Professional Documents
Culture Documents
MAJOR ARTICLE
Background. The design of diagnostic and preventive strategies have been prevented by gaps in knowledge of the epidemiology
of congenital cytomegalovirus (cCMV) with the type of maternal infection as well as the lack of large-scale neonatal screening tools.
Methods. In sum, 11 715 consecutive newborns were screened for cCMV by polymerase chain reaction (PCR) in saliva.
Prevalence, type of maternal infection, sociodemographic, obstetrical, and serological data were analyzed.
Results. Positive predictive value of CMV PCR in saliva was 59%; false positive results were associated with lower viral loads
(P < .001). Maternal seroprevalence was 61%, birth prevalence was 0.37%, resulting from primary and nonprimary infections in 52%
and 47.7% of cases, respectively. The risk to deliver an infected baby after primary infection was increased in younger (OD = 7.9),
parous (OD = 4.1) women born in high resources countries (OD = 5.2) and from higher income groups (P = .019). The only 2 risk
factors to deliver an infected baby after nonprimary infection were to be young (OD = 4.6) and unemployed (OD = 5.8). The risk to
deliver an infected baby was 4-fold higher in women seronegative before their pregnancy (P = .021).
Conclusions. A positive CMV PCR in newborns’ saliva should always be confirmed in a repeat-sample. Sociodemographic
characteristics of women giving birth to an infected baby after primary and nonprimary infection are different. Seronegative, parous
women represent the highest risk population for cCMV in countries with low to intermediate seroprevalence. Urgent action is
needed to stop the cCMV’s epidemic, particularly in this population easily identifiable by maternal serology and amenable to pre-
vention messages.
Clinical Trials Registration. NCT01923636.
Keywords. cytomegalovirus; congenital; screening; saliva; nonprimary infection.
Congenital CMV (cCMV) is the main cause of neurological Large scale screening studies have been published, but none
handicap of infectious origin with a birth prevalence of 0.65% has reported birth prevalence in the light of sociodemograph-
worldwide [1]. Although cCMV meets most criteria for screen- ics, obstetrical and serological data within the same population
ing programs, none have been implemented in any country. [2–7]. We undertook universal screening of cCMV infection in
This is explained by gaps in knowledge on several fronts. A bet- live-born infants delivered in 2 French maternities located in the
ter understanding of the epidemiology of cCMV related to Paris area between September 2013 and August 2015. Screening
both maternal primary and nonprimary infection is needed to was performed by CMV DNA detection and quantification by
evaluate the impact of universal maternal serological screening. real-time polymerase chain reaction (PCR) in saliva [5, 7]. All
A better knowledge of the performance of broad screening tools positive screening tests were controlled in repeat samples allow-
is mandated before considering universal neonatal screening. ing calculation of the positive predictive value of the screening
test in saliva. For each infected newborn, the type of maternal
Received 27 January 2017; editorial decision 24 March 2017; accepted 10 April 2017; infection was characterized between primary or nonprimary.
published online April 17, 2017.
Maternal demographics have rarely been considered and never
Correspondence: M. Leruez-Ville, Laboratoire de Microbiologie Clinique, Hôpital Necker-
E.M., 149 rue de Sèvres, 75015 Paris, France (marianne.leruez@aphp.fr). accordingly to the type of maternal infection within the same
Clinical Infectious Diseases® 2017;65(3):398–404 study population [8–10]. We found that the type of maternal
© The Author 2017. Published by Oxford University Press for the Infectious Diseases Society
infection is closely related to sociodemographic characteristics
of America. All rights reserved. For permissions, e-mail: journals.permissions@oup.com.
DOI: 10.1093/cid/cix337 with implications for prevention programs to be based upon.
