The cumulative burden of double-stranded DNA virus detection after allogeneic HCT is associated with increased mortality

Abstract

Strategies to prevent active infection with certain double-stranded DNA (dsDNA) viruses after allogeneic hematopoietic cell transplantation (HCT) are limited by incomplete understanding of their epidemiology and clinical impact. We retrospectively tested weekly plasma samples from allogeneic HCT recipients at our center from 2007 to 2014. We used quantitative PCR to test for cytomegalovirus, BK polyomavirus, human herpesvirus 6B, HHV-6A, adenovirus, and Epstein-Barr virus between days 0 and 100 post-HCT. We evaluated risk factors for detection of multiple viruses and association of viruses with mortality through day 365 post-HCT with Cox models. Among 404 allogeneic HCT recipients, including 125 cord blood, 125 HLA-mismatched, and 154 HLA-matched HCTs, detection of multiple viruses was common through day 100: 90% had ≥1, 62% had ≥2, 28% had ≥3, and 5% had 4 or 5 viruses. Risk factors for detection of multiple viruses included cord blood or HLA-mismatched HCT, myeloablative conditioning, and acute graft-versus-host disease (P values < .01). Absolute lymphocyte count of <200 cells/mm³ was associated with greater virus exposure on the basis of the maximum cumulative viral load area under the curve (AUC) (P = .054). The maximum cumulative viral load AUC was the best predictor of early (days 0-100) and late (days 101-365) overall mortality (adjusted hazard ratio [aHR] = 1.36, 95% confidence interval [CI] [1.25, 1.49], and aHR = 1.04, 95% CI [1.0, 1.08], respectively) after accounting for immune reconstitution and graft-versus-host disease. In conclusion, detection of multiple dsDNA viruses was frequent after allogeneic HCT and had a dose-dependent association with increased mortality. These data suggest opportunities to improve outcomes with better antiviral strategies.

Figure 1.

Proportion of patients with dsDNA virus detection. (A) Proportion of patients with individual viruses detected through day 100 post-HCT, as well as the proportion with additional viruses detected. (B) Venn diagram depicting the number of patients with detection of dsDNA viruses alone and in combination; ovals are not to scale. (C) Proportion of patients with detection of each virus by week post-HCT.

Figure 2.

Cumulative incidence plot of time to any dsDNA virus detection by day 100 post-HCT. (A) The cumulative incidence of each category of the number of viruses detected is demonstrated in the entire cohort. The cumulative incidences are higher in each category among cord blood HCT recipients (B) and HLA-mismatched HCT recipients (C) than among HLA-matched HCT recipients (D).

Figure 3.

Histogram of the weekly proportion of patients with concurrent detection of multiple dsDNA viruses. The proportion of patients with multiple viruses detected per week in the entire cohort (A), cord blood HCT recipients (B), HLA-mismatched related or unrelated HCT recipients (C), and HLA-matched related or unrelated HCT recipients (D).

Figure 4.

Plots of time to overall mortality and NRM through day 365 post-HCT among day-100 survivors (n = 358). Kaplan-Meier plots of overall mortality stratified by the cumulative number of different viruses detected at any time by day 100 (A) and cumulative maximum number of viruses detected at the same time by day 100 (B). Cumulative incidence curves of NRM stratified by the cumulative number of different viruses detected at any time by day 100 (C) and cumulative maximum number of viruses detected at the same time by day 100 (D).

Figure 5.

Forest plot demonstrating the association of the cumulative viral load AUC with overall mortality from adjusted Cox models. (A) Models of the association of the cumulative viral load AUC with overall mortality in the entire cohort and among day-30 survivors with adjustment for immune reconstitution parameters. (B) Models that considered the effect of the cumulative viral load AUC within strata of acute GVHD to account for effect modification. In all models, viral detection was included as a time-dependent variable for overall mortality by day 100 and a time-invariant variable for overall mortality between day 101 and day 365 (because virus testing stopped at day 100). ^aAdjusted for age, HCT comorbidity index, underlying disease risk, conditioning regimen intensity, and acute GVHD (time dependent), and cumulative steroid dose AUC. ^bAdjusted for age, HCT comorbidity index, underlying disease risk, conditioning regimen intensity, and acute GVHD (time dependent), day 30 neutrophil engraftment, day 30 ALC < 200 cells/mm³, day 30 AMC < 300 cells/mm³, and cumulative steroid dose AUC. ^cAdjusted for age, HCT comorbidity index, conditioning regimen intensity, and underlying disease risk.