Cytomegalovirus Viremia Predicts Postdischarge Mortality in Kenyan HIV-Exposed Uninfected Children

Abstract

Background: Cytomegalovirus (CMV) viremia is associated with mortality in severely ill immunocompetent adults and hospitalized children with HIV (CWH). We measured CMV viremia in HIV-exposed and -unexposed Kenyan children aged 1-59 months discharged from hospital and determined its relationship with postdischarge mortality.

Methods: CMV DNA levels were measured in plasma from 1024 children (97 of which were HIV exposed uninfected [HEU], and 15 CWH). Poisson and Cox proportional hazards regression models were used to identify correlates of CMV viremia ≥ 1000 IU/mL and estimate associations with 6-month mortality, respectively.

Results: CMV viremia was detected in 31% of children, with levels ≥ 1000 IU/mL in 5.8%. HIV infection, age < 2 years, breastfeeding, and midupper arm circumference < 12.5 cm were associated with CMV viremia ≥ 1000 IU/mL. Among HEU children, CMV ≥ 1000 IU/mL (hazard ratio [HR] = 32.0; 95% confidence interval [CI], 2.9-354.0; P = .005) and each 1-log increase in CMV viral load (HR = 5.04; 95% CI, 1.7-14.6; P = .003) were associated with increased risk of mortality. CMV viremia was not significantly associated with mortality in HIV-unexposed children.

Conclusions: CMV levels at hospital postdischarge predict increased risk of 6-month mortality in Kenyan HEU children. CMV suppression may be a novel target to reduce mortality in HEU children.

Clinical trial registration: NCT02414399.

Keywords: HIV-1; azithromycin; child mortality; cytomegalovirus; postdischarge mortality.

Figure 1.

Cytomegalovirus (CMV) detection and levels in Kenyan infants, toddlers, and children at hospital discharge. A, Proportion of children with CMV detection above the limit of detection (black bars, P < .001 for comparison across ages) and ≥ 1000 IU/mL (gray bars, P < .001 for comparison across ages). B, Boxes show CMV viral load medians and interquartile ranges among children with detectable viremia at hospital discharge. Whiskers show upper adjacent values and outside values. P < .0001 for younger than 1 year versus 1–2 year olds; P = .07 for 1–2 versus 2–5 year olds.

Figure 2.

Six-month postdischarge mortality in cytomegalovirus (CMV) viremic and aviremic children stratified by HIV exposure. Kaplan-Meier functions are shown for children grouped by CMV viremia and HIV exposure status. P values were determined by log-rank test for equality of survivor function. CMV viremia was categorized as a binary variable ≥ 1000 IU/mL (dashed line) versus aviremic or < 1000 IU/mL (solid line) for (A) all children, n = 1024, (B) HIV-unexposed children, n = 906, (C) HIV-exposed uninfected children, n = 103, and (D) children with HIV, n = 15.