Validation of a novel and more accurate endpoint for GVHD clinical trials

Icahn School Of Medicine At Mount Sinai, New York NY

Investigators

Abstract

ABSTRACT Acute graft vs. host disease (GVHD), attacks the skin, liver, and/or GI tract, is treated with high doses of steroids, and is the primary cause of non-relapse mortality (NRM) in patients undergoing allogeneic hematopoietic cell transplantation. The overall response rate (ORR) to treatment that combines complete (CR) and partial response (PR) at day 28 (D28) was accepted by the FDA as evidence of therapeutic benefit in a clinical trial based on both accuracy and timeliness: it was the earliest time point that stratified patients for NRM (and survival) among several evaluated. Disadvantages to D28 ORR as the primary endpoint for clinical trials include: (1) lack of inclusion of severity at onset in calculating response; (2) all target organ responses are weighted equally despite the fact that NRM is driven by GI GVHD; (3) symptom severity often reflects the presence of confounding factors; and (4) CR and PR response groups have the same long-term outcomes and thus cannot support different approaches to GVHD of differing severity. This application proposes to validate a new model that overcomes these disadvantages. The Mount Sinai Acute GVHD International Consortium (MAGIC) algorithm probability (MAP) uses serum biomarkers to calculate a risk score that predicts long-term outcomes after treatment. Our preliminary “real world” data show that (1) a D14 MAGIC clinical model that accounts for symptom severity both at onset and D14 is as accurate a surrogate for long-term outcome as D28 ORR; and (2) a new MAGIC integrated response that adds the D14 MAP to the new D14 MAGIC clinical response more accurately predicts long-term outcomes than D28 ORR. The D14 MAGIC integrated response also creates three response groups with distinctly different outcomes. These findings require confirmation in patients who were treated in the context of clinical trials. We propose to use existing clinical and laboratory data from five multicenter GVHD treatment trials (n=682 patients) to test the hypothesis that novel D14 clinical trial endpoints are accurate surrogates for one-year NRM. In Specific Aim (SA) 1, we will validate the D14 MAGIC clinical response an accurate surrogate for 1 year NRM if both sensitivity and specificity are non-inferior to D28 ORR. Other performance measures such as balanced accuracy (average of sensitivity and specificity), PPV, NPV, and AUC for NRM will provide a comprehensive comparison between the two clinical endpoints. In SA2, we will validate the D14 MAGIC integrated response as the best surrogate for 1y NRM if its balanced accuracy is significantly better than D28 ORR. We will also analyze the performance of different definitions of treatment response when there are three, rather than two, distinct response categories. If successfully validated these novel endpoints will be the first advance in GVHD clinical trial design in 15 years and provide the ability to tailor trial endpoints to different treatment goals as conceived when CR and PR groups were defined.