Lead Investigator: Martin Okun, Fort HealthCare
Title of Research Proposal: Geospatial and Seasonal Variation in Psoriasis Severity Analysis of Placebo Response Data from Phase 3 placebo-controlled trials in moderate-severe psoriasis.
Vivli Data Request: 4316
Funding Source: None
Potential Conflicts of Interest: A potential conflict of interest is that I am a former employee of Abbott Laboratories and AbbVie and serve as a consultant and on the speaker’s bureau for AbbVie. However, as I am accessing only placebo data, the relevance of my past and current work experience with AbbVie is uncertain.
Summary of the Proposed Research:
In placebo controlled randomized clinical trials of new therapeutics for moderate to severe psoriasis patients, some patients assigned to the placebo arm experience clinical improvement in their psoriasis signs. The basis for this observed improvement is not clear. Possible explanations include: random fluctuations in disease severity, increased attention to skin care instigated by study personnel translating into more diligent application of emollients by the subject, or the subject experiencing the psychological benefit of inaccurately believing that he or she is receiving effective experimental therapy (i.e., the true “placebo” effect). In addition to these explanations, it is also possible that seasonal fluctuations in ambient UV exposure may account for some of the response observed among placebo patients. The purpose of the proposed research is to investigate if and to what extent fluctuations in ambient UV exposure could be causing improvements in psoriasis among patients being treated with placebo.
There is evidence from the dermatologic literature for seasonal variation in psoriasis severity: Pascoe and Kimball (2015) collected psoriasis PGA (Physician’s Global Assessment) scores from psoriasis patients in a large New England-based healthcare system, analyzed the data for seasonal variations, and noted that a higher percentage of psoriasis patients were clear or almost clear in the summer than in the winter, with statistically significant variation in the percentage of patients who were clear or almost clear across the different seasons. This outcome is biologically plausible, because UV (ultraviolet) light exposure is greater in the summer than in the winter, with the difference in exposure more marked at higher latitudes, and because UV exposure is known to ameliorate psoriasis. However, the results from Pascoe and Kimball may be confounded by the concomitant treatments that the psoriasis patients were receiving at their office visits. To eliminate the effect of this confounding, analysis of controlled clinical trial data would be valuable.
Statistical Analysis Plan
A total of ~600 placebo-treated subjects across the 3 requested clinical trials will be divided randomly into a training set and a validation set.
Estimates of the erythemally weighted daily dose for each placebo-treated subject at baseline and at Week 16 will be generated:
Each subject’s location will be assigned to the location of his/her investigational site.
Each investigational site will be geocoded (an advantage of using the investigational site is that subject privacy is protected).
The glUV dataset (http://www.ufz.de/gluv) contains monthly mean values of the UV-B erythemally weighted daily dose values, averaged across the years 2004-2013.
Dataset has a spatial resolution of 15 arc-minutes.
Data from where the investigator site is located will be extracted for these analyses:
Baseline and Week 16 values will be assigned based on the monthly mean values for the months in which these visits fall.
(Week 16 PASI [Psoriasis Area and Severity Index score] – Baseline PASI) values will be plotted against (Week 16 UV exposure – Baseline UV exposure), to check visually for evidence of correlation.
Proposed Linear Regression Model: (Week 16 PASI – Baseline PASI) = f(Week 16 UV exposure – Baseline UV exposure).
R2 value, regression coefficient, and p value of regression model will be determined.
It is possible that the effects of UV on PASI are time-lagged. This will be checked as a sensitivity analysis:
Proposed Linear Regression Model: (Week 16 PASI – Baseline PASI)=f(Week 12 UV exposure – (Week -4 UV exposure)).
As a further sensitivity analysis, the monthly mean temperature minimum at baseline and at Week 16 (derived from public NOAA databases) for the zipcode of the investigational site for the placebo-treated subjects will be checked for correlation with monthly mean values of the UV-B erythemally weighted daily dose values. In the absence of evidence of high correlation (i.e., minimal confounding), the monthly mean temperature will be checked as an independent variable, and a multivariate model including both mean temperature minimum and monthly mean UV-B values will also be tested.
The validation set will be checked using the same model if the training set demonstrates significant correlation between independent and dependent variables.
The rationale for using these studies is that they have a relatively large (~600) placebo population and the placebo period lasts 16 weeks. Subjects with missing Baseline or Week 16 PASI scores will be excluded from the analysis.
A Phase III, Multicenter Study of the Efficacy and Safety of Adalimumab Treatment in Subjects With Moderate to Severe Chronic Plaque Psoriasis
Study ID: NCT00237887
Sponsor ID: M03-656
BI 655066/ABBV-066 (Risankizumab) Versus Ustekinumab and Placebo Comparators in a Randomized Double Blind trIal for Maintenance Use in Moderate to Severe Plaque Type Psoriasis (UltIMMa-1)
Study ID: NCT02684370
Sponsor ID: M16-008
BI 655066 Versus Ustekinumab and Placebo Comparators in a Randomized Double Blind trIal for Maintenance Use in Moderate to Severe Plaque Type Psoriasis-2 (UltIMMa-2)
Study ID: NCT02684357
Sponsor ID: M15-995
268 Analysis of association between variation in ambient solar ultraviolet exposure and disease severity for patients with moderate-severe psoriasis
Okun, T.S. et al.
Journal of Investigative Dermatology, Volume 141, Issue 5, S48