CENTER FOR DRUG EVALUATION AND
`RESEARCH
`
`APPLICA TION NUMBER:
`
`21-372/S008/S010
`
`STATISTICAL REVIEW! S)
`
`

`

`
`
`US. Department of Health and Human Services
`Food and Drug Administration
`Center for Drug Evaluation and Research
`Office of Translational Science
`Office of Biostatistics
`
`STATISTICAL REVIEW AND EVALUATION
`CLINICAL STUDIES
`
`NDA/Ser Nr:
`
`21372/8008;
`
`('0) (4)
`
`DATE RECEIVED:
`PDUFA DATE:
`DRUG NAME:
`INDICATION:
`
`APPLICANT:
`REVIEW PRIORITY:
`
`April 27, 2007
`March 4, 2008
`Aloxi (palonosetron HCL) Intravenous Injection
`Prevention of postoperative nausea and vomiting
`following elective abdominal or gynecological
`laparoscopic surgery
`Helsinn Healthcare SA
`Standard
`
`BIOIVIETRICS DIVISION:
`STATISTICAL REVIEWER:
`
`Division of Biometrics 3 (HEB-725)
`Wen-Jen Chen, PhD
`
`CONCURRING REVIEWER:
`
`Mike Welch, PhD, Dep. Dir., DB3
`
`MEDICAL DIVISION:
`CLINICAL TEAM:
`PROJECT MANAGER:
`
`Gastroenterology Products (HFD-180)
`Nancy Snow, MD, Hugo Gallo—Torres, MD (TL)
`Jagjit Grewal, MPH
`
`STATISTICAL KEYWORDS:
`
`Clinical studies; NDA review; Multiplicity
`adjustments; Baseline adaptive randomization.
`
`(b) (4)
`
`

`

`TABLE OF CONTENTS
`
`1.0
`
`EXECUTIVE SUMMARY OF STATISTICAL FINDINGS
`
`Conclusions and Recommendations
`Brief Overview of Clinical Studies
`
`Statistical Issues and Findings
`1.3.1
`Study PALO-04-06
`1.3.2
`Study PALO-O4-07
`
`2.0
`
`INTRODUCTION
`
`2.1 Overview
`
`2.1.] Outline for Study PALO-04-O6
`2.1.2 Overview for Study PALO—04-07
`
`2.2 Data Sources
`
`3.0
`
`STATISTICAL EVALUATION
`
`3.1 Evaluation of Efficacy for Study PALO-04-06
`3.1.1
`Study Design and Endpoints
`3.1.2
`Statistical Methodologies
`3.1.3 Applicant’s Efficacy Analysis
`
`and Conclusion
`
`19
`
`3.1.3.1 Co-primary Endpoint Analysis
`3.1.3.2 Secondary Endpoint Analysis
`3.1.4 Statistical Reviewer’s Analysis and Comments
`
`3.2
`
`Evaluation of Safety for Study PALO—04-O6
`
`4.0
`
`FINDINGS IN SPECIAL/SUBGROUP POPULATIONS
`
`4.1 Gender, Race, and Age for Study PALO-04-06
`4.2 Other Special/Subgroup Populations
`
`5.0
`
`SUMMARY AND CONCLUSIONS
`
`5.1 Statistical Issues and Collective Evidence
`
`5.1.1
`5.1.2
`
`Study PALO-04-06
`Study PALO-O4-07
`
`5.2 Conclusions and Recommendations
`
`oxA-hww
`
`OO\]O\
`
`10
`
`10
`10
`13
`
`19
`22
`28
`
`35
`
`35
`
`35
`37
`
`37
`
`37
`37
`38
`
`38
`
`

