms
`(male and female with 81.4% vs. 18.6% and 81.5% vs. 18.5% in LY/cis and
`Cisplatin groups, respectively). The multivariate analysis for the treatment
`and gender showed that the interaction between treatment and gender had a
`small p-value (p-value=0.072) for the RT population and was statistically
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1120-0005
`
`

`
`STATISTICAL REVIEW AND EVALUATION
`
`significant for the FS population (p-value=0.035). Therefore, the influences
`of treatment were depended on the subgroups of gender.
`¯ The subgroup analysis of gender showed that the new treatment was
`significant for the RT population and FS population in female patients (p-
`value=0.012 and 0.010, respectively) and was not significant for PS+NS
`population (p-value=0.878). The analyses within the subgroup of male
`showed that the new treatment group was not statistically significant for the
`RT, FS, and PS+NS populations (p-value=0.176, 0.388, 0.219, respectively).
`¯ The hazard ratios showed the consistency of the magnitude of survival benefit
`in both the RT population and subgroup alike. The efficacy analyses of
`secondary endpoints, q"I’PD, TTTF and response rate, showed the consistency
`to primary endpoint.
`
`Findings:
`
`Table 1 gives the summary of efficacy results of primary endpoint, survival time
`(months), for the RT population. A total of 226 patients on the LY/cis arm and
`222 patients on the Cisplatin alone arm were included in the survival analysis.
`The median survival time for patients treated with LY/cis was longer than for
`patients treated with Cisplatin alone: 12.1 versus 9.3 months. There was a
`statistically significant difference (p=0.021) between the two treatment groups.
`
`Table 1.
`
`Patien~ dead"
`
`Primary Endpoint: Survival for RT Population (FDA Analysi.s)
`PS+NS Population
`RT Population
`FS Population
`(N=448)
`(1’4=117)
`(N---.331)
`Cisplatin
`Cisplatin
`LYicis
`Cispiatin
`LY/cis
`LY/cis
`(N=168)
`(N=!63)
`(N=58)
`(N=59)
`(N=226)
`(N=222)
`n (%)
`n (%)
`n (~)
`n (~)
`n (%)
`n (%)
`159 (72)
`103 (63)
`50 (86)
`145 (64)
`95 (57)
`56 (95)
`
`Sun, ival ti~ne (months)
`Median
`(95% Cl)
`p-value~
`Long-rank
`Wilcoxon
`
`12.1
`(10.0,14.4)
`
`9.3
`(7.8, 10.7)
`
`13.3 10.0
`(11.4,14.9) (8.4, 11.9)
`
`9.5
`(8.1, 10.8)
`
`7.2
`(6.5, 9.9)
`
`0.021
`0.028
`
`0.051
`0.039
`
`0.253
`0.440
`
`Hazard Ratio’
`0.758
`0.766
`95% CI for Hazard Ratioc
`(0.57, 1.0)
`(0.61, 0.96)
`Statistical reviewer’s results based on the analysis data sets provided by the sponsor.
`~ Patients were died for different reasons: study disease related, study toxicity, and other causes.
`b P-value is based on the test results for the two treatment groups.
`c Hazard Ratio is based on the proportional-hazards model with the treatment as single independent variable.
`
`0.798
`(0.54, 1.177
`
`Among the 448 patients of RT population, 331 were FS and 117 were PS+NS.
`The analysis of the FS subgroup indicated that the median survival for patients
`treated with LY/cis was 13.3 months vs. 10.0 months, a difference with small p-
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1120-0006
`
`

`
`STATISTICAL REVIEW AND EVALUATION i~ ~!~:.i~ i~. :.]
`
`value (0.051). The analysis of the PS+NS subgroup indicated that the median
`sur¢ival time for patients treated with LY/cis was 9.5 months vs. 7.2 months, but
`this difference did not reach the overall significance level (p-value=0.253). The
`hazard ratio for the RT population and for the FS and PS+NS subgroups were
`0.766, 0.758, and 0.798, respectively, indicating the consistency of the magnitude
`of the survival benefit in both the RT population and subgroup alike.
