Conclusion
This analysis confirmed several previously reported risk factors for grade 2 to 4 acute GVHD. However, several new factors were identified whereas others are no longer significant. These new data may facilitate individualized risk estimates and raise several interesting biologic questions.
Methods
Acute GVHD risk factors were analyzed in 1,960 adults after HLA-identical sibling myeloablative transplant for acute myeloid leukemia (AML), acute lymphocytic leukemia (ALL), or chronic myeloid leukemia (CML) reported by 226 centers worldwide to the Center for International Blood and Marrow Transplant Research from 1995 to 2002. Outcome was measured as time from transplant to onset of grade 2 to 4 acute GVHD, with death without acute GVHD as a competing risk.
Purpose
Acute graft-versus-host disease (GVHD) causes substantial morbidity and mortality after human leukocyte antigen (HLA)-identical sibling transplants. No large registry studies of acute GVHD risk factors have been reported in two decades. Risk factors may have changed in this interval as transplant-related techniques have evolved. Patients and
Results
Cumulative incidence of grade 2 to 4 acute GVHD was 35% (95% CI, 33% to 37%). In multivariable analyses, factors significantly associated with grade 2 to 4 acute GVHD were cyclophosphamide + total-body irradiation versus busulfan + cyclophosphamide (relative risk [RR] = 1.4; P < .0001), blood cell versus bone marrow grafts in patients age 18 to 39 years (RR = 1.43; P = .0023), recipient age 40 and older versus age 18 to 39 years receiving bone marrow grafts (RR = 1.44; P = .0005), CML versus AML/ALL (RR = 1.35; P = .0003), white/Black versus Asian/Hispanic race (RR = 1.54; P = .0003), Karnofsky performance score less than 90 versus 90 to 100 (RR = 1.27; P = .014), and recipient/donor cytomegalovirus-seronegative versus either positive (RR = 1.20; P = .04). Stratification by disease showed the same significant predictors of grade 2 to 4 acute GVHD for CML; however, KPS and cytomegalovirus serostatus were not significant predictors for AML/ALL.