Graft-Versus-Tumor Effect In Patients with Graft-Versus-Host Disease Defined by NIH Consensus Criteria

Abstract 2343 Background. The National Institutes of Health consensus criteria (NCC) for chronic graft-versus-host disease (GVHD) are based on clinical manifestations rather than onset time after transplantation. Previous studies about the feasibility of the new criteria have been performed in patients who survived beyond 100 days after transplantation and had GVHD presenting after day 100. In order to investigate and compare the clinical impact of acute and chronic GVHD by NCC on survival, a cohort including patients with GVHD presenting ‘before' as well as ‘after' day 100 is needed. Additionally, a proper statistical tool should be applied to clarify the effect of GVHD on survival outcomes, because the occurrence of GVHD is a well known time-dependent event. In this context, we examined a cohort including all patients who underwent allogeneic stem cell transplantation (SCT) in order to investigate clinical impact of acute or chronic GVHD by NCC on survival outcomes. Methods. We retrospectively investigated 771 patients who underwent allogeneic SCT between January 2002 to December 2008. To study a homogenous cohort, patients received 2 allogeneic SCT and/or donor-lymphocyte-infusion was excluded. We used time-dependent analyses to reveal the effect of GVHD on survival outcomes. In particular, to approach the effects of GVHD as a time-dependent covariate on competing risks [relapse or transplant-related mortality (TRM)], we analyzed cause-specific hazards by Cox proportional hazards regression model (Cartese G, Andersen PK. Biometrical Journal 2009;51:138–158). Results. The median age was 36 years (range, 15–68). Patients had various hematologic malignancies (AML/ALL/CML/MDS/MM, 361/226/86/67/31, respectively) and were transplanted from matched sibling (n=485), well-matched unrelated donors (n=154), partially-matched unrelated donors (n=101), and mismatched unrelated donors (n=31). Conditioning regimens consisted of myeloablative (n=536) and reduced-intensity regimens (n=235). GVHD prophylaxis consisted of cyclosporine and short-course methotrexate for related SCT and tacrolimus and short-course methotrexate for unrelated SCT. Among 771 patients, 540 patients were diagnosed with GVHD after transplantation. According to onset time of GVHD, in 348 patients GVHD developed within 100 days after SCT, whereas in 156 patients GVHD occurred more than 100 days after SCT. Using the NCC, we classified patients as three categories regardless of onset time: (1) acute GVHD (n=215), if patients had only acute features (no chronic features) during the course of GVHD, (2) acute GVHD with following chronic GVHD (n=118), if patients had any chronic features after the occurrence of acute GVHD, and (3) chronic GVHD (n=207), if patients had any chronic features at the onset of GVHD. Multivariate analyses using cause-specific hazards revealed that acute GVHD was significantly associated with lower relapse incidence [HR (95% CI) 0.65 (0.44–0.94), P=0.023]. However, it did not influence disease-free survival (DFS, P=0.602) and overall survival (OS, P=0.294) due to higher TRM in patients with acute GVHD [HR (95% CI) 1.74 (1.15–2.61), P=0.008]. Acute GVHD with following chronic GVHD was also associated with lower relapse incidence [HR (95% CI) 0.31 (0.15–0.66), P=0.002] and higher TRM [HR (95% CI) 2.89 (1.65–5.04), P<0.001], which did not influence DFS (P=0.567) and OS (P=0.820). On the other hand, the occurrence of chronic GVHD significantly reduced relapse rates [HR (95% CI) 0.46 (0.29–0.75), P=0.002] without increasing TRM (P=0.177). Indeed, this effect was translated into improved DFS [HR (95% CI) 0.67 (0.48–0.95), P=0.023]. Conclusions. Our data show that both acute and chronic GVHD by NCC reduced the risk of relapse, demonstrating the presence of graft-versus-tumor (GVT) effect by the occurrence of GVHD and no difference according to the clinical features (acute or chronic) by NCC. However, survival benefit was only observed in chronic GVHD by NCC compared to acute GVHD with/without following chronic GVHD by NCC due to higher TRM. This study represents the first one demonstrating the GVT effect of GVHD defined by NCC in a cohort including GVHD presenting ‘before' as well as ‘after' day 100 using appropriate time-dependent analyses by cause-specific hazards by Cox proportional hazards regression model. Disclosures: No relevant conflicts of interest to declare.