`
`Reviewer’s Comments:
`
`1) The sponsor’ efficacy claim was based on RT population which was by not
`including 8 patients randomized but died before any dosing. The sponsor also
`used the efficacy results of the FS subgroup to support to their efficacy claim.
`2) The median survival times for the RT population and for the FS and PS+NS
`subgroups show the consistency of the pattern of the survival difference in
`both the RT population and subgroups alike.
`3) The hazard ratios for the RT population and for the FS and PS+NS subgroups
`show the consistency of the magnitude of the survival benefit in both the RT
`population and subgroups alike.
`
`2 Introduction
`
`2.1 Overview
`
`The beneficial effect of LY231514 is a novel anti folate that can inhibit multiple
`tumor targets involved in both purine and pyrimidine pathways of DNA synthesis.
`LY231514 exhibits, highly cytotoxic in vitro activity against the CCRF-CEM
`human leukemia cell line. LY231514 has also shown significant anti.an-nor
`activity against thymidine- and hypoxanthine-deficient murine tumor cell lines as
`well as two human colon xenografts resistant to methotrexate. Phase 1 studies
`were conducted exploring three treatment schedules: once daily times 5 every 3
`weeks (Study H3E-BP-001); once weekly times 4 every 6 weeks (Study H3E-
`MC-JMAB); and once every 3 weeks (Study H3E-MC-JMAA). In Study JMAA,
`LY231514 was achninistered to 37 patients as a 10-minute infusion once every 3
`weeks at doses ranging from 50 to 700 mg/m2 (Rinaldi et al. 1996). Based on this
`study, the recommended dose for Phase 2 studies was 600 mg/m2.
`
`2.1.1 Background
`
`Malignant mesothelioma is a rare, seldom curable, tumor of the pleura or the
`peritoneum whose origin has generally been linked to asbestos exposure. The
`most common sites of origin are the pleura, accounting for 80% of cases,
`followed by peritoneum, pericardium, and tunica vaginalis testes (Sterman et al.
`1999). Survival of untreated patients is poor, with a median survival of usually 6
`
`4
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1120-0007
`
`

`
`to 8 months, though the range may be much wider, depending on the
`characteristics and selection of the population ofmesothelioma patients (Antman
`et al. 1997). Most studies in published literature have involved MPM, including
`those characterizing the disease and its treatment. A number of factors including
`histologic subtype, performance status, disease extent at baseline, presence of
`chest pain, gender, and white blood cell count, among others, have been suggested
`as predictors of outcome, including survival (Curran et al. 1998; Herndon et al.
`1998).
`
`MPM is a difficult tumor to treat. In general, neither surgery nor radiotherapy
`results in increased survival (Antman et al. 1997). A wide variety of
`chemotherapeutic agents have shown modest activity in single-agent or
`combination Phase 2 trials, including gemcitabine (GEMZAR~), doxorubicin,
`cisplatin, ifosfamide, methotrexate, edatrexate, mitoxantrone, epirubicin,
`etoposide, and paclitaxel. Response rates have seldom exceeded 20% in single-
`agent Phase 2 trials (van Breukelen et al. 1991; Mattson et al. 1992; Solheim et al.
`1992; Belani et al. 1994; van Meerbeeck et al. 1996; Millard et al. 1997; Sahrnoud
`et al. 1997).
`
`The registration Study JMCH was a multi-nation, multi-center, single-blind, and
`parallel-arm Phase III trial with MPM patients randomized to LY231514 plus
`Cisplatin and Cisplatin alone treatment arms. A total of 574 patients were entered
`into the study; 456 of these patients were randomized to a treatment arm; 448 of
`these patients were treated and constitute the randomized and treated (RT)
`population. The study period was from April 1999 to February 2002.
`
`2.1.2 Major Statistical Issues
`
`The major statistical issues can be found in Section 1.3.
