Breastfeeding in the time of Zika: a systematic literature review

Search strategy

We searched the following electronic databases: Cochrane Central Register of Controlled Trials (CENTRAL), EBSCO, Gale, Science Direct, Scopus, US National Library of Medicine (PubMed) and Web of Science. Other sources employed were the World Health Organization and Centers for Disease Control and Prevention web pages between the 11th and 16th of July 2018. We designed search terms for each database according to the level of indexation. These search terms were: Zika, Zika virus, breast milk, human milk, breastfeeding, lactation, nursing, lactating women, lactating mother, nursing women and nursing mother (Supplementary material I). There were no restrictions in terms of study design, date or language. We registered this systematic review in PROSPERO, the international prospective register of systematic reviews funded by the National Institute for Health Research and the Centre for Reviews and Dissemination of the University of York, under the number CRD42018102055. The recommendations of the PRISMA group were followed in terms of identification, screening, eligibility and inclusion criteria (Moher et al., 2009).

Selection of studies

Both authors assessed the titles and abstracts of references of relevance to the objective of this review. For each reference, primary data pertaining to the participants, exposures, outcomes, methods and results were reviewed. The references cited in the studies included were also reviewed. There was no disagreement regarding the eligibility of studies.

Type of studies included

We included all studies that comply with the following criteria: (1) studies must comprise primary data produced from any type of study design, (2) participants: mother with infant(s), mother only or infant only, (3) data of breast milk as a potential fluid for the transmission of ZIKV, or (4) primary or secondary data of follow-up of infants with at least one previously published study that complies with criteria 1, 2 or 3 (Table 1). Applying these criteria, nine studies were ultimately included in the qualitative synthesis (Fig. 1).

Data extraction and synthesis

Based on the review objective, we extracted data pertaining to the history of suspected, probable or confirmed maternal Zika disease during pregnancy or the postnatal period, according to the PICOS model. According to the adapted Grading of Recommendations Assessment, Development and Evaluation (GRADE) for observational studies, the overall quality of all proposed outcomes was very low (Balshem et al., 2011; Guyatt et al., 2011; Johnson et al., 2016). All of the corresponding authors of the original studies included in the qualitative synthesis of this review were contacted in order to request additional information regarding the studies.

Subgroup I

The studies included in this review indicate that five mothers had the onset of symptoms of ZIKV infection prior to the birth of their babies: case 1 (C1), two days before birth (Besnard et al., 2014); C2, on arrival at the hospital for the birth (Dupont-Rouzeyrol et al., 2016); C3, week 28 of gestation (Cavalcanti et al., 2017); C4, week 36 of gestation (Sotelo et al., 2017) and C5, week 9 of gestation (Giovanetti et al., 2018). For this reason, where there was ZIKV infection in the newborns, the probable routes of infection were transplacental or during delivery.

The incubation period for ZIKV, estimated by the Center for Disease Control and Prevention among 197 symptomatic travelers probably infected via mosquito bites, is from 3 to 14 days, with symptoms expected within 1 week of infection in 50% of cases and within 2 weeks in 99% (Krow-Lucal, Biggerstaff & Staples, 2017). This incubation period did not differ qualitatively between pregnant and non-pregnant travellers (Krow-Lucal, Biggerstaff & Staples, 2017). The incubation period in a case of probable ZIKV infection through unprotected sexual intercourse in a pregnant woman at 15 weeks of gestation was reported at 22 days, since ZIKV RNA was detected in plasma and urine samples by real time reverse transcription polymerase chain-reaction (RT-PCR). At delivery, testing was negative for ZIKV in breast milk, umbilical cord, placenta, membranes and serum of the newborn (Desclaux et al., 2018). Moreover, the sequences of NS4 and NS5 ZIKV genes were identical in the plasma of the pregnant woman and semen of her husband (the index case), which is consistent with direct transmission according authors (Desclaux et al., 2018).

In the subgroup I of the reviewed cases (Besnard et al., 2014; Dupont-Rouzeyrol et al., 2016; Cavalcanti et al., 2017; Sotelo et al., 2017; Giovanetti et al., 2018), four newborns were identified born after the full term of pregnancy. For newborn C3, the gestational week at birth was not reported. In three of these cases, the delivery mode is known and in two cases the birthweight is known. Severe congenital abnormalities were reported in newborn C5 only, including microcephaly and alteration on left hand, but it was not specified whether the phase of acute infection by ZIKV in the mother had been confirmed during gestation, moreover, the mother was IgG positive for rubella and cytomegalovirus. The rest of the newborns are referred to as unremarkable or asymptomatic and follow-up is known in two newborns, C1 and C2, to 32 and 8 months, respectively, with both reported as being free of complications (Besnard, Dub & Gerardin, 2017; Colt et al., 2017).

