Assessing Treatment Response in Cutaneous Sclerosis after Allogeneic Hematopoietic Cell Transplantation: A Chronic Gvhd Consortium Study

Cutaneous sclerosis, a highly morbid subtype of chronic graft vs. host disease (cGVHD), is associated with significant morbidity, disability, and poor response to available therapeutic agents. This limited response may be driven both by limited activity of available therapeutic agents and by limited sensitivity of NIH Response Measures for this cGVHD subtype. Given this need for novel sclerosis-specific response measures, we leveraged data from the national Chronic GVHD Consortium to examine which measures of skin and joint/fascia involvement were associated with treatment response.A training cohort included patients with cutaneous sclerosis from (a) a randomized trial of imatinib vs. rituximab, and (b) a Consortium observational study. The validation cohort was a different Consortium observational study. Clinician-reported (PROM score, total BSA involved with sclerosis, NIH 0-3 scores, Hopkins skin score, Fascia scale, and Vienna skin scale) measures (baseline, and baseline to 6-month change) were examined for association with 6-month clinician-reported response (better vs. not better). Patient-reported (SHAQ, modified HAP, SF-36, Lee symptom scale, FACT-BMT) measures (baseline and baseline to 6-month change) were studied for association with 6-month patient-reported response (better vs. not better). In univariate analysis, we examined association of each variable with response outcome. Multivariate analysis by logistic regression used stepwise regression with entry/retention criteria of p < 0.05. The final model performance was assessed through AUC values, and calculated sensitivity, specificity, positive and negative predictive values (PPV, NPV).A total of 357 subjects were included (training 183, validation 174). While multiple skin and joint measures were associated with clinician-reported response on univariate analysis, PROM total score, PROM total score change, and NIH 0-3 skin change were retained in the final multivariate model (AUC 0.83 training, 0.75 validation). Similarly, many patient-reported measures were associated, but final multivariate analysis retained the HAP AAS, SF36 vitality change, LSS skin, and LSS skin change in the model (AUC 0.86 training, 0.77 validation), see fig 1-3.We identified which sclerosis measures have greatest association with 6-month clinician- and patient-reported treatment response, a previously unstudied area. However, given the observed performance in the validation cohorts, we conclude that further work is needed. We suggest that novel response measures (e.g. skin thickness measures, novel imaging modalities, tissue and/or blood biomarkers) may be needed to optimally assess treatment response in cutaneous sclerosis.