OSI EX. 2004 - 0001
`OSI EX. 2004 - 0001
`
`CENTER FOR DRUG EVALUATION AND RESEARCH
`
`APPROVAL PACKAGE FOR:
`
`APPLICATION NUMBER
`
`2 1 -743
`
`Statistical Review(s)
`
`OSI EXHIBIT 2004
`OSI EXHIBIT 2004
`APOTEX V. OSI
`APOTEX V. OSI
`IPR2016-01284
`IPR2016-01284
`
`

`
`OSI EX. 2004 - 0002
`OSI EX. 2004 - 0002
`
`
`
`U5. Dcpartmt of Health and Human Services
`Food and Drug Administration
`Center for Drug Evaluation Research
`Office of Pllarmacocpidemlology and Statistical Science
`Office of Biostatistics
`
`STATISTICAL REVIEW AND EVALUATION
`
`CLINICAL STUDIES
`
`ADDENDUM 1
`
`NDA /Serial Number:
`
`21-743 /N000
`
`Drug Name:
`
`Applicant:
`
`Indication(s):
`
`Date(s):
`
`Tarcevam (erlotinib hydrochloride, OSI-774)
`
`OSI Pharmaceuticals
`
`Metastatic Non-small Cell Lung Cancer
`
`Submission Date: July 30, 2004
`
`PDUFA Date: January 30, 2005
`Review Date: November 1, 2004
`
`Review Priority:
`
`Priority
`
`Biometrics Division:
`
`Division of Biometrics I (HFD-710)
`
`Statistical Reviewers:
`
`Rajeshwari Sridhara, Ph.D.
`
`Concurring Reviewer:
`
`Yeh-Fong Chen, Ph.D.
`Kooros Mahjoob, Ph.D., Acting Director
`
`Medical Division:
`Clinical Team:
`
`Oncology Drug Products (HFD-150)
`
`Martin Cohen, MD. & John Johnson, M.D.
`
`Project Manager:
`
`Mr. Paul Zimmerman
`
`Keywords:
`
`Superiority, log-rank test, Cox regression, QoL
`
`1
`
`

`
`OSI EX. 2004 - 0003
`OSI EX. 2004 - 0003
`
`In this addendum additional exploratory analyses with respect to smoking history
`in the Study BR.2l are presented. These analyses do not change the conclusions
`and recommendations of the review.
`
`The following table shows the baseline characteristics in patients with smoking
`history and without smoking history.
`
`Table 1: Demographics and Baseline Characteristics of Patients by Smoking
`History
`
`
`
`
`
`
`=358
`
`=13
`
`== 78
`
`N = 49
`
`
`
`3 -
`
`EE{
`
`ositive
`
`neative
`
`.
`
`Unknown— 25 ‘W’
`
`IJ
`
`
`
`
`
`
`
`
`
`
`
`
`
`

`
`OSI EX. 2004 - 0004
`OSI EX. 2004 - 0004
`
`Reviewer ‘.9 Comments:
`
`1. The patients characteristics appear to be balanced between the treatment
`arms within the subgroup of patients with smoking history (except for age
`group), and within the subgroup of patients with no smoking history
`(except for EGFR positive status, performance status and response to prior
`therapy).
`,
`2. There is however a difference between the subgroups with respect to the
`distribution of gender, race and histology.
`3. Results from analyses adjusting for imbalances within each of the two
`subgroups were similar to the unadjusted analyses:
`Smoking Group: HR = 0.865, 95% CI: 0.713, 1.050 Unadjusted analysis;
`HR = 0.866, 95% CI: 0,713, 1.052 Adjusted for age group analysis.
`Non-smoking Group: HR = 0.422, 95% CI: 0.278, 0.640 Unadjusted
`analysis;
`HR = 0.422, 95% CI: 0.296, 0.645 Adjusted for performance status,
`response to prior therapy and EGFR status.
`
`Further analyses of difference between patients with smoking history versus no
`smoking history in each of the treatment arms are presented below.
`
`Table 2: Survival Analyses Results in Tarceva Treated Patients
`
`Known Po - ulation
`
`N=358
`
`N=l04
`
`10.6, 16.1
`4.7, 6.5
`months 95% CI
`
`Hazard Ratio = Smokers I Non—smokers; Unadjusted, log-rank test.
`
`12.3
`
`95% C
`1.860
`(1.413,2.441)
`
`<0.0001
`
`
`
`
`
`
`
`
`Med. Survival in
`
`
`
`Table 3: Survival Analyses Results in Placebo Treated Patients
`
`
`
`
`Known Po - ulation
`N=l87
`N-=42
`95% C
`-51 0.989
`
`Med. Survival in
`4.6
`(0.691, 1.417)
`
`
`3.9, 6.2
`months 95% CI
`3.5, 8.0
`1 Hazard Ratio = Smokers I Non-smokers;
`2 Unadjusted, log-rank test.
`
`
`
`0.9532
`
`Reviewer 's comment:
`
`The nonsmokers appear to benefit more from Tarceva compared to smokers.
`
`