`
`2.2 Data Sources
`
`Data used for review is from the electronic submission received on October 24,
`2002. The efficacy analysis data were submitted by the sponsor on December 6,
`2002. All data sets analyzed are electronic documents and are located in the
`Electronic Document Room (EDR) of CDER of FDA under the Letter Date "24-
`OCT-2002" and "6-DEC-2002", respectively. The major data set for the efficacy
`analysis is "SURVLOCK" which defines the survival time and events.
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1120-0008
`
`

`
`3 Statistical Evaluation
`
`3.1 Evaluation of Efficacy
`
`The registration Study JMCH was used for efficacy evaluation. The primary
`analyses of the study were performed on an RT basis. The RT population was
`defined as all patients randomly assigned to a treatment arm, who received study
`drag (LY/cis or Cisplatin). Of the 456 patients randomly assigned to a treatment
`arm, 448 (98.2%) received LY/cis or Cisplatin monotherapy. These patients
`constituted the RT population for this study. Among the 448 patients in the RT
`population, sub-populations defined by supplementation status (FS, PS, and NS)
`were considered in key additional analyses and presented in this review.
`
`3.1.1 Study JMCH
`
`3.1.1.1 Introduction
`
`Study JMCH was a multi-nation, multi-center, single-blind, and parallel-arm
`Phase Ill trial with MPM patients randomized to LY/cis and Cisplatin alone
`treatment arms. A total of 574 patients were entered into the study (that is, signed
`the Informed Consent Document); 456 of these patients were randomized to a
`treatment arm; 448 of these patients were treated and constituted the RT
`population.
`
`...
`
`3.1.1.2 StatisticalIssues
`
`The major statistical issues can be found in Section 1.3.
`
`As we stated in the first bullet of statistical issues in the section 113, there were
`456 patients randomized to treatment arms and 8 of these patients died from study
`disease before any dosing. Table 2 gives the detailed list for those died patients.
`The sponsor’s primary efficacy analysis was based on the RT population which
`did not include those 8 patients.
`
`Table 2. ,List of Randomized Patients before Dosing (FI)A Anal~,sis)
`Enroll Date End Date Treatment Primary Reason Discontinue
`Patient
`ID
`Protocol entry criteria not met
`1999-09-09 Cisplatin
`1999/09/06
`1342
`Personal conflict or other patient decision
`2000-11-02
`Cisplatin
`2000-11-01
`1472
`Personal conflict or other patient decision
`1634
`2001-03-27 Cisplatin
`2001-03-26
`Protocol entry criteria not met
`2000-05-25
`Cisplatin
`2000-05-09
`2133
`Personal conflict or other patient decision
`Cisplatin
`2000-09-12
`2200
`2000-09-13
`Adverse event
`2000-02-04
`LY/Cis
`2000-02-03
`3161
`Death from study disease
`2000-06-06 2000-06-12
`LY/cis
`5109
`2000-12-01
`2000-12-07
`6014
`Personal conflict or other patient decision
`Cislglatin
`Statistical reviewer’s results based on the analysis data sets provided by the sponsor.
`
`6
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1120-0009
`
`

`
`3.1.1.3 Study Objectives
`
`The primary objective of this study was to compare survival in patients with
`MPM when treated with LY231514 plus Cisplatin combination therapy to
`survival in the same patients population when treated with Cisplatin alone.
`
`*
`,
`
`The secondary objectives of this study were to compare the follows between the
`two treatment arms:
`¯
`time-to-event efficacy measures:
`duration of response for responding patients
`time to progressive disease
`time to treatment failure
`tumor response rate
`clinical benefit response rate (pain intensity, analgesic consumption,
`dyspnea, performance status)
`, Lung Cancer Symptom Scale (LCSS) patient and observer scores
`* pulmonary function test scores (forced vital capacity, vital capacity, forced
`expiratory volume)
`¯
`lung density determinations in approximately 170 patients
`¯
`relative toxicities.
`Additional secondary objectives of this study were:
`¯
`to assess toxicity experienced in cycles in which patients did receive folic
`acid aad vitamin B~2 supplementation and toxicity experienced in cycles in
`which patients did not receive folic acid and vitamin Bl2 supplementation
`to assess PK effects
`to collect information regarding vitamin deficiency markers status in this
`patient population.
