Background The source for late-onset neonatal infections (LONI) due to group B Streptococcus (GBS) has not been fully explored. We reviewed GBS LONI cases associated with contaminated breast milk to determine whether breast milk was a possible route for neonatal infection.
Data sources A PubMed search from January 1977 to March 2013 was performed with MeSH words “Streptococcus agalactiae”, “group B Streptococcus”, “infection”, “milk”, “human”, “late-onset infection” and/or “neonate”; relevant cross references were also reviewed.
Results Forty-eight documented cases of GBS LONI matched our search criteria and were retrieved from the literature. When performed, molecular typing identified clonal isolates in the neonate and milk samples taken after LONI in all cases, with the hypervirulent sequence type 17 (ST-17) clone identified in two of these cases. Caesarean delivery combined with the absence of GBS recovery from maternal samples other than milk was noted for four cases. The rate of recurrent infections was high (35%) and, together with the data reviewed, points to a potential role of breast milk in GBS LONI.
Conclusions The cases reviewed here, together with the evidence of breast milk transmission for other pathogens, suggest that breast milk, which would account for repeated GBS transmission to the neonate, may favour gut translocation and subsequent LONI. Further investigations are nevertheless needed to study the relative importance of this contamination route compared with persistent postnatal gut colonisation and the dynamics of milk and neonatal gut colonisation.
Statistics from Altmetric.com
What is already known on this topic
Breast milk represents a source of pathogen transmission to the neonates.
Source of Group B Streptococcus (GBS) involved in late-onset neonatal infections (LONI) has not been yet fully elucidated.
Cases of GBS LONI were associated with contaminated breast milk.
What this study adds
Summarised information is given for 48 Group B Streptococcus (GBS) late-onset neonatal infections (LONI) cases associated with contaminated milk.
GBS LONI may occur in the absence of identified risk factors other than GBS milk contamination and high recurrence rate is revealed.
Breast milk may be heavily GBS-contaminated and the role of contaminated milk in GBS LONI has to be considered beside other sources of neonatal contamination.
The composition of bacterial gut microbiota and its evolution show notable differences in term and preterm infants,1–3 as the diversity and dynamics of intestinal microbiota establishment depend on multiple factors.4 ,5 A significant factor is the first feeding practice.6 ,7 Breast milk is widely acknowledged to be beneficial for newborn health by preventing digestive bacterial overgrowth.8–12 However, many pathogens can contaminate human milk, including viruses, bacteria and parasites. Some of them have been implicated in the neonatal infections related to contaminated milk,6 ,13–16 the most striking example nowadays being HIV.17 Breast milk is usually not sterilised but is given raw, frozen or pasteurised to keep the maximum benefits of breast feeding18 and thus may be a route for neonatal contamination by pathogenic microorganisms.
Streptococcus agalactiae, also called group B Streptococcus (GBS), is involved in about one-tenth of late-onset neonatal infections (LONI), and the hypervirulent clone of sequence type 17 (ST-17) identified among serotype III GBS (S3GBS) is significantly predominant in these LONI.19 The average incidence of GBS LONI, 0.24 per 1000 live births (95% CI 0.17 to 0.30),20 has remained stable in the past several years.20 GBS LONI are thought to occur after gut translocation following neonatal contamination at birth from maternal vaginal and/or digestive colonising bacteria,21 and contamination mediated via other routes has received little attention. However, contaminated breast milk, which would account for repeated GBS transmission, should be considered. Moreover, identifying the relative contributions of all contamination routes is essential to better understanding the pathophysiology of LONI and devising appropriate prevention strategies.
We herein review the literature on GBS LONI cases linked to contaminated milk. Our twofold aim is to present the clinical spectrum and bacteriological and epidemiological characteristics of these infections and to evaluate the evidence for and against the role of contaminated breast milk in the occurrence and/or recurrence of LONI.
Data source and definition
A PubMed search from January 1977 to March 2013 was performed using MeSH words “S agalactiae”, “group B Streptococcus”, “infection”, “milk”, “human”, “late-onset infection” and/or “neonate”; relevant cross references were also reviewed.