For all of the mothers, except C5, the acute phase of infection by ZIKV was confirmed, with positive results in the serum or blood found using real time reverse transcription quantitative polymerase chain reaction (RT-qPCR) within a period no greater than 3 days from the onset of the symptoms. In mother C5, approximately 29 weeks after the report of symptoms compatible with infection by ZIKV and after delivery of the child, authors report “low viral loads” in the serum that “prevented virus detection by RT-qPCR”, and that the “serum yielded no detectable products” with conventional nested PCR for NS5. In breast milk, they also report that these “prevented virus detection by RT-qPCR”, although there was a positive result using conventional nested PCR for NS5 (Giovanetti et al., 2018).

The presence of ZIKV was confirmed through RT-qPCR and conventional nested PCR for NS5, respectively, in newborns C1 (Besnard et al., 2014) and C5 (Giovanetti et al., 2018) only. For newborn C2, an “ambiguous” result was reported, without specifying the results of other molecular or serological analyses for ZIKV in this child (Dupont-Rouzeyrol et al., 2016). In newborns C3 (Cavalcanti et al., 2017) and C4 (Sotelo et al., 2017), the results for tests for ZIKV disease were reported as negative.

In the breast milk of mothers C1 (Besnard et al., 2014), C2 (Dupont-Rouzeyrol et al., 2016), C4 (Sotelo et al., 2017) and C5 (Giovanetti et al., 2018), the presence of ZIKV was confirmed through RT-qPCR or conventional nested PCR for NS5, with newborns C1 and C2 breastfed from the day of delivery, breastfeeding avoided in newborn C4 and breastfeeding unspecified in newborn C5.

Subgroup II

Following analysis of the studies included in this review, we classified one case in this subgroup: C6 (Besnard et al., 2014), a mother with ZIKV detected using a RT-qPCR conducted the day after cesarean delivery, with symptoms appearing two days afterwards. The newborn C6 was born at the end of a full term of pregnancy, but its birthweight is not known. Serum testing of the newborn produced a negative result for ZIKV on the first and third days after birth.

Breastfeeding of newborn C6 began on the third day after birth and, on that day, a breast milk sample tested positive for ZIKV by RT-PCR, but no replicative ZIKV were detected in the cell culture. One day after beginning breastfeeding, newborn C6 developed a rash and its serum tested positive for ZIKV by RT-PCR. By day eight after delivery, mother C6 tested negative for ZIKV in serum and positive in urine, her breast milk sample had a positive result for ZIKV, but no replicative ZIKV was detected in cell culture. That same day, newborn C6 tested positive for ZIKV in the serum, but tested negative one day afterwards in the urine.

While there were no results reported for the presence of ZIKV RNA in the placenta, cord blood or amniotic fluid of C6, the authors of the study suggest that the mother “was viraemic or at least incubating ZIKV at the time of delivery”. This interpretation concurs with the incubation period estimated in the travellers (Krow-Lucal, Biggerstaff & Staples, 2017). Moreover, using pooled data of 25 travellers in an endemic area, a mean period of 9.9 days (95% credible interval, 6.9–21.4) was estimated in order to have no detectable ZIKV RNA in the blood following infection, and 95% of cases would be expected to have no detectable virus by 18.9 days (95% CrI: 13.6–79.4) after infection (Lessler et al., 2016). This estimation was made with 11 women, comprising one case of probable sexual transmission, while the rest were presumed to have been infected through mosquito bites, and 14 men, one infected experimentally and the rest presumed to have been infected through mosquito bites (Lessler et al., 2016). It has also been reported in 6 patients that ZIKV RNA was detectable in urine samples at ≤15 days (range 10 days to >20 days) after the onset of symptoms; however, this particular publication did not specify subject sex, age or probable route of transmission (Gourinat et al., 2015).

Considering these estimates of the periods of incubation and viral clearance in the blood, as well as the persistence of ZIKV RNA in urine, it is highly likely that mother C6 acquired the infection by ZIKV during pregnancy, and thus her baby acquired the infection via the transplacental route. The authors state that transmission to newborn C6 via infected mosquito bite “seems fairly improbable because of the air-conditioned rooms in the hospital”, although the temperature inside the hospital was not specified. Newborn C6 evolved favorably, was breastfed for 2 months and, at 32 months, was reported to have no complications (Besnard et al., 2014; Besnard, Dub & Gerardin, 2017).

Subgroup IV

Following analysis of the studies included in this review, in this subgroup were classified two cases: C7 (Cavalcanti et al., 2017) and C8 (Blohm et al., 2017; Blohm et al., 2018), both mothers with ZIKV infection confirmed in the urine at day five after the onset of symptoms. The delivery mode, birthweight and gestational week at birth of newborns C7 and C8 were not specified.

