Response Endpoints for Acute Graft-Versus-Host Disease Treatment Trials

Background: We evaluated short-term response endpoints for clinical trials testing initial treatment of acute graft-versus-host disease (GVHD). We postulated that the definition of response should include a reduced symptom burden and improved failure-free survival (FFS), with failure defined as death, recurrent malignancy or initiation of second-line systemic treatment. We also postulated that the response endpoint should include an upper limit of the steroid dose at the time of assessment.

Methods:  In a cohort of 227 adult patients who received initial systemic steroid treatment for grades IIB-IV acute GVHD between 2000 and 2005, treatment response was evaluated at days 28 and 56 after starting treatment. Standard definitions were used to define complete response (CR) and traditional partial response (PR). Very good partial response (VGPR) was defined as a subcategory of traditional PR when patients had minor rash, minor elevations of total serum bilirubin or minor gastrointestinal symptoms but otherwise met criteria for CR (2009 Joint Statement, BBMT). Sensitivity and specificity analyses were used to evaluate each response definition in predicting FFS at 6 months after initial treatment. Averages of sensitivity and specificity were used to evaluate the trade-off between sensitivity and specificity.

Results: Response rates were 33% CR, 24% VGPR and 10% other PR at day 28, and 40% CR, 14% VGPR and 7% other PR at day 56. Residual symptom burden at days 28 and 56 after treatment was lower in patients with VGPR than in those with other PR, indicating that VGPR is preferred over traditional PR. In evaluating day 28 and day 56 response endpoints as predictors of FFS at 6 months, loss of sensitivity outweighed the gain of specificity with CR compared to CR+VGPR (Table). Sensitivity for CR+VGPR was similar at days 28 and 56, but specificity was higher at day 56 than at day 28. Since CR+VGPR at day 56 still had a 26% false-positive rate (i.e., 74% specificity), we evaluated response definitions that incorporated an upper limit of the steroid dose as an additional criterion of success (Figure). As shown by the averages, sensitivity and specificity showed a balanced trade-off at prednisone limits between 2.0 and 0.5 mg/kg/day, but loss of sensitivity outweighed the gain of specificity at limits below 0.5 mg/kg/day. Incorporation of prednisone doses ≤0.5 mg/kg/day in the CR+VGPR endpoint definition at day 56 decreased the false-positive rate to 15%. When recurrent malignancy was excluded from the definition of failure after day 56, this endpoint had a false-positive rate of only 4%.

Conclusion: Our results support the use of CR+VGPR with an upper limit of the steroid dose at day 56 as the short-term response endpoint for acute GVHD treatment trials.

Table

	Response definition	Sensitivity	Specificity	Average
Day 28	CR	48%	81%	65%
	CR+VGPR	82%	66%	74%
Day 56	CR	64%	82%	73%
	CR+VGPR	83%	74%	78%

Figure