Bidirectional Temporal Association Between Cytomegalovirus Reactivation and Graft-Versus-Host Disease Following Allogeneic Hematopoietic Stem Cell Transplantation: A Single-Center Real-World Cohort Study

Background: Cytomegalovirus (CMV) reactivation and graft-versus-host disease (GVHD) are major causes of morbidity and non-relapse mortality following allogeneic hematopoietic stem cell transplantation (allo-HSCT). Although their association has been extensively investigated, the temporal relationship between these complications and their impact on survival remain incompletely understood, particularly in centers where routine letermovir prophylaxis is unavailable and CMV is managed using a pre-emptive treatment strategy. This study evaluated the bidirectional temporal association between CMV reactivation and GVHD using time-dependent analyses and assessed their impact on survival in a real-world allo-HSCT cohort. Methods: We retrospectively analyzed 100 consecutive adult patients who underwent allo-HSCT for hematologic malignancies between January 2016 and February 2025. CMV surveillance was performed by weekly quantitative CMV-DNA PCR, and reactivation was managed using a standardized pre-emptive treatment strategy. Patients who relapsed or died within the first 100 days after transplantation were excluded. The incidence, temporal sequence, risk factors, and prognostic impact of CMV reactivation and GVHD were evaluated. Overall survival (OS) and relapse-free survival (RFS) were estimated using the Kaplan–Meier method, while temporal associations were assessed using time-dependent Cox regression analyses. Results: CMV reactivation occurred in 72 patients (72%), whereas GVHD developed in 51 patients (51%). Among patients who experienced both complications, CMV reactivation preceded GVHD in 32 patients, while GVHD preceded CMV reactivation in 14 patients. Older recipient age (42.3 ± 13.5 vs. 35.4 ± 11.2 years, p = 0.018) and older donor age (37.2 ± 10.9 vs. 32.5 ± 9.7 years, p = 0.048) were associated with CMV reactivation. Older donor age was also associated with GVHD development (38.1 ± 10.7 vs. 33.6 ± 10.5 years, p = 0.037). The median time to CMV reactivation was 37 days, and CMV end-organ disease occurred in 13.9% of affected patients. No significant differences in OS or RFS were observed according to CMV reactivation or GVHD status in the day-100 landmark cohort. In the AML subgroup, both CMV reactivation and GVHD showed a trend toward inferior OS and RFS without reaching statistical significance. Time-dependent Cox regression demonstrated that CMV reactivation independently increased the subsequent risk of GVHD (HR 2.93, 95% CI 1.51–5.69; p = 0.001), whereas GVHD also independently increased the subsequent risk of CMV reactivation (HR 1.98, 95% CI 1.06–3.71; p = 0.032). Conclusions: CMV reactivation was highly prevalent after allo-HSCT and frequently preceded GVHD, supporting a bidirectional temporal relationship between these complications. Older donor age was associated with both CMV reactivation and GVHD. Although survival did not significantly differ according to CMV or GVHD status in this day-100 landmark cohort, clinically important CMV-related complications, including end-organ disease and platelet engraftment failure requiring eltrombopag, remained common. These findings indicate that a standardized pre-emptive strategy did not completely prevent CMV-associated morbidity. Because letermovir was not evaluated in this cohort, any potential benefit of prophylaxis should be interpreted as an inference from external evidence rather than as a direct finding of this study. Prospective multicenter studies incorporating immune reconstitution analyses are warranted.