CENTER FOR DRUG EVALUATION AND
`
`RESEARCH
`
`APPLICA TION NUMBER:
`
`NDA 22-249
`
`STATISTICAL REVIEWgSQ
`
`

`

` US Department of Health and Human Services
`
`Food and Drug Administration
`Center for Drug Evaluation and Research
`Office of Translational Sciences
`Office of Biostatistics
`
`STATISTICAL REVIEW AND EVALUATION
`
`CLINICAL STUDIES — TEAM LEADER’S MEMO
`
`NDA/Serial Number:
`
`22—249
`
`Drug Name:
`
`Indication(s):
`
`Applicant:
`
`Date(s):
`
`Bendamustine (Treanda)
`
`patients with (Binet Stage B/C) B-CLL requiring therapy
`
`Cephalon
`
`Submission date: September 20, 2007
`
`PDUFA due date: March 20, 2008
`
`Review Priority:
`
`Priority (pilot for GRMP)
`
`Biometrics Division:
`
`Division of Biometrics 5 (HFD-71 1)
`
`Primary Statistical
`Reviewer:
`
`Shenghui Tang, Ph.D.
`
`Secondary Reviewer:
`
`Rajeshwari Sridhara, Ph,D., Team Leader/Deputy Division Director
`
`Concurring Reviewer:
`
`Aloka Chakravarty, Ph.D., Division Director
`
`Medical Division:
`
`Clinical Team:
`
`Oncology Drug Products (HFD-ISO)
`
`Qin Ryan, M.D., Ph.D., Virginia Kwitkowski,, MS, RN., CRNP
`
`Amna Ibrahim, M.D. (CDTL)
`
`Project Manager:
`
`Ms. Dorothy Pease
`
`Keywords: Co-primary endpoints, objective response rate, progression-free survival
`
`

`

`Conclusion and Recommendation
`
`This is Team Leader’s memo of the New Drug Application (NDA) submission seeking approval
`for bendamustine (Treanda) as the first line treatment of chronic lymphocytic leukemia (CLL)
`based on one randomized study comparing to chlorambucil in previously untreated adults with
`symptomatic Binet stage B or stage C CLL requiring treatment. I concur with the primary
`reviewer, Dr. Tang’s conclusion that the data submitted supports the claim that bendamustine has
`demonstrated superior overall response rate (OR) and progression—free survival (PFS)
`compared to chlorambucil (OR of 59% vs. 26% and PFS HR = 0.52, p—value < 0.0001). Please
`refer to the primary review by Dr. Tang for the details of the study and the results.
`
`Progression-free survival was assessed by a panel of three independent expert hematologic
`oncologists and also objectively calculated using an algorithm based on NCI working group
`criteria. According to the sponsor, in performing the review the members ofthe independent
`panel were allowed to exercise clinical judgment in determining response and did not include
`bone marrow evaluations as required by the NCI working group criteria. The FDA reviewers
`were able to verify the calculated response rates and PFS, but could not verify the same as
`determined by the independent panel due the subjective nature of the independent evaluation.
`Therefore, it is recommended that the calculated response rates and PFS estimates be included in
`the product label.
`
`

`

`This is a representation of an electronic record that was signed electronically and
`this page is the manifestation of the electronic signature.
`
`Rajeshwari Sridhara
`2/25/2008 10:20:09 AM
`BIOMETRICS
`
`Aloka Chakravarty
`2/25/2008 12:04:52 PM
`BIOMETRICS
`
`

