Feasibility, performances and predictive value of congenital CMV neonatal screening on saliva swabs

Background

Early diagnosis of congenital CMV infection (cCMVI) allows for early intervention and follow-up to detect delayed hearing loss. While CMV PCR in urine is the gold standard for cCMVI diagnosis, saliva testing is often performed.

Objectives

Our aim was to determine (i) if swab saliva sampling needed standardization, (ii) if a threshold value in “virus copies per million cells (Mc)” in saliva samples could improve clinical specificity, and (iii) to establish a correlation between viral load in saliva and symptomatology/outcome of cCMVI.

Materials and methods

In our institution, universal newborn screening is performed on saliva swabs at delivery or until day 3 of life. If positive, CMV PCR in urine is done within 2 weeks to confirm or exclude cCMVI.

Results

Cell quantification showed that saliva swab sampling was well performed as 95.4 % samples had more than 100 cells/10 µL. There was a good correlation between saliva viral load in copies/mL and in copies/Mc (Pearson's r = 0.96, p < 0.0001). However, threshold values, established to determine a viral load level at which we could confidently identify infected newborns, did not improve positive predictive value (21.8 % for copies/mL and 21 % for copies/Mc vs 25.4 % without threshold) but instead reduced sensitivity (88 % and 85% vs 100 % without threshold). Samples collected on day 2 or 3 had better positive predictive value (38.7 %) compared to those collected on day 1 (23.8 %). Symptomatology at birth was not significantly associated with viral load in saliva at diagnosis. However, sequelae occurrence was associated with viral load in saliva (copies/Mc).

Discussion

Our results confirm that saliva swab is a suitable sample for universal neonatal screening. However, identifying newborns that will develop sequelae remains an issue in the management of cCMVI.