`
`OSI EX. 2004 - 0005
`OSI EX. 2004 - 0005
`
`
`
`This is a representation of an electronic record that was signed electronically and
`this page is the manifestation of the electronic signature.
`-canuuuu-no-n---------—-----uu----u----oaunnn—-—--—-—-q-yu--u-------nun.¢-—--——--——.qo----------na—--
`
`Raj eshwari Sridhara
`11/1/04 10:43:09 AM
`BIOMETRICS
`
`

`
`OSI EX. 2004 - 0006
`OSI EX. 2004 - 0006
`
`
`
`U.S. Department of Health and Human Services
`Food and Drug Administration
`Center for Drug Evaluation Research
`Office of Pharlnacoepidemiology and Statistical Science
`Office of Biostatistics
`
`STATISTICAL REVIEW AND EVALUATION
`CLINICAL STUDIES
`
`NBA [Serial Number:
`
`21-743 {N000
`
`Drug Name:
`
`Applicant:
`
`Indication(s):
`
`Date(s):
`
`Tarcevam (erlotinib hydrochloride, OSI-774)
`
`OSI Pharmaceuticals
`
`Metastatic Non-small Cell Lung Cancer
`
`Submission Date: July 30, 2004
`
`PDUFA Date: January 30, 2005
`
`Review Completion Date: October 1, 2004
`
`Review Priority:
`
`Priority
`
`Biometrics Division:
`
`Division of Biometrics I (I-IFD-710)
`
`Statistical Reviewers:
`
`Rajeshwari Sxidhara, Ph.D.
`
`Concurring Reviewer:
`
`Kooros Mahjoob, Ph.D., Acting Director
`
`Yeh-Fong Chen, Ph.D.
`
`Medical Division:
`
`Clinical Team:
`
`Oncology Drug Products (HFD-150)
`
`Martin Cohen, M.D. & John Johnson, M.D.
`
`Project Manager:
`
`Mr. Paul Zimmerman
`
`Keywords:
`
`Superiority, log-rank test, Cox regression, QoL
`
`

`
`OSI EX. 2004 - 0007
`OSI EX. 2004 - 0007
`
`Table of Contents
`
`1 Executive Summary ...................................................................................... 3
`1.1
`Conclusions and Recommendations...................... 3
`1.2
`Brief Overview of Clinical Studies
`3
`
`2
`
`1.3
`
`2.1
`
`Statistical Issues and
`
`Overview
`
`3
`5
`
`5
`
`2.1.1
`2.1.2
`2.2
`
`Background ................................................................................ .. 5
`Statistical Issues ......................................................................... .. 5
`Data
`6
`
`3
`
`Statistical
`
`6
`
`3.1
`3.1.1
`3.1.1.1
`3.1.1.2
`3.1.1.3
`3.1.1.4
`3.1.1.5
`3.1.1.6
`3.1.1.7
`
`Evaluation of Efficacy............................................................... 6
`Study BR.2l ............................................................................... ..7 '
`Study
`7
`Study Objectives ........................................................................ .. 8
`Efficacy Endpoints ..................................................................... .. 8
`Sample Size Considerations ....................................................... .. 9
`Interim Analysis ......................................................................... .. 9
`Eflicacy Analysis Methods ........................................................ .. 9
`Sponsor’s Results and Statistical Rcviewer’s Findings!
`Comments ................................................................................ .. ll
`3.l.l.7.1 Baseline Characteristics ....................................................... .. ll
`
`3.l.1.7.2 Primary Efficacy Analyses .................................................. .. 12
`3.1.1.7.3 Exploratory Survival Analyses ............................................ .. 16
`3.1.1.7.4 Secondary Efficacy Analyses ................................................ 25
`3.1.1.7.-4.1
`Progression-free Survival .............................................. .. 25
`3.1.1.7.4.2 Objective Response Rate................................................. 26
`3.1.1.7.4.3 Quality of Life Endpoints.............................................. .. 28
`Study A248-1007 ....................................................................... 33
`Studies 0SI2298g and BO164l1 ............................................. .. 34
`Evaluationof 35
`Findings in Speciallsubgroup Populations............................ 35
`Gender, Race andAge 35
`Other Speciallsubgroup Populations
`36
`Summary and Conclusions...................................................... 38
`Statistical Issues and Collective Evidence............................. 38
`
`Conclusions and Recommendations ....................................... 39
`
`3.1.2
`3.1.3
`3.2
`
`4
`
`5
`
`4.1
`4.2
`
`5.1
`
`5.2
`
`APPENDICES
`
`..............................................................................
`
`............ 40
`
`Appendix 1: Measurement Schedule (Excerpt from Sponsor protocol)
`Appendix 2: Response Criteria (Excerpt from Sponsor’s Protocol)
`Appendix 3: QoL Variables (excerpt from sponsor’s Statistical Analysis
`.
`-oooanaocouo-----a-cu-cu
`42
`Plan).
`Appendix 4: QoL Questionnaire (Excerpt from spo§§3}7§';;3i3ZL1) ........ 44
`
`40
`41
`
`

