Placental infection by Zika virus in French Guiana

Study population

Details of the study protocol are described elsewhere^{14, 15}. Briefly, the study was conducted at the French Guiana Western Hospital Center (Centre Hospitalier de l'Ouest Guyanais (CHOG)) at the beginning of the ZIKV epidemic from 1 January to 15 July 2016. Initial inclusion of ZIKV-infected pregnant women and controls occurred either through routine serological testing (performed in each trimester of pregnancy and at birth) of all pregnant women admitted to the prenatal diagnosis unit of the CHOG or through serological and molecular testing in the presence of maternal symptoms. Molecular and serological testing were performed at the French Guiana National Reference Center for arboviruses (Pasteur Cayenne), using the Realstar Zika Kit (Altona Diagnostics GmbH, Hamburg, Germany) for reverse transcription polymerase chain reaction (RT-PCR), and in-house M antibody-capture enzyme-linked immunosorbent assay and micro-neutralizing assays for serological testing.

Patients were monitored according to the clinical standard of care in France, with the exception of prenatal ultrasound being performed monthly in patients who were positive for ZIKV, and two supplementary ultrasound examinations being performed in patients who were negative for ZIKV (at 26–28 weeks' gestation and 36–38 weeks' gestation), as recommended by the French authorities and others^16-18. Data regarding demographic, medical and obstetric characteristics and possible risk factors for congenital diseases were collected prospectively. The study protocol was approved by the institutional review board of the CHOG.

Study groups

Pregnant women were defined as ZIKV positive based on either a positive RT-PCR result in blood and/or urine, or by the presence of ZIKV-specific immunoglobulin-M (IgM) after anti-ZIKV antibody detection in the blood, confirmed by a micro-neutralizing assay in cases of suspected coinfection with other arboviruses (expected to be minimal, as circulation of Dengue virus has been low in French Guiana since 2014).

Confirmed congenital ZIKV infection was defined as either ZIKV-RNA amplification by RT-PCR from at least one fetal/neonatal sample (placenta, amniotic fluid, cerebrospinal fluid, urine or blood) or identification of ZIKV-specific IgM in umbilical cord/neonatal blood or in cerebrospinal fluid.

Placental ZIKV-infection status was classified into three categories as follows: (1) non-exposed control placentae from women who tested negative for ZIKV up to delivery; (2) exposed placentae from women with proven ZIKV infection without reported congenital infection in the newborn; (3) infected placentae from women with proven ZIKV infection and proven congenital infection in the newborn.

Infected placentae were then classified into those resulting in CZS, fetal loss or asymptomatic congenital infection at birth.

Placental analysis

Sonographers, clinicians and pathologists were aware of the ZIKV status of the pregnant women, as well as potential fetal malformations at the time of prenatal and postnatal placental examinations, but were unaware of the fetal virological status for ZIKV, which was defined after postnatal testing.

Prenatal placental examination

All prenatal ultrasound examinations were performed by two experienced sonographers (V.L., L.P.) using E8 and E10 Voluson ultrasound machines with transabdominal (RM6C) and transvaginal (RIC5-9-D) transducers (GE Healthcare, Zipf, Austria). Placental maximum vertical thickness was measured at each prenatal ultrasound examination, by longitudinal (non-oblique) scanning with the ultrasound probe and beam perpendicular to the chorionic plate. The measurements were performed avoiding areas of amniochorionic detachment and periods of uterine contractility. Placentomegaly was defined as a placental thickness greater than 40 mm^{19, 20}. Umbilical artery Doppler was performed on a free vertical loop of cord at each prenatal ultrasound examination in ZIKV-infected pregnant women. Umbilical artery resistance index (RI) was reported and correlated with gestational age. Abnormal umbilical artery Doppler was defined as a RI greater than the 95^th percentile²¹.

Postnatal placental examination

Routine examination of the placenta at the time of delivery was completed by a midwife and/or an obstetrician following live birth or intrauterine death (IUD). We intended to offer anatomopathological examination of the placenta to all patients with confirmed ZIKV infection during pregnancy, but during the peak of the epidemic this was limited owing to a lack of storage capacity, fixation material and availability of pathologists. The placentae were fixed in 10% buffered formalin for 48 h. Placental weight was measured and birth weight/placental weight ratio was calculated using the method of Thompson et al.²². Low and high birth weight/placental weight ratios were defined as a value < 3^rd percentile and > 97^th percentile, respectively (according to gestational age and sex)²². Samples from the umbilical cord, membrane and placental parenchyma, including the decidua and chorionic plates, were collected. Specific immunochemistry examination was offered for cases when CZS was identified prenatally.

Anatomopathological anomalies included abnormal birth weight/placental weight ratio, signs of placental inflammation (chorioamnionitis lesions, villitis and intervillositis, calcifications), infarcts, ischemic necrosis with fibrin deposits (INFD), thrombosis, leukocytic infiltration and Hofbauer cell hyperplasia (Appendix S1).