The literature proposes two lower age limits to define LONI: 3 or 7 days.20 ,22 In this review, LONI between 72 h and 3 months of life were included. Only LONI proved by GBS recovery from a usually sterile site were included. Breast milk was assumed to be contaminated by GBS in cases of either positive culture or positive PCR.
Epidemiology, clinical spectrum and bacteriological features of GBS LONI associated with contaminated breast milk
Severe infections with high recurrence rate
Forty-eight cases of GBS LONI related to GBS-positive breast milk were reported in the literature (tables 1 and 2).21 ,23–45 Nine cases occurred after the first month of life (between the 33rd and 74th day), and three cases occurred between the 3rd and 7th days. Infection occurred in term (54%) and preterm (46%) neonates, and caesarean delivery was noted for 15 of the 27 neonates with documented delivery mode (tables 1 and 2). Despite appropriate antimicrobial therapy (data not shown), 13 of the 37 neonates whose outcome was described (35%) presented one (n=10) or two (n=3) episodes of recurrent infection, occurring from 10 to up to 77 days after the first early-onset infection (EOI) (n=4) or LONI (n=9) episode (table 1). For 37 cases, the site of GBS isolation was reported. GBS was mainly recovered from blood cultures (n=30, 81% of the neonates) and cerebrospinal fluid (n=16, 43%). More rarely, urinary tract infections (n=5, 13.5%) and osteoarticular infections (n=1, 2.7%) were described, at times associated with bacteraemia and meningitis. For the 25 cases with available long-term follow-up, outcome was mostly favourable, except for the death of one preterm and one term neonate (4.2% fatality rate) and two cases of recovery with neurological sequelae (4.2% of the infants).29 ,31 ,43
Prevalence of serotype III GBS and demonstration of the hypervirulent ST-17 clone
The characteristics of the GBS strains were available for 21 cases (table 1). S3GBS was identified in samples from 18 of these infants (86%). S3GBS is known to be predominantly involved in GBS LONI, being recovered in approximately 50% up to 83% of LONI cases.20 Multilocus sequence typing identified the hypervirulent clone ST-17 S3GBS in three cases, including a case of recurrent infection. Cases with sequelae or deceased cases involved one ST-17 S3GBS and three GBS without specification (table 1).
Molecular epidemiology showed clonal strains in milk and neonatal samples
Molecular typing of maternal milk and neonatal GBS isolates showed that they were clonal in all 20 cases for which comparison was performed.26 ,27 ,31 ,35 ,37 ,38 ,40 ,42 ,45 The hypervirulent ST-17 clone was present in breast milk and neonatal samples in two cases (table 1).43
Critical review of the role of breast milk in GBS LONI pathophysiology
Vaginal carriage, a well-known but inconsistently detected condition for GBS transmission to the neonate
To prevent EOI, the Centers for Disease Control and Prevention (CDC) recommends screening for GBS colonisation in pregnant women with isolation of the bacterium from vaginal and rectal swabs by growth in a selective enrichment medium, followed by subculture on sheep blood agar.46 ,47 Carriage may be intermittent and detection is less reliable if not performed within 35–37 weeks of gestation.48 Berardi et al44 reported data for 98 neonates with LONI. The GBS culture was positive during prenatal screening for 13/40 of the screened cases and positive after diagnosis for 30/47 of the screened cases. However, despite appropriate detection procedures, some cases of neonatal infection remained unrelated to GBS maternal carriage, as 10 mothers were always culture negative.44 In a cohort of 1569 mother/infant pairs for which the sampling protocol followed the CDC recommendations, GBS colonisation was positive in 44 infants (3.6%) despite no detection of maternal carriage.49 Similarly, a low percentage of colonised neonates (1%) was born from permanent non-carrier women in the study of Hansel et al.50
Although these findings should be carefully interpreted because of intermittent maternal carriage—evaluated at 19%50 and the lack of systematic and complete information on sampling time and screening method (vagina, rectum, enrichment or not) (table 2), the absence of detected maternal carriage (n=10; 59% of the screened women), caesarean delivery (n=15; 56% of the documented cases) or combined observations (n=4; 67% of cases for which both data were available) suggest that another route for GBS contamination of neonates may be possible.