(88% [5328/6030] in Necker and 79% [6529/8145] in Poissy). had profound unilateral hearing loss with threshold of 100 db
Members of staff were either too busy or forgot to propose the in both cases. Whereas among the 2 neonates infected after
study to 2318 (16%) mothers. In sum, 76 newborns were not nonprimary infection, one had moderate unilateral hearing loss
included because their mothers refused the study. And 41 saliva (threshold 60 db), and the other had moderate to severe bilat-
samples were defective (not labeled, broken vial…). Among eral hearing loss with respective threshold of 60 db in the best
the 11 740 newborns screened, 25 had a CMV PCR result not ear and 70 db in the worst ear.
interpretable because of a negative GAPDH PCR suggesting
improper sampling. In total, 11 715 neonates had an interpret- Positive Predictive Value of Congenital Cytomegalovirus (cCMV)
Screening With CMV Polymerase Chain Reaction in Saliva
able CMV PCR results, and 51 were screened and confirmed
CMV PCR was positive in the first saliva sample in 87 new-
infected. Seven of these 51 newborns were excluded from the
borns. All 87 had subsequent saliva and blood control samples
prevalence calculation because their mothers were referred with
with positive CMV PCR confirmed in only 51 following resa-
a known CMV fetal infection.
mpling. PPV of the screening test was 58.6% (95% CI = 47.5–
The prevalence of cCMV was 0.37% (44/11 715) (95% CI:
69.1). Median viral load in true positive samples was much
0.246, 0.606) and was similar in Necker and Poissy: 0.39%
higher than in false positive samples (7.6 [inter quartile {IQ}
(21/5296) (95% CI: 0.246, 0.606) and 0.36% (23/6419) (95% CI:
6.3, 8.4] vs 2.2 [IQ 1.8, 2.4] P < .001) (Figure 2).
0.227, 0.537), respectively (P = .848). And 23 (52% 95% CI: 36.7,
67.6) and 21 (48%, 95% CI: 32.5, 63.3) cases followed maternal
Comparison of Socio-Demographic, Obstetrical and Serologic Maternal
primary or nonprimary infection, respectively; similar propor- Data Between Uninfected and Infected Newborns Following Maternal
tions were found in Necker 57% (12/21) vs. 43% (9/21) and in Primary or Nonprimary Infection (Table 1)
Poissy 48% (11/23) vs. 52% (12/23) (P = .623). The 48 mothers of the 51 infected neonates (44 plus the 7
Among the 44 infected neonates (51 minus 7 born from infected newborns from referred mothers) were classified
mothers referred for CMV infection), 9 (21%) were sympto- depending on the type of maternal infection. Retrospective
matic including: 4 (9%) with hearing loss, 4 (9%) with isolated CMV serology testing was performed in maternal stored sera
SGA (small for gestational age), and 1 with isolated thrombo- for these 51 infected neonates. This allowed classification into
cytopenia. The proportions of neonates with symptoms and primary and nonprimary group. Seroconversion during preg-
hearing loss were similar for both types of maternal infection in nancy was demonstrated in 16 women, the presence of positive
Necker and Poissy with 21% vs. 19% and 8.6% vs. 9.5%, respec- IgG, positive IgM, and low or intermediate IgG avidity was evi-
tively (P > .99). The 2 neonates infected after primary infection denced in 12 women; these 28 women composed the primary
Uninfected Newborns. Infected Newborns After Primary Infected Newborns After Nonprimary
Mothers N = 11 088 Infection. N = 28 P Infection. N = 20 P
Data are mean ± standard deviation, n (%), or n/N (%) where data are missing.
Countries of birth of mothers from the first group: France (87%), other European Countries (12%), North American countries (1%).
Countries of birth of mothers from the second group: North African countries (39%), non-North African countries (26%), East Asian countries (19%), Middle East countries (6%), South
American countries (5%), others (5%).