`

` US. Department of Health and Human Services
`
`Food and Drug Administration
`Center for Drug Evaluation Research
`Office of Translational Science
`Office of Biostatistics
`
`STATISTICAL REVIEW AND EVALUATION
`CLINICAL STUDIES
`
`NDA /Serial Number:
`
`22-249
`
`Drug Name:
`
`Applicant:
`
`Indication(s):
`
`Treanda
`
`Cephalon
`
`Patients with (Binet Stage B/C)
`B-CLL Requiring Therapy
`
`Date(s) :
`
`Submission Date: September 20, 2007
`
`PDUFA Date: March 20, 2008
`
`Review Completion Date: February 19, 2008
`
`Review Priority:
`
`Priority
`
`Biometrics Division:
`
`Statistical Reviewer:
`
`Concurring Reviewer:
`
`Medical Division:
`
`Clinical Team:
`
`Division of Biometrics V (HFD-711)
`
`Shenghui Tang, Ph.D.
`
`Rajeshwari Sridhara, Ph.D., Team Leader
`Aloka Chakravarty, Ph.D., Director
`
`Oncology Drug Products (HFD-150)
`
`Qin Ryan, M.D., Virginia Kwitkowski, M.D.
`
`Amna Ibrahim, M.D.
`
`Project Manager:
`
`Ms. Dorothy Pease
`
`Keywords:
`
`Objective response rate, Duration of response, PFS
`
`

`

`Table of Contents
`
`1
`
`2
`
`3
`
`Executive Summary ....................................................................................... 2
`1.1
`Conclusions and Recommendations....................................................... 2
`1.2 Brief Overview of Clinical Studies ......................................................... 2
`
`Statistical Issues and Findings ................................................................ 3
`1.3
`Introduction .................................................................................................... 6
`2.1 Overview................................................................................................... 6
`2.1.1
`Background ......................................................................................... 6
`2.1.2
`Statistical Issues .................................................................................. 7
`
`2.2 Data Sources ............................................................................................. 9
`Statistical Evaluation ...............................................-...................................... 9
`3.1
`Evaluation of Efficacy ............................................................................. 9
`3.1.1.
`Study Design ................................................................................ 9
`3.1.2
`Study Objectives ........................................................................ 10
`3.1.3
`Efficacy Endpoints ..................................................................... 10
`3.1.4
`Sample Size Considerations ....................................................... 12
`3.1.5
`Interim Analyses ...................................................... 13
`3.1.6
`Efficacy Analysis Methods ........................................................ 13
`3.1.7
`Sponsor’s Results and Statistical Reviewer’s Findings/
`Comments ................................................................................................... 14
`3.1.7.1
`Baseline Characteristics ............................. '............................ 14
`
`Primary Efficacy Analyses .................................................... 17
`3.1.7.2
`Secondary Efficacy Analyses ................................................ 24
`3.1.7.3
`Evaluation of Safety............................................................................... 27
`3.2
`Findings in Special/Subgroup Populations ................................................ 27
`4.1 Gender, Race and Age ........................................................................... 27
`Summary and Conclusions .......................................................................... 29
`5.1
`Statistical Issues and Collective Evidence ........................................... 29
`5.2 Conclusions and Recommendations..................................................... 30
`
`4
`
`5
`
`

`

`1 Executive Summary
`
`1.1 Conclusions and Recommendations
`
`The sponsor submitted this application to evaluate the efficacy of Treanda
`(Bendamustine) compared with chlorambucil in the initial treatment of patients
`with chronic lymphocytic leukemia (CLL) in Binet stage B or Binet stage C
`requiring treatment. The applicant is seeking approval based on the co-primary
`efficacy endpoints-overall response rate (OR) and progression-free survival
`(PFS). OR was the proportion of patients in each treatment group with a best
`response of CR, nPR, or PR. Progression—free survival (PFS) was defined as the
`time from randomization to progressive disease (PD) or death for any cause,
`whichever occurred first. The primary analyses were based on the Independent
`Committee for Response Assessment (ICRA) adjudicated responses and
`adjudicated event time points. This application was based primarily on data from
`an open-label, randomized, phase 3 study (02CLLIII). Patients were randomly
`assigned (with stratification by Binet stage and study center) to either the
`bendamustine or chlorambucil treatment group at a ratio of
`l: 1.
`
`A total of 302 patients were screened and 301 were randomly assigned to
`treatment (1 patient was not assigned to a treatment group due to refusal) at 45
`centers throughout 8 countries. Per the sponsor’s statistical analysis plan, in order
`to account for the multiplicity of endpoints, superiority of bendamustine on PFS
`would not be claimed unless the 2-sided p-value for OR and PFS were both less
`than or equal to 0.016. The sponsor reported that the proportion of patients with
`ORR (determined by ICRA) was 62% in the bendamustine treatment group
`compared with 33% in the chlorambucil treatment group (p<0.0001). The primary
`PFS analysis showed that the bendamustine treatment was superior to
`chlorambucil treatment (median 21 vs. 9 months, hazard ratio (HR) 0.23,
`p<0.0001). Based on the data submitted bythe sponsor these results were
`confirmed by this reviewer and the data support the efficacy claim.
`
`Whether the endpoints and the sizes of the effects on these two endpoints in this
`phase III study are adequate for approval is a clinical decision.
`
`1.2 Brief Overview of Clinical Studies
`
`Study 02CLLIII was a phase 3, randomized, open-label, multicenter study to
`evaluate the clinical efficacy and safety of bendamustine compared with
`chlorambucil in the treatment of previously untreated adults with symptomatic
`Binet stage B or stage C CLL requiring treatment. A 5-stage adaptive standard
`group sequential procedure was applied with a maximum of 4 planned interim
`analyses. The number of patients to be enrolled was assumed to be approximately
`350 patients. This statistical design allowed closing study enrollment as soon as
`
`