`

`1.0
`
`EXECUTIVE SUMMARY OF STATISTICAL FINDINGS
`
`1.1
`
`Conclusions and Recommendations
`
`Based on review of the pivotal Study Palo—04-06 and the supportive Study Palo-04-07,
`palonosetron 0.075 mg is supported by substantial evidence of efficacy for use in prevention of
`postoperative nausea and vomiting for the first 24 hour postoperative observation period
`following elective abdominal or gynecological laparoscopic surgery.
`
`1.2
`
`Brief Overview of Clinical Studies
`
`Two phase 3 Studies PALO-04-06 and PALO-04-07 were submitted by the applicant to support
`Aloxi for the prevention of postoperative nausea and vomiting (PONV)
`(b) (4)
`(b) (4)
`
`Supplement 008 refers to the PONV indication
`(b) (4)
`
`(D) (4)
`
`(b) (4)
`
`For the PONV indication, due to partial imhlinding of treatment codes, Study PALO-04-07 is
`considered a supportive study, and Study PALO-04-06 is considered a single principal study and
`thus required to show substantial evidence of efficacy, with the uncompromized data from the
`PALO-04-07 trial providing supportive evidence. However, based on that data, the applicant’s
`efficacy analysis for study PALO—04-07 failed to show effectiveness of palonosetron over
`placebo for either the 0-24 or 24-72 hour periods for any of the dose groups (0.025 mg, 0.05 mg,
`and 0.075 mg). Thus the data from Study .PALO-04-07 do not provide positive support for the
`pivotal Study PALO-04—06. (For more detail review of Study 07, refer to Section 2.1.2.)
`
`Study PALO-04-06 was designed as a randomized, double-blind, multi-center, parallel group,
`placebo-controlled phase 3 trial to evaluate the efficacy and safety of single i.v. doses of
`palonosetron (0.025 mg, 0.050 mg, or 0.075 mg) versus placebo for the prevention of PONV
`from 0 to 24 hours and 24 to 72 hours in the post-operative period in male and female outpatients
`undergoing elective laparoscopic abdominal or gynecological surgery with general anesthesia. In
`this study, outpatients were defined as those expected to go home on the same day as surgery (no
`overnight stay). All laparoscopic procedures took place under general anesthesia.
`
`(b) (4)
`
`(b) (4)
`
`(b) (4)
`
`(b) (4)
`
`(b) (4)
`
`

`

`4
`
`The population for this study consisted of eligible male or female outpatients with age 2 18,
`undergoing elective laparoscopic abdominal or gynecological surgery, who met the inclusion and
`not the exclusion criteria for the study and gave their informed consent to participate in the
`study.
`
`The duration of this study was 12 and one-half months: first patient enrolled on 23 May 2005
`and last patient completed on 12 June 2006. A total of 43 investigators from two countries (USA
`- and Romania) participated in the study: 36 from USA and 7 from Romania.
`
`The study drug was administered as a single i.v. dose of palonosetron or placebo for the
`prevention of PONV immediately (no more than 5 minutes) prior to induction of anesthesia
`(Study Day 1). In addition, efficacy and safety were assessed on the day of study drug
`administration at Visit 2 (Study Day 1) and at a final visit (Study Days 6 to 10) following the
`surgical procedure. The primary endpoint for this study was Complete Response (CR) defined
`as the absence of nausea and vomiting during the specified period.
`
`The study had two co-primary objectives as follows:
`
`1. To compare the effect of a single IV dose of palonosetron (0.025 mg, 0.050 mg or 0.075 mg)
`versus a single IV dose of placebo on CR during the first 24-hour postoperative observation
`period.
`2. To compare the effect of a single IV dose of palonosetron (0.025 mg, 0.050 mg or 0.075 mg)
`versus a single IV dose of placebo on CR during the first 24-72 hour postoperative observation
`period.
`
`A total of 639 patients were screened for this study, of which 574 patients were randomized into
`one of the four treatment arms: single i.v. doses of palonosetron (0.025 mg, 0.050 mg, or 0.075
`mg) or placebo. A total of 547 patients were treated. The applicant used an adaptive
`randomization method to stratify and assign subjects to the four treatment groups in a 1:1:121
`ratio.
`
`1.3
`
`Statistical Issues and Findings
`
`1.3.1
`
`Study PALO-04-06
`
`For the first 24 hour observation period, superiority of palonosetron over placebo was
`demonstrated for the 0.05 and 0.075 mg dose groups showing efficacy evidence in prevention of
`postoperative nausea and vomiting following elective abdominal or gynecological laparoscopic
`surgery. However, the efficacy comparisons performed by the applicant for the 0.025 mg dose
`for the 0-24 hour period did not show superiority; and for the 24-72 hour postoperative
`observation period, all three palonosetron doses (0.025 mg, 0.05 mg, and 0.075 mg) failed to
`demonstrate superiority to placebo.
`
`