`
`¯
`¯
`
`3.1.1.4 Efficacy Endpoints
`
`The primary efficacy endpoint was survival. Survival was defined as the time
`from study enrollment (randomization) to time of death from any cause.
`
`The key secondary efficacy endpoints were time to progressive disease, time to
`treatment failure, tumor response rate, and duration of tumor response.
`
`Time to progressive disease (’I’I’PD) was defined as the time from randomization
`to the first observation of disease progression or death because of any cause.
`
`Tumor response rate was defined as the ratio ofresponders over the total number
`of patients qualified for rumor response assessment times 100 to quote the rate as
`a percentage. A responder was defined as any patient who had a complete
`response (CR) or a partial response (PR). All responses were documented by
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1120-0010
`
`

`
`using appropriate diagnostic tests that were repeated approximately every 6 weeks
`to continue evaluation.
`
`The duration ofa CR or PR was defined as the time from first objective status
`assessment of CR or PR to the first time of disease progression or death because
`of any cause.
`
`Time to treatment failure (TTTF) was def’med as the time from study enrollment
`(randomization) to the first observation of disease progression, death because of
`any cause, or discontinuation because of any other reason.
`
`3.1.1.5 Sample Size Considerations
`
`¯
`
`The primary objective of this study was to compare survival of patients with
`MPM receiving LWcis combination therapy versus those receiving Cisplatin
`monotherapy. The sponsor described the sample size considerations as follows.
`
`During the conduct of this study, a programmatic change was made by the
`sponsor in the clinical development of L¥231514 whereby every patient treated
`with LY231514 must be supplemented with folic acid and vitamin B~2 to improve
`patient safety. Initiation of supplementation in this study was done in both
`treatment arms and at the same time point to preserve study blinding at the patient
`level. This programmatic change was implemented in this study beginning with
`Protocol Amendment (C). The decision to extend enrollment so that a planned
`280 FS patients would be randomized to this trial is documented in Protocol
`Amendment (E). The following describes the statistical properties associated
`with this sample size.
`
`A planned 280 qualified patients receiving vitamin supplementation during every
`cycle of their study therapy were to be randomized to this trial. A treatment was
`judged superior if it is associated with a 33% reduction in the hazard ratio of the
`two treatments by median survival time. Assuming an exponential survival, 15-
`month patient accrual, and an additional minimum 9-month follow-up for all
`patients and a censoring rate of 30% or less after the 24 month accrual and
`follow-up period, the procedure described above gives at least an 81% chance
`.(Power) to detect a 33% shift in hazard ratio as reflected by a 63% survival
`probability on the best treatment arm by the time only 50% of patients are still
`alive (median time) on the least efficacious treatment arm. In terms of event rate,
`a total of 197 deaths in the FS sub-population would yield approximately 80%
`power to detect a hazard ratio of 67%. These calculations use a two-sided log
`rank test with a 0.05 chance of rejecting the null hypothesis H0 of no difference in
`survival between the two treatment arms when H0 is actually true.
`
`8
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1120-0011
`
`

`
`After the conduct of the planned interim analysis and before the final database
`lock, the sponsor and the agency confn-med that the primary survival analysis
`would be conducted on the entire patient population (RT population) as stated in
`Protocol Amendment (C).
`
`3.1.1.6 Stratification
`
`The study was stratified by some prognostic factors which the sponsor chose as
`potential confounders to survival and other study outcomes as suggested by the
`literature. The statistical reviewer’s comments about the stratification is in
`Section 3.1.1.9.1.
`
`3.1.1.7 Interim Analysis
`
`A planned interim analysis was Conducted and presented to the Data Safety
`Monitoring Board for resulting hi a decision to continue the trial to planned
`completion.
`
`3.1.1.8 Efficacy Analysis Methods
`
`The primary analysis was comparison of survival time between the two treatment
`arms in the RT population. Differences were assessed using a two-sided log-rank
`test. Because an interim analysis was conducted, the comparison of survival was
`tested at the cz=0.0476 level. Comparison of survival was also tested using the
`Wilcoxon test.