Covariates

Gestational age was recorded at each prenatal measurement of placental thickness and at the time of delivery (after live birth or IUD). Maternal comorbidities (diabetes, vascular pathologies, thrombophilia, severe anemia, lead poisoning, alcohol consumption, malnutrition, vitamin-K deficiency) or coinfections (TORCH group infections) and risk of aneuploidy > 1/250 were evaluated as effect modifiers for the association of placental thickness, Doppler abnormalities or pathological results with the ZIKV status of the placenta.

Statistical analysis

Baseline maternal and fetal/neonatal characteristics were compared according to the ZIKV status of the placenta. Categorical variables were compared using the chi-square or Fisher's exact test, while the distribution of continuous variables was compared using the Kruskal–Wallis test.

Placental thickness measurements were compared every month between the three groups (infected placentae, exposed placentae and controls) using ANOVA. The weekly incidence of placentomegaly (> 40 mm thickness) was compared between the three groups using Kaplan–Meier analysis, and the cumulative incidence was compared using the logrank test. The weekly incidence of placentomegaly in the infected group was also compared between those with CZS, fetal loss or asymptomatic congenital infection at birth, using the same statistical methods. Factors associated with the incidence of placentomegaly throughout pregnancy were evaluated using the Cox proportional hazards model. Factors associated with abnormal umbilical artery Doppler measurements or anatomopathological findings, including placental ZIKV status, were evaluated using a logistic regression model. Abnormal Doppler measurements and anatomopathological findings in the infected group were also compared between those with CZS, fetal loss or asymptomatic congenital infection at birth, using the same statistical methods. CIs for relative risks (RR) and hazard ratios (HR) were calculated using the exact method. When data are missing, denominators are presented. Statistical analysis was conducted using Stata version 14 (StataCorp., College Station, TX, USA).

Association between prenatal ultrasound findings and placental ZIKV status

Placental thickness was measured at least once during pregnancy, with a median number of four, four and three measurements in infected, exposed and control placentae, respectively. Infected placentae tended to be thicker than exposed placentae between 18 and 22 weeks' gestation (P = 0.05) and were significantly thicker after 26 weeks (P < 0.001). Similarly, infected placentae were significantly thicker than control placentae after 18 weeks (P < 0.001) (Figure 2). The kinetics of placental thickness according to gestational age, between control, exposed and infected placentae is shown in Figure S1.

Placentomegaly (> 40 mm thickness) was observed more frequently in infected placentae (39.5%) (Figure S2) than in exposed placentae (17.2%) or controls (7.2%), even when adjusting for gestational age at diagnosis and comorbidities (2.20% vs 0.79% and 0.43% per week; adjusted HR (aHR), 2.02 (95% CI, 1.22–3.36); P = 0.007, and aHR, 3.23 (95% CI, 1.86–5.61); P < 0.001, respectively) (Figure 3). Placentomegaly was no more frequent in exposed placentae than in controls after adjusting for gestational age and comorbidities (aHR, 1.34 (95% CI, 0.78–2.29); P = 0.288). Placentomegaly appeared earlier in infected placentae (at a median of 30 weeks and as early as 18 weeks) compared with exposed placentae (at a median of 33 weeks and as early as 23 weeks) and with controls (at a median of 34 weeks and as early as 22 weeks) (P < 0.001).

When considering infected placentae, placentomegaly was observed more frequently in those from pregnancies with CZS (10/16 (62.5%)) or fetal loss (5/11 (45.5%)) than in those from pregnancies with asymptomatic congenital ZIKV infection (15/49 (30.6%)), even when adjusting for gestational age at diagnosis and comorbidities (4.09% and 3.14% vs 1.48% per week; aHR, 5.43 (95% CI, 2.17–13.56); P < 0.0001, and aHR, 4.95 (95% CI, 1.65–14.83); P = 0.004, respectively) (Figure 4). Placentomegaly appeared earlier in placentae from pregnancies with CZS (as early as 18 weeks) or fetal loss (as early as 19 weeks) compared with those from pregnancies with asymptomatic congenital infection (as early as 26 weeks) (P = 0.044 and P = 0.048, respectively).

When considering placentomegaly > 40 mm as a predictor of congenital ZIKV infection, sensitivity was 39.5%, specificity was 88.8%, positive predictive value (PPV) was 32.3% and negative predictive value (NPV) was 91.3%. As a predictor for CZS, sensitivity was 62.5%, specificity was 86.7%, PPV was 11.0% and NPV was 98.9%. As a predictor of fetal loss related to congenital ZIKV infection, sensitivity was 45.5%, specificity was 86.0%, PPV was 5.5% and NPV was 98.9%.