`

`the required level of significance was reached. Patients were randomized and
`prospectively stratified by study center and Binet stage (Binet B or Binet C). The
`recruitment period for the study was approximately 4 years and the follow-up
`period ends 1 year after the last enrolled patient completes treatment.
`
`The data presented in this submission werethose for the patients with data that
`were included in the third interim analysis. Following the third interim analysis
`the Independent Data Monitoring Committee (IDMC) made a recommendation
`that enrollment be stopped and the final analysis performed. The first patient was
`enrolled on November 5, 2002. The data were cleaned for the final analysis with a
`cut-off date of 26 March 2006.
`
`1.3
`
`Statistical Issues and Findings
`
`Study 02CLLIII was designed to evaluate the efficacy of Treanda (Bendamustine)
`compared with chlorambucil in the initial treatment of patients with chronic
`lymphocytic leukemia (CLL) in Binet stage B or Binet stage C requiring
`treatment. The applicant is seeking approval based on the co-primary efficacy
`endpoints-overall response rate (OR) and progression-free survival (PFS).
`
`Statistical Issues:
`
`.
`
`1. This study was planned as a 5-stage adaptive standard group sequential design
`with a Pocock-type boundary and a rule for adaptively recalculating the
`sample size in the next stage. According to the protocol and statistical analysis
`plan, both primary endpoints, overall response rate (OR) and progression-
`free survival (PFS), were analyzed at each interim analysis. In order to
`account for the multiplicity of endpoints, superiority of bendamustine on PFS
`would not be claimed unless the 2-sided p-value for OR and PFS are both
`less than or equal to 0.016.
`
`2. Because 3 interim analyses were performed, patients in this study could fall
`into the one of the following 3 segments: lSt interim analysis (n=87), 2nd
`interim analysis (n=77), or 3rd interim analysis (n=137). Table 1 shows that
`the final p—values for ORR and PFS from the combined results of all 3
`segments were less than 0.0001.
`
`
`Table 1. Interim Analfls for ORR and PFS
`
`Segment 1
`Segment 2
`Segment 3
`Combined P-Value
`
`87
`77
`137
`Sample Size
`
`P-value for ORR
`0.0007
`0.0043
`0.0305
`<0.0001
`
`j_ <0.0001P-Value for PW <0.0001 <0.0001 0.0295
`
`
`1: 2-sided Fisher’s Exact Test; 2: 2-sided log-rank test
`
`