`
`Key secondary analyses were conducted to assess the impact of supplementation
`on survival in the LY/cis arm. The Kaplan-Meier subgroup analyses of survival
`were conducted on FS and on PS+NS patients. Also, survival time was analyzed
`with a Cox proportional hazards model including treatment arm, supplementation
`group, and the treatment-by-supplementation interaction. The interaction term
`was evaluated to assess the impact of supplementation on the survival benefit
`associated with LY/cis.
`
`Other time-to-event measures were analyzed by using the same method as
`described for survival time. Comparisons of the tumor response rates between the
`two treatment arms (in the RT, FS, and PS+NS populations) were made by using
`the Fisher’s Exact test with 95% CI calculated using the method of Leemis and
`Trivedi. Tumor response was also analyzed with a logistic regression model
`including treatment arm, supplementation group, and the treatment-by-
`supplementation interaction. The interaction term was evaluated to assess the
`impact of supplementation on the survival benefit associated with LY/cis. Time-
`to-event and tumor response measures were also analyzed to assess the effect of
`potential prognostic factors. Subgroup analyses were conducted on statistically
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1120-0012
`
`

`
`STATISTICAL REVIEW AND EVALUATION
`
`significant factors (p<0.05). Repeated measures analyses were conducted on
`LCSS patient scale and PFT parameters by using linear mixed models. Clinical
`benefit response was analyzed by using the Fisher’s Exact test. LCSS observer
`scale data were analyzed by the Mantel-Haenszel chi-square test and also assessed
`by using simple analysis of variance (ANOVA) techniques. Simple summary
`statistics by treatment arm and by cycle were calculated for lung density
`measurements. Analyses ofLCSS, PFTs, and CB data were conducted in the RT,
`FS, and PS+NS populations.
`
`3.1.1.9 Sponsor’s Results and Statistical Reviewer’s Findings/Comments
`
`This section will summarize the results of intent to treat analysis for Study JMCH.
`In this study a total of 456 patients were randomized to a treatment arm; 448 of
`these patients were treated and constitute the randomized and treated (RT)
`population, where 226 patients were enrolled into the LY/cis arm and 222 patients
`were enrolled into the Cisplatin arm.
`
`3.1.1.9.1 Baseline Characteristics
`
`Table 3 shows key baseline demographic characteristics for the RT population by
`treatment ann and further by supplementation status. All characteristics showed
`balance between the two treatment arms.
`
`Table 3.
`
`Age"
`< 65 years
`> 65 years
`
`Sex
`Male
`Female
`
`Baseline Characteristics of RT Population (FDA Anal~’sis)
`PS+NS Populatid~
`RT Population
`FS Population
`(N:117)
`(N=448)
`(N=331),
`Cisplatin
`Cisplatin
`Cisplatin
`LY/cis
`LY/cis
`LY/cis
`(N=59)
`(N=222)
`(N=168)
`(N=163)
`(N=58)
`(N=226)
`n (%)
`n (%)
`n (%)
`n (%)
`n (%)
`n (%)
`
`143 (63)
`83 (37)
`
`136 (61)
`86 (39)
`
`107 (64)
`61 (36)
`
`97 (60)
`66 (40)
`
`36 (62)
`22 (38)
`
`39 (66)
`20 (34)
`
`184 (81)
`42 (19)
`
`181 (82)
`41 (18)
`
`136 (81)
`32 (19)
`
`134 (82)
`29 (18)
`
`48 (83)
`"10 (17)
`
`47 (80)
`12 (20)
`
`Origin
`Caucasian
`Hispanic
`Asianb
`African
`¯ Statistical reviewer’s results.
`b Western and East/Southeast Asian have been combined.