Professional occupation: 1 = independent retailers, entrepreneurs, 2 = executives, intellectual professions, 3 = intermediate professions, 4 = employees, 5 = workers, 6 = unemployed.
respective burden of cCMV arising from primary and non- risk to deliver an infected baby than women seropositive before
primary infections [19–21]. Applying these models to our pregnancy.
population bearing a 61% seroprevalence leads to an expected In our study, the proportion of symptomatic infected neo-
proportion of congenital infections attributable to nonprimary nates was higher than reported in a meta-analysis (20% vs 13%).
infections of over 70% and not under 50% as observed [21]. This may be at least partly due to the definition of a symptomatic
In light of our data, these models seem therefore to be over- neonate that has evolved over time [15]. The proportion of neo-
estimating the proportion of congenital infections following nates with hearing loss at birth was 10%, and this was similar in
nonprimary infections. Moreover, in our population, women neonates infected after maternal primary or nonprimary infec-
seronegative before pregnancy had a 4-fold higher individual tions. Neonatal hearing loss has long been thought to mainly
Table 2. Risk Factors to Deliver an Infected Newborn After Maternal Primary Infection or After Maternal Nonprimary Infection
Risk to Deliver an Infected Neonate After Risk to Deliver an Infected Neonate After
Maternal Primary Infection Maternal Nonprimary Infection
follow primary rather than nonprimary maternal infection [22]. than 25 years, with lower economic status, caring for preschool
However, our results together with other recent studies suggest children and living in a household with more than 3 people
that hearing loss related to maternal nonprimary infection is as [8–10]. Conversely, one study conducted in Australia reported
severe as that related to maternal primary infection [4, 6, 23]. an association between higher socioeconomic status and cCMV
CMV detection in urine has been the gold standard for cCMV [28]. However, in these studies, sociodemographic character-
diagnosis for years. However, a recent large prospective study of istics of the women were not studied accordingly to the type
cCMV screening showed that CMV PCR in saliva has excellent of maternal infection. Our study provides a thorough analysis
sensitivity and specificity compared to standard saliva rapid cul- of sociodemographic data associated with cCMV and demon-
ture; and that saliva could therefore be used for cCMV screen- strates that mothers of infected neonates belong to 2 markedly
ing [5]. Similar performance was reported between urine and different sociodemographic groups in the 2 types of maternal
saliva in 2 small studies [24, 25]. However, false positive cCMV infection. Although maternal age below 25 was a risk factor for
diagnoses in saliva could be an issue because CMV is frequently both types of infection, the risk to deliver an infected infant fol-
present in breast milk [26] and in the birth canal [27], and lowing primary infection was higher in parous women born in
therefore contamination of the newborn’s saliva samples could high-resource country. Higher income was also associated with
happen throughout the peripartum period. We systematically the birth of an infected infant after primary infection. The risk
controlled positive results in repeat saliva and blood samples. to deliver an infant infected following nonprimary infection
Among the 11 715 samples tested, 51 were true-positive and 36 was higher in unemployed women, who also belonged to low
were false-positive giving a PPV for cCMV screening in saliva socioeconomical groups, whereas to be parous was not a risk
samples of only 55%. CMV DNA quantification helped discrimi- factor. Altogether, these results suggest that the source of con-
nate between false and true positive results since high viral loads tamination for primary infection is mainly the mother’s older
(>4.0 log copies/mL) were only found in true positive samples. child or children. However, women with lower socioeconomic
However, low viral loads were found both in true and false posi- status have a higher likelihood of being exposed to individuals
tive samples. Because the neonates were screened at birth in the shedding CMV in addition to their own children and therefore
delivery room, false positive samples could result from contam- have a higher risk of reinfection. Because this study showed that
ination by CMV DNA shed in vaginal secretions. These results seronegative, parous women represent the highest risk group
indicate that a positive CMV PCR in saliva obtained from a neo- of congenital CMV infection, we advocate that these women
nate should be controlled in a subsequent sample particularly should be identified by CMV testing before pregnancy and tar-
when the amount of CMV DNA detected is low. geted by prevention messages. Indeed, educational and hygiene
Studies conducted in the United States reported higher birth measures aiming to avoid intimate contacts with young chil-
prevalences of cCMV in African American women, younger dren are efficient to prevent primary maternal infection [29].