Umbilical artery Doppler measurements were available for 75/76 infected placentae (not performed in a case of IUD at 18 weeks' gestation) and for all exposed placentae (Figure S3). Umbilical artery RI > 95^th percentile tended to be more frequent in those with infected placentae than in those with exposed placentae, although not statistically significantly so (12.0% vs 6.0%, adjusted for maternal comorbidities: adjusted RR (aRR), 1.95 (95% CI, 0.85–4.19); P = 0.0945). When considering infected placentae, umbilical artery RI > 95^th percentile was more frequent in those from pregnancies with fetal loss (3/10 (30.0%)) compared with those from pregnancies with asymptomatic congenital infections (3/49 (6.1%); aRR, 4.83 (95% CI, 1.09–20.64); P = 0.0292), but not compared with cases of CZS (3/16 (18.8%); aRR, 3.0 (95% CI, 0.60–13.09); P = 0.13).

Association between histopathological findings and placental ZIKV status

Pathological examinations were available for 43/76 (56.6%) infected placentae and 171/215 (79.5%) exposed placentae (Figure 1). No differences in baseline maternal characteristics or pregnancy outcome were observed between these two groups, apart from more frequent fetal loss in the infected group (P < 0.001) (Table S1). No differences in baseline maternal characteristics or pregnancy outcome were observed according to whether pathological examination was available (data not shown).

Among infected placentae, 27/43 (62.8% (95% CI, 48.3–77.2%)) demonstrated pathological anomalies: five chorioamnionitis lesions, nine infarcts, 14 cases with INFD, eight cases with chronic villitis or intervillositis, two cases with subchorionic thrombosis and two calcifications. Leukocytic infiltration and Hofbauer cell hyperplasia were demonstrated by immunohistochemistry in two placentae of fetuses with CZS (following termination of pregnancy at 27 weeks and stillbirth at 33 weeks, respectively).

Among exposed placentae, 37/171 (21.6% (95% CI, 15.5–27.8%)) demonstrated anomalies on pathological examination: four chorioamnionitis lesions, 10 infarcts, five cases with INFD, five cases with chronic villitis or intervillositis, five cases with subchorionic thrombosis and eight calcifications.

Infected placentae exhibited a higher risk of any pathological anomaly than did exposed placentae (RR, 2.90 (95% CI, 2.01–4.19); P < 0.0001), even when adjusting for gestational age at delivery and comorbidities (aRR, 2.60 (95% CI, 1.40–4.83); P = 0.002). Frequencies of anatomopathological findings are presented in Table 2. Low birth weight/placental weight ratio was also more frequent in infected placentae than in exposed placentae (aRR, 2.78 (95% CI, 1.12–8.78); P = 0.035).

When considering infected placentae, pathological anomalies were no more frequent in those from pregnancies with CZS (8/12 (66.7%)) or fetal loss (8/10 (80.0%)) than in those from pregnancies with asymptomatic congenital infection (11/21 (52.4%)) after adjusting for gestational age at birth/IUD and comorbidities (aRR, 1.22 (95% CI, 0.66–2.17); P = 0.42 and aRR, 1.49 (95% CI, 0.82–2.48); P = 0.14, respectively).

Among infected placentae, positive RT-PCR of the placenta was found in 51/58 (87.9% (95% CI, 76.7–95.0%)) placentae tested and was not associated with a higher risk for any placental pathology (P = 0.80) or low birth weight/placental weight ratio (P = 0.69).

Principal findings

In this study, ZIKV-infected placentae were thicker after 26 weeks' gestation, and placentomegaly was more prevalent and presented earlier in gestation, particularly in cases of fetal/neonatal adverse outcome (CZS and/or fetal loss), than in non-infected placentae. Infected placentae also exhibited a higher risk of anatomopathological anomalies than did non-infected placentae.

Limitations

While the accuracy of molecular and serological testing for maternal and congenital ZIKV infection is still under debate, we assessed ZIKV status at multiple time points and in numerous samples¹⁵, thus reducing the risk of false-negative results. Non-infected pregnant women remained negative on ZIKV testing during all trimesters of pregnancy and at delivery¹⁴. The ZIKV status of infants born to ZIKV-positive and -negative pregnant women was assessed in multiple samples, including umbilical-cord and neonatal blood, placenta, urine, amniotic fluid and/or cerebrospinal fluid for symptomatic cases, limiting the risk of misclassification¹⁵. Misclassification would result in potential underestimation of the observed differences between the three groups. The risk of a false-positive ZIKV result was limited because pregnant women were tested in the early stage of the epidemic, without significant circulation of other Flaviviridae during this period, and possible cross-reactions were minimized using a micro-neutralizing assay¹⁵.

