Maternal-fetal transmission and adverse perinatal outcomes in pregnant women infected with Zika virus: prospective cohort study in French GuianaBMJ 2018; 363 doi: https://doi.org/10.1136/bmj.k4431 (Published 31 October 2018) Cite this as: BMJ 2018;363:k4431
All rapid responses
Re: Maternal-fetal transmission and adverse perinatal outcomes in pregnant women infected with Zika virus: prospective cohort study in French Guiana
Pomar et al. reported a 26% risk of maternal-fetal transmission of Zika virus (ZIKV), and a greather than 4-fold increase in the risk of adverse outcomes in ZIKV-infected fetuses/neonates, in a cohort of women from French Guiana. These findings are likely a consequence of methodological flaws in their study.
They recruited women with ZIKV infection symptoms admitted to the prenatal unit (PU) or the Department of Obstetrics and Gynaecology (DOBGYN) of the CHOG. Admission diagnoses were not reported, making it uncertain to whom study findings might apply. Women with pregnancy complications, abnormal fetal ultrasound, and at high risk of being ZIKV-infected were likely oversampled. This explains a prevalence of maternal ZIKV infection (MZVI) 7.60 times higher than in the general population (29.5% vs. 3.9%), and a prevalence of central nervous system (CNS) fetal ultrasound abnormalities in non-infected women (4.3%) at least six times higher than in other developing countries.[3, 4, 5] Concerns about the role of MZVI on fetal abnormalities made women with both conditions more likely admitted and included in the study than those with only one condition. Consequently, the risk of congenital ZIKV infection (CZVI) and the association between MZVI or CZVI and fetal abnormalities were necessarily overestimated in the cohort, even if these conditions were not associated in the whole population.
Investigators used an in-house MAC-ELISA assay with a sensitivity of 94.4% (95% confidence interval: 86.2 to 98.4) and a specificity of 60.0% (48.4 to 70.8) for serologic diagnosis of ZIKV infection. Re-testing individuals with other flavivirus infections for neutralizing antibodies could have increased specificity up to a maximum of 80% (65.4 to 90.4). Therefore, even if 30% to 50% of the participants were true cases, 1/4 to 1/6 of serologically-diagnosed cases should have been false positive. Though this limitation did not apply to RT-PCR tests, RT-PCR was used only in mothers admitted to the DOBGYN.
Unfortunately, the impact of errors in ZIKV infection diagnosis is uncertain, because authors failed to report the distribution of participants by recruitment source (PU/ DOBGYN) and diagnostic test (IgM/RT-PCR).
CZVI was defined as a positive RT-PCR test in placenta, amniotic fluid, or fetal/neonatal cerebrospinal fluid, urine, or blood. However, RT-PCR in amniotic fluid and placenta is unreliable for a diagnosis of CZVI.[7, 8, 9] CZVI cases were also detected through IgM for ZIKV in umbilical cord/neonatal blood or cerebrospinal fluid. Positive cases were retested at age 3-days, to exclude maternal contamination of umbilical cord blood. Nevertheless, 20% (10 to 35) of ZIKV IgM positive neonates, without other flavivirus infections, would still be false positive.
In a previous report based on the first 110 births in this cohort the risk of CZVI was 10.9% (5.8 to 18.3). Therefore, 64 new CZVI cases from 181 mothers were added in the current report, for a risk of 35.4% (28.4 to 42.8). The probability of this increase in risk of CZVI happening by chance is about 1 in 241,000. Authors provided no explanation for this inconsistency.
The authors’ definition of fetal/neonatal adverse outcome rested solely on the assumption that a congenital ZIKV syndrome (CZVS) exists and is identifiable through fetal ultrasound.[2, 10] However, it is uncertain which clinical findings constitute the syndrome, and no study has shown that among fetus/neonates of ZIKV-infected mothers those findings cluster more frequently than in fetuses/neonates of non-infected mothers. Moreover, claims that neuroimaging findings are highly suggestive of CZVS come from studies in women with rash, but without confirmed ZIKV infection.[13, 14, 15, 16, 17] More important, an association between specific neuroimaging findings and CZVS was untestable in those studies, because none included a control group.
Ultrasound findings used to define “any adverse outcome” (AAO), the main study outcome, frequently occur as isolated conditions, resolve after a variable period, and have good prognosis and no obvious cause.[18, 19, 20, 21] Authors provided no rationale for combining clinical findings to define AAO. It seems findings were labeled as "major" or "minor" depending on their prognostic value. However, findings with good prognostic value may contribute little to an accurate diagnosis of CZVS/AAO, or to identifying their etiology, particularly if they tend to concur. Indeed, there is no reason to expect cases with three "minor signs" to have similar prognosis or be more likely caused by CZVI than those with one "major sign". Additionally, it seems authors departed from their own definition of “severe symptoms”, since cases 185, 216, 205, 266, and 297 were considered severe, but had none of the major signs in Appendix 1.