`
`OSI EX. 2004 - 0008
`OSI EX. 2004 - 0008
`
`Appendix 5: Evaluation of Baseline Prognostic Factors — Univariate
`Analyses
`
`49
`
`

`
`OSI EX. 2004 - 0009
`OSI EX. 2004 - 0009
`
`1 Executive Summary
`
`1.1 Conclusions and Recommendations
`
`In this reviewer's opinion the study results from a single, randomized, multicenter,
`double-blinded, placebo-controlled phase III trial support the claim of efficacy
`based on overall survival of Tarceva'"" (erlotinib hydrochloride) for patients with
`locally advanced or metastatic non-small cell lung cancer after failure of at least
`one prior chemotherapy regimen.
`
`1.2 Brief Overview of Clinical Studies
`
`This submission consists of results of one phase III, randomized, placebo-
`controllcd, double-blinded clinical trial (registration trial BR.2l, referred as
`BR.2l here after) comparing OSI-774 (Tarcevam, referred as erlotinib here after)
`versus placebo in patients with incurable stage IIIB/IV non-small cell lung cancer
`(NSCLC) who have failed standard therapy for advanced or metastatic disease.
`The sponsor has also provided supportive efficacy data from a phase II, single
`arm study (A248-1007) of erlotinib following failure of platinum based
`combination chemotherapy in patients with advanced NSCLC. In addition the
`sponsor has also submitted results of two phase IH, randomized, double-blinded,
`multicenter trials (Study OSI2298g and Study B01641 1) of erlotinib plus
`chemotherapy (carboplatin + paclitaxel, and cisplatin + gemcitabine, respectively)
`vs. chemotherapy alone in patients with advanced (stage IIIB/IV) NSCLC who
`had not received prior chemotherapy. The addition of erlotinib to chemotherapy
`in both the studies did not demonstrate additional benefit with respect to overall
`survival compared to chemotherapy alone.
`
`Study BR.2I was a phase IH, comparative international study conducted in 731
`patients from 86 study centers in 17 countries. Patients 2 18 years old with
`histologically or cytologically confirmed diagnosis of incurable stage IIIB/IV
`NSCLC who have received at least one but no more than two prior regimens of
`which at least one had to be combination chemotherapy (if 2 70 years old), who
`had ECOG performance status of 0 to 3, had adequate renal and hepatic functions
`were randomized in 2:1 ratio to receive either erlotinib (150 mg tablets orally) or
`placebo. In this study, patients were stratified at randomization by center, number
`of prior regimens, prior platinum therapy, best response to prior therapy, and
`ECOG performance status.
`
`1.3
`
`Statistical Issues and Findings
`
`This NDA submission is to support administration of erlotinib in patients with
`advanced or metastatic NSCLC who have failed at least one prior chemotherapy.
`
`