`

`The following comments on the efficacy assessments are for the superiority of palonosetron 0.05
`mg and 0.075 mg to placebo based upon the primary endpoint - complete response (no nausea,
`no vomiting) for the first 24 hour postoperative observation period.
`
`5
`
`First, the estimated proportion of complete response using data from baseline adaptive
`randomization may not be an unbiased estimate of the population proportion. It follows
`that the estimated proportion should be used with caution in estimating the effect size of
`study drug versus placebo. In general, simple randomization with permuted blocks is
`currently the favored allocation method in clinical trials. Accordingly, based upon the
`number of 574 patients, the applicant should have applied simple randomization with
`random permuted blocks to allocate patients to different treatment groups and applied the
`proposed logistic regression analysis method to adjust the covariate effect on the efficacy
`comparisons assessed by the two co—primary endpoints (complete response in the 0-24
`and 24-72 hours of postoperative periods).
`
`for palonosetron
`Second, based upon the efficacy comparison by investigate-site,
`0.075mg, the proportions of complete response in most sites are much greater than that in
`the placebo group. No particular sites are identified to have unusually large proportions
`of complete response to dominate the superiority of palonosetron 0.075 mg to placebo.
`However, for palonosetron 0.05mg, although the proportions of complete response in
`most sites are greater than that of placebo, it seems unusual that the complete responses
`for all nine patients in Site 182 are assessed as success. The concern of 9 out of 9 (100%)
`patients in Site 182 of palonosetron 0.05 mg assessed as success in complete response
`can not be ruled out. In addition, it is also noted that the p-value of testing the superiority
`of palonosetron 0.05 mg to placebo is on the border-line close to the nominal significance
`level proposed by the applicant’s multiplicity adjustment method.
`
`Thirdly, based upon the results of the sensitivity analyses, one learns that in the efficacy
`analysis of palonosetron 0.05 mg to placebo, a small Site 182 (total patients of this site
`only covered 5% patient of the study) has influence on the superiority testing of
`palonosetron 0.05 mg. It is also noted that the superiority of palonosetron 0.05 mg to
`placebo is no longer held when the status of the complete responses of two patients
`switching from success to failure. Accordingly, the superiority of palonosetron 0.05 mg
`to placebo is not robust. In addition, since the supportive Study PALO-04-07 did not
`replicate the superiority of palonosetron 0.05 mg to placebo using primary analysis data
`set (PFAS), palonosetron 0.05 mg is considered not supported by substantial evidence in
`prevention of postoperative nausea and vomiting following elective abdominal or
`gynecological laparoscopic surgery.
`
`Finally, for the efficacy assessment of palonosetron 0.075 mg during the first 24 hour
`postoperative observation period, based upon the applicant’s logistic regression analysis
`and this reviewer’s non-parametric covariate analysis, the superiority of palonosetron
`0.075 mg versus placebo is demonstrated and is a robust result. In addition, the secondary
`endpoint analyses on the complete control, emetic episodes, severity nausea, and quality
`
`