Persistent neonatal gut colonisation, a rarely monitored factor favouring GBS digestive translocation
Intestinal GBS persistence is assumed to be the source for LONI. As the properties of the newborn's mucosal surface and immunity are not fully mature, gut translocation is thought to precede neonatal septicaemia.51 ,52 Prolonged intestinal carriage of GBS—up to 1 year—was observed in about 10% of infants whose mothers carried GBS and may favour gut translocation.50 In addition, the hypervirulent ST-17 clone of S3GBS displayed high invasive potential53 and a high propensity for intestinal tract persistence at high load for prolonged periods, at least 14 days, and translocation.54 However, the global burden of neonatal intestinal carriage of GBS remains unknown.
GBS colonisation of neonatal gut has rarely been investigated in LONI cases. The reports indicate differences in the timing, but most investigations were conducted after a LONI episode. In the cases reviewed here, GBS digestive carriage was investigated in eight neonates and was found positive in two of them during LONI (cases 27 and 44) (tables 1 and 2). Data on the dynamics of GBS gut colonisation in neonates in correlation with LONI are also lacking, particularly because (i) very few prospective cultivation-dependent studies have been conducted,50 (ii) cultivation-independent studies based on the 16S rRNA gene have usually reported on global Streptococcus spp. representation, without discriminating between species in the genus3 and, last but not least, (iii) few LONI cases were included in these studies.
Breast milk is rarely but possibly heavily GBS-contaminated
GBS was found in 0 to 2% of raw milk samples,55–61 and Serafini et al59 found GBS in 1.4% of 144 pasteurised milk samples. Cultivation-independent studies of milk bacterial communities have generally shown complex communities with several genera representing more than 5% of the relative community abundance; these major phylotypes belonged to the genera Streptococcus and Staphylococcus, but no complementary data on the Streptococcus species were available.62
The results of breast milk analysis are not systematically reported in cases of GBS LONI.19 ,54 For example, Tazi et al54 presented the case of a fatal LONI due to ST-17 S3GBS in a breastfed infant but did not report on the bacteriological findings in the milk. When milk was investigated, however, 25% of the cases showed contamination.21 Similarly, the type of milk given may not be recorded, even in LONI linked to contaminated milk (10 out of the 48 cases in table 2). The time of milk sampling for bacteriological analyses was available in 21 of the 48 reviewed cases, and contaminated breast milk was documented before the onset of neonatal infection in two of these cases (cases 1 and 31). In the 13 cases of recurrent infections, breast milk was cultured only at the time of the second LONI episode, except for two infants (cases 19 and 21). The GBS count in breast milk was available for 14 of the 48 cases and ranged between 103 and >109 CFU/mL after LONI onset (table 2).
Milk contamination may originate in the neonate or in maternal gut colonisation
Maternal milk ducts are generally assumed to be infected during breast feeding by bacteria that colonise the neonate's oropharyngeal mucosa in the perinatal period because of the high degree of retrograde flow back into the mammary ducts during suckling.63 However, recent studies using animal models or finding lactobacilli in the milk after oral administration suggest that bacteria of maternal digestive origin may also colonise the breast64–66 (for a recent review on milk contamination origin, ie, contamination vs active migration of bacteria, see Jeurink et al67). Vertical transfer of bacteria from the mother's gut to her milk and through the milk to the infant's gut has also been hypothesised for lactic acid bacteria.68 In this review, maternal gut colonisation was never specifically documented but only explored during rectovaginal carriage detection in four cases (table 2). Whatever the source, further multiplication of bacteria increases the microbial concentration in the milk,31 which would account for the repeated intake of highly contaminated milk, favouring persistence and increasing the GBS inoculum in neonatal gut.
