FDA Statistical Review and Evaluation

         

Document for the Vaccines and Related Biological Products Advisory Committee

(VRBPAC)

 

December 15, 2005

 

 

 

 

 

 

 

 

 

ZOSTAVAX™: Zoster vaccine live (Oka/Merck)

 

Indication: Prevention of herpes zoster (shingles), prevention of postherpetic neuralgia (PHN), and reduction of acute and chronic zoster-associated pain for individuals 50 years of age or older.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Sang Ahnn, Ph.D.

FDA/CBER/OBE

Introduction

 

In the application for licensure of ZOSTAVAXä, Merck submitted information from 6 randomized studies and one open-label study in which a total of ~ 20,000 subjects received zoster vaccine live (Oka/Merck).  Merck also submitted information from a ProQuadä study and a VARIVAXä study.  Protocol 004 is the pivotal phase III clinical trial for this application.  This statistical briefing document covers protocol 004.

 

 

Protocol 004

 

This was a multi-center (22 sites), double-blind, placebo-controlled, randomized study to evaluate the efficacy, immunogenicity, and safety of administration of zoster virus vaccine.  A total of ~38,500 varicella history-positive and HZ history-negative subjects 60 years of age or older were randomized (1:1 ratio, stratified by site and age group: 60 to 69 years and ≥70 years) to receive a single dose of zoster virus vaccine or placebo.

 

Initial (release) potency of the vaccine ranged from 44,000 PFU/mL to 114,184 PFU/mL.   Approximately 96% of the vaccinees received heat-aged vaccine with release doses ranging from 44,000 PFU/mL to 79,200 PFU/mL. 

 

 

 

Merck’s analyses

 

Efficacy

 

The primary endpoint for this study is the Herpes Zoster Burden Of Illness (HZ BOI) during the 6 months (Day 0 to 182) following HZ rash onset.  The HZ BOI is a severity-by-duration measure defined by the HZ incidence, severity, and duration of HZ-associated pain and discomfort.  HZ BOI (in a subject) is defined as “the area under the worst pain and discomfort response (rated on a 0 to 10 scale) versus time curve during the 6-month period following HZ rash onset in an individual subject who develops HZ.”  For an evaluable HZ case (Merck stated “The results of the Clinical Evaluation Committee review were combined with the results of the Central PCR Laboratory, as well as virus culture” to determine the evaluable case), computation of HZ BOI score is as above.  Subjects who did not develop HZ were assigned an HZ BOI score of zero.   Then Merck defines VEBOI (vaccine efficacy based on BOI) as VEBOI  = 1 - RBOI where RBOI is the ratio of mean HZ BOI (stratified by age) of the vaccine relative to the placebo group (see Appendix for details of the calculation of VEBOI). 

 

The original secondary endpoint is the incidence of PHN.

 

For regulatory submission purposes, Merck proposed to treat these two endpoints as co-primary and to define success of the study as showing efficacy in at least one of the two endpoints.  An adjustment for multiple hypothesis testing to control the overall type I error rate at the two-sided 0.05 level was implemented.

 

The two primary hypotheses are

 

“Zoster vaccine will reduce ‘Burden of Illness’ (BOI) associated with herpes zoster (HZ)”; and

 

“Zoster vaccine will reduce the incidence of postherpetic neuralgia (PHN).” 

 

The success criteria are “VEBOI (vaccine efficacy based on BOI)  > 47% and lower bound of the 95% CI  > 25%,” and “VEPHN (vaccine efficacy based on PHN incidence) > 62% and lower bound of the 95% CI  > 25%,” respectively. 

 

Merck concludes that

“Compared with placebo, the zoster vaccine reduced the BOI related to HZ pain. The estimated VEBOI was 61.1% with 95% CI of (51.1%, 69.1%)”; and

 

“Compared with placebo, the zoster vaccine reduced the incidence of PHN. The estimated VEPHN was 66.5% with 95% CI of (47.5%, 79.2%).”

 

The three additional secondary hypotheses, originally designated as tertiary hypotheses are

 

“Zoster vaccine will reduce the incidence of HZ”;

 

“Zoster vaccine will reduce the duration of HZ pain”; and

 

“Zoster vaccine will reduce ADLI (Activity of Daily Living Interference).” 