Investigators reviewed and classified all fetal/neonatal outcomes blinded to MZVI status. However, blinding was not mentioned in two previous reports,[2, 10] and the approach used for blinding was not described. More important, blinded review of findings was inconsequential, because investigators themselves ascertained those findings during the follow-up and, as treating physicians, were ethicaly bound to know the mother’s infection status. Moreover, blinding could have been broken just by counting ultrasound exams, since ZIKV-infected mothers had monthly ultrasound exams (an average of four), while other women “were monitored as clinically indicated”. In that regard, it is unclear why only one ultrasound exam, and which one, was reviewed, since only two cases in Appendix 4 have more than one exam.
Authors reported fetal findings in 69 RT-PCR positive mothers (Appendix 4). The association between CZVI and AAO could be estimated from these data while avoiding selection bias, because selection was independent of CZVI (request details and analysis code to email@example.com). When fetal IgM and RT-PCR in any tissue/fluid were used to define CZVI, the odds ratio (OR) for AAO was 0.95 (0.21 to 4.28). When RT-PCR tests in placenta and amniotic fluid were not used to define CZVI, the OR was 1.54 (0.23 to 10.56). Corresponding ORs for “any major CNS finding” (Appendix 1) were 1.86 (0.67 to 5.12) and 2.17 (0.60 to 8.33), respectively. Thus, study data provides little evidence of a CZVI-AAO association. Finally, uncertainty about which one happened first made it impossible to know if CZVI was a cause or a consequence of fetal abnormalities.
1. Pomar L, Vouga M, Lambert V, et al. Maternal-fetal transmission and adverse perinatal outcomes in pregnant women infected with Zika virus: prospective cohort study in French Guiana. Br Med J. 2018;363. doi: 10.1136/bmj.k4431
2. Pomar L, Malinger G, Benoist G, et al. Association between Zika virus and fetopathy: a prospective cohort study in French Guiana. Ultrasound Obstet Gynecol. 2017;49(6):729-736. doi: 10.1002/uog.17404.
3. Hassan SM, Ramadan W, Kamel AM. Prevalence of central nervous system malformations in Egypt: a hospital-based study. Journal of Evidence-Based Women’s Health Journal Society. 2016;6(3):106-109. doi: 10.1097/01.EBX.0000488792.53166.aa.
4. Ghavami M, Abedinzadeh R. Prevalence of perinatal central nervous system anomalies in East Azarbaijan-Iran. Iran J Radiol. 2011;8(2):79-81.
5. Deepali Onkar PO, Kajal Mitra. Evaluation of fetal central nervous system anomalies by ultrasound and its anatomical correlation. J Clin Diag Res. 2014;8(6):AC05-AC07. doi: 0.7860/JCDR/2014/8052.4437.
6. Flamand C, Fritzell C, Matheus S, et al. The proportion of asymptomatic infections and spectrum of disease among pregnant women infected by Zika virus: systematic monitoring in French Guiana, 2016. Eurosurveillance. 2017;22(44):17-00102. doi: 10.2807/1560-7917.ES.2017.22.44.17-00102.
7. Theel ES, Hata DJ. Diagnostic testing for Zika virus: a postoutbreak update. J Clin Microbiol. 2018;56(4). doi: 10.1128/JCM.01972-17.
8. George KS. Molecular diagnosis of Zika virus infections. Rev Med Microbiol. 2018;29(1):8-16. doi: 10.1097/MRM.0000000000000125.
9. Reagan-Steiner S, Simeone R, Simon E, et al. Evaluation of placental and fetal tissue specimens for Zika virus infection - 50 States and District of Columbia, January-December, 2016. Morb Mortal Wkly Rep. 2017;66(24):636-643. doi: 10.15585/mmwr.mm6624a3.
10. Lambert V, Pomar L, Malinger G. The Zika virus epidemic in French Guiana: proposition of an ultrasound-based score for the diagnosis of fetal congenital Zika virus syndrome (Abstract). Ultrasound Obstet Gynecol. 2017 2017/09/01;50(S1):19-19. doi: 10.1002/uog.17617.
11. Costa F, Sarno M, Khouri R, et al. Emergence of congenital Zika syndrome: Viewpoint from the front lines. Ann Intern Med. 2016;164(10):689-691. doi: 10.7326/M16-0332.
12. Bautista LE. Maternal Zika virus infection and newborn microcephaly —An analysis of the epidemiological evidence. Ann Epidemiol. 2017;28(2):111-118. doi: 10.1016/j.annepidem.2017.11.010.
