Early versus standard anti-thymocyte globulin administration in haploidentical transplantation using post-transplant cyclophosphamide: a pilot comparative study

Combining rabbit anti-thymocyte globulin (ATG) with post-transplant cyclophosphamide (PTCy) reduces graft-versus-host disease (GVHD) after haploidentical peripheral blood stem cell transplantation (haplo-PBSCT) but may increase cytomegalovirus (CMV) reactivation due to intensified immunosuppression. Because ATG exposure at the time of graft infusion is influenced by administration timing, we evaluated whether earlier ATG delivery modifies CMV outcomes while maintaining comparable observed transplant outcomes in this small pilot cohort. In this single-center, non-randomized pilot study, 36 adults with acute leukemia undergoing first myeloablative haplo-PBSCT with a uniform ATG/PTCy platform were analyzed. The intervention cohort received rabbit ATG 2.5 mg/kg/day on days - 9 to - 7 (n = 18), while the control cohort received the same total dose on days - 3 to - 1 (n = 18). All patients received busulfan/cyclophosphamide conditioning, PTCy (40 mg/kg on days + 3 and + 4), and cyclosporine. Letermovir was not used. The primary endpoint was CMV reactivation (> 500 copies/mL) within 90 days. The primary endpoint was not met: The 90-day cumulative incidence of CMV reactivation was similar between early and standard ATG cohorts (55.6% vs. 59.3%; p = 0.94). Early ATG was associated with numerically lower peak CMV viral loads (median 995 vs. 1,456 copies/mL; p = 0.86) and fewer cases of high-level viremia ≥ 1,000 copies/mL (22.2% vs. 33.3%). Neutrophil and platelet engraftment occurred earlier with early ATG (median 12.5 vs. 15 days, p < 0.001; and 14 vs. 17 days, p = 0.042). Rates of grade III-IV acute GVHD (11.1% in both groups), relapse, non-relapse mortality, and two-year overall survival were comparable. Exploratory analyses suggested numerically lower CMV viral burden and faster hematopoietic recovery in the early ATG cohort, but these observations are hypothesis-generating only. Given the non-randomized design, historical control, and limited sample size, no conclusions regarding efficacy preservation or superiority can be drawn. Confirmation in larger, prospective, ideally randomized studies incorporating immune monitoring is warranted.