`
`OSI EX. 2004 - 0010
`OSI EX. 2004 - 0010
`
`In this NDA submission, study BR.2l is the only randomized pivotal study
`conducted to establish efficacy. This study enrolled a total of 731 patients with
`488 patients who received erlotinib and 243 patients who received placebo. The
`primary efficacy endpoint of this study was survival. The applicanthas submitted
`this application claiming efficacy based on overall survival. There was a
`statistically significant difference between the two treatment arms with respect to
`overall survival in the ITT population (log-rank test, P-value = 0.002, stratified
`log-rank test, P-value= < 0.0001).
`
`-
`
`Statistical Issues:
`
`1. The primary analysis of the primary endpoint overall survival was based
`on stratified log-rank test including randomization stratification factors
`and EGFR status. In 67% of the patients EGFR status was not evaluated.
`An adjusted analysis including these 67% patients with missing data on
`EGFR status is questionable.
`2. The results of exploratory analyses in the subgroups suggest a significant
`survival benefit due to erlotinib in the EGFR positive patients and suggest
`no survival benefit in the EGFR negative population.
`
`Findings:
`
`The protocol specified primary analysis was stratified log-rank test in the intent-
`to-treat (ITT) population to compare overall survival between the two treatment
`arms. This study demonstrates efficacy based on overall survival as presented in
`the following Table A.
`
`Table A: Primary Efficacy of Overall Survival Analysis in the ITT
`Population
`
`
`
`ITT Population
`
`Placebo
`N=243
`
`
`
`Tarceva
`N=488
`
`Hazard Ratio
`(95% CI)
`
`
`
`# of Deaths
`Med. Survival in
`4.7
`6.7
`
`months 95% C
`4.1, 6.3
`5.5, 7.8
`Hazard Ratio = Tarceva I Placebo; Unadjusted, log-rank test, adjusted (for
`randomization stratification factors) analysis p-value < 0.0001
`
`
`
`P-value
`
`0-0018
`
`
`
`0.764
`(0.645, 0.905)
`
`

`
`OSI EX. 2004 - 0011
`OSI EX. 2004 - 0011
`
`2
`
`Introduction
`
`2.1 Overview
`
`Lung cancer is a common disease in U.S. Currently the treatments approved for
`the first line therapy of non-small cell lung cancer (NSCLC) Stage IIIB/IV
`patients are paclitaxel/cisplatin, gemcitabine/cisplatin, and vinorelbine i cisplatin.
`Approved treatments for the second line therapy ofNSCLC Stage IIIB/IV patients
`are docetaxel (approval based on demonstration of survival benefit over best
`supportive care) and alimta (accelerated approval based on response rate). Iressa
`was granted accelerated approval for the third line setting of NSCLC Stage
`IIIB/IV patients based on observed response rate.
`
`2.1.1 Background
`
`Epidermal growth factor receptor (EGFR) and its ligands are overexpressed or
`involved in autocrine growth loops in a number of tumor types, including
`NSCLC. EGFR is considered an important prognostic indicator in patients with
`epithelial malignancies. Increased EGFR expression is correlated with aggressive
`morphology, poor outcome in NSCLC, and poor response to therapy. 'Erlotinib
`acts through direct and reversible inhibition of the EGFR tyrosine kinase, and it
`inhibits the EGF-dependent proliferation of cells at nanomolar concentrations and
`blocks cell cycle progression at the G1 phase.
`
`In this application the sponsor has submitted results of 4 studies in NSCLC
`patients and 5 studies in other tumor types. Study BR.2l is submitted as the
`registration study.
`
`2.1.2 Statistical Issues
`
`1. The primary analysis of the primary endpoint overall survival was based
`on stratified log-rank test including randomization stratification factors
`and EGFR status. In 67% of the patients EGFR status was not evaluated.
`An adjusted analysis including these 67% patients with missing data on
`EGFR status is questionable.
`2. The results of exploratory analyses in the subgroups suggest a significant
`survival benefit due to erlotinib in the EGFR positive patients and suggest
`no survival benefit in the EGFR negative population.
`
`

