Identification of High-Risk Patients and Establishment of a Clinical Prediction Model for Frontline Steroid Failure in Chronic Graft-Versus-Host-Disease

Introduction: Contemporary practice for frontline chronic graft-versus-host disease (cGvHD) treatment predominantly involves the use of systemic corticosteroids. Treatment with newer agents for cGvHD including ruxolitinib, belumosudil or axatilimab may benefit patients at high risk of treatment failure. We aimed to delineate high risk characteristics associated with frontline systemic steroid treatment failure in patients with cGvHD.

Methods: We conducted a retrospective review of 267 consecutive patients who were diagnosed with cGvHD and were treated with frontline systemic corticosteroids after allogenic stem cell transplant (allo-SCT) at Princess Margaret Cancer Centre in Toronto, Canada. We collected data including demographics, disease and transplant characteristics as well as characteristics and classification of cGvHD using NIH Consensus Criteria. The primary outcome of this analysis was failure free survival (FFS); failure was defined as 1) systemic treatment change, 2) relapse or 3) non-relapse mortality. We used the Kaplan-Meier method with log rank test for univariate analysis and the Cox proportional hazard regression model for multivariate analysis using a modified version of R.

Results: The median age of the 267 included patients was 51 (range: 19-71); 107 patients (40%) were female. 186 patients (70%) had a myeloid malignancy. With regards to transplant characteristics, 171 patients (64%) had a matched related donor, and 249 patients (93%) received stem cell transplant from a peripheral blood source. 48 patients (20%) underwent T-cell depletion and 94 patients (35%) underwent reduced intensity conditioning. The proportion of patients with mild, moderate and severe cGvHD according to NIH criteria was 82 (31%), 140 (52%) and 45 (17%) respectively. More than one organ was involved in 226 patients (85%), with the skin (n=167, 63%), liver (n=163, 61%) and mouth (n=131, 49%) being the most common organs involved.

The median follow-up duration was 2.6 years amongst survivors. FFS was 61.7% [95% CI: 55.6-67.3] at 6 months, 37.8% [31.7-43.8] at 1 year, and 26.6% [20.9-32.5] at two years; median time to failure among the 267 patients was 252 days [217, 321]. Age, sex, and type of underlying condition were not significant predictors of FFS. With regards to transplant characteristics, no significant association was observed between FFS and any of: stem cell source (bone marrow vs. peripheral blood), donor-to-recipient sex or donor/recipient CMV status, T-cell depletion, reduced-intensity conditioning or type of GvHD prophylaxis. On univariate analysis, history of acute GvHD grade III-IV, severe cGvHD NIH global score at time of initial diagnosis, involvement of more than one organ, and mouth or GI involvement were each associated with shorter median time to failure; of these, acute GvHD grade III-IV (140 days vs. 291 days; HR: 1.46, [1.01, 2.13], p=0.047), severe cGvHD at initial diagnosis (164 days vs. 294 days, HR: 1.75, [1.20, 2.55], p=0.004) and involvement of more than one organ (226 days vs. 508 days, HR: 1.75, [1.11, 2.75], p=0.02) were each associated with significantly shorter median time to failure on multivariate analysis. Matched related donor status was associated with significantly longer median time to failure on both univariate and multivariate (329 days vs. 189 days, HR: 0.63, [0.47, 0.85], p=0.02) analysis. We developed a risk score, assigning one point for each of the high-risk characteristics associated with frontline steroid treatment failure (severe acute or chronic GvHD, involvement of more than one organ and unrelated donor). With increasing points on the score, a progressively shorter median time to failure was observed: risk score 0-1 (n=136: 346 days [291, 673]), 2 (n=85: 217 days [164, 304]), 3-4 (n=36: 103.5 days [59-179], p<0.001). Increasing points on the score were also associated with reduced FFS at 1 (data not shown) and 2 years: risk score 0-1 (n=136: 38.3% [29.5, 47.0]), 2 (n=85: 16.1% [8.3-26.2]), 3-4 (n=36: 8.5% [1.8-22.1], p<0.001).

Conclusion: Unrelated donor use, severe acute/chronic GvHD and ≥2 involved organs are high risk characteristics for frontline steroid failure. Future research should assess interventions including incorporation of newer agents such as ruxolitinib, belumosudil or axatilimab in addition to corticosteroids in these high-risk patients.