Statistical Review and Evaluation - ARTISS
SPONSOR: Baxter BioScience
TITLE: FS VH S/D 4 Fibrin Sealant (Human)
DOCUMENT REVIEWED: BLA
FROM: Boris Zaslavsky, Ph.D, Dr.Sc. (HFM-219)
THROUGH: Ghanshyam Gupta, Ph.D., Chief, TEB (HFM-219)
TO: Pratibha Rana, Kimberly Lindsey, M.D.
FS 4IU VH S/D is intended to affix skin grafts to surgically prepared wound beds resulting from burns
BRANDNAME: FS VH S/D 4 Fibrin Sealant (Human)
In CBER letter of 11/21/07 we requested the 2 x 2 table used to calculate the confidence intervals for efficacy endpoints in the pivotal study (Study 550201). The table for the intent-to-treat (ITT) was provided, the table for the per protocol (PP) population was not submitted.
A Study to Evaluate the Safety and Efficacy of Fibrin Sealant with 4IU/mL Thrombin, Vapor Heated, Solvent Detergent Treated (FS 4IU VH S/D) to Adhere Tissues and Improve Wound Healing. The primary efficacy objective was to test noninferiority of FS 4IU VH S/D compared to the current standard of care (staples). Evaluation of complete (100%) wound closure by Day 28 should indicate whether the use of FS 4IU VH S/D in grafting procedures results in non-inferior graft adherence and wound healing when compared to the use of staples.
The primary efficacy endpoint is complete wound closure by Day 28 after treatment with either FS 4IU VH S/D or staples as determined by a blinded independent review of the Day 28 photographs.
Primary safety endpoint is adverse experiences (AEs) deemed possibly or probably related to treatment with FS 4IU VH S/D
This was a Phase 3, multi-centered, prospective, evaluator blinded, randomized study comparing FS 4IU VH S/D to staples for use in skin graft adherence and wound healing in subjects with burn wounds. Eligible subjects were required to have deep partial thickness or full thickness burn wounds that could be designated as a test area, and which could be split into 2 comparable test sites. The test sites had to be either a single wound measuring between 2% and 8% total burn surface area (TBSA) that could be split into 2 halves or 2 comparable wounds each measuring between 1% and 4% TBSA. Both test sites were required to receive autologous sheet skin grafts with a thickness of 8/1000 of an inch to 16/1000 of an inch. In accordance with the predetermined randomization scheme, FS 4IU VH S/D was to be used to affix skin grafts at one test site (treatment), and staples used to affix skin grafts at the other test site (control). Each subject was to serve as his/her own control. The postoperative follow-up was planned for 1 year. Subjects were required to undergo evaluations and study procedures at Screening and on Day 0, 5, 14, and 28, and Month 3, 6, 9, and 12 visits. The study was divided into two reporting parts:
Part A: A 28-day postoperative follow-up including the primary efficacy measurement, Part B: A cosmetic outcome follow-up through Month 12. This report covers data collected in both Part A and Part B.
SAMPLE SIZE: 138
Analyzed: Safety = 138, Intent-to-treat (ITT) = 127, Per protocol (PP) =106.
Part A. The primary efficacy endpoint is complete wound closure by Day 28 as determined by the majority opinion of a panel of 3 experts evaluating photographs of the test sites. Photographs were presented to the panel members in a random, blinded manner, with each of the 3 reviewers attending separate sessions to avoid bias. The primary efficacy analysis was performed on the intent-to-treat population. The intent-to-treat population consisted of all subjects with at least one available primary endpoint assessment. In addition, an efficacy analysis was done on those subjects deemed evaluable per protocol (i.e., met all inclusion/exclusion criteria, randomized and treated correctly, and adhered to study procedures). The primary statistical analysis was used to test non-inferiority of FS 4IU VH S/D compared to staples with respect to complete wound closure at Day 28 after surgery. To assess the non-inferiority of success rate in wound closure with FS 4IU VH S/D as compared to that with staples, a one-sided 97.5% confidence interval for the difference of correlated proportions was calculated. The non-inferiority margin was set at 10%. If the lower limit of the one-sided confidence interval of the difference in closure rate between FS 4IU VH S/D treated and staples sites is above zero, it is considered to be proven that FS 4IU VH S/D treatment is superior to grafting with staples in terms of statistical significance at the 0.025 alpha level (p < 0.025). Median, 95% confidence interval of the median and range (proportions and 95% confidence interval) are presented in tables of summary statistics for the secondary endpoints. Secondary endpoints measured by planimetry (% areas of hematoma/seroma, % area of engraftment, % area of wound closure) were analyzed twice: once without the regrafted subjects and subjects who received additional staples, and once with the regrafted sites and sites that received additional staples imputed to a 'worst case' result.
Part B. The Vancouver Scar Scale was evaluated with non-parametric signed rank test separately for each individual evaluation; scar maturation assessments were only performed in Part B of this study (Months 3, 6, 9, and 12). All subjects receiving FS 4IU VH S/D were included in the safety analysis. Probability of occurrence and 95% confidence interval of the probability of occurrence of AEs possibly or probably related to FS 4IU VH S/D was calculated. Occurrence of local AEs is tabulated for FS 4IU VH S/D and staples separately. In agreement with the assessment of the primary efficacy endpoint, a 95% confidence interval was carried out for the ifference of local AEs between FS 4IU VH S/D and the control. The number, proportion and 95% confidence interval of the proportions are presented for categories of local AEs. All AEs are presented in incidence tables by MedDRA system organ class and preferred term. Experiences classified as possibly related or probably related to the FS 4 IU VH S/D administrations are presented separately and combined with all other experiences. Frequency and percent of subjects with serious adverse experiences (SAEs) are also displayed. All AEs, vital signs and concomitant medications data are provided in individual subject listings.