`
`OSI EX. 2004 - 0012
`OSI EX. 2004 - 0012
`
`2.2 Data Sources
`
`Data and reports used for review are from the electronic submission received on
`5/i2l04, 6/22/04 and 7129/04 The network paths are:
`\\Cdsesub l\N2 I 743\N 0O0\2004—05- l 2\ciinstat\lung ,
`\\Cdsesub l\N2 l743\N 0O0\2004-06-22\c1't\datasets\BR2l ,
`\\Cdsesub1\N21743\N 000\2004—06—22\clinstat\lung\BR2l ,
`\\Cdsesub1\N2l743\N 000\2004-07-29\clinstat\lung\ise , and
`\\Cdsesub l\N2 1 743\N 000\2004-09—l 7 .
`
`3 Statistical Evaluation
`
`3.1 Evaluation of Efficacy
`
`The sponsor has submitted efficacy results from the following 4 studies conducted
`in NSCLC patients:
`
`(a) The registration Study BR.2l was a multicenter, international, double-blinded,
`randomized, phase III trial of erlotinib 150 mg tablet/day versus placebo
`tablet/day (matched to erlotinib in color, shape, size and packaging) for locally
`advanced or metastatic NSLC patients who had failed at least one but no more
`than two prior chemotherapy regimen. The trial treatment was continued until
`disease progression or unacceptable toxicity. First patient was entered in this
`Study BR.2l on November 1, 2001 and the last patient was entered on January
`31, 2003. The data cut-ofi' date for this application was January 30, 2004. A
`detailed statistical evaluation of efiicacy evidence of this study is presented in
`section 3.1.1 of this review.
`
`(13) Study A248-I007 was a multicenter, open-label, phase II single arm trial of
`erlotinib 150 mg tablet/day following failure of platinum based combination
`chemotherapy in EGFR-positive patients with advanced NSCLC. The trial
`treatment was taken until disease progression, or unmanageable toxicity. First
`patient was entered in this Study A248-1007 on January 25, 2000 and the last
`patient was entered on February 14, 2001. The data cut-off date for this
`application was January 27, 2003. Results fi’om Study A248-I007 were submitted
`as supportive evidence to Study BR.2l. A summary of the efficacy findings of
`this study is presented in section 3.1.2 of this review.
`
`(0) Study 0SI2298g was a randomized, double-blinded, multicenter, Phase III
`comparative trial of erlotinib in combination with chemotherapy (paclitaxel and
`carboplatin) versus chemotherapy alone in patients with advanced (stage IIIB or
`IV) NSCLC who have not received prior chemotherapy. This study was initiated
`
`

`
`OSI EX. 2004 - 0013
`OSI EX. 2004 - 0013
`
`on July 18, 2001 and completed on July 1 1, 2003. A summary of the eflicacy
`findings of this study is presented in section 3.1.3 of this review.
`
`(d) Study B0l641l was a randomized, double-blinded, placebo-controlled,
`multicenter, Phase III comparative trial of erlotinib in combination with
`chemotherapy (gemcitabline and cisplatin) versus chemotherapy alone in patients
`with advanced (stage IIIB or IV) NSCLC who have not received prior
`chemotherapy. A summary of the efficacy findings of this study is presented in
`section 3.1.3 of this review.
`
`3.1.1 Study BR.2I
`
`3.1.1.1 Study Design
`
`Study BR.2l was a phase III, randomized, placebo-controlled, double-blinded
`clinical trial comparing erlotinib to placebo. Patients 2 18 years old with
`histologically or cytologically confirmed diagnosis of incurable stage IIIB/IV
`NSCLC who have received at least one but no more than two prior regimens of
`which at least one had to be combination chemotherapy (if _>_ 70 years old), who
`had ECOG performance status of 0 to 3, had adequate renal and hepatic functions
`were randomized in 2:1 ratio to receive either erlotinib (150 mg/day tablets orally)
`or placebo (“150 mg”/day tablets matched to erlotinib in color, shape, size and
`packaging). In this study, patients were stiatifiedaat randomization by center,
`number of prior regimens, prior platinum therapy, best response to prior therapy,
`and ECOG performance status. Patients were treated until documented
`progressive disease or until development of intolerable toxicity. Dose escalation
`was not permitted. The prescribed dose was self—ad1ninistered, taken in the
`morning with up to 200 mL of water at least 1 hour before or 2 hours after
`ingesting any food or other medications.
`
`Efficacy was evaluated by periodic assessments (Appendix 1) of survival and
`QoL scores. Tumor measurements were evaluated every 8 weeks. Safety was
`assessed every 4 weeks.
`
`31
`_
`_
`_
`‘
`_
`‘J_forrnerly E”
`ll
`3 interactive voice
`generated randomization codes and managed the ‘I:
`response system (IVRS). The IVRS minimized potential imbalances between
`treatment arms, based on the above stated 5 stratification factors using a dynamic
`minimization technique. Therapy began within 2 working days after
`randomization and patients were considered on treatment until study drug was
`discontinued.
`
`,_
`
`

