Lead Investigator: Amit Garg, Northwell Health
Title of Proposal Research: Association between skin-related quality of life and race in patients with hidradenitis suppurativa, and Repeatability/Intra-rater agreement of lesion counts in hidradenitis suppurativa
Vivli Data Request: 9241
Funding Source: None
Potential Conflicts of Interest: Dr. Garg has received personal fees from AbbVie, Aclaris Therapeutics, Anaptys Bio, Aristea Therapeutics, Boehringer Ingelheim, Bristol Myers Squibb, Incyte, InflaRx, Insmed, Janssen, Novartis, Pfizer, Sonoma Biosciences, UCB, Union Therapeutics, and Viela Biosciences, and receives honoraria. Dr. Garg receives research grants from Abbvie, Pfizer, UCB, and National Psoriasis Foundation. He is co-copyright holder of the HS-IGA and HiSQOL instruments.
Interests will be disclosed when the research is published and presented.
Summary of the Proposed Research:
Study 1: Association between skin-related quality of life and race in patients with hidradenitis suppurativa
Hidradenitis suppurativa (HS) is a chronic inflammatory disorder that affects approximately 0.1 to 1.0% of the population worldwide. HS is characterized by inflamed nodules (lumps of tissue), abscesses (swollen areas containing pus), and tunnels which result from skin wounds that have created pathways underneath the skin. Additionally, HS can cause significant scarring in the breast folds, axillary and groin regions.
HS typically occurs in the second to fourth decades of life, and disproportionately affects women and black patients. Patients with HS experience a significant decrease in quality of life due to pain, drainage, odor, disfigurement, embarrassment, and shame. No singular treatment is uniformly effective for the management of the disease and its consequences on patient quality of life.
The life impact of other common inflammatory diseases of the skin has been shown to have disparate racial outcomes. A recent study evaluating the disease burden of psoriasis showed that Black patients suffer from more severe disease with greater psychological impact and impaired quality of life compared to White patients. We will compare patient-reported quality of life at baseline between White and Black patients in two clinical trials of HS patients using the validated Dermatology Life Quality Index. Our evaluation of life impact is critical to understand potential differences in HS disease burden based on race, considering 1 in 3 HS patients in the United States is African American. Additionally, understanding these differences will foster a patient-centered approach and aid in establishing appropriate goals of treatment.
Study 2: Repeatability/Intra-rater agreement of lesion counts in Hidradenitis Suppurativa
Measurement of disease outcomes and assessment of treatment effectiveness in HS are often based on counts of distinct lesion types reported by clinicians. The standard for efficacy in HS trials has been the Hidradenitis Suppurativa Clinical Response (HiSCR), defined as > 50% reduction in total abscess (swollen areas containing pus) and inflammatory nodule (lumps of tissue) count (AN) relative to baseline and no increase in abscess or draining tunnel (pathway underneath the skin created by skin wounds) count. Using this outcome, clinical trials for HS drug development frequently result in high placebo (treatment/substance with no therapeutic effect) group response rates. Potential explanations for this high placebo response rate include natural variation in disease activity as well as poor repeatability of clinicians’ lesion counts (differences in clinicians’ counts of lesions on separate occasions when the true lesion count has not changed). While the statistical reliability of repeated lesion counts by the same rater has been investigated, reliability is a relative measure and does not describe the absolute differences in lesion counts across repeated assessments. These absolute differences are most relevant when evaluating change over time, as in a clinical trial. We aim to characterize the repeatability and intra-rater agreement of clinician reported lesion counts using measurements recorded at screening and baseline visits in two clinical trials in HS patients. The results of this study may indicate a need to improve the training of clinicians when evaluating lesion counts in the clinical trial setting. Alternatively, our results may demonstrate the adequacy of current practices.
Statistical Analysis Plan:
General Analysis Considerations
Specific details of the analysis for the two proposed studies are described in greater detail below (study 1 = skin-related quality of life according to race; study 2 = repeatability of lesion count measurements). For the first study of skin-related quality of life according to race, separate analyses will be performed for the PIONEER I and PIONEER II datasets, and study-level results will be pooled using fixed-effects meta-analysis. For the second study of lesion count repeatability, separate analyses will be performed for the PIONEER I and PIONEER II datasets. For this second study, results from PIONEER I and PIONEER II will not be pooled quantitatively, but rather discussed qualitatively.
