The Bidirectional Relationship Between Cytomegalovirus Replication and Graft-Versus-Host Disease - a Retrospective Single Center Study.

Abstract 2236

Poster Board II-213

Cytomegalovirus (CMV) infection and graft-versus host disease (GVHD) are important complications after allogeneic HSCT with a clear link. Multiple studies show that GVHD and its treatment put patients at risk for CMV reactivation. Data on CMV infection as a cause of GVHD, in contrast, are controversial. The association of pre-transplant CMV serology with GVHD development and reduced rates of chronic GVHD after preemptive CMV treatment are indicative of such an association. However, analyses of the direct impact of CMV infection on GVHD are rare; a recent small study found no effect of CMV replication on subsequent development of acute GVHD (Wang et al, BMT 2008). We analyzed in a single centre study the association of CMV reactivation with acute GVHD in 517 patients treated between 1993 and 2008. 59% of patients were male, median age was 42 years (range 16 to 70). Diagnoses were AML (31%), ALL (16%), CML (15%), MDS/MPN (13%), lymphoma (21%), and other (4%). Conditioning regimens were Cy/TBI ±/- etoposide (49%), Cy/Bu (17%), fludarabine and TBI (16%), or others (18%). GVHD prophylaxis consisted of CyA/MTX (78%) or CyA/MMF (21%). Donors were HLA-identical siblings (65%), other family members (4%), or unrelated donors (31%). We made use of a standardized CMV policy over the last decades. CMV reactivation was monitored using real-time polymerase chain reaction or pp65 antigenemia assay weekly in patients without infection, twice weekly in patients with CMV replication. CMV was preemptively treated with gancyclovir or foscarnet. To determine the correlation of CMV infection with acute GVHD, we used a stringent Cox regression model, in which CMV replication was modeled as a time-dependent covariate becoming positive on the day of the first detection of CMV and negative on the first negative assay thereafter. Multiple episodes of CMV replication were considered. Acute GVHD was modeled as a time-dependent covariate in models with CMV infection as endpoint. Hazard ratios (HR) were adjusted for patient age, disease, disease stage, donor type, stem cell source, conditioning regimen and GVHD prophylaxis. The analysis was restricted to the time from transplant to day 100. CMV reactivation was detected at least once in 16% (84/517) of patients at a median of 33 (range 1-95) days after HSCT. Median duration of CMV reactivation was 8.5 days (range 2-62). 19 patients showed multiple episodes of CMV replication. Donor and recipient serostatus significantly influenced the day 100 cumulative incidence of CMV infection: D-/R- (N=173) 6%; D±/R- (N=61) 10%; D±/R± (N=128) 25%; and D-/R± (N=99) 37%, p<0.001. The cumulative incidence for any acute GVHD (grade I-IV) was 67% (95% CI 56-78%) with a median onset time at day 14 (range day 5-94); the cumulative incidence for severe acute GVHD (grade II-IV) was 48% (95% CI 40-56%). When both endpoints (CMV, GVHD) were combined, 150 patients (29%) experienced neither GVHD nor CMV reactivation, 281 (54%) GVHD only, 19 (4%) CMV reactivation only, and 67 (13%) both CMV reactivation and GVHD. Of the 67 patients with both GVHD and CMV, 46 (69%) developed GVHD prior to CMV reactivation, 17 (25%) developed GVHD during CMV reactivation, and 4 (6%) developed GVHD after CMV reactivation. Cox modeling revealed that presence of GVHD grade II-IV increased the risk of CMV infection (HR 1.61, 95% CI 1.03-2.52, p=0.04). Similarly, patients were at increased risk of developing acute GVHD during phases of CMV replication (grades I-IV: HR 2.23, 95% CI 1.39-3.81, p=0.001; grades II-IV: HR 2.00, 95% CI 1.08-3.72, p=0.03). GVHD grade was not influenced by concomitant CMV reactivation (median grade II, in patients with or without CMV reactivation). The overall proportion of GVHD that occurred during phases of CMV replication was small (3% versus 64% occurring in CMV non replicating patients). Even if GVHD occurring after resolution of CMV reactivation was additionally taken into account, the majority of GVHD occurred without preceding CMV infection (63% versus 4%). These data describe the complex relationship between CMV infection and GVHD. We confirm previous studies that GVHD (and GVHD therapy) can induce CMV infection. We describe as well that patients with active CMV replication have a significantly higher risk of developing GVHD compared to patients without CMV replication. However, the proportion of GVHD that could be linked to CMV reactivation was small in this population with a low overall incidence of CMV reactivation.