Interpretation

Placentomegaly was observed in 39.5% of infected placentae and was associated significantly with congenital ZIKV infection. Low birth weight/placental weight ratio was also associated with congenital ZIKV infection. Placentomegaly identified on ultrasound within a few weeks after infection may be the consequence of placental inflammation due to recent maternal and transplacental infection¹⁰. The higher risk of low birth weight/placental weight ratio in the infected group is probably related to the increase in placental weight associated with placentomegaly, which could result from fibrinoid deposition and small vascularized villi that form to compensate for in-utero hypoxia, described recently in congenital ZIKV infection and other congenital infections^{9, 23}. Placentomegaly appeared earlier in the infected group than in the exposed and control groups, and may be an early sign of transplacental infection. Placentomegaly was observed in cases of both asymptomatic and symptomatic congenital infection. In fetuses with CZS, placentomegaly appeared after the fetal anomalies in one case, at the same time in two cases and before the appearance of CZS in seven cases (from 2 to 13 weeks before).

The relatively low rate of low birth weight/placental weight ratio compared with the rate of placentomegaly could also indicate that some of the instances of placentomegaly observed were transient and did not affect placental volume at birth. These transient cases of placentomegaly may be associated with other comorbidities and natural growth of the placenta, which may be why placentomegaly was identified in the third trimester in the exposed and control groups, and was thus unrelated to ZIKV. Abnormal umbilical artery Doppler measurements tended to occur more frequently in infected than exposed placentae, but to a lesser extent than placentomegaly. These results may indicate that placental infection resulting in placentomegaly does not always lead to placental dysfunction. However, abnormal umbilical artery Doppler associated with placentomegaly in infected placentae may predict an acute risk of fetal loss, even if the fetus does not present with growth restriction at diagnosis^{24, 25}.

The sensitivity of placentomegaly in the prediction of congenital ZIKV infection is debatable, but is similar to that for other congenital infections²³. The sensitivity is, however, higher when predicting CZS or fetal loss. The NPV of > 90% for congenital infection and of > 98% for adverse outcome of placental thickness < 40 mm may help to reassure pregnant patients in low-resource areas in which molecular and serological testing are not available during an epidemic peak. Moreover, placental thickness measurement by ultrasound is easy to perform and does not require advanced expertise, as compared with neurosonograms.

It is important to keep in mind that increased placental thickness is not specific to congenital ZIKV infection and is associated with other congenital infections, chromosomal and fetal abnormalities and maternal pathologies^26-29.

Compared to exposed placentae, infected placentae exhibited a 2.6-fold increased risk of any histopathological anomaly. Placental inflammation consisted of diffuse vascular inflammation, chorioamnionitis lesions, villitis and calcifications, as described in congenital cytomegalovirus (cCMV) infection³⁰. It has been shown that ZIKV is able to specifically infect human placental macrophages and trophoblasts³¹, but placental inflammation has been described inconsistently^{11, 32, 33} and reported to occur only in the early stages of congenital infection³⁴. Our study highlights a higher rate of INFD in ZIKV-infected placentae, which could increase placental growth, as in congenital syphilis infection³⁵. Only excessive INFD was considered and was found in all cases of fetal loss; however, this finding does not seem to be specific to ZIKV infection, as it is also often found in fetal loss of other origin³⁶. Subchorionic thrombosis and congestive capillaries, which have been described previously in congenital ZIKV infection, were observed more often in infected placentae, and could contribute to fetal hypoxia³⁷. We observed Hofbauer cell hyperplasia and leukocytic infiltration in ZIKV-infected placentae, which has been described by others as a specific finding in ZIKV-infected placentae^{11, 24, 38}.

Histopathological examination of the placenta highlights the potential long-term consequences of placental disorders during pregnancy, and the anomalies found may be the result of different pathologies impacting on the placenta²⁷. We cannot exclude the possibility that some of the placental signs observed may be partially caused by other comorbidities; however, the similarities seen in maternal comorbidities and delivery characteristics between the exposed and infected groups point to a reduced influence of other factors. The main difference between these groups is the rate of fetal loss, but this may be due to the fact that fetal loss is a known consequence of fetal and placental infection by ZIKV^{5, 15}.

Overall, ZIKV infection of the placenta precedes viral transmission to the fetus. We cannot exclude the possibility that ZIKV infection may be restricted to the placenta with no further involvement of the fetus, as described in some cases of CMV infection³⁹. Placentomegaly may represent an early sign of congenital infection, before CZS or other adverse outcomes become apparent⁴⁰.

Conclusions

Our study provides a comprehensive description of the placental consequences of ZIKV infection. Infected placentae exhibited a higher risk for placentomegaly, low birth weight/placental weight ratio and histopathological anomalies. Although placentomegaly is non-specific for ZIKV infection, early placentomegaly may represent the first sign of congenital ZIKV infection and should lead to enhanced prenatal follow-up of ZIKV-exposed pregnancies.