`
`OSI EX. 2004 - 0014
`OSI EX. 2004 - 0014
`
`3.1.1.2 Study Objectives
`
`The primary objective of this study was to compare overall survival between the
`two treatment arms (erlotinib vs. placebo).
`
`The secondary objectives included comparison of (1) progression-free survival
`(PFS), (2) response rates (RR), (3) response duration, (4) nature, severity, and
`frequency of toxicities, and (5) quality of life (QoL) as measured by the European
`Organization for the Research and Treatment of Cancer (EORTC) quality of life
`questionnaires QLQ-C30 and the lung cancer module QLQ-LC13. The objectives
`also included to correlate the expression of tissue EGFR levels (at diagnosis) with
`outcomes and response to treatment, and to measure and correlate trough levels of
`erlotinib with clinical responses and /or adverse events.
`
`3.1.1.3 Efficacy Endpoints
`
`Pgn_fi Efficacy Endmint of this study was survival (OS) defined as the time
`fi-om the date of randomization to the date of death from any cause. Survival time
`was censored at the date of last post-therapy follow-up visit for patients who were
`still alive.
`
`Secondary Efficacy Endpoints included:
`(1) PFS defined as the length of time from randomization to the first observation
`of disease progression or death due to any cause.
`(2) Objective response, determined using RECIST criteria (Appendix 2).
`(3) Duration of response was measured fi'om the time measurement criteria for
`CR/PR were first met until the first date that recurrent or progressive disease or
`death was objectively documented.
`(4) QoL, assessed by the EORTC QLQ—30 and QLQ«LCl3, with the emphasis on
`the three pre-specified symptoms (cough, dyspnea, and pain) (Appendix 3).
`
`Reviewer ’s Commenst:
`
`1. All patients who had measurable lesions and who had at least one
`objective tumor assessment after baseline were considered as evaluable for
`.response.
`2. All patients who had completed quality of life assessments were evaluable
`for QoL.
`'
`3. The protocol does not specifically emphasize on the three QoL symptoms,
`cough, dyspnea and pain. These were added as pre-specified symptoms in
`the statistical analysis plan and study report.
`
`

`
`OSI EX. 2004 - 0015
`OSI EX. 2004 - 0015
`
`3.1.1.4 Sample Size Considerations
`
`The study was planned to enroll 700 patients. With this sample size, the study
`was powered to detect a 33% increase in overall survival time (median survival of
`4 months to 5.3 months, hazard ratio (HR) of 1.33) in the erlotinib arm compared
`to placebo arm with 90% power and maintaining a family-wise two-sided type I
`error rate of 0.05. The final analysis was planned to be conducted when 582
`deaths were observed.
`
`Reviewer ’s Comments:
`
`1.
`
`in the original protocol (dated September 10, 2001) the sample size was
`determined to be 330 patients with the planned final survival analysis
`when 256 deaths occurred. This calculation was based on detecting a 50%
`improvement in survival for erlotinib (6 months vs. 4 months, HR = 1.5)
`with 90% power using two-sided 5% level of significance. The sample
`size was amended (amendment dated August 29, 2002) to be 700 patients
`in order to detect a 33% improvement instead of 50% improvement. It
`was stated that because of the lack of demonstrable benefit in adding
`Iressa to standard chemotherapy for NSCLC in the first line setting as
`reported by Astraleneca in their press release dated August 19, 2002, and
`the rapid accrual to study BR.2i, that it was decided to increase the
`sample size to be able to detect a smaller but clinically relevant
`improvement in survival.
`2. The actual number of patients entered on this study was 731 patients.
`
`3.1.1.5 Interim Analysis
`
`No interim analysis was planned for this study.
`
`3.1.1.6 Efficacy Analysis Methods
`
`The primary efiicacy analysis‘ was to compare overall survival time using log-
`rank test stratified by all stratification factors except center plus patient’s EGFR
`status (positive/mutated vs. unlcnown vs. negative) at baseline in all the
`randomized patients (ITT population). Kaplan-Meier curves were to be used to
`display the survival curves and 95% confidence intervals for the median survival
`computed using the method of Brookmeyer and Crowley. In addition, the effect
`of study center and other potential prognostic factors on overall survival was
`planned to be assessed using Cox regression. Additional supporting analysiswas
`to include Kaplan-Meier estimation.
`
`Progression-free survival (PFS) was specified as one of the secondary endpoints.
`PFS was defined as the time from randomization to the first observation of
`
`