 

The success criteria are “The lower bound of the 95% CI of VEHZ (vaccine efficacy based on HZ incidence) > 25%,” and “p-value based on the stratified log-rank test < 0.05 to compare the duration of HZ pain between the two groups,” and “The lower bound of the 95% CI of VEADLI:HZ (vaccine efficacy based on substantial ADLI beyond that of VEHZ) > 0%.”

 

Merck concludes that

 

“Compared with placebo, the zoster vaccine reduced the incidence of HZ.  The estimated VEHZ was 51.3% with 95% CI of (44.2%, 57.6%);” and

 

“Compared with placebo, the zoster vaccine reduced the duration of clinically significant pain associated with HZ.  The median durations in vaccine and placebo groups are 20 days vs. 22 days respectively, and p-value = 0.04;” and

 

 “The lower bound of the 95% CI of VEADLI:HZ = -9.4%.”

 

 

Immunogenicity

 

The Cell Mediated Immunity (CMI) substudy consists of a randomly selected subgroup of subjects (N=1,395) from those who were enrolled at the San Diego and Denver sites.  Blood samples were collected from participants in this substudy.  There is no formal statistical hypothesis associated with immunogenicity.  Merck reported that at 6 weeks postvaccination, VZV IFN-γ ELISPOT counts were significantly higher in the zoster vaccine group than in the placebo group [2.2-fold difference with 95% CI of (1.9, 2.5)].

 

 

Safety

 

Merck reported that “Four vaccine-related serious adverse experiences occurred within 42 days postvaccination (2 in each group), among the entire study population.” Polymyalgia and asthma occurred in the zoster vaccine group, and anaphylaxis and polymyalgia rheumatica occurred in the placebo group.

 

 

 

Merck’s Overall Conclusion

 

Merck concluded “In subjects 60 years of age or older who received 1 dose of zoster vaccine or placebo, (compared to placebo) the zoster vaccine significantly reduced the burden of illness related to HZ pain, the incidence of PHN, the incidence of HZ, and the median duration of clinically significant pain associated with HZ.  Zoster vaccine did not reduce the risk of having substantial ADLI.  The zoster vaccine was generally well tolerated.” 

 

 

 

Reviewer’s Comments

 

 

1.      As shown in the following Table 1, the vaccinees are very heterogeneous with respect to release potency of the vaccine, ageing process of the vaccine (not aged vs. heat-aged), and the duration of follow-up time.  The reviewer does not find any clear association between the release potency of the vaccine and the HZ incidence among vaccinees (see Figure 1).  As Merck stated, “No dose response was observed in vaccine efficacy against HZ and PHN across the potencies used in the trial.”  It is unclear, however, whether release potency of 100,000 PFU/mL (for example) without ageing is comparable to the release potency of 100,000 PFU/mL after heat-ageing.

  

 

 

Table 1.  Incidence of Evaluable HZ Cases by Clinical Lot (MITT* population)

 

PLACEBO

(N=19247)

Zoster Vaccine

(N=19254)

 

 

# Subjects

 

#

HZ

Cases

 

Total Follow-up Time (yrs)

 

Incidence Rate

(per 1000 person yrs)

 

 

Average

Follow-up Time

(days)

 

LOT #

 

Release Potency

(PFU/mL)

 

 

# Subjects

 

#

HZ

Cases

 

Total Follow-up Time (yrs)

 

Incidence Rate

(per 1000 person yrs)

 

 

Average

Follow-up Time

(days)

 

 

 

 

 

 

 

 

 

 

19247

 

 

 

 

 

 

 

 

 

 

642

 

 

 

 

 

 

 

 

 

 

57736

 

 

 

 

 

 

 

 

 

 

11.120

 

 

 

 

 

 

 

 

 

 