`

`This reviewer also calculated the unadjusted p-values for OR and PFS. Both
`unadjusted p—values for OR and PF S were also less than 0.0001.
`
`The hazard ratio obtained for the combined ratio is 0.23 with a 95%
`
`confidence interval of 1.34 to 0.39 which was adjusted for repeated testing.
`Both point estimate and confidence interval were based on estimates within
`each study segment that were then combined across the segments. The
`unadjusted hazard ratio for PF S was 0.22 with a 95% confidence interval of
`0.14 to 0.33.
`
`3.
`
`Initial observations within the dataset received from Ribosepharrn led to
`further quality control (QC) review of study center 1 in Bulgaria. During the
`QC process, a number of centers were reviewed in addition to study centers 1
`and 2 in Bulgaria. The findings at center 1 in Bulgaria indicated that the center
`had not followed all the procedures in accordance with the protocol, ie, the
`data collected could not always be substantiated in the patient's medical charts
`or source data available for review. For center 2 in Bulgaria, the documents
`supporting the informed consent process were not in accordance with GCP. In
`order to ensure the consistency of the findings between these 2 centers and the
`other centers in the study, the sponsor also provided the analyses of the
`primary endpoint analyses with both centers 1 and 2 excluded from the
`analyses. A total of 54 patients were excluded (28 from center 1 and 26 from
`center 2). The analyses on OR and PFS with Centers 1 & 2 Excluded
`showed the results were similar to those seen in the total population.
`
`4. The log—rank test showed that there was no difference between two
`distributions of time to assessment, except the 1St assessment. The median
`difference in the 1St assessment was less than a week. With PF S medians of 21
`
`months in the bendamustine arm and 9 months in the chlorambucil arm, these
`small differences in time to assessment is unlikely to influence the final
`outcome of the study (Table 11).
`
`5. The investigator and the ICRA assessment showed an agreement in 258 (86%)
`patients; 38% of patients scored as progressors by both, and 48% of patients
`censored by both. In 14.3% of patients the results were discordant (Table 12).
`
`Findings:
`
`A total of 302 patients were screened and 301 were randomly assigned to
`treatment (1 patient was not assigned to a treatment group due to refusal) at 45
`centers throughout 8 countries. The sponsor reported that the proportion of
`patients with ORR was 62% in the bendamustine treatment group compared with
`33% in the chlorambucil treatment group (p<0.0001) as determined by the ICRA.
`The primary PFS analysis showed that the bendamustine treatment was superior
`
`

`

`to Chlorambucil treatment (median 21 vs. 9 months, hazard ratio (HR) 0.23,
`p<0.0001). For patients in the ITT analysis set with ICRA responses of CR, PR,
`or nPR, the median duration of response was 16 months for the 95 responders in
`the bendamustine‘ treatment group and 6 months for the 49 responders in the
`Chlorambucil treatment group.
`
`
`Table 2. Response Analysis (ITT Population)
`Bendamustine
`Chlorambucil
`
`N=153
`N: 148
`
`Complete response (CR)
`42 (27%)
`3 (2%)
`
`odular partial response (nPR)
`15 (10%)
`4 (3%)
`
`artial response (PR)
`38 (25%)
`42 (28%)
`
`nconfirmed response
`9 (6%)
`8 (5%)
`
`Stable disease
`22 (14%)
`37 (25%)
`
`Progressive disease
`4 (3%)
`26 (19%)
`
`0t examined
`23 (15%)
`28 (19%)
`
`
`
`
`
`
`Overall Response Rate (OR
`95 (62%)
`49 (33%)
`
`95% CI
`(54.4%, 69.8%)
`(25.5%, 46.7%)
`P-value for comparing ORR
`P<0.0001
`
`isher’ s exact test (adjusted)
`P<0.0001
`-value for comparing ORR
`Fisher’s exact test (unadjusted)
`
`
`Table 3. PFS Analysisin ITTPopulation
`
`I‘
`
`Bendamustine
`
`Chlorambucil
`
`__l
`
`N=148
`N=1 53
`
`atients with events
`47 (31%)
`66 (45%)
`
`edian TTP in months (95% CI)
`21. 1 (17.7, 25.6 )
`9.4 (8.7, 11.7)
`
`djusted P-value (log-rank test)
`P<0. 0001
`
`nadj usted P-value (log-rank test)
`P<0.0001
`
`djusted Hazard ratio (BEN/CLB)
`0,23
`
`(95% CI)
`(0.13, 0.39)
`
`nadjusted Hazard ratio (CLB/BEN)
`0,22
`
`(95% CI)
`(0.14, 0.33)
`
`
`