`
`OSI EX. 2004 - 0016
`OSI EX. 2004 - 0016
`
`disease progression or death due to any cause. A patient who stopped treatment
`with study drug and went on to receive alternative therapy for NSCLC, prior to
`documentation of disease progression, was planned to be censored on the date
`alternative therapy began. Same methods as used in overall survival analysis
`were to be employed to analyze PFS data.
`
`The secondary endpoint of response rate was planned to be estimated as the
`proportion of patients evaluable for response who met the criteria of complete or
`partial response. A Cochran-Mantel-Haenzel test was to be used to compare the
`tumor response rate between the two treatment arms adjusting all stratification
`factors, except center plus patient’s EGFR status. Duration of response was
`planned to be analyzed using similar methods as described for overall survival.
`
`The EORTC QLQC30 (Appendix 4) that was used in this study is a self-
`administered cancer specific questionnaire with multi-dimensional scales. It
`consists of both rnulti-item scales and single item measures, including five
`functioning domains, a global quality of life domain, three symptom domains and
`six single items. For each domain or single item measure a linear transformation
`was to be applied to standardize the raw score to range between 0 and 100. The
`QLQ-LC13 (Appendix 5) lung cancer module which was also used in this study
`includes questions assessing lung cancer-associated symptoms (cough,
`haemoptysis, dyspnea, and site~specific pain), treatment-related side effects (sore
`mouth, dysphagia, peripheral neuropathy and alopecia) and pain medication. The
`protocol specified that the questionnaires would be scored as described in QLQ-
`C30 manual, and analyzed accordingly. The protocol specified that the method of
`analysis of variance for repeated measures was planned to be used for domains
`represented by aggregate scores. Questionnaires for patients had to be completed
`at baseline and every 4 weeks while on study drug. A final questionnaire had to be
`completed within 2 weeks of progressive disease, or it would be considered
`completed at the 4-week visit afier the end of treatment if it had not already been
`completed within 2 weeks of progressive disease.
`
`In the statistical analysis plan it was specified that the primary endpoints in the
`quality of life analysis were defined as the time from randomization to
`deterioration in the following three QoL symptoms: cough (Question 1 in QLQ-
`LCI3), dyspnea (Question 8 in QLQ-C30) and pain (Questions 1 and 19 in QLQ-
`C30). Patients were considered as deteriorated for a given symptom if their
`change of score from the baseline on the domain/single item defining this
`symptom was 10 points or higher at any time-point after the baseline assessment.
`It was stated in the statistical analysis plan that the value of 10 points on a 100
`scale was chosen as previous studies had indicated that a 10% change of highest
`possible score are perceived as clinically significant.
`
`10
`
`

`
`OSI EX. 2004 - 0017
`OSI EX. 2004 - 0017
`
`The statistical analysis plan stated that for each symptom, all patients who had a
`baseline and at least one of the follow-up QoL assessments for the symptom
`would be included in the time to deterioration analysis. Patients would be
`censored at the time of the last QoL questionnaire completion if they had not
`deteriorated before that. Unstratified log-rank test would be used as the primary
`method to compare the time to deterioration in each symptom between the two
`treatment arms. The Hochberg procedure was planned to be used to adjust the p-
`values of the log-rank tests for these three comparisons.
`
`Reviewer '5' Comment:
`
`The primary endpoints and the method of QoL analysis was specified differently
`in the statistical analysis plan which was finalized on June 11, 2003, after all
`patients had been entered on the study. Given that this was a secondary endpoint
`and the analysis plan was finalized after the completion of accrual to the study
`(last patient entered on January 31, 2003) analysis of QoL data can only be
`considered as exploratory.
`
`3.1.1.7 Sponsor’s Results and Statistical Reviewer’s Findings! Comments
`
`In the BR.2l study, a total 731 patients with advanced or metastatic NSCLC were
`entered into the study across 86 study sites, 27 in Canada, 1 in US and 58
`internationally (rest of the world). The overall survival efficacy analysis
`submitted in this NDA was based on 488 patients in erlotinib treatment arm and
`243 patients in placebo arm. A total of 7 patients were declared as lost to follow-
`up (4 patients in the erlotinib arm and 3 patients in the placebo arm). The sponsor
`has reported some discrepancies between the data provided by the center to obtain
`randomization and the actual baseline data. Of note, response to prior
`chemotherapy at baseline was better than was reported at randomization for 26
`patients (5%) in the erlotinib arm and for 9 patients (4%) in the placebo arm, and
`it was worse for 57 patients (12%) and 22 patients (9%) in the erlotinib and
`placebo 311115 respectively.
`
`3.l.1.7.l Baseline Characteristics
`
`The baseline Characteristics of the overall population are presented in Table 1.
`
`Reviewer ‘s Comment:
`
`In the overall patient population the baseline characteristics appear to be balanced
`between the two treatment arms.
`
`

