T Cell Costimulation Blockade with CTLA4-Ig (Abatacept) for Acute Gvhd Prevention in HLA Matched and Mismatched Unrelated Donor Transplantation: Results of the First Phase 2 Trial

Abstract

We performed a Ph2 trial in adults and children to test abatacept for AGVHD prevention (‘ABA2', Clinicaltrials # NCT01743131), based on our promising preclinical and pilot patient data. ABA2 had 2 cohorts: A) HLA-mismatched (‘7/8', n = 43), a single-arm study with pre-specified CIBMTR matched analysis (vs CNI+MTX or CNI+MTX+ATG). B) HLA-matched (‘8/8', n = 142), randomized double-blind, comparing CNI+MTX+placebo vs CNI+MTX+ABA (‘ABA'). For each ABA arm, patients received 4 doses of 10mg/kg on d -1,5,14,28. ABA2 was designed as a screening Ph2 trial, with relaxed Type 1 error (0.2) and standard Type 2 error (0.2). Power analysis assumed ABA would decrease Gr 3-4 AGVHD from 30%–>10% in 7/8s and from 20%–>10% in 8/8s. Here we report top-line results for the 7/8s (median f/u = 708d, 264-1491) and 8/8s (median f/u = 369d, 180-1175). Reduced Grade 3-4 AGVHD: ABA was associated with decreased d180 Gr 3-4 AGVHD. In 7/8s, Gr 3-4 AGVHD = 2.5% in ABA (1 event/43 patients) vs 31% (CNI+MTX) and 22% (+ATG), 1 sided p = 0.001, 0.005,). In 8/8s, Gr 3-4 AGVHD = 6.85 % (5 events/73 patients) in ABA vs 14.6% (10 events) in placebo, 1 sided p = 0.068). Reduced Grade 2-4 AGVHD in 8/8s: ABA was associated with decreased d180 Gr 2-4 AGVHD. In 7/8s, Gr 2-4 AGVHD = 42% (ABA) vs 54% (CNI+MTX) and 45% (+ATG, 1 sided p = 0.098, 0.25). In 8/8s, Gr 2-4 AGVHD = 44.5% in ABA vs 62.3% in placebo (1 sided p = 0.004). Chronic GVHD: For 7/8s, 1 yr CGVHD = 38.8% (ABA) vs 43.5% (CNI+MTX) and 35.5% (+ATG, p = 0.4, 0.99). For 8/8s, follow up is still too short to adequately evaluate, and adjudication is ongoing. No Increase In Relapse: In 7/8s, relapse = 9.37% at 1 y vs 12.9% in CNI+MTX and 13.6% in +ATG (p = 0.115 and 0.085). In 8/8s, relapse = 13.8% at 1y (ABA) vs 20.5% (placebo, p = 0.7). Remarkably, in 7/8s, where median follow up=708d, there have been no further relapses reported in ABA2, with 9.37% relapse at 2 yr vs 20.63% (CNI+MTX) and 23.4% (ATG), despite matched disease risk. Statistically significant survival advantage in 7/8s: TRM: For 7/8s, 1 yr TRM = 10.5% (ABA) vs 32.7% (CNI+MTX) and 26% (+ATG, p = 0.024, 0.365). For 8/8s, TRM = 7.1% vs 14.6% at 1 yr (p = 0.5). Relapse-Free Survival (RFS): For 7/8s, RFS = 73.7% (ABA) vs 38.7% in CNI+MTX and 48.7% in +ATG (p = 0.001 and 0.027). For 8/8s, RFS = 79.1% for ABA vs 64.9% (placebo, p = 0.38). OS: For 7/8s, OS = 71% (ABA) vs 47.5% (CNI+MTX) and 58% (+ATG, p = 0.01 and 0.145). For 8/8s, OS = 83.2% (ABA) vs 76.6 (placebo, p = 0.32). Our results suggest that short-course (4 doses) ABA can safely prevent AGVHD without compromising relapse. While this was a modestly sized study (7/8s: n = 43, 8/8 n = 142), the comparative event size for 7/8s was high enough that the protective effect of ABA was highly statistically significant. For 8/8s, there was a statistically significant improvement for Gr 2-4 GVHD and a trend toward an advantage for ABA in all other parameters. These results are the first to demonstrate efficacy of in vivo T cell costimulation blockade in preventing AGVHD.