`

`2
`
`Introduction
`
`2.1 Overview
`
`The applicant has submitted this application to evaluate the efficacy of Treanda
`(Bendamustine) compared with chlorambucil in the initial treatment of patients
`with chronic lymphocytic leukemia (CLL) in Binet stage B or Binet stage C
`requiring treatment.
`
`2.1.1 Background
`
`Chronic lymphocytic leukemia (CLL) is a chronic lymphoproliferative disorder
`that is characterized by a progressive accumulation of functionally incompetent
`lymphocytes of monoclonal origin. The disorder is considered under the current
`World Health Organization (WHO) classification to be identical to small
`lymphocytic lymphoma (SLL), representing a different stage of that disease. CLL
`is the most common form of leukemia in Western industrialized nations with an
`annual incidence of 3 to 3.5 cases per 100000. It is predominantly a disease of the
`elderly with a median age of diagnosis of 72 years. CLL is approximately twice as
`common in men as in women.
`
`Two staging systems for CLL have been widely adopted. The Rai staging system
`is based on the progressive accumulation of malignant cells with physical signs of
`progression (eg, lymphadenopathy) and eventual compromise of bone marrow
`function. The Binet staging system is similar but places more emphasis on the
`number of involved sites. The Binet staging system is more commonly applied in
`Europe and was used in this study.
`
`CLL typically follows an indolent course and the recommended clinical approach
`to patients with Binet stage A disease, with no specific risk factors or evidence of
`progression, is watchful waiting. Treatment is generally initiated for
`patients with symptomatic Binet stage B disease (3 or more enlarged nodal areas)
`or for those patients with stage C disease (disease-related anemia or
`thrombocytopenia). For patients under 65 years of age the therapeutic objective is
`to achieve long-lasting remissions, while in older patients the treatment is largely
`palliative with a goal of maintaining a high quality of life.
`
`For these older patients with CLL, continuous or intermittent oral administration
`of chlorambucil, either alone or in combination with glucocorticoids, has been
`considered a principal treatment option (CLL Trialists' Collaborative Group)
`999). This regimen has no impact on the natural history of CLL. However,
`alkylating agents such as chlorambucil are particularly suitable for the treatment
`of lymphocytosis in this indolent disease because they have cytotoxic effects on
`leukemic cells independently of cell division. In younger or healthier patients
`
`

`

`fludarabine and fludarabine-based regimens have proven successful in achieving a
`higher proportion of durable responses than chlorambucil. However, this
`increased efficacy was achieved with a cost in the tolerability of the regimens, and
`the long-term benefits in overall survival remain to be determined.
`
`Bendamustine is a cytotoxic compound with an alkylating nitrogen mustard group
`and a purine-like benzimidazole ring. Bendamustine appears to act primarily as an
`alkylating agent inducing extensive and durable deoxyribonucleic acid (DNA)
`breaks, which result in inhibition of DNA replication, repair, and transcription,
`and cell cycle arrest. The presence of the benzimidazole ring structure of
`bendamustine may explain differences between bendamustine and other
`alkylating agents, such as slower repair of damaged DNA following exposure,
`activity against multi drug resistant cells, and only partial cross-resistance with
`otheralkylating agents.
`
`Bendamustine has been studied as a treatment for patients with CLL in a number
`of clinical studies, and is approved for the treatment ofthis disease in Germany.
`In a study of bendamustine in a mixed population of chemona'fve (previously
`untreated) and previously treated patients with CLL, Kath et al demonstrated a
`high rate of durable response with doses of bendamustine of 50 to 60 mg/m2,
`days 1 through 5 every 28 days, with a complete response (CR) rate of25% and an
`overall response rate 0175% in the 12 previously untreated patients. A regimen
`of70 to 110 mg/m2 bendamustine on days 1 and 2 of a 3-week cycle was also
`shown to be efficacious in studies of patients with relapsed/refractory CLL.
`
`Study 02CLLHI was a phase 3, randomized, open-label, multicenter study to
`evaluate the clinical efficacy and safety of bendamustine compared with
`chlorambucil in the treatment of previously untreated adults with symptomatic
`Binet stage B or stage C CLL requiring treatment. A 5-stage adaptive standard
`group sequential procedure was applied with a maximum of 4 planned interim
`analyses. The final number of patients to be enrolled could not be calculated
`apriori, but it was assumed to be approximately 350 patients. This statistical
`design allowed closing study enrollment as soon as the required level of
`significance was reached, but only at 1 of the prespecified interim analyses.
`Patients were randomized and prospectively stratified by study center and Binet
`stage (Binet B or Binet C). The recruitment period for the study was
`approximately 4 years and the follow-up period ends 1 year after the last enrolled
`patient completes treatment.
`
`2.1.2 Statistical Issues
`
`1. This study was planned as a 5-stage adaptive standard group sequential design
`with a Pocock-type boundary and a rule for adaptively recalculating the
`sample size in the next stage. According to the protocol and statistical analysis
`
`