`

`6
`
`of life indicated that Palonosetron 0.075 mg are numerically better than placebo. As a
`consequence, the superiority of palonosetron 0.075 mg to placebo can be considered
`supported by substantial evidence in use of the prevention of postoperative nausea and
`vomiting during the first 24 hour postoperative observation period.
`
`1.3.2
`
`Study PALO-04-O7
`
`Due to the issue of unblinding the treatment codes and based upon the Agency’s response
`to this issue, Study PALO-04-07 is treated as a supportive study. In addition, in order to
`respond to the Agency’s concern regarding data integrity, the applicant specified the
`primary full analysis set (PFAS) excluding the 175 patients involved in the unblinding
`problem as the primary analysis data set.
`
`However, using this PFAS population, the applicant’s efficacy analysis showed that for
`both periods of 0-24 hours and the 24-72 hours, comparisons of the single dose groups of
`palonosetron with placebo did not result
`in significant differences for any of the
`palonosetron treatment groups (0.025, 0.05, and 0.075 mgs) versus placebo after
`multiplicity adjustments proposed by the applicant. In this reviewer’s opinion, data from
`Study PALO-04—07 do not support the efficacy results of the pivotal Study PALO-04-06.
`
`2.0
`
`INTRODUCTION
`
`2.1
`
`Overview
`
`In the introduction of the clinical study report for Study PALO-04-07, the applicant made the
`following comments with regard to Aloxi:
`
`is a frequent complication of surgery under general
`Postoperative nausea and vomiting (PONV)
`anesthesia, with considerable medical and economic impact, and associated with high levels of patient
`discomfort and dissatisfaction. To many patients, PONV is a distressing adverse event (AB), which is
`reportedly feared more than postoperative pain.
`(b) (4)
`(b) (4)
`
`(b) (4)has been directed at blocking neurotransmitter receptors
`Research into preventing PONV
`located either centrally or peripherally such as histamine Hi, dopamine D2, muscarinic acetylcholine, and
`5-hydroxytryptamine sub-receptor 3 (5-HT3). The 5-HT3 antagonists such as ondansetron, dolasetron and
`granisetron have proven effective for the prevention and treatment of PONV, with minimal adverse
`effects and no dose-related sedation or extrapyramidal reactions and no effect on vital signs. Patients at
`substantial risk of PONV are expected to benefit from multiple anti-emetic interventions including a 5—
`HT3 antagonist. Palonosetron hydrochloride (Aloxi) is a potent and selective 5-HT3 receptor antagonist
`that blocks the binding of serotonin to 5-HT3 receptors. The compound has a higher receptor binding
`affinity and a longer plasma half-life (40 hours) compared to ondansetron, dolasetron, and granisetron.
`
`Two phase 3 studies (PALO-04-06 and PALO-04—07) were submitted by the applicant to support
`Alnxi in the nreventinn nfnns‘rnnerative nausea and vnmitino (PONV\
`(b) (4)
`
`(b) (4)
`
`(b) (4)
`
`(b) (4)
`
`(b) (4)
`
`(b) (4)
`
`

`

`(b) (4)
`
`For the PONV indication, due to the issue of partial unblinding of treatment codes, Study PALO-
`04-07 was to be considered as a supportive study. This would imply that a positive finding in the
`pivotal Study PALO—04-06, could then be supported by the remaining data in PALO-04-07 trial.
`
`2.1.1 Overview of Study PALO-04—06
`
`The study was designed as a randomized, double—blind, multi—center, parallel group, balanced,
`stratified, placebo-controlled phase III trial to evaluate the efficacy and safety of single i.v. doses
`of palonosetron (0.025 mg, 0.050 mg, or 0.075 mg) versus placebo for the prevention of PONV
`from 0 to 24 hours and 24 to 72 hours in the post—operative period in male and female outpatients
`undergoing elective laparoscopic abdominal or gynecological surgery with general anesthesia. In
`this study, outpatients were defined as those expected to go home on the same day as surgery (no
`overnight stay). All laparoscopic procedures took place under general anesthesia.
`
`The population for this study consisted of eligible male or female outpatients with age 2 18,
`undergoing elective laparoscopic abdominal or gynecological surgery, who met the inclusion and
`not the exclusion criteria for the study and gave their informed consent to participate in the
`study.
`
`The duration of this study was 12.5 months: first patient enrolled on 23 May 2005 and last
`patient completed on 12 June 2006. A total of 43 investigators from two countries (USA and
`Romania) participated in the study: 36 from USA and 7 from Romania.
`
`The study drug was administered as a single i.v. dose of palonosetron or placebo for the
`prevention of PONV immediately (no more than 5 minutes) prior to induction of anesthesia
`(Study Day 1).
`In addition, efficacy and safety were assessed on the day of study drug
`administration at Visit 2 (Study Day 1) and at a final visit (Study Days 6 to 10) following the
`surgical procedure. The primary endpoint for this study was Complete Response (CR) defined
`as the absence of nausea and vomiting during the specified period.
`The study had two co—primary objectives, as follows:
`1. To compare the effect of a single IV dose of palonosetron (0.025 mg, 0.050 mg or 0.075 mg)
`versus a single IV dose of placebo on CR during the first 24-hour postoperative observation
`period.
`
`(b) (4)
`
`

