Statistical Review and Evaluation

 

 

NDA #: 21-321

Related IND #: 51,898

Applicant: Baxter Healthcare Corporation

Name of Drug: ExtranealTM (Icodextrin)

Indication: Treatment of chronic renal failure

Document reviewed: Volume 1-155

Date of submission: December 22, 2000

Statistical Reviewer: John Lawrence, Ph.D. (HFD710)

Medical Reviewer: Stephen Fredd, M.D. (HFD110)

 

 

1. Introduction

 

On October 19, 2000, a closed session of the Cardio-Renal Advisory Committee was held and the 12-month safety study RD-97-CA-131 was discussed. The protocol had required only 30-day follow-up for patients who had withdrawn from the study. The committee asked to see the results of the mortality analysis for the 12-month intended duration of treatment and for 30 days thereafter. The document under review here relates to this analysis in the addendum to the clinical report for this study.

 

2. Study Design

 

Study RD-97-CA-131 was a 52-week prospective, double-blind, randomized, active-controlled, parallel group study comparing the safety of Extraneal with dextrose for the long dwell in Continuous Ambulatory Peritoneal Dialysis (CAPD) and Automated Peritoneal Dialysis (APD) patients. 287 patients (112 control and 175 Extraneal) were enrolled in the study.

 

3. Planned Statistical Analysis

 

The original analysis of mortality required 30-day follow-up for patients who had withdrawn or completed the study. Based on the comments from the Advisory Committee, patients' status for the intended 12-month duration of the study plus 30 days thereafter was analyzed.

 

4. Results

 

The mortality status of 17 patients was not known after 360 days from the start of the study (6 control and 11 Extraneal). The status of 161 patients was not known 375 days from the start of the study (59 control and 102 Extraneal). The status of 168 patients was not known 395 days from the start of the study (63 control and 105 Extraneal). The Kaplan-Meier estimates of the probability of having follow-up at each time point for each group appears in the appendix.

 

Using the mortality data for the 12-month intended treatment period plus 1 month additional follow-up, a total of 29 patients died. This includes 20 patients from the Extraneal group (20/175 = 11.4%) and 9 patients from the control group (9/118 = 8.0%). The Kaplan-Meier estimates of the survival curves appear in Figure 4.1.

 


 


Figure 4.1 Kaplan-Meier estimates of survival curves using available data with

13-months of follow-up.

 

 

The value of the logrank statistic is not significant (p=0.305). However, insufficient evidence to rule out equality of the two curves is not the same thing as proving that there is no difference in the curves. Numerically, the estimated hazard ratio for mortality in the Extraneal group relative to the control group was 1.51 with a 95% confidence interval of (0.686, 3.30). This estimate is from a Cox regression model with one term for treatment group. In order to subjectively evaluate the two survival curves, the Medical Officer wanted to examine the 80% confidence intervals for mortality rates at each month. The choice of 80% was pre-specified according to the Medical Officer. The point estimates and lower and upper bounds of the 80% confidence intervals for each group are in Table 4.1.

 


Table 4.1 Point estimates and 80% confidence interval for survival in each group.

Month

Extraneal

Control

Lower bound

Point estimate

Upper bound

Lower bound

Point estimate

Upper bound

1

0.987

0.994

1.000

 

1.000

 

2

0.970

0.983

0.995

 

1.000

 

3

0.963

0.977

0.992

 

1.000

 

4

0.963

0.977

0.992

0.980

0.991

1.000

5

0.941

0.960

0.979

0.980

0.991

1.000

6

0.913

0.936

0.960

0.954

0.973

0.993

7

0.906

0.930

0.956

0.954

0.973

0.993

8

0.893

0.919

0.946

0.942

0.964

0.987

9

0.886

0.913

0.941

0.942

0.964

0.987

10

0.873

0.901

0.931

0.931

0.955

0.981

11

0.866

0.896

0.926

0.919

0.946

0.974

12

0.852

0.883

0.915

0.896

0.928

0.960

The 80% confidence intervals overlap except at months 5 and 9 where the upper limit of the confidence interval for the Extraneal group is 0.001 below the lower limit for the control group. The standard error (and consequently, the confidence limits) could not be estimated for the first three months in the control group because all patients survived at least 3 months in that group.

