4F-04 - Clinically Recognized Congenital Cytomegalovirus Infection and the Risk of Neurodevelopmental Disorders

Background: Congenital cytomegalovirus (cCMV) can cause sensorineural hearing loss and short-term developmental delays, which antiviral treatment has been shown to improve, as demonstrated in a 2015 clinical trial. However, the risk of long-term neurodevelopmental disorders (NDDs) in cCMV-infected infants has not been thoroughly evaluated.

Objectives: To assess the risk of NDDs among infants with clinically recognized cCMV.

Methods: Using Marketscan (2003-22) and Medicaid (2000-18), infants with cCMV were identified based on ≥1 CMV diagnostic code and antivirals (i.e., ganciclovir and valganciclovir) were determined based on ≥1 dispensation within 45 days after birth. The proportion of treated infants was estimated. NDDs included autism, attention-deficit/hyperactivity disorder, learning disorder, speech/language disorder, coordination disorder, intellectual disability, and behavioral disorder. Infants with risk factors for NDDs (e.g, hearing loss) are more likely to undergo cCMV testing, creating a spurious association between cCMV and NDDs. Restricting analysis to tested infants mitigates this bias but may underestimate risk, as NDD risk factors that prompt cCMV testing may mediate the cCMV-NDD relationship. We therefore compared infected infants to (a) all uninfected infants (upper bound of risk estimates) and (b) uninfected infants with testing (lower bounds). We estimated cumulative incidences, and hazard ratios (HRs) of any and individual NDDs using Cox proportional hazards models, adjusting for year, region, race/ethnicity, multiples, and other congenital infections.

Results: Among 23,209,467 infants, 6,445 (3 per 10,000) were diagnosed with cCMV. Antiviral use in infected infants increased from 5.0% before 2015 to 21.9% after, with > 90% receiving valganciclovir. By age 8, the crude cumulative incidence of NDDs was 31.3% in infected infants vs. 13.1% in uninfected infants (without cCMV testing: 9.2% vs. with testing: 31.1%) in MarketScan, and 46.1% vs. 20.8% (without testing: 20.8% vs. with testing: 33.2%) in Medicaid. Compared to uninfected infants, the pooled adjusted HR (aHR) for NDDs was 2.83 (95% CI: 2.78-2.89), with the point estimates for individual NDDs ranging from 1.44 (behavioral disorder) to 10.04 (intellectual disability). Compared to tested uninfected infants, aHR for any NDD remained elevated but attenuated (1.85; 1.69-2.02), as did individual NDDs.

Conclusions: Clinically recognized cCMV is associated with a higher risk of NDDs. Estimating an accurate risk is challenging because either restricting or not restricting to tested infants could lead to bias. Future studies in populations with universal cCMV and NDD screening are needed to confirm the causal effect, and the impact of antivirals warrants further investigation.