1096

1535W-E046

Not aged

97,821

278

6

1097

5.469

1441

1536W-E047

Not aged

114,184

278

7

1119

6.256

1470

1537W-E048

Not aged

92,581

279

8

1133

7.061

1483

1553W-E462

Heat aged

66,000

326

4

1277

3.132

1430

1554W-E463

Heat aged

79,200

326

8

1282

6.240

1436

1555W-E464

Heat aged

65,400

326

7

1266

5.529

1418

1562W-E471

Heat aged

54,600

2906

56

9764

5.735

1227

1563W-E472

Heat aged

54,300

2903

63

9766

6.451

 1229

1564W-E473

Heat aged

67,000

2901

53

9745

5.439

 1227

1588W-G479

Heat aged

44,000

2912

35

7271

4.814

912

1589W-G480

Heat aged

53,700

2908

38

7238

5.250

909

1590W-G481

Heat aged

51,100

2911

30

7247

4.140

909

19247

642

57736

11.120

1096

 

19254

315

58203

5.412

1104

*The modified intent-to-treat (MITT) population included all subjects randomized in the study who were followed for at least 30 days postvaccination and did not develop an evaluable case of HZ within the first 30 days postvaccination.

 

 

 

 

 

 

Figure 1.

 

 

2.      It is not clear from the submission whether subjects were randomly assigned to vaccine lots, and if so, how.

 

3.      The HZ BOI (Burden of Illness) is a composite endpoint that incorporates the HZ incidence, severity, and duration of HZ-associated pain.  Each subject’s BOI is determined by an HZ pain curve based on IZIQ (Initial Zoster Impact Questionnaire) and ZBPI (Zoster Brief Pain Inventory) over a 182-day period.  While the reviewer agrees that Merck’s method of deriving BOI from the area under the HZ pain curve could be one of several useful measures to quantify the disease burden associated with HZ, the reviewer also perceives that the BOI index could be misleading.  For example, let’s suppose that 2,000 subjects are randomized either to zoster vaccine or placebo. Two years after the study starts, there are 10 HZ cases in the vaccine group and 20 in the placebo group.  If a subject with HZ in the placebo group says his/her worst pain score is 3 (out of ten) on IZIQ and all ZBPI questionnaires, then his/her BOI is 3×182(days) = 546.  Let’s assume all twenty HZ cases have the same BOI. On the other hand, if a subject with HZ in the vaccine group says his/her worst pain score is 2 (out of ten) on the IZIQ and all ZBPI questionnaires, then his/her BOI is 2×30(days) = 60 (pain score under 3 does not contribute to BOI after 30 days from rash onset).  Let’s assume all ten HZ cases have the same BOI.  Then VEBOI=1–[(10×60)/(20×546)]=94.5%.  In this case, the vaccine reduces the incidence rate of HZ by half (vaccine efficacy based on HZ incidence is 50%).  However, using the composite endpoint combining HZ incidence and severity, the BOI vaccine efficacy is 94.5.

 

4.      As is shown in Table 2 below, at any time point (from Day 0 to Week 26), the difference in medians of worst pain scores (based on the ZBPI questionnaire) between the two groups does not exceed 1, except at Day 0 (difference in medians is 2).  The mean difference does not exceed 0.4 at any time point except at Day 0 (mean difference is 1.4).

 

Table 2.  Comparison of Worst Pain Scores on ZBPI (MITT population)

among HZ cases

 

 

 

 

 

Time after HZ rash onset

Placebo

(642 HZ cases)

Zoster vaccine

(315 HZ cases)

 

 

# of HZ cases who took ZBPI

 

Mean worst pain

 

 

Median worst pain

 

# of HZ cases who took ZBPI

 

Mean

worst pain

 

 

Median worst pain

Day 0  (HZ rash onset)

58

5.03

5

28

3.64

3

Day 1

158

4.25

4

72

4.06

3.5

Day 2

242

4.16

4

114

4.28

4

Day 3

239

4.51

4

122

4.63

4

Day 4

219

4.24

4

98

4.48

5

Day 5

175

3.95

4

94

3.97

4

Day 6

211

3.86

3

84

3.93

3

Day 7

189

3.85

3

84

4.04

4

Day 8

168

3.43

3

65

3.84

4

Day 9

163

3.79

3

92

3.37

3

Day 10

201

3.66

3

87

3.83

3

Week 2 (Day 12 ~ Day 16)

581

3.18

3

278

3.09

2

Week 3 (Day 19 ~ Day 23)

497

2.56

2

236

2.15

1

Week 4 (Day 26 ~ Day 30)

509

2.02

1

248

1.85

0

Week 5 (Day 33 ~ Day 37)