Study 1: Association between skin-related quality of life and race in patients with hidradenitis suppurativa
Patients and Studies
The source population for this analysis will include all patients in the placebo and adalimumab arms of the PIONEER I (NCT 01468207) and PIONEER II (NCT 01468233) Phase 3 trials of adalimumab for patients with moderate-to-severe hidradenitis suppurativa. These studies were selected because they include a relatively large, racially diverse sample of HS patients with longitudinal data on overall and disease-related quality of life using validated measurement instruments. Eligibility criteria for this analysis will include: 1) self-reported race of White or Black; 2) available data for Dermatology Life Quality Index (DLQI) at baseline.
For our primary aim, we will compare the DLQI total score at baseline (Week 0) between White and Black patients.
Descriptive Statistics
We will summarize patient characteristics at baseline using descriptive statistics, stratified by race and study (PIONEER I vs PIONEER II). Quantitative variables will be described using the mean, standard deviation, median, and interquartile range, while categorical variables will be described using counts and percentages. Descriptive statistics will be provided for demographic characteristics (e.g., age, sex), disease-related characteristics (body mass index, Hurley stage, disease duration, total abscess/nodule (AN) count, individual counts for abscesses, inflammatory nodules, and draining fistulas, prior surgery for HS, previous systemic treatment for HS), as well as medical history variables (BMI, current smoking status, anemia, asthma, other comorbidities present in > 5% of patients, prior surgeries).
Primary outcome analysis
We will compare the mean DLQI total score at baseline between white and black patients in each trial individually, and synthesize these results using a fixed-effect meta-analysis. The meta-analysis will be performed according to the two-stage approach, in which study-level “effects” are estimated separately within each study in step 1 and combined using fixed-effect meta-analysis with inverse variance weights in step 2. Patients missing baseline DLQI information will be excluded. We will perform an unadjusted comparison in which a simple mean difference is calculated in each trial and combined. We will also calculate an adjusted mean difference between black and white patients using a multiple linear regression model with baseline DLQI as the dependent variable and the following independent variables: race, age, sex, total AN count (baseline), total draining fistula count (baseline). The model will assume a linear relationship between DLQI and each of age, AN count, and draining fistula count. The purpose of the multivariable analysis is to assess whether disease-related quality of life differs between black and white patients with similar levels of disease activity. The regression coefficient for race (black vs. white) will be calculated in each trial individually and combined using inverse-variance fixed effect meta-analysis.
Power and sample size for primary outcome
Based on the formula for statistical power for the fixed effects mean effect provided by Valentine (2010), the meta-analysis will have > 99% power at the 0.05 significance level to detect a mean baseline DLQI difference of 4 units between white and black patients. This estimated power is based on detecting a standardized mean difference of 0.571 (=4/7). This corresponds to a mean difference of 4 (minimal clinically important difference for the DLQI as stated in Basra (2015)) divided by a pooled standard deviation of 7 (approximate standard deviation of the DLQI as reported in Kimball (2016)). For power calculations, the variance of the pooled effect size was estimated based on the standardized mean difference above, and the average number of black and white patients in PIONEER I and PIONEER II (average # white ≈ 253, average # black ≈ 45).
Secondary outcomes
We will report group means at baseline stratified by race (white, black) and study (PIONEER I, II) for the following secondary patient-reported outcomes: Patient Global Assessment of Skin Pain-Numeric Rating Scale (Worst), subscale scores of the Work Productivity and Activity Impairment Questionnaire: Specific Health Problem (WPAI:SHP), subscale scores of the Short Form-36 (SF-36) Health Status Survey. SF-36 will only be described for PIONEER I as it was not collected in PIONEER II. Patients with missing data for individual variables at baseline will be excluded from the analysis of that variable. Formal statistical inference for the baseline variables above will not be performed.
Aim 2 Analysis
A secondary aim of this quality of life study is to assess whether the change in DLQI between baseline and week 12 differs according to race among patients who achieve HiSCR response. Patients who self-report race as white or black and have available DLQI measurement at baseline are eligible for this analysis. We will compare change in DLQI between white and black patients using a multiple linear regression model with change in DLQI (week 12 – baseline) as the dependent variable, and the following independent variables: baseline DLQI, baseline abscess/nodule count, baseline draining fistula count, race indicator variable (black vs white), HiSCR responder indicator variable (yes/no), and an interaction term between race indicator and HiSCR indicator. The interaction between race and HiSCR response status will allow us to estimate adjusted differences in the mean DLQI change between white and black patients for HiSCR responders and non-responders separately. The adjusted difference in mean DLQI change between white and black patients based on this model will be pooled using inverse variance fixed effect meta-analysis (two-stage approach described above for primary analysis).