`
`OSI EX. 2004 - 0018
`OSI EX. 2004 - 0018
`
`Table 1: Demographics and Baseline Characteristics of Patients
`
`
`
` -T
`
`
`
`
`
`
`
`Baseline PS, prior response, prior number of therapy, prior platinum treatment
`were stratification factors at baseline (characteristics in blue; using data variables
`alaecog, alabrsp, alareg and alaplat in patientxpt data set).
`
`3.1.1.12 Primary Efficacy Analyses
`
`Primary eficacy analysis was overall survival analysis using stratified log-rank
`test. At the time of this analysis presented in this NDA a total of 587 deaths were
`observed (final analysis planned with 582 deaths). The results of the survival
`analysis based on ITT population using unadjusted log-rank test are presented in
`
`12
`
`

`
`OSI EX. 2004 - 0019
`OSI EX. 2004 - 0019
`
`Table 2 (same as reported by the sponsor). The results of the stratified log-rank
`test, primary analysis as specified in the protocol, were similar to the unadjusted
`analysis (adjusted p-value < 0.0001). The results of adjusted analysis using Cox
`regression as specified in the protocol are presented in Table 3. The Kaplan-
`Meier curves of the overall survival in the ITT population are illustrated in Figure
`1.
`
`Table 2: Primary Efficacy of Overall Survival Analysis in the ITT
`Population
`
`ITT Population
`
`Placebo
`N=243
`
`Tarceva
`N=488
`
`Hazard Ratio
`
`P-value
`
`0.764
`
`0.0018
`
`(95% CI) # of Deaths
`
`Med. Survival in
`
`4.7
`
`months 95% C
`4.1, 6.3
`5.5, 7.8
`Hazard Ratio = Tarceva I Placebo; Unadjusted, log-rank test, adjusted (for
`randomization stratification factors) analysis p—value < 0.0001
`
`(0.645, 0.905)
`
`Figure 1: Kaplan-Meier Survival Curves in the ITT Population
`
`
`
`ProportionSurvivingOOOin_I:In
`
`
`' *1:-'r
`'
`‘
`iii
`(0 = placebo and = erlotinib)
`
`0
`
`13
`
`

`
`
`
`
`
`
`
`
`
`0.0002
`
`<0.000l
`
`0.3725
`0.0037
`0.1413
`0.0105
`
`0.3695
`0.1814
`
`OSI EX. 2004 - 0020
`OSI EX. 2004 - 0020
`
`Table 3: Cox Regression Analysis in the ITT Population Adjusting for
`Randomized Stratification Factors and Baseline EGFR Status (Protocol
`Specified Analysis)
`
`Treatment Tarceva vs Placebo
`
`Baseline ECOG PS 2-3 vs. 0-1
`Response to prior therapy
`(SD vs. CR/PR + PD)
`PD vs. CR/PR + SD
`Number of riorthera
`2vs. 1
`
`EGFR Status
`
`
`
`'
`
`Hazard Ratio
`0.726
`
`1.895
`
`0.915
`1.356
`H36
`0.663
`
`95% C-L
`0.612 0 861
`
`1.593 2 254
`’
`.
`0 752, 1.113
`1.104, 1.666
`0.959 1 346
`0.484. 0 909
`
`0.854, 1.527
`1.142
`Negative vs. Positive + Unknown
`
`0.930, 1.470
`Unknown vs. Neative + Positive
`1.169
`‘P-value by stratified log-rank including stratification factors and EGFR status was also < 0.0001.
`
`
`
`
`
`
`
`
`
`
`
`Reviewer ‘.9 Comments.‘
`
`1. The final analysis of the overall survival demonstrates superiority of
`erlotinib over placebo with respect to overall survival (Table 2 and Figure 0
`1 ).
`It should be noted that there were no events observed in some of the strata
`
`2.
`
`and thus the results of the adjusted analyses should be interpreted with
`caution.
`
`3.
`
`4.
`
`In about 67% of the patients EGFR status was unknown at baseline.
`Including EGFR status in the model amounts to categorizing the unknown
`group as another category (as if it was an intermediate category between
`positive and negative status). However the unknown EGFR status
`includes patients who were not assessed for EGFR status (missing data)
`and would in fact be either EGFR positive or negative. Thus the results of
`the model including EGFR status are not interpretable.
`It should also be noted that more than 90% of the patients had received
`prior platinum therapy in both the treatment arms and thus adding it as a
`covariate in the mod