There have been considerable changes in the practice of allogeneic transplantation over the last several years. In this study, we evaluate the impact of these changes on chronic graft-versus-host disease (cGVHD), the most important long-term complication. Our study utilized the central database of the CIBMTR to describe time trends for cGVHD incidence, presentation, nonrelapse mortality and overall survival from 1995-2007. The 12-year period was divided into three time periods- 1995-1999, 2000-2003, 2004-2007 to allow for the systematic analysis and description of trends. The trends of cGVHD severity (mild, moderate, severe), type of onset (progressive, interrupted, de novo), and major organ involved over the past years were analyzed using chi-square tests. Cumulative incidence was used to estimate cGVHD incidence and nonrelapse mortality. Death/ 2nd transplant/donor cell infusion/relapse were treated as competing risks for estimation of cumulative incidence of cGVHD. A test of trend was used to estimate HR across time periods. Multivariate logistic regression analysis was used to determine clinical variables affecting the incidence of cGVHD, including the time period of transplant. The study included 26,563 patients with acute leukemia (AML= 10,737, ALL= 6756), chronic myeloid leukemia (n=6341) and myelodysplastic syndrome (n=2729). In the univariate analysis, the incidence of cGVHD was significantly increased in the recent time periods (HR= 1.14, p for test of trend <0.0001). This trend persisted when the incidence was evaluated by donor type (HLA identical sibling-HR=1.17; unrelated donor-HR=1.07; cord blood-HR=1.24, all p < 0.01), graft type (PBSC-HR=1.19; cord blood-HR=1.24, p < 0.01), or conditioning intensity (myeloablative-HR=1.13; reduced intensity-HR=1.16, p< 0.01). In mismatched-related donors (HR=1.08, p=0.24) and bone marrow grafts (HR=1.01,p=0.54) there was no significant change in the incidence of cGVHD. Progressive cGVHD was found to be less frequently diagnosed over time, possibly from the recognition of the late acute classification introduced in 2005. Extensive, moderate and severe categories of cGVHD were more frequent in the two most recent time periods (2000-2003 and 2004-2007), as compared to the earliest time period (1995-1999). Skin was more frequently involved in the recent time period (2004-2007), with greater association of skin involvement at maximum severity reported in peripheral blood (33%), compared with bone marrow (25%), thus graft type appears to have impacted cGVHD presentation. In multivariate analysis, the recent time period (2004-2007) was associated with higher risk of cGVHD when compared to the two earlier time periods (2004-2007 versus 1995-1999, OR 1.19, p<0.0001; 2004-2007 versus 2000-2003, OR 1.13, p=0.002). Use of bone marrow with an unrelated donor (matched or mismatched), and peripheral blood graft with all categories of donor group were associated with higher risk of cGVHD, as compared to the use of bone marrow with a matched sibling donor. The risk of cGVHD was similar between marrow with a matched sibling donor and cord blood (matched or mismatched). In conclusion, this analysis of cGVHD trends over time demonstrates an increased cGVHD incidence in recent years despite controlling for factors in the donor, the graft and the conditioning that are associated with that trend, and serves as a useful reference for future research in the management of cGVHD.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.