`

`8
`
`2. To compare the effect of a single IV dose of palonosetron (0.025 mg, 0.050 mg or 0.075 mg)
`versus a single IV dose of placebo on CR during the first 24-72 hour postoperative observation
`period.
`
`A total of 639 patients were screened for this study, of which 574 patients were randomized into
`one of the four treatment arms: single i.v. doses of palonosetron (0.025 mg, 0.050 mg, or 0.075
`mg) versus placebo. A total of 547 patients were treated. The sponsor used an adaptive
`randomization method to stratify and assign subjects to the four treatment groups in a 1:1:1:1
`ratio.
`
`2.1.2 Overview of Study PALO-04-07
`
`PALO-04-07, similar in design to Study PALO-04-06, was conducted in Europe including
`Germany, Poland, and the Czech Republic. This study enrolled 674 PONV patients from 28
`recruiting centers, and enrollment was completed in April 2006. Two separate potential
`unblinding incidents occurred during this study:
`(1) A clinical label supply issue resulted in
`possible unblinding of the first 130 patients enrolled; and (2) after completion of the clinical
`phase of the study, a problem was found with vial labels at the Germany sites, which resulted in
`possible unblinding of 178 subjects. Meetings were held with the sponsor to discuss the
`unblinding problems and seek resolution.
`(Refer to Agency Meeting Minutes dated February 1,
`2006 and October 24, 2006, respectively.)
`
`In partial resolution of the first incident, the applicant enrolled an additional 130 subjects into the
`study, and redefined the efficacy analysis data based to exclude those subjects potentially
`unblinded. Regarding the second incident, the applicant indicated that in Germany, 261 patients
`enrolled at 13 sites participated in the study, and of these, 178 subjects were potentially affected
`by an unblinding of treatment codes. That is, drug accountability forms were erroneously
`affixed with the adhesive tear—off treatment vial label. The applicant fiirther indicated that this
`active treatment label did not identify the active treatment doses being evaluated, namely, 0.025
`mg, 0.05 mg or 0.075 mg, but it identified whether the treatment was palonosetron or placebo.
`The concern was that the clinical staff responsible for signing the drug accountability forms may
`have become unblinded when they signed the forms.
`
`In order to reduce the impact of the second unblinding issue on the efficacy assessments, the
`Agency responded that it had significant concerns regarding the data integrity from the affected
`clinical sites.
`It was agreed that patients from the 13 sites in Germany should be excluded from
`the full efficacy analysis data set. The Agency then, agreed that a positive finding in the PALO-
`04—06 study, if robust, could then be supported by data from the remaining subjects who were
`not involved in the unblinding of treatment codes.
`'
`
`In order to explore the impact of the unblinding issue on the efficacy analysis for the supportive
`Study PALO—04-O7, this reviewer calculated the proportions of complete responses for the four
`treatment groups (three palonosetron treatments plus placebo). Since none of the efficacy for the
`three palonosetron treatments (palonosetron 0.025 mg, 0.05 mg and 0.075 mg) in the pivotal
`
`

