Identification of Symptomatic Fetuses Infected with Cytomegalovirus Using Amniotic Fluid Peptide Biomarkers

Abstract

Cytomegalovirus (CMV) is the most common cause of congenital infection, and is a major cause of sensorineural hearing loss and neurological disabilities. Evaluating the risk for a CMV infected fetus to develop severe clinical symptoms after birth is crucial to provide appropriate guidance to pregnant women who might have to consider termination of pregnancy or experimental prenatal medical therapies. However, establishing the prognosis before birth remains a challenge. This evaluation is currently based upon fetal imaging and fetal biological parameters, but the positive and negative predictive values of these parameters are not optimal, leaving room for the development of new prognostic factors. Here, we compared the amniotic fluid peptidome between asymptomatic fetuses who were born as asymptomatic neonates and symptomatic fetuses who were either terminated in view of severe cerebral lesions or born as severely symptomatic neonates. This comparison allowed us to identify a 34-peptide classifier in a discovery cohort of 13 symptomatic and 13 asymptomatic neonates. This classifier further yielded 89% sensitivity, 75% specificity and an area under the curve of 0.90 to segregate 9 severely symptomatic from 12 asymptomatic neonates in a validation cohort, showing an overall better performance than that of classical fetal laboratory parameters. Pathway analysis of the 34 peptides underlined the role of viral entry in fetuses with severe brain disease as well as the potential importance of both beta-2-microglobulin and adiponectin to protect the injured fetal brain infected with CMV. The results also suggested the mechanistic implication of the T calcium channel alpha-1G (CACNA1G) protein in the development of seizures in severely CMV infected children. These results open a new field for potential therapeutic options. In conclusion, this study demonstrates that amniotic fluid peptidome analysis can effectively predict the severity of congenital CMV infection. This peptidomic classifier may therefore be used in clinical settings during pregnancy to improve prenatal counseling.

Conflict of interest statement

The authors have declared that no competing interests exist.

Figures

Fig 1. Detected peptides in amniotic fluid and differences between symptomatic and asymptomatic cases in the discovery cohort.

(A) Representation of 4076 peptides, detected in all amniotic fluid samples (n = 47) by CE-MS. Each peptide was identified by a unique identifier based on the migration time (min) and specific mass (kDa), with a peak height representing the relative abundance. (B) In the discovery cohort, 76 amniotic fluid peptides were identified as differentially secreted between symptomatic and asymptomatic patients. (C) Cross-validation score of a SVM peptide classifier called CMV34 consisting of 34 of the 37 sequenced peptides obtained from the analysis of the discovery cohort. ***p<0.0001, Mann-Whitney test for independent samples. Asympt., asymptomatic; Sympt., symptomatic.

Fig 2. Performance of the CMV34 classifier in the validation cohort.

(A) Correlation analysis of the CMV34 classifier and gestational age. (B) ROC curve for the CMV34 classifier. (C) Box-whisker plot for classification of symptomatic and asymptomatic patients in the validation set according to the CMV34 score. **p<0.01, Mann-Whitney test for independent samples.

Fig 3. Performance of other frequently used parameters in the validation cohort.

ROC curves for CMV DNA levels in amniotic fluid (A) and fetal blood (B) and the fetal platelet count (C) in the combined discovery and validation cohort. AF, amniotic fluid; FB, fetal blood; Plat., platelet.

Fig 4. Classification of primary CMV infection with moderately symptomatic neonates.

(A) The amniotic fluid peptide content of CMV-infected fetuses with moderate neonatal symptoms (hearing loss (HL) or hearing loss and vestibular dysfunction (HL+VD)) was scored with the CMV34 classifier. The scores allowed a nearly significant difference after analysis by Mann-Whitney test for independent samples (p = 0.06). CMV DNA levels in amniotic fluid and fetal blood (B and C, respectively) and the fetal platelet count (D) were clearly not different between hearing loss versus hearing loss and vestibular dysfunction (p values of 0.32, 0.19 and 0.73, respectively, Mann-Whitney test for independent samples).

Fig 5. Alteration of the neurological disease pathway.

Ingenuity Pathway Analysis software showed a significant activation of neurological disease pathway in symptomatic fetuses compared with asymptomatic, with 7 out of 13 non-collagen peptide parent proteins being associated to the network. Red: increased abundance. B2M, beta-2-microglobulin; HBD, hemoglobin delta subunit; ADIPOQ, adiponectin; SCRIB, protein scribble homolog; CACNA1G, T calcium channel alpha-1G; MCM6, DNA replication licensing factor; TSC2, tuberin.