Lead Investigator: Emanual Maverakis, University of California Davis
Title of Proposal Research: Developing Models from Existing Atopic Dermatitis Outcome Data
Vivli Data Request: 7243
Funding Source: Dr. Maverakis is funded through an NIH mid-career award, K24AR077313
Potential Conflicts of Interest: None
Summary of the Proposed Research:
Background
Atopic dermatitis (atopic eczema) is an inflammatory skin disease affecting children and adults. Usually, the disease presents in childhood with red itchy skin lesions. Later in life these same children often develop asthma and/or allergic rhinitis. Atopic dermatitis can also present in adulthood or persist from childhood to adulthood in some cases. Currently, around 16.5 million adults have atopic dermatitis, with the disease incidence reportedly being on the rise. Patients with atopic dermatitis also suffer from other comorbidities including skin infections, anxiety and depression.
Necessity of the research
In order for new drugs to be developed or the efficacy of one treatment to be compared to another, investigators need accurate tools to assess disease severity. This will allow them to determine how much the patient improves and how fast they improve after starting the treatment.
How the research will add to medical science or patient care
The ultimate goal of the type of research proposed herein is to develop improved tools or to simplify existing tools to assess skin disease severity. Such tools can then be used in the development of new treatments for skin disease.
How the research will be conducted/What design and methods you have you chosen and why
The proposed research will study the relationships between the individual components of the various efficacy assessment tools used to determine skin disease severity. To accomplish this we will employ various statistical methodologies.
We hope to conduct this analysis using data from both atopic dermatitis and psoriasis clinical trial datasets but other well-curated clinical trial datasets are also of interest, including trials outside the field of dermatology.
Statistical Analysis Plan:
To select a trial for this study we performed a search of the Vivli database using the keywords “Atopic Dermatitis” and Study Phase – Phase 2/Phase 3 as a modifier. This yielded 4 clinical trials. From this list this trial was selected because it had the largest sample size.
Statistical analysis. All statistical analysis will be performed using R software. A multivariate linear regression model will be used for the analysis of each variable’s contribution to the index. As a measure of the variable importance the absolute value of the t-statistic for each model parameter will be used. Collinearity of variables in the multivariate models will be examined by calculating variance inflation factor (excessive if > 2.5) with R package “car”. Nonlinear relationships between the analytes and the outcome will be evaluated with R package “mfp” using a multiple fractional polynomial method. The association between variables will also be analyzed using multinomial logistic regression models with R package “nnet”. A restricted maximum-likelihood estimator will be used to estimate between-study variance. To present the correlations between all variables simultaneously the dimensionality reduction algorithm “t-distributed stochastic neighbor embedding” (t-SNE) will be used, implemented in the R package “Rtsne”.
Handling of missing data. Reasons for missingness will be analyzed using multinomial logistic regression with treatment group as one of the covariates. If necessary, missing data imputation by predictive-matching algorithms will be conducted. The results of imputation methods vs. missing data exclusion will be compared in a sensitivity analysis.
Requested Studies:
Study 203121: A Randomized, Blinded, Vehicle-Controlled, Dose-Finding Study of GSK2894512 Cream for the Treatment of Atopic Dermatitis
Data Contributor: GlaxoSmithKline
Study ID: NCT02564055
Sponsor ID: 203121
Summary of Results:
One of our research aims was to identify the redundancies in the existing dermatology disease assessment tools and provide insight into how these tools can be simplified. We analyzed results from 363 patients with atopic dermatitis. We assessed the collinearity between the variables within each patient’s baseline EASI score. No variables were highly collinear or indicative that there was any redundancy in the current version of the EASI.