13. de Fatima Vasco Aragao M, van der Linden V, Brainer-Lima AM, et al. Clinical features and neuroimaging (CT and MRI) findings in presumed Zika virus related congenital infection and microcephaly: retrospective case series study. Br Med J. 2016;353:i1901-i1901. doi: 10.1136/bmj.i1901
14. Schuler-Faccini L. Possible association between Zika virus infection and microcephaly -Brazil, 2015. Morb Mortal Wkly Rep. 2016;65(3):59-62. doi: 10.15585/mmwr.mm6503e2.
15. Turchi Martelli CM, Albuquerque MdF, Araújo TVB, et al. Microcephaly in infants, Pernambuco State, Brazil, 2015. Emerg Infect Dis. 2016;22(6):1090-1093. doi: 10.3201/eid2206.160062.
16. Hazin AN, Poretti A, Di Cavalcanti Souza Cruz D, et al. Computed tomographic findings in microcephaly associated with Zika Virus. N Engl J Med. 2016;374(22):2193-2195. doi: 10.1056/NEJMc1603617.
17. Melo A, Aguiar R, Amorim M. Congenital zika virus infection: Beyond neonatal microcephaly. JAMA Neurology. 2016;73(12):1407-1416. doi: 10.1001/jamaneurol.2016.3720.
18. Larcos G, Gruenewald SM, Lui K. Neonatal subependymal cysts detected by sonography: prevalence, sonographic findings, and clinical significance. AJR. 1994;162(4):953-956. doi: 10.2214/ajr.162.4.8141023.
19. Kelly EN, Allen VM, Seaward G, et al. Mild ventriculomegaly in the fetus, natural history, associated findings and outcome of isolated mild ventriculomegaly: a literature review. Prenat Diagn. 2001 2001/08/01;21(8):697-700. doi: 10.1002/pd.138.
20. Haimovici JA, Doubilet PM, Benson CB, et al. Clinical significance of isolated enlargement of the cisterna magna (> 10 mm) on prenatal sonography. J Ultrasound Med. 1997;16(11):731-734. doi: 10.7863/jum.19220.127.116.111.
21. Epelman M, Daneman A, Blaser SI, et al. Differential diagnosis of intracranial cystic lesions at head US: Correlation with CT and MR imaging. Radiographics. 2006;26(1):173-196. doi: 10.1148/rg.261055033.
Competing interests: No competing interests
Re: Adverse perinatal outcomes in pregnant women infected with Zika virus in French Guiana: missing confounders?
I read with interest the recent article by Pomar et al. which indicates the incidence of Zika virus (ZIKV) related congenital disease in French Guiana was lower than that reported elsewhere in the early days of the South American ZIKV epidemic.
I am curious to know about some possible potential confounding factors apparently not mentioned in this paper. I note that although alcohol consumption states was reported, tobacco smoking/use was not, even though smoking is well-known to have deleterious effects on pregnancy e.g. intra-uterine growth retardation (IUGR). Did the authors list alcohol consumption but not tobacco smoking because the former is a prevalent habit in the French Guianan pregnant population but not the latter? It would be helpful to understand why this important risk factor for adverse pregnancy outcomes was not presented along with alcohol consumption.
The other potential confounding factor which could affect pregnancy & fetal outcomes is consanguinity. I have never been to French Guiana and am not familiar with it is demographics. However, I note the Centre Hospitalier de l'Ouest Guianais (CHOG) is in Saint Laurent du Maroni. I understand this is near the Maroni River area of French Guiana. I have gleaned from a quick search of the literature that the Maroni River area is populated by descendants of escaped African slaves, a people known as Maroons. The Maroons are apparently quite a genetically distinctive population with little mixing with French Guiana's other ethnic groups. This has led to a relatively high incidence of some genetic diseases like Infantile Pompe's Disease in this population.
Although Pomar et al comment on the socio-economic challenges facing the Maroni River population, they do not mention consanguinity or the limited genetic diversity of this population, which could affect the rates of congenital anomalies. It would be interesting to hear the CHOG group's opinion on this point as they undoubtedly have the local expertise to make a more informed commentary on this than I can.
Without further information on these important confounding factors, it is difficult to dispel a "Devil's Advocate" suspicion that the observed adverse fetal outcomes attributed to congenital Zika syndrome (CZS) are due in part to factors other than CZS in the distinctive Maroni River population.
 Maternal fetal transmission and adverse perinatal outcomes in pregnant women infected with Zika virus: prospective cohort study in French Guiana. Pomar L et al. BMJ 2018';363k4431 https://www.bmj.com/content/363/bmj.k4431
 Smoking and pregnancy - A review on the first major environmental risk factor of the unborn. Mund M et al. Int J Environ Res Public Health 2013 Dec; 10(12):6485-6499 https://dx.doi.org/10.3390%2Fijerph10126485
 Incidence of Infantile Pompe Disease in the Maroon population of French Guiana. Elenga N et al. BMJ Paediatrics Open 2017; 2(1) http://dx.doi.org/10.1136/bmjpo-2017-000182
Competing interests: No competing interests