By day 5, mother C7 had negative results for ZIKV RNA in the serum and breast milk. Her infant “rejected the formula” and continued breastfeeding during the symptomatic phase, and was reported as asymptomatic (Cavalcanti et al., 2017). No studies were carried out to detect ZIKV RNA in the infant, for which reason there is no information with which to determine probable routes of transmission. Moreover, the breast milk, which was the only test conducted, was negative for ZIKV RNA. Over the same period, the husband of mother C7 had negative results for ZIKV RNA in both serum and urine (Cavalcanti et al., 2017).

On day 5 after the onset of symptoms, mother C8 was IgM positive, IgG “borderline” and ZIKV RNA positive in urine, with no results provided for analysis of ZIKV RNA in serum. The authors interpreted the serological data as being indicative of seroconversion. Mother C8 practiced full breastfeeding with her 5 month-old infant, which was reported as asymptomatic throughout the study period (Blohm et al., 2017; Blohm et al., 2018).

The optimal period for IgM detection by enzyme-linked immunosorbent assay (ELISA) and immunofluorescent assay (IFA) has been reported to be from 7 days to 2 months after onset of ZIKV disease (Atif et al., 2016), and for IgG by ELISA, IFA and neutralization test, from 10 days to 6 months after onset of ZIKV disease (Atif et al., 2016). In this sense, it is plausible that mother C8 could have undergone a “rapid seroconversion” due to the physiology of lactation itself.

Specific immunomodulatory constituents of human milk have been reported in maternal human immunodeficiency virus (HIV) infection (Pedersen et al., 2016), maternal breast infections (Hassiotou & Geddes, 2015) and other maternal organ-specific infections (Hassiotou et al., 2013). Moreover, infant infection stimulates a rapid leukocyte response in human milk, where exclusive breastfeeding is associated with a greater baseline level of leukocyte (Hassiotou et al., 2013).

Authors (Blohm et al., 2017) have reported the full genomic sequence of ZIKV RNA, obtained from the spent media of LLC-MK2 cells inoculated with lipid-enriched fraction breast milk of mother C8 and the urine of her child. Both sequences revealed 99% identity between the strains isolated from the mother and child. The authors do not discuss the probable source of infection for mother C8, and there are no references to clinical results in her epidemiological encounters. It is reported that mother C8 and her infant had no recent travel history and lived in Barquisimeto Venezuela in an air-conditioned, screened house, where the infant spent most of its time, without specifying the indoor temperature of the house. The authors (Blohm et al., 2017; Blohm et al., 2018) therefore state that the risk of mosquito transmission to infant C8 would be minimal.

Aedes aegypti mosquitoes present a high effectiveness of virus transmission in nature although they tolerate only a narrow range of temperatures (Centers for Disease Control and Prevention, 2018a; Brady et al., 2013; Marinho et al., 2015; Tsai et al., 2018). However, Aedes albopictus can tolerate a broader temperature range and cooler temperatures, which enables them to persist in more temperate climates (Brady et al., 2013; Marinho et al., 2015). Moreover, Aedes albopictus (Richards et al., 2006) and Aedes aegypti mosquitoes are capable of biting more than one person within a household (Chow-Shaffer et al., 2000).

Subgroup V

Following analysis of the studies included in this review, two cases were classified in this subgroup: C9 (Cavalcanti et al., 2017) and C10 (Cavalcanti et al., 2017), both mothers with ZIKV infection confirmed in serum by real time RT-PCR, but with no specification of delivery mode, birthweight or gestational week of birth of the children.

By day 4 after onset of symptoms, the breast milk of mother C9 was positive for ZIKV RNA and infective particles were detected in the cell culture analysis. Breastfeeding of her infant of 11 months of age was avoided after the onset of the maternal rash and resumed after this maternal rash disappeared. The only test of the infant, conducted in the urine on day 10 after onset of the maternal symptoms, was negative and the infant was reported as asymptomatic at least by day 17 (Cavalcanti et al., 2017).

The breast milk of mother C10 (Cavalcanti et al., 2017), in the only test conducted for detection of ZIKV by day 9 after onset of symptoms, the viral genome was not amplified by real time RT-PCR and her infant of 10–11 months of age, breastfed during the symptomatic maternal period, was found non-reactive for ZIKV, but positive for CHIKV in the only test conducted on the serum. Infant C10 presented cutaneous exanthema and a fever that lasted for 2 days. It should be noted that mother C10 was one month pregnant. Prolonged serum detection of ZIKV RNA in symptomatic pregnant women has been documented at 14 or more days after the onset of symptoms, with a range of 17–46 days (Meaney-Delman et al., 2016).