Missing data for this analysis will be handled using non-responder imputation for HiSCR status at week 12 and last observation carried forward for DLQI total score at week 12. These were the primary approaches used for imputation of categorical and continuous efficacy variables in the main trial efficacy analysis. Baseline DLQI scores will not be carried forward.
Study 2: Repeatability/Intra-rater agreement of lesion counts in Hidradenitis Suppurativa
The purpose of this analysis is to assess the absolute differences between repeated measures of lesion counts by the same rater/observer in the absence of true change. This measurement property is often referred to as “repeatability” when describing deviations of repeated measures by the same rater, and “agreement” when describing deviations between measurements between two different raters or methods.
Patients and Studies
The source population for this analysis will include all patients in the placebo and adalimumab arms of the PIONEER I (NCT 01468207) and PIONEER II (NCT 01468233) Phase 3 trials of adalimumab for patients with moderate-to-severe hidradenitis suppurativa. These studies were selected because they include repeated measures of clinician-reported lesion counts for specific types of lesions over a relatively short time period (between screening and baseline visits).
All analyses described below will be performed separately for PIONEER I and PIONEER II data. Results from each trial will not be combined quantitatively.
Descriptive Statistics
We will summarize patient characteristics at baseline using descriptive statistics. Quantitative variables will be described using the mean, standard deviation, median, and interquartile range, while categorical variables will be described using counts and percentages. Descriptive statistics will be provided for demographic characteristics (e.g., age, sex, race) and disease-related characteristics (e.g., Hurley stage, disease duration, total abscess/nodule (AN) count, individual counts for abscesses, inflammatory nodules, and draining fistulas), and the time between screening and baseline.
Analysis of Primary Outcomes
We will repeat the analyses described below for each of the following types of lesion counts: 1) total abscess and inflammatory nodule (AN) count; 2) abscess count; 3) inflammatory nodule count; 4) draining fistula count.
For each patient, the difference between lesion count at baseline and screening will be calculated (baseline minus screening), and the average and standard deviation of these differences will be calculated. A histogram of the differences in lesion counts will be plotted. To visualize the magnitude of these differences and their relation with the total lesion count, we will create a Bland-Altman plot, including the average lesion count on the x-axis, and the difference value on the y-axis. Given the large number of patients with 0 abscesses and draining fistulas, the distribution of differences between baseline and screening is not expected to be normally distributed. Therefore, methods to quantify repeatability and agreement which rely on normally distributed data are not applicable. Accordingly, we will estimate non-parametric limits of agreement using sample quantile estimators (2.5% and 95% percentiles) (Frey et al., 2020). 95% of differences between screening and baseline are expected to lie between these limits. We will also calculate alternative quantiles of the distribution of differences, including the 10% and 90% percentiles.
Subgroup Analysis
We will stratify patients according to the median AN count (i.e., baseline and screening AN count will be averaged, and the median of this value will be used to stratify patients into “high” and “low” AN count groups). The above analysis of non-parametric limits of agreement for total AN count between screening and baseline will be repeated in these two groups.
Sensitivity Analysis
We will repeat the analyses described above in the subgroup of patients whose screening and baseline visits occurred within 14 days. The assumption of the stability of true lesion counts, which is a prerequisite for evaluating repeatability/agreement, is more likely to hold within shorter time intervals.
Requested Studies:
A Phase 3 Multicenter Study of the Safety and Efficacy of Adalimumab in Subjects With Moderate to Severe Hidradenitis Suppurativa – PIONEER I
Data Contributor: AbbVie
Study ID: NCT01468207
Sponsor ID: M11-313
A Phase 3 Multicenter Study of the Safety and Efficacy of Adalimumab in Subjects With Moderate to Severe Hidradenitis Suppurativa – PIONEER II
Data Contributor: AbbVie
Study ID: NCT01468233
Sponsor ID: M11-810
Public Disclosures:
- Midgette, B., Strunk, A. and Garg, A., 2024. Association between skin-related quality of life and race in patients with moderate-to-severe hidradenitis suppurativa: Analysis of two phase 3 clinical trials. Journal of the American Academy of Dermatology. Doi : 10.1016/j.jaad.2024.07.1508
- Midgette B, Strunk A, Garg A. Association Between Skin-Related Quality of Life and Race in Patients with Hidradenitis Suppurativa. Annual Meeting of the Medical Dermatology Society. Poster. 2024.