Temitope Fisayo, Ceri Evans, Kuda Mutasa, Dagmar Alber, Moira Spyer, Sandra Rukobo, Mutsa Bwakura-Dangarembizi, Victor Musiime, Kusum Nathoo, Adeodata Kekitiinwa, Cissy Kityo, Diana M Gibb, NJ Klein, A Sarah Walker, Andrew J. Prendergast
BACKGROUND: Cytomegalovirus (CMV) co-infection is associated with mortality in adults with HIV, but whether CMV is associated with mortality in children with HIV remains uncertain. METHODS: In 498 children (median age 6.3 years, interquartile range 2.3-9.6) enrolled in the ARROW trial (ISRCTN24791884) in Uganda (336/498) and Zimbabwe (162/498) selected using a case-cohort design, CMV was quantified using real-time polymerase chain reaction at initiation of non-nucleotide reverse transcriptase inhibitor-based antiretroviral therapy (ART), 12-weeks post-ART, and 84-weeks post-ART. Associations between baseline CMV viraemia and mortality were evaluated using multivariable models, adjusting for baseline HIV viral load, CD4+ percentage, and IL-6 concentrations. RESULTS: Baseline CMV viraemia was associated with mortality, with relationships differing by country and assay. In Zimbabwe (assay limit of detection 20 copies/mL), 119/162 (73%) children had detectable CMV, and each log10 higher CMV viral load was associated with over 2-fold higher mortality (adjusted hazard ratio (aHR)=2.74; 95% confidence interval (CI) 1.57-4.77). In Uganda (assay limit of detection 120 copies/mL), 89/336 (26%) children had detectable CMV viraemia, which was associated with 3-fold higher mortality compared to undetectable CMV (aHR=3.15; 95%CI 1.11-8.93). In a subset of 48 children with immunophenotyping data, we found no evidence that CMV was associated with immune activation. CONCLUSIONS: CMV viraemia is independently associated with mortality in children with HIV starting ART in sub-Saharan Africa. Future studies should define the underlying mechanisms and evaluate whether suppressing CMV viraemia reduces mortality in children with HIV.