Role of mastitis in pathogen transmission to the neonate: a controversial issue
Subclinical mastitis has been identified as a factor that favours milk-mediated GBS transmission.31 Mastitis was detected for 21 of the reviewed cases (table 2). Berardi et al44 suggested that the highest inocula of GBS were observed in milk from women with mastitis. The data retrieved for cases 16 and 23 indicate that the association between high GBS inoculum in milk and mastitis may not be strict, however, as a low bacterial load was observed in milk from one mother with mastitis and, conversely, a GBS inoculum >106 CFU/mL was found in milk from another mother without mastitis (table 2).
Eradication of GBS in milk: inconsistent results of antibiotic therapy
In some cases, successive antibiotic courses failed to produce a negative breast milk culture,43 suggesting that either antibiotics have poor diffusion in maternal colonisation sites or suckling leads to recontamination of milk ducts. In any case, bacteriological testing of milk after maternal treatment is not routinely performed to ensure the absence of GBS. In case 47, a term neonate who presented two LONI episodes following EOI, GBS was cultured from breast milk at the time of the first LONI episode; after the mother's antibiotic treatment, milk sterilisation was not controlled and the child was given pasteurised milk, which did not prevent the second GBS LONI episode.40 In case 19, the expressed breast milk was contaminated by GBS at the time of the first LONI episode (day 10 of life). The mother received adapted antibiotic treatment but the breast milk was not tested for bacteria post-treatment. The infant went to breast at 17 days of life, LONI recurred 8 days later (day 25), and the milk was found positive for GBS 48 h after the second hospital admission (day 27) (table 2).33
In 18 cases, the mothers chose to stop breast feeding because of positive milk cultures at the time of the first LONI episode (n=12), after LONI recurrence (n=4), after the second LONI recurrence (n=1) or after LONI following EOI (n=2) (table 2).
In contrast to GBS EOI, strategies to prevent GBS LONI have not yet been established because their pathophysiology is still not fully understood. The route for GBS transmission to the neonate in particular remains unclear. Although long-term GBS persistence in the neonatal gut has been demonstrated after colonisation at birth, other GBS sources may account for subsequent digestive translocation and LONI.50 ,54 Despite incomplete data, this review of LONI cases associated with contaminated breast milk argues for a potential role of breast feeding in GBS transmission to the neonate and further LONI occurrence and/or recurrence. First, LONI cases were reported in the absence of maternal GBS detection in neonates delivered by caesarean section, although this should be cautiously interpreted because of intermittent GBS carriage and the possibility for GBS transmission from maternal digestive and/or vaginal origin after birth. Second, a strikingly high rate of recurrent infection, 35%, was observed, although this may partly reflect a preference for publication of unusual cases with multiple episodes of infection. Of note, the overall rate of recurrent GBS invasive infections was recently estimated as 4.5%.19 However, the role of breast milk in pathogen transmission to the neonate cannot be definitively established until longitudinal prospective studies are conducted to evaluate and compare the dynamics of breast milk and neonatal gut colonisation and determine their respective chronologies. Such studies would also be helpful in clarifying the transmission mode for other pathogens more frequently involved in LONI, like coagulase negative staphylococci, which are the predominant contaminant of breast milk and the main bacteria involved in LONI in preterm neonates. In the meanwhile, because it is clear that raw milk is a vehicle for GBS, it should at least be considered as an amplification factor of neonatal gut colonisation.
Of all the milk given to infants, raw maternal milk is the only type not subjected to microbial control.69 Very recently, best practices on the use of breast milk in the context of prevention of GBS neonatal disease have been proposed, including the search for GBS in milk at the time of recurrent GBS neonatal disease, as previously suggested,43 ,45 and in cases of mastitis in mothers of high-risk preterm neonates admitted to neonatal intensive care units.45 For these most vulnerable neonates, bacteriological control of raw milk has also been proposed in the absence of mastitis.16 ,45 However, whatever the strategy may be, in cases of milk positivity, the best response has yet to be defined because a recent randomised controlled trial showed that pasteurising maternal milk does not appear to significantly change the overall incidence of LONI in preterm infants.70
Thanks are extended to C Carmeni for English editing of the manuscript.
Contributors AF, FL and MR: review of the literature, AF, FL and HM: manuscript preparation, AJ, GC, MR and EJB: critical reading of the manuscript.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.