Celiac Disease Risk Determined By HLA-DQ Genotype Protects Against Gvhd in a Dose-Responsive Manner

Graft versus host disease (GVHD) causes significant morbidity and mortality after allogeneic stem cell transplant (SCT). Little is known about genetic determinants of GVHD risk, including human leukocyte antigen (HLA) genotypes outside of HLA-mismatch. HLA-DQ2 and DQ8 genotypes are associated with celiac disease (CD) and are present in ≥90% of patients with CD. Both GVHD and CD are systemic inflammatory conditions that may have T cell-mediated enteropathy. We hypothesized HLA-DQ alleles associated with CD risk would confer increased risks of acute (a) and chronic (c) GVHD in pediatric SCT recipients. We performed a retrospective cohort study of children aged 0-21 years undergoing first allogeneic SCT at the Children's Hospital of Philadelphia from 10/1/12 — 7/1/16. Patients with primary graft failure or missing HLA-DQ genotypes were excluded. We grouped patients by low, moderate or high CD risk (Clin Gastroenterol Hepatol 2009, 7:966). Primary outcomes were 1) 100-day aGVHD incidence; 2) day 100 — 1-year cGVHD incidence; and 3) time to GVHD with relapse, death, secondary graft failure, second SCT, or donor lymphocyte infusion as competing risks. Logistic regressions were used to calculate odds ratios (OR) and 95% confidence intervals (CI). Sub-distribution hazard models were used to estimate crude and adjusted sub-distribution hazard ratios (sHR) in time-to-event analyses. Multivariate models were adjusted for race-ethnicity, malignant/nonmalignant SCT indication, donor/mismatch, and graft source. We identified 167 patients (mean age 9.2 ±6.2 years). Proportions by CD risk were: low-risk n=108 (64.7%), moderate n=33 (19.8%), and high n=26 (15.6%). Day 100 incidence of ≥grade 2 aGVHD was 14.4% (8.4% had ≥grade 3). cGVHD 1-year incidence was 42.7% (14% had extensive). CD risk was protective against GVHD but did not reach statistical significance (Table 1). Risk of clinically significant GVHD (≥grade 2 or any chronic) was reduced in children with CD risk in the adjusted time-to-event model (moderate: sHR 0.42, 95% CI 0.18 — 0.95, p=0.037; high: sHR 0.31, 95% CI 0.12 – 0.8, p=0.016; Figure 1A). The protective effect of high CD risk was more pronounced for severe GVHD (≥grade 3 or chronic extensive) but did not reach statistical significance (adjusted sHR 0.14, 95%CI 0.02-1.09, P = 0.061; Figure 1B). In contrast to our hypothesis, CD risk by HLA-DQ is protective against GVHD in a dose-responsive manner. Reports of HLA-DQ2 and DQ8 protecting against other non-CD autoimmune diseases (rheumatoid arthritis and inflammatory bowel disease) support our observations. These results require confirmation and evaluation in a larger patient sample to define this novel GVHD risk modifier more robustly.