`

`study PALO—04-06 showed superior to that of placebo during the 24-72 hour post—operative
`period, following the comments regarding the role of the supportive study stated in the meeting
`minutes, efficacy data from the supportive study for the 24—72 hour post-operative period can not
`be used to support the pivotal study. Accordingly, the proportions of complete responses for this
`supportive study are calculated only for the postoperative period of 0-24 hours. Table 2.1.2
`presents the proportions of complete responses during 0-24 hour postoperative period using
`PFAS set for the supportive Study PALO-04-07.
`
`9
`
`Table 2.1.2 (Reviewer’s) Proportions of complete responses during 0-24 hour postoperative period
`using PFAS set
`175 atients involved in unblindin_ uroblem
`
`
`ENDPOINT
`Complete response in first
`24
`hour
`postoperative
`observation mod.
`
`
`PLACEBO
`n/N (%)
`
`»
`12/48 (26.0%)
`
`
`
`
`
`PALT 0.025 MG
`PAL 0.05 MG
`PAL 0.075 MG
`n/N (%)
`n/N (%)
`n/N (%)
`
`13/48 (27.1%)
`
`18/41 (44.0%)
`
`23/40 (58.0%)
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`369 atients (PFAS) not involved in unblindin; roblem
`
`
`
`
`
`
`ENDPOINT
`Complete response in first
`24
`hour
`postoperative
`
`observation eriod.
`
`T: Palonosetron
`
`PLACEBO
`n/N (%)
`
`PALT 0.025 MG
`n/N (%)
`
`PAL 0.05 MG
`n/N (%)
`
`37/90 (41.0%)
`
`
`
`50/88 (57.0%)
`
`46/96 (48.0%)
`
`PAL 0.075 MG
`n/N (%)
`
`53/95 (56.0%)
`
`
`
`
`it is noted that the percentage of complete responders for placebo using
`From Table 2.1.2,
`patients involved in the unblinding is 15% less than that of patients not
`involved in the
`unblinding. The applicant indicated that due to the unblinding of treatment codes, the clinical
`staff may be able to identify whether the treatment is palonosetron or placebo. It appears that the
`result shown by this table for placebo lends support to that assessment. In addition, this result
`also reinforces the Agency’s concern regarding loss of data integrity due to the unblinding.
`
`In addition, using the PFAS population (data not involved in the unblinding), the applicant’s
`efficacy analysis showed that for study PALO-04-07, during the periods of 0-24 hours and the
`24-72 hours, comparisons of the single dose groups of palonosetron with placebo resulted in no
`significant differences
`for any of the palonosetron treatment groups, after multiplicity
`adjustments as proposed by the applicant were applied to adjust the p-values. It follows that data
`from Study PALO-04-07 does not lend positive support for the pivotal Study PALO-04—06.
`Therefore, Study ~07 is not further detailed in this review.
`'
`
`2.2
`
`Data Sources
`
`To assess the clinical efficacy of Aloxi in prevention of postoperative nausea and-vomiting, I
`reviewed the NDA submission hard copies Volumes 211 to 332, dated April 27, 2007.
`In
`addition,
`I reviewed the data sets and documents submitted by the applicant and located at
`\\CDSESUB1\NONECTD\N21372\S_008\2007-12-06.
`
`

`

`10
`
`3.0
`
`STATISTICAL EVALUATION
`
`3.1
`
`Evaluation of Efficacy for Study PALO-04-06
`
`3.1.1
`
`Study Design and Endpoints
`
`The study was designed as a randomized, double-blind, multi—center, parallel group, balanced,
`stratified, placebo-controlled phase 3 trial to evaluate the efficacy and safety of single i.v. doses
`of palonosetron (0.025 mg, 0.050 mg, or 0.075 mg) versus placebo for the prevention of PONV
`from 0 to 24 hours and 24 to 72 hours in the post—operative period in male and female outpatients
`undergoing elective laparoscopic abdominal or gynecological surgery with general anesthesia. In
`this study, outpatients were defined as those expected to go home on the same day as surgery (no
`overnight stay). All laparoscopic procedures took place under general anesthesia.
`
`The population for this study consisted of eligible male or female outpatients with age 2 18,
`undergoing elective laparoscopic abdominal or gynecological surgery, who met the inclusion and
`not the exclusion criteria for the study and gave their informed consent to participate in the
`study.
`
`The duration of this study was 12.5 months: first patient enrolled on 23 May 2005 and last
`patient completed on 12 June 2006. A total of 43 investigators from two countries (USA and
`Romania) participated in the study: 36 from USA and 7 from Romania.
`
`_
`
`The study drug was administered as a single i.v. dose of palonosetron or placebo for the
`prevention of PONV immediately (no more than 5 minutes) prior to induction of anesthesia
`(Study Day 1).
`In addition, efficacy and safety were assessed on the day of study drug
`administration at Visit 2 (Study Day 1) and at a final visit (Study Days 6 to 10) following the
`surgical procedure.
`
`Patients were contacted on three occasions by telephone (Days 2, 4 and 15) to check on any AEs.
`[Patient diaries were used to record emetic episodes, use of rescue medication, the severity of
`nausea, the severity of postoperative pain experienced by patients and Quality of Life (QoL).
`QoL evaluations for nausea and emesis were assessed according to the Osoba questionnaire,
`which was partially implemented in the study to investigate the interference of nausea and
`vomiting on patients' daily activities. Rescue medication for the treatment of nausea and
`vomiting after surgery, with the exception of palonosetron and droperidol, was permitted at the
`discretion of the investigator.
`The investigational drug and the comparator were administered in a blind fashion according to
`specific procedures described below (Table 3.1.1.1). The pre-filled syringes containing the study
`drug supplied by the unblinded research pharmacist or unblinded designated person at site were
`to be administered within 3 hours of preparation on Study Day 1.
`
`

