Objectives: Current grading systems in acute GVHD can not effectively identify patients with poor prognosis at time of GVHD diagnosis. The aim of this study was to evaluate clinical or biological parameters at the onset of GVHD associated with poor prognosis of acute GVHD.

Methods: Among 257 patients (pts) who received an allogeneic stem cell transplant (SCT) after a myeloablative conditioning regimen between 1993 and 1999, 146 pts developed acute GVHD. One hundred-day cumulative incidence of acute GVHD was 57.3% (95%CI 51.1–63.4) whereas 100-day cumulative incidence of death non related to GVHD was 5.3% (95%CI 2.5–8.1). Patients with acute GVHD were analyzed for risk factors associated with non-relapse-mortality (NRM) using proportional cause-specific hazards models.

Characteristics of patients: Median age was 32 years (from 4 to 60 years). Sex ratio men/women was 1,4. Disease was defined as low risk (n=43) (acute leukemia in first remission, aplastic anemia without more than 20 blood transfusions before transplantation and chronic myeloid leukemia in first chronic phase) and high risk (n=99, all other diagnosis). Source of stem cell was bone marrow in 106, peripheral blood stem cells in 23 and cord blood in 13 pts. 87 patients received SCT from a sibling donor and 55 patients from an unrelated donor. 85 patients were HLA identical with their donor. Median follow-up was 76 months.

Results: At time of diagnosis, GVHD involved a single organ in the majority of the patients (109 pts, 78%): skin for 63, gut for 35, liver for 11 pts. 37 patients had multiple organs affected by GVHD: 34 patients had 2 organs and 3 had 3 organs (liver, skin and gut). The Table 1 shows acute GVHD grade from diagnosis to maximal grade.

Table 1

Maximal grade
Initial gradeNo GVH1234Total
No GVH 111 111 (43%) 
31 16 10 60 (23%) 
56 13 14 83 (32%) 
3 (1%) 
Total 111 (43%) 31 (12%) 72 (28%) 18 (7%) 25 (10%) 257 
Maximal grade
Initial gradeNo GVH1234Total
No GVH 111 111 (43%) 
31 16 10 60 (23%) 
56 13 14 83 (32%) 
3 (1%) 
Total 111 (43%) 31 (12%) 72 (28%) 18 (7%) 25 (10%) 257 

Overall 1-year NRM was 54.9% (95%CI 46.6–63.2). 57 patients experienced a failure to corticosteroid treatment and were considered as “non-responders”. Covariates tested for NRM predictive factors were gender, CMV status, conditioning regimen, source of SCT, HLA match, hematological disease, initial characteristics of GVHD and response to treatment after 7 days. Significant (p < 0.05) risk factors for NRM were non sibling donor (NRM: 64.5% [51.5–77.5]), absence of methotrexate in GVHD prophylaxis (NRM: 67.3 [50.2–84.5]), initial liver involvement (NRM: 80.2% [62.2 to 98.2]) and acute GVHD resistant to corticosteroids (NRM: 78.9 % [68.1–89.8]). Patients who were responder to corticosteroids had same NRM than patients who did not develop any GVHD. Albumin level (alb) was associated with NRM in patients with initial gut involvement (NRM 21.4% [0.0–44.0] in pts with alb > or = 35 and 61.1% [37.4–84.8] in pts with alb < 35 gr/L, p=0.035). Initial grade was not significantly associated with NRM: 1-year cumulative incidence was 52.4% (39.3–65.6), 56.3% (45.3–67.4) and 66.7% (0.0–00) in patients with initial grade I, II and III, resp. as compared to 26.2% (17.4–5.0) in patients without GVHD. In multivariate analysis, initial liver involvement (HR: 2.45 (1.46 to 4.12), p=0.0007) and non sibling donor (HR: 1.58 (1.02 to 2.43, p= 0.0039) were both associated with NRM.

Conclusion: In this study, classification of GVHD at time of diagnosis was not predictive for NRM. Initial liver involvement was the most important clinical predictor and may be considered in clinical management and need prospective clinical trials.

Disclosures: No relevant conflicts of interest to declare.

Author notes

Corresponding author

Sign in via your Institution