AI-generated research brief — verify at source
Empirical Valganciclovir Shows Signals of Early Mortality Reduction in HIV-Associated Infant Pneumonia
In a multicentre randomised trial of 558 HIV-positive infants hospitalised with severe pneumonia across six African countries, empirical valganciclovir did not significantly reduce all-cause mortality at day 15 (23% vs 27%; rate ratio 0.81, 95% CI 0.61–1.08, p=0.15) or at 12 months (43% vs 48%; HR 0.88, 95% CI 0.74–1.05, p=0.15) in the primary intention-to-treat analysis. However, a time-varying effects model suggested approximately 40% lower hazard of death during the first 15 days of treatment (adjusted HR 0.60, 95% CI 0.41–0.87, p=0.0063), indicating a potential early survival benefit that the trial’s primary analysis was not able to confirm conclusively.
What Was Studied
The EMPIRICAL trial investigated whether adding empirical oral valganciclovir — targeting cytomegalovirus (CMV) as a suspected but unconfirmed co-pathogen — to standard pneumonia care reduces all-cause mortality in infants living with HIV who were admitted to hospital with severe pneumonia in sub-Saharan Africa. Mortality in this clinical group remains very high, and the potential contribution of CMV infection to this burden provided the rationale for testing antiviral treatment even in the absence of confirmed CMV diagnosis.
How It Was Studied
This was a multicentre, open-label, 2×2 factorial, randomised controlled superiority trial conducted across 19 hospitals in Côte d’Ivoire, Malawi, Mozambique, Uganda, Zambia, and Zimbabwe. A total of 558 infants aged 28 to 365 days (median age 4.4 months) living with HIV and hospitalised with severe pneumonia were centrally randomised in a 1:1:1:1 allocation to one of four arms: standard of care (SOC) alone, SOC plus 15 days of oral valganciclovir (16 mg/kg every 12 hours), SOC plus six months of empirical tuberculosis treatment, or both interventions combined. The primary outcomes were all-cause mortality assessed at day 15 and over 12 months of follow-up in the intention-to-treat population. Enrolment ran from March 2020 to January 2024.
What Was Observed
- Day 15 mortality (primary endpoint, not significant): 23% of infants in the valganciclovir groups died within 15 days, compared with 27% in groups without valganciclovir — a difference that did not reach statistical significance (rate ratio 0.81, 95% CI 0.61–1.08, p=0.15), meaning the trial cannot rule out chance as an explanation for this numerical difference.
- 12-month mortality (primary endpoint, not significant): Cumulative mortality at one year was 43% in valganciclovir-treated infants versus 48% in controls — again a numerical trend favouring valganciclovir but falling short of conventional significance (HR 0.88, 95% CI 0.74–1.05, p=0.15).
- Time-varying model — early mortality hazard: In a pre-specified time-varying effects analysis, valganciclovir was associated with approximately 40% lower hazard of death at day 15 (adjusted HR 0.60, 95% CI 0.41–0.87, p=0.0063), and about 24% lower hazard over one year when deaths within the first 48 hours were excluded (HR 0.76, 95% CI 0.58–1.00, p=0.050).
- Safety: Severe adverse events were not more frequent in valganciclovir-treated infants than in controls — if anything, the point estimate favoured valganciclovir (OR 0.64, 95% CI 0.35–1.16), though this confidence interval is wide and includes no difference.
Why This Matters
This trial provides the first randomised evidence that empirical CMV treatment may reduce early mortality in infants with severe HIV-associated pneumonia, a population in which overall one-year mortality exceeded 40% even with standard care. The findings position CMV as a plausible and potentially actionable therapeutic target in this high-mortality clinical setting. The absence of excess harm from valganciclovir further supports the feasibility of evaluating this approach in future, better-powered trials.
How to Read This Result
While this is a well-designed, large multicentre randomised trial, the primary intention-to-treat endpoints did not reach statistical significance, and the most compelling survival signal emerged only from a secondary time-varying model — meaning these findings should be interpreted as hypothesis-generating rather than definitive evidence of benefit.
Limitations
The primary endpoints at both day 15 and 12 months failed to reach conventional statistical significance in the intention-to-treat analysis (p=0.15 at both timepoints), which limits the strength of conclusions that can be drawn. The open-label design introduces the possibility of performance or assessment bias. Loss to follow-up was notably higher in the valganciclovir group (9%) than in the control group (5%), which could influence survival estimates in either direction. The key statistically significant result supporting benefit was derived from a time-varying effects model rather than the pre-specified primary analysis, warranting cautious interpretation.