`

`11
`
`Table 3.1.1.1 (Applicant’s) Treatment Group
`Gran
`vDose
`Volume for IV incl-1&3}? A-.. W _.__V
`Group 1
`0.025 mg
`0.5 mL aiiquot of paionosetron + 1.5 mL saline
`Group 2
`0.050 mg
`1.0 mL aliquot of paionosetron + 1.0 mL saline
`Group 3
`0.075 mg
`1.5 mL aliquot of palonosetron + 0.5 mL saline
`Gmup 4
`Placebo
`2.0 mL saline
`
`(b) (4)
`
`prior to commencing the study
`A kit randomization list
`to allow for blinding 01 mt: lour LI'cuuucuL groups. we UJUUK size was 4. According to this
`randomization list, sealed cartons containing the study drugs were prepared and supplied to the
`unblinded research pharmacist or designated person at the investigative sites. Patients meeting
`the inclusion and exclusion criteria were randomized to receive a single IV dose for one of the
`four treatments (palonosetron: 0.025 mg, 0.050 mg, 0.075 mg, or placebo).
`
`The applicant indicated that randomization was stratified by gender, history of PONV and/or
`currently prone to motion sickness, and smoking status. The stratified randomization scheme
`produced eight strata. Considering each stratum, patients who met the inclusion and exclusion
`criteria were assigned to one of four I.V. treatments, following an adaptive randomization
`algorithm. Randomization was performed to assign patients to the four treatment groups in the
`ratio of 1:1:1:1. [For full details of the adaptive randomization algorithm, see Appendix I of the
`protocol, Section 16.1.1.]
`
`The time frame of the study included Visit 1 (Screening/Baseline) from Study Day -7 to 0, Visit
`2 (Randomization/Treatment) at Study Day 1, Study Days 2 and 4 scheduled for Telephone
`Contact 1 and 2, Final Visit (Visit 3) from Study Day 6 to 10, and Study day 15 scheduled for
`third telephone contact.
`
`The efficacy of I.V. palonosetron or placebo was assessed by measuring complete response (CR,
`defined as no emetic episode, no rescue medication), complete control (CC, defined as CR and
`no more than mild nausea), number of emetic episodes, patients with emetic episodes, time to
`first emetic episode, time of administration and need for rescue medication, time to treatment
`failure (TTF) (time to first emetic episode, or administration of rescue medication, whichever
`occurred first), time to first fluid intake, time to first solid meal intake, and severity of nausea.
`
`For the efficacy parameters recorded up to 72 hours after T0, T0 was defined as the time when
`the patient woke up and was able to respond to verbal commands post operatively.
`
`The study had two co-primary objectives, as follows:
`
`1. To compare the effect of a single lV dose of palonosetron (0.025 mg, 0.050 mg or 0.075 mg)
`versus a single IV dose of placebo on CR during the first 24-hour postoperative observation
`period.
`
`(b) (4)
`
`

`

`l2
`
`2. To compare the effect of a single IV dose of palonosetron (0.025 mg, 0.050 mg or 0.075 mg)
`versus a single IV dose of placebo on CR during the first 24—72 hour postoperative observation
`period.
`
`Patients with missing or incomplete data were handled as follows (worst case scenario):
`
`A patient with missing date regarding emetic episodes and/or rescue medication was
`defined as non-responder for each time interval.
`‘
`A patient for whom the duration of retching or relaxation could not be calculated, e.g.
`due to incomplete onset or end time, was defined as non-responder for the affected time
`interval.
`
`If no retching and no vomiting were ticked, but date and/or time was provided, the
`episode was considered as a vomiting episode.
`A patient who discontinued during the first 24 hours of the 72 hours observation period
`was defined as a non-responder for both primary efficacy variables.
`A patient who discontinued during the 24—72 hours was defined as a non-responder for
`the second primary efficacy variable.
`
`The following parameters were evaluated as secondary efficacy measures during the 72 hours
`postoperative observation period:
`
`CR at 0-2 hours, 0-6 hours, 2-6 hours, 6—24 hours, 24-48 hours, 48-72 hours, 0-48 hours
`and 0-72 hours
`
`CC (CR and no more than mild nausea) at 0-24 hours, 24—72 hours, 0-2 hours, 0—6 hours,
`2-6 hours, 6-24 hours, 24—48 hours, 48-72 hours, 0-48 hours and 0-72 hours
`
`The number of emetic episodes at 0-24 hours, 24-72 hours, 0-2 hours, 0-6 hours, 2-6
`hours, 6-24 hours, 24-48 hours, 48-72 hours, 0-48 hours and 0-72 hours
`
`Severity of nausea, measured on a 4-point Likert scale at 0-24 hours, 24-72 hours, 0—2
`hours, 0-6 hours, 2-6 hours, 6—24 hours, 24-48 hours, 48-72 hours, 0-48 hours and 0-72
`hours
`
`Time to first emetic episode
`Time to first administration and need for rescue medication
`
`time to first emetic episode or
`TTF (Time to Treatment Failure:
`administration of rescue medication, whichever occurred first)
`Time to first fluid intake
`
`time to first
`
`Time to first solid meal intake
`
`In addition to the efficacy assessments, the following endpoints for quality of life were also
`collected in the patient diary:
`
`Interference of nausea and vomiting symptoms on patients' daily life activities.
`
`Severity of post-operative pain.
`
`