`

`plan, both primary endpoints, overall response rate (OR) and progression-
`free survival (PFS), were analyzed at each interim analysis. In order to
`account for the multiplicity of endpoints, superiority of bendamustine on PFS
`would not be claimed unless the 2—sided p-value for OR and PFS are both
`less than or equal to 0.016.
`
`. Because 3 interim analyses were performed, patients in this study could fall
`into the one of the following 3 segments: 1St interim analysis (n=87), 2nd
`interim analysis (n=77), or 3rd interim analysis (n=137). Table 1 shows that
`the final p-values for OR and PFS from the combined results of all 3
`segments were less than 0.0001 .This reviewer also calculated the unadjusted
`p-values for OR and PFS. Both unadjusted p-values for OR and PFS were
`also less than 0.0001.
`
`The hazard ratio obtained for the combined ratio is 0.23 with a 95%
`confidence interval of 1.34 to 0.39 which was adjusted for repeated testing.
`Both point estimate and confidence interval were based on estimates within
`each study segment that were then combined across the segments. The
`unadjusted hazard ratio for PFS was 0.22 with a 95% confidence interval of
`0.14 to 0.33.
`
`.
`
`Initial observations within the dataset received from Ribosepharm led to
`further quality control (QC) review of study center 1 in Bulgaria. During the
`QC process, a number of centers were reviewed in addition to study centers 1
`and 2 in Bulgaria. The findings at center 1 in Bulgaria indicated that the center
`had not followed all the procedures in accordance with the protocol, ie, the
`data collected could not always be substantiated in the patient's medical charts
`or source data available for review. For center 2 in Bulgaria, the documents
`supporting the informed consent process were not in accordance with GCP. In
`order to ensure the consistency of the findings between these 2 centers and the
`other centers in the study, the sponsor also provided the analyses of the
`primary endpoint analyses with both centers 1 and 2 excluded from the
`analyses. A total of 54 patients were excluded (28 from center 1 and 26 from
`center 2). The analyses on OR and PFS with Centers 1 & 2 Excluded
`showed the results were similar to those seen in the total population.
`
`. The log-rank test showed that there was no difference between two
`distributions of time to assessment, except the lSt assessment. The median
`difference in the lSt assessment was less than a week. With PFS medians of 21
`months in the bendamustine arm and 9 months in the chlorambucil arm, these
`small differences in time to assessment is unlikely to influence the final
`outcome of the study (Table 11).
`
`