`
`204 (90)
`11 (5)
`10 (4)
`1 (0.4)
`
`206 (93)
`12 (5)
`4 (2)
`0
`
`150 (89)
`10 (6)
`7 (4)
`1 (0.6)
`
`153 (94)
`7 (4)
`3 (2)
`0
`
`54 (93)
`1 (2)
`3 (5)
`0
`
`53 (90)
`5 (9)
`I (0.7)
`0
`
`Table 4 and 5 summarize stratification factors and baseline disease characteristics
`for the RT population, respectively.
`
`10
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1120-0013
`
`

`
`Table 4.
`
`Baseline Stratification Factors Used for Randomization of RT
`Population (Sponsor Anal,vsis)
`RT Population FS Population
` =448) =331)
`Cisplatin
`LY/cis
`Cisplatin
`LY/cis
`(N=168)
`(N=226)
`(N=222)
`(N=163)
`n (%)
`n (%)
`n (%)
`n (%)
`
`PS+NS Population
`(N=117)
`LY/cis
`Cisplatin
`(N=58)
`(N=59)
`n (%)
`n (%)
`
`KPS
`Low (_< 80)
`High (_> 90)
`
`Degree of Meaurability~
`Unidimensional
`Bidimensional
`
`Histologic Subtype
`EPithelial
`Mixed
`Sarcomatoid
`Other
`
`WBC
`Low (< 8.3 GUL)
`High (> 8.3 GUL)
`
`Pain Intensityb
`Low (< 20 nun)
`High (> 20 mm)
`
`Analgesic Consumption
`Low (< 60 mg morp
`eqiday)
`High (_> 60 mg morp
`eq/day)
`
`Dyspneab
`Low (< 20 ram)
`High(> 20 ram)
`
`Homocysteine
`Low (< 12 umoUL)
`High (>12 umol/L)
`
`109 (48) 97 (44) 83 (49) 69 (42) 26 (45) 28 (47)
`117 (52) 125 (56) 85 (51) 94 (58) 32 (55) 31 (53)
`
`73 (32) 73 (33) 61 (36) 62 (38) 12 (21) 11 (19)
`152 (68) 149 (67) 106 (64) 101 (62) 46 (79) 48 (81)
`
`154 (68)
`18(8)
`37 (16)
`17 (8)
`
`152 (69)
`25(11)
`36 (16)
`9 (4)
`
`117 (70)
`25(15)
`14 (8)
`12 (7)
`
`113 (69)
`25(15)
`17 (10)
`8 (5)
`
`37 (64)
`12(21)
`4 (7)
`5 (9)
`
`39 (66)
`11 (19)
`8 (14)
`1 (2)
`
`97 (43) 91 (41) 72 (43) 68 (42) 25 (43) 23 (39)
`129 (57) 131 (59) 96 (57) 95 (58) 33 (57) 36 (61)
`
`112 (50) 113 (51) 82 (49) 80 (49) 30 (52) 33 (56)
`112 (50) 109 (49) 84 (51 ) 83 (51 ) 28 (48) 26 (44)
`
`173 (77)
`
`170 (77)
`
`129 (77)
`
`124 (76)
`
`44 (76)
`
`46 (78)
`
`53 (23)
`
`52 (23)
`
`39 (23)
`
`39 (24)
`
`14 (24)
`
`13 (22)
`
`91 (41)
`133 (59)
`
`92 (41)
`130 (59)
`
`66 (40)
`100 (60)
`
`68 (42)
`95 (58)
`
`25 (43)
`33 (57)
`
`24 (41)
`35 (59)
`
`155 (69) 156 (70) 119 (71) 118 (72) 36 (62) 38 (64)
`71 (31) 66 (30) 49 (29) 45 (28) 22 (38) ¯
`21 (36)
`
`Male
`Female
`
`134 (82)
`29 (l 8)
`
`48 (83)
`lO (17)
`
`136 (81)
`181 (82)
`184 (81)
`41 (18)
`42 (19)
`32 (19)
`Sponsor’s results confirmed by this statistical reviewer.
`¯ A single patient was missing their evaluable disease measurement at baseline.
`~’ Patients 302-3025 and 720-7209 completed the patient LCSS at baseline, but outside of the
`protocol defined window; those da