`

`13
`
`The sample size was based on the assumption of a responder rate of 60% in the palonosetron
`group and a responder rate of 40% in the placebo group. Using a two-sided test of difference,
`with a = 0.0166 (obtained as Type I error divided by the number of comparison = 0.05/3) and B
`= 0.2 for each comparison, the sample size was estimated at 131 evaluable patients per group.
`The number was rounded up to a planned sample size of 136 patients per group. For the multiple
`secondary endpoints comparisons, the sponsor did not specify a procedure to control the Type I
`error rate.
`
`3.1.2 Statistical Methodologies
`
`The applicant indicated that analysis of data in this study was based on the final protocol dated
`28 February 2005 and protocol amendments 1 and 2 dated 18 October 2005 and 06 April 2006,
`respectively, and the final statistical analysis plan (SAP) dated 06 September 2006; A blind
`review meeting for the definition of analysis sets was held on 07 September 2006. The database
`was closed on 11 October 2006 and data were unblinded on 12 October 2006.
`
`The co-primary efficacy hypotheses relevant to the co-primary endpoints were that at least one
`dose of palonosetron was superior to placebo, considering the CR rate, in the 0-24 hour period
`and in the 24-72 hour period. These were tested using a closed testing procedure. The 0-24 hour
`period was tested first; if at least one comparison result was statistically significant, the 24-72
`hour interval can be tested, using the same method and the same alpha as further detailed.
`
`For each of the two primary efficacy variables three comparisons were considered:
`0
`To assess the superiority of a single i.v. dose of palonosetron 0.025 mg over placebo.
`0
`To assess the superiority of a single i.v. dose of palonosetron 0.050 mg over placebo.
`0
`To assess the superiority of a single i.v. dose of palonosetron 0.075 mg over placebo.
`
`To control Type I error for the three primary treatment comparisons (0.025 mg vs. placebo,
`0.050 mg vs. placebo and 0.075 mg -vs. placebo) a multiple type-I error level was guaranteed by
`the Holm-Bonferroni method: in order to claim a significant difference, the smallest of the three
`2-sided p-values must not exceed 0.0166 (0.05/3). If this was achieved, the second smallest p-
`value must not exceed 0.025 (0.05/2) in order to be significant, and if this was also achieved, the
`significance threshold for the third p-value was 0.05. This sequential procedure was to be
`stopped, if the respective threshold was exceeded. The procedure guarantees the multiple type-l
`error level of 0.05.
`
`Each primary test was performed using a multiple logistic regression adjusted for covariates,
`where each dose of palonosetron was compared with placebo. The covariates included into the
`model were the stratification criteria, i.e. gender, history of PONV and/or currently prone to
`motion sickness, and smoking status. Odds ratio, 95% CI for the odds ratio and p—values derived
`from the logistic regression were summarized. The same logistic regression model was used to
`compare each dose of palonosetron to each other. This was done on a descriptive level,
`i.e.
`without any further adjustment for multiplicity.
`
`

`

`14
`
`It is noted that a dynamic adaptive treatment allocation procedure including a randomization
`component was applied in this study. The applicant i