`

`5. The investigator and the ICRA assessment showed an agreement in 258 (86%)
`patients; 38% of patients scored as progressors by both, and 48% of patients
`censored by both. In 14.3% of patients the results were discordant (Table 12).
`
`2.2 Data Sources
`
`Data used for review is from the electronic submission received on September 20,
`2007. The data is in the network path \\Cdsesub1\nonectd\N22249\N 000\2007—
`09—19.
`
`3 Statistical Evaluation
`
`3.1 Evaluation of Efficacy
`
`A total of 302 patients were screened and 301 were randomly assigned to
`treatment (1 patient was not assigned to a treatment group due to refusal) at 45
`centers throughout 8 countries (non US.) as follows: Germany (126 patients at 22
`centers), Bulgaria (1 17 patients at 8 centers), Italy (19 patients at 5 centers),
`France (16 patients at 2 centers), Spain (15 patients at 3 centers), Sweden (4
`patients at 2 centers), Austria (3 patients at 2 centers), and the UK (1 patient). All
`randomized patients were evaluated for efficacy. The first patient was enro11ed on
`November 5, 2002. The data were cleaned for the final analysis with a cut-off date
`of 26 March 2006.
`
`3.1.1 Study Design
`
`This is a phase 3, randomized, open-label, multicenter study to evaluate the
`clinical efficacy and safety of bendamustine compared with chlorambucil in the
`treatment of previously untreated adults with symptomatic Binet stage B or stage
`C CLL requiring treatment. A 5-stage adaptive standard group sequential
`procedure was applied with a maximum of 4 planned interim analyses. The final
`number of patients to be enrolled could not be calculated a priori, but it was
`assumed to be approximately 350 patients.
`
`Patients were randomized and prospectively stratified by study center and Binet
`‘ stage (Binet B or Binet C). Patients who met all inclusion and exclusion
`criteria were randomly assigned (1: 1) to receive either bendamustine at 100
`mg/m2 administered by continuous intravenous (iv) infusion over a period of30
`minutes on days 1 and 2 of each cycle or chlorambucil at 0.8 mg/kg (Broca's
`normal weight) administered orally on days 1 and 15 of each cycle (or as divided
`closes on days 1 and 2 and days 15 and 16 of each cycle). The recruitment period
`for the study was approximately 4 years and the follow-up period ended 1 year
`after the last enrolled patient completed treatment.
`
`

`

`3.1.2 Study Objectives
`
`The objective of this study was to demonstrate superior efficacy of bendamustine
`compared to chlorambucil in the initial treatment of patients with CLL in Binet
`stage B or Binet stage C requiring treatment.
`
`The primary endpoints of this study were to compare overall response rate (OR)
`and progression-free survival (PF S) between the bendamustine group and the
`chlorambucil group.
`
`The secondary endpoints of the study were as follows:
`. time to progression (TTP)
`. duration of response
`. overall survival (OS)
`. infection rate
`
`. quality of life
`. toxicities
`
`3.1.3 Efficacy Endpoints
`
`The primary efficacy endpoints were overall response rate (OR) and
`progression-free survival (PFS) assessed for the intent—to-treat (ITT) population
`using adjudicated responses and dates of progression from the ICRA.
`
`Overall response rate (ORR) was defined as the proportion of patients in each
`treatment group with a best response of CR, nPR, or PR to treatment.
`
`Progression—free survival (PFS) was defined as the time from randomization to
`progressive disease (PD) or death for any cause, whichever occurred first. The
`primary analysis of progression-free survival was based on the ICRA adjudicated
`responses and adjudicated event time points.
`
`OR was to be determined using the NCI- WG criteria for response and these
`were detailed in the protocol as follows:
`
`Complete response/remission (CR)
`
`CLL response was considered a CR if all of the following criteria were met for at
`least 8 weeks:
`
`- enlarged lymph nodes no longer detectable by palpation (x-ray or ultrasound
`were optional)
`- absence of hepatomegaly or splenomegaly confirmed by palpation (computed
`- tomography (CT) and ultrasound were optional)
`- no disease symptoms (ie, B symptoms) present
`
`10
`
`