Part A. A total of 138 subjects were randomized and treated at 13 study sites. All 138 subjects were treated at one test site with FS 4IU VH S/D and a separate test site with staples. Of the 138 treated subjects, 94 (68.1%) were male and 44 (31.9%) were female. The mean ± SD age was 30.8 ± 17.6 years; 19 (13.8%) subjects were ≤ 6 years old, 21 (15.2%) subjects were 7 to 18 years old, and 98 (71.0%) were > 18 years old. The mean ± SD estimated TBSA for all burn wounds was 13.6 ± 9.2%. The mean ± SD estimated TBSA requiring skin grafting was 8.0 ± 6.9%. The mean ± SD estimated TBSA for the FS 4IU VH S/D test sites was 1.7 ± 0.8% and for the stapled test sites was 1.7 ± 0.7%. Burn wound thickness was classified as full thickness in 106 (76.8%) of the 138 treated subjects, and partial thickness in 32 (23.2%) subjects. Of the 138 treated subjects, 131 completed the Day 28 visit and 7 subjects were lost to follow-up prior to the Day 28 visit. The intent-to-treat (ITT) population included 127 of the treated subjects. The per protocol (PP) population consisted of the 106 subjects who met the criteria defined for the PP population in the a priori statistical analysis plan.
Complete wound closure by Day 28 was achieved in 55 (43.3%) of the FS 4IU VH S/D test sites and 47 (37.0%) of the stapled test sites in the 127 ITT subjects. The lower limit of the 97.5% confidence interval of the difference between FS 4IU VH S/D and staples was -0.029. A similar result was obtained with the PP population, complete wound closure by Day 28 was achieved in 48 (45.3%) of the FS 4IU VH S/D test sites and 42 (39.6%) of the stapled test sites in the 106 PP subjects. The lower limit of the 97.5% confidence interval of the difference between FS 4IU VH S/D and staples was -0.041. Therefore, FS 4IU VH S/D is non-inferior to staples at the 97.5% one-sided level for complete wound closure by Day 28 in the ITT and PP populations because the lower limit of the confidence interval of the difference between FS 4IU VH S/D and staples success rates was greater than the predefined limit of -0.1.
Part B. Scar maturation was assessed by blinded Vancouver Scar Scale evaluations on Months 3, 6, 9, and 12. Overall, the Vancouver Scar Scale scores were similar between the FS 4IU VH S/D test sites and stapled test sites on Months 3, 6, and 9 for both the ITT and PP populations. However, on Month 12, hyperpigmentation was observed in more stapled sites than FS 4IU VH S/D-treated sites. Analysis of the Vancouver Scar Scale scores showed a difference in pigmentation score between FS 4IU VH S/D and stapled test sites on Month 12 (p=0.0192 for ITT analysis and p=0.0129 for PP analysis). However, no adjustments were made for multiple testing, and therefore, statistical significance cannot be declared. Outcomes of scar maturation assessments excluding regrafted subjects and subjects with unblinded evaluations were similar.
COMMENT TO CBER:
The sponsor used SAS and asymptotic formulae to calculate confidence intervals. I checked the confidence interval for the ITT population using the more conservative exact formulae (Cytel Studio 8). My estimate of the lower limit is -0.031. Therefore, FS 4IU VH S/D is non-inferior to staples at the 97.5% one-sided level for complete wound closure by Day 28 in the ITT and PP populations: the lower limit of the confidence interval of the difference between FS 4IU VH S/D and staples success rates for the ITT subjects was greater than the predefined limit of -0.1.
Part A. Of the SAEs occurring at test sites, 3 incidences of infection and 1 incidence of staphylococcal infection occurred at both the FS 4IU VH S/D test site and the stapled test site. There were also 4 incidences of skin graft failure occurring at the FS 4IU VH S/D test site and 3 incidences at the stapled site. In all cases, skin graft failure was due to infection or mechanical forces (scratching or shearing).. Part B. Of the 3 SAEs that occurred at test sites, there was 1 incidence of skin graft failure at both the FS 4IU VH S/D-treated site and stapled site, 1 wrist fracture at the stapled site, and 1 joint contracture at the stapled site. In both Part A and Part B of the study, none of the SAEs occurring in either the FS 4IU VH S/D test site or the stapled test site were considered related to FS 4IU VHS/D by the investigator. Likewise, none of the non-test site SAEs were considered related to FS 4IU VH S/D.
REVIEWER'S COMMENTS TO CBER:
Overall, the data collected and analyzed during Baxter clinical study demonstrate that FS 4IU VH S/D is safe and efficacious for the attachment of sheet skin grafts in subjects with deep partial thickness or full thickness burn wounds. No superiority of the FS 4IU VH S/D treatment over control has been proved using the one-sided confidence level of 97.5%.
On Dr. Lindsey request, I analyzed the sensitivity of the result to a possible imputation of 11 subjects which were randomized but were excluded from the ITT population. If all 11 subjects were successes (the best case scenario) with the FS 4IU VH S/D treatment and failures with the staples treatment, the superiority could be claimed (the lower confidence limit is 0.042 (asymptotic) and 0.034 (exact)). If all 11 subjects were failures with the FS 4IU VH S/D treatment and success with the control (the worst case scenario), the non-inferiority could be rejected based on -0.1 margin: the lower confidence limit is - 0.12 (asymptotic and exact).