 

The amendment to the study report states that exploratory analyses (Cox regression models) were done to determine whether any baseline characteristics could explain the numerical difference in mortality between the two groups. The covariates that were studied included: age, race, gender, diabetes, time on chronic dialysis therapy, and serum albumin. The report indicates that although there was a significant relationship between some of these covariates and mortality, there was no significant treatment by covariate interaction that would help explain the numerical difference in mortality rates. This reviewer believes the purpose of the amendment is to answer a narrower question regarding mortality, viz. is there an overall difference in the mortality rates between the two groups. Also, there doesn't seem to be an adequate amount of information to answer questions about the treatment by covariate interactions. For these reasons, this will not be investigated in further depth in this review.

 

5. Conclusions

 

Since the mortality status of over half (161/289) of the patients was not known 375 days from the start of the study, this reviewer doubts that the questions raised by the Advisory Committee can be answered from the data provided. The data provided seems to indicate that there is insufficient evidence to rule out the equality of the two survival curves. Numerically, the estimated hazard ratio for mortality in the Extraneal group relative to the control group was 1.51 with a 95% confidence interval of (0.686, 3.30). Moreover, the rate of loss to follow-up in the last month is high and the Extraneal group has more patients lost to follow-up. This might induce bias in favor of the Extraneal group. Hence, the excess risk could be much higher than observed.

 

 

Appendix

 

This reviewer estimated the proportion of patients in each group that survive and the mortality status is known to the investigator at each time point. In this analysis, a month is defined as (#days from start of study)*12/365. A patient who was known to be alive x number of days from the start of the study (with no further follow-up) is counted as having an event at that time. A patient who died is counted as censored at the time of death. Figure A1 shows the resulting Kaplan-Meier estimates and Table A1 shows the actual numerical estimates for each group. Both the figure and the table illustrate that a relatively high number of patients in each group did not have follow-up for the intended 13-month period. Roughly 40% in the control group and 32% in the Extraneal group did have complete follow-up for this period of time.

 


 


Figure A1. Kaplan-Meier estimates of probability of remaining in study over time. Patients who died are counted as censored in this analysis. Months = Days*12/365.

 

 


Table A1. Kaplan-Meier estimates of proportion of patients who survive and the mortality status is known by the investigator.

 

Control group

Months from

start of study

# at risk

# of events

Survival

8.88

108

1

0.991

11.08

105

1

0.981

11.74

104

2

0.962

11.80

102

1

0.953

11.84

100

1

0.943

11.87

99

1

0.934

11.93

97

9

0.847

11.97

88

4

0.809

12.00

84

11

0.703

12.03

73

11

0.597

12.07

62

5

0.549

12.10

57

1

0.539

12.16

56

2

0.520

12.20

54

3

0.491

12.23

51

2

0.472

12.26

49

2

0.453

12.30

47

1

0.443

12.33

46

1

0.433

12.36

45

1

0.424

12.43

44

1

0.414

12.46

43

1

0.404

12.72

42

1

0.395

13.02

40

40

0.000

 


Table A1 (continued).

 

Extraneal group

Months from

start of study

# at risk

# of events

Survival

0.329

175

1

0.99429

3.058

170

1

0.98844

3.682

169

1

0.98259

11.507

154

1

0.97621

11.540

153

1

0.96983

11.704

152

1

0.96345

11.770

151

2

0.95069

11.803

149

1

0.94431

11.836

148

2

0.93154

11.901

146

1

0.92516

11.934

145

8

0.87412

11.967

136

9

0.81627

12.000

126

18

0.69966

12.033

108

14

0.60897

12.066

94

8

0.55714

12.099

86

8

0.50531

12.132

78

4

0.47940

12.164

74

5

0.44701

12.197

69

3

0.42757

12.230

66

10

0.36279

12.263

56

1

0.35631

12.296

55

2

0.34335

12.460

53

1

0.33688

12.592

52

1

0.33040

12.658

51

1

0.32392

13.019

50

46

0.02591

13.151

4

1

0.01944

13.381

3

1

0.01296

13.512

2

1

0.00648

13.644

1

1

0.00000


 

 

 

_________________________

John Lawrence, Ph.D.

Mathematical Statistician

 

 

This review consists of 1 pages of text, tables, and figures.

 

 

 

Concur: James Hung, Ph.D.

Acting Team Leader, Biometrics I

 

 

 

George Chi, Ph.D.

Division Director, Biometrics I

 

cc: NDA # 21-321

HFD-110/Dr. Lipicky

HFD-110/Dr. Fredd

HFD-110/Mr. Guzman

HFD-700/Dr. Anello

HFD-710/Dr. Chi

HFD-710/Dr. Hung

HFD710/chron

 

 

LAWRENCEJ/5945375/report.doc/4/23/2003