“…Recent data suggest that, although CMV-I may be protective against relapse, particularly in HSCT recipients with AML undergoing myeloablative allogeneic HSCT, this effect is superseded by an increased non-relapse mortality risk. 2,4,6,8,9,30 When analyzed using a robust Markov multi-state model, accounting for AML and myeloablative conditioning as well as time-dependent GVHD, we observed that the directionality of the effect of CMV-I on relapse favored a greater hazard of relapse, although the CI was greater than one. Using the same Markov multi-state model to account for the effect of CMV-I on mortality independent of the competing risk of relapse, we noted a plausible association, although again this did not reach statistical significance (lower limit of the CI was 0.9).…”