`

`- lymphocytes of 4.0x] 09/L or less
`- neutrophils of 1.5x109/L or more
`- platelets greater than 100x109/L
`- Hgb greater than 11 g/dL (without blood transfusion)
`- Bone marrow biopsy (histology and cytology) was to be performed 8 weeks
`after meeting the above criteria. The bone marrow must have been at least
`normocellular for age with less than 30% lymphocytes.
`
`Nodular partial response/remission gnPR)
`
`Patients fulfilling all of the above criteria for CR with lymphocytes less than 30%
`in the bone marrow sample but still showing focal infiltration were assessed as
`having a response of nPR. These patients seem to have a shorter PFS than patients
`with confirmed CR and, therefore, were to be documented and analyzed
`separately.
`
`Partial response/remission (PR)
`
`CLL response was considered a PR if the following criteria were met for at least
`8 weeks:
`
`- at least a 50% decrease in peripheral blood lymphocyte counts from the
`pretreatment baseline value
`and at least 1 of the following 2 criteria:
`- at least a 50% reductiQn of enlarged lymph nodes (total of affected lymph
`nodes)
`- a 50% reduction of hepatomegaly and/or splenomegaly (if enlarged at baseline)
`and at least 1 of the following 3 criteria
`— neutrophil count 1 .5x1 09/L or more or 50% improvement compared with the
`baseline value
`
`- platelet count greater than 1 00x1 09/L or 50% improvement compared with the
`baseline value
`.
`
`- Hgb greater than 11 g/dL or 50% improvement compared with the baseline
`value (without a blood transfusion)
`
`Progressive disease (PD)
`
`A patient had PD if at least 1 of the following criteria was met:
`- at least a 50% lymph node enlargement (from the nadir) (total of enlargement of
`at least 2 lymph nodes) (one of the enlarged lymph nodes was to have a
`diameter of at least 2 cm) on 2 consecutive occasions at least 2 weeks apart
`and/or new palpable lymph nodes
`- at least a 50% increase (from baseline) in liver or spleen size, as determined by
`measurements under the respective costal arch; occurrence of definite
`hepatomegaly or splenomegaly that had not previously been detectable
`
`11
`
`

`

`- at least a 50% increase in absolute lymphocyte count (ALC) (from the nadir) to
`At least 5xio9/L
`
`- transformation to a more aggressive histology (Richter or PLL with more than
`55% pro lymphocytes)
`
`Stable Disease (SD)
`
`A patient had SD if CR, nPR, PR, and PD criteria were not met.
`
`The ICRA were to measure the patient's response to treatment by response criteria
`outlined above, based on a review of clinical data listings provided by
`Ribosepharrn. At the ICRA meeting on 24 and 25 August 2006, the ICRA
`adopted the following conventions:
`
`- Disease progression was achieved if at least 1 parameter worsened by 50%
`compared to the best response during study conduct.
`- For the evaluation of hepatomegaly and splenomegaly, results from palpation
`were used. Only ifno palpation data were available, were data from imaging
`techniques used for the response assessment.
`- Calculations of the reduction in lymph node size took into account all enlarged
`nodes reported.
`- Patients who had CR for all other parameters but had no bone marrow biopsy
`were considered PR. Patients with PR or CR but missing response confirmation
`due to the date of the data cut-off on 27 February 2006 were classified as
`"unconfirmed" and entered the analysis as nonresponders. Patients who had PR
`or CR but progressed prior to the first follow—up visit were considered
`nonresponders.
`- Since no threshold for baseline ALC was given in the inclusion criteria the
`ICRA agreed to consider patients with a baseline ALC of less than 5x 109/L to
`be eligible. These patients may have had a diagnosis of SLL and this disease is
`now recognized as the same disease as CLL but at a different stage.
`
`3.1.4 Sample Size Considerations
`
`The previous studies suggested that the anticipated effects of bendamustine
`versus chlorambucil might be an OR of approximately 60% versus 30% (initial
`primary endpoint) and a median PFS of approximately 20 months versus 14
`months (second primary endpoint), respectively. Under these assumptions, the
`sample size required to provide 80% power with a 2—tailed test at a=0.05 was
`estimated at 42 patients per arm for the first primary endpoint of OR, and at a
`total of 326 patients for the second primary endpoint of PFS. Both of these
`estimates were based on a fixed sample design with a single primary endpoint and
`no interim analyses. This study used a 5-stage adaptive standard group sequential
`design. The protocol provided for a maximum of 4 interim analyses and 1 final
`
`12
`
`

`

`efficacy analysis. When the proposed 5-stage adaptive standard group se