503

1.50

0

227

1.41

0

Week 6 (Day 40 ~ Day 44)

466

1.21

0

216

1.13

0

Week 7 (Day 47 ~ Day 51)

489

1.14

0

219

0.92

0

Week 8 (Day 54 ~ Day 58)

481

1.05

0

229

0.74

0

Week 10 (Day 68 ~ Day 72)

449

0.85

0

226

0.56

0

Week 12 (Day 82 ~ Day 86)

440

0.77

0

214

0.56

0

Week 16(Day 110 ~ Day 114)

416

0.59

0

197

0.32

0

Week 20(Day 138 ~ Day 142)

389

0.49

0

190

0.29

0

Week 24(Day 166~ Day 170)

390

0.42

0

167

0.17

0

Week 26(Day 180 ~ Day 184)

361

0.38

0

144

0.17

0

 

 

5.      It appears that efficacy of the vaccine with respect to BOI beyond the efficacy on the HZ incidence is minimal (see the reviewer’s exploratory comparison of median HZ BOI among HZ cases in the following Table 3).  However, it is clear that efficacy regarding HZ incidence met the pre-specified success criterion, assuming surveillance and ascertainment of the HZ cases were appropriate.

 

Table 3.  Comparison of BOI between Vaccine and Placebo Groups

 

 

Zoster vaccine

Placebo

 

# subjects

19254

19247

 

# HZ cases

315

642

 

Total follow-up

time (yrs)

 

58203

 

57736

 

mean follow-up

per subject (yrs)

 

3.02

 

3.00

 

 

HZ incidence rate

Per 1000 person-yrs

 

5.41

 

11.12

VEHZ = 51.3%

(44.3%, 57.4%)

HZ incidence rate

(crude rate)

 

1.64%

 

3.34%

VEHZ = 51.0%

(44.0%, 57.1%)

 

Sum of HZ BOI

46341

114057

 

mean HZ BOI

per HZ case

 

147.1

 

177.7

 

 

median HZ BOI

among HZ cases

 

82.50

 

87.75

 

p-value (Wilcoxon) = 0.25

mean HZ BOI

per subject

 

2.41

 

5.93

VEBOI = 61.1%

(51.1%, 69.1%)

 

 

6.      It also appears that efficacy of the vaccine on PHN beyond the efficacy of the vaccine on the HZ incidence is minimal (see the reviewer’s exploratory comparison of percent of PHN among HZ cases in the following Table 4).

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 4.  Comparison of PHN Incidence between Vaccine and Placebo Groups

 

 

Zoster vaccine

Placebo

 

# subjects

19254

19247

 

# HZ cases

315

642

 

Total follow-up

time (yrs)

 

58203

 

57736

 

mean follow-up

per subjects (yrs)

 

3.02

 

3.00

 

 

# PHN cases

27

80

 

percent of PHN

among HZ cases

 

8.57%

 

12.5%

 

p-value (Fisher) = 0.08

PHN incidence rate

per1000 person-yrs

 

0.464

 

1.384

VEPHN = 66.5%

(48.4%, 78.3%)

 

7.      Efficacy of the vaccine on the duration of clinically significant pain beyond the efficacy of the vaccine on the HZ incidence appears to be minimal even though it is statistically significant after age adjustment (see Table 5 below for an exploratory analysis by the reviewer). 

 

Table 5.  Comparison of Duration of Clinically Significant Pain between the Vaccine and Placebo Groups

 

 

Zoster vaccine

Placebo

 

# subjects

19254

19247

 

# HZ cases

315

642

 

 

Median duration of clinically significant pain (days)

 

19

 

22

 

p-value (log-rank) =0.10

p-value based on the protocol-specified stratified (by age) log-rank test = 0.04

 

8.      The BOI index is a composite measure that consists of three components (HZ incidence, severity, and duration of pain).  Thus, PHN (a severe case of HZ) incidence is an element of the BOI (thus not independent of the BOI).  Likewise, HZ incidence and duration of pain are elements of BOI (not independent of the BOI).  Considering this lack of independence in the metrics used to evaluate efficacy, the appropriateness of separate claims of efficacy in reducing HZ BOI and in reducing the incidence of HZ (or the incidence of PHN or the duration of pain) may warrant consideration.

 

 

 

 

 

Appendix