We retrospectively analyzed data from a network of hospitals in Italy. Episodes of LOD GBS are anonymously reported on a monthly basis to the coordinating center, Azienda Ospedaliero—Universitaria of Modena (Modena, Italy) (10). Hospitals participating in the network follow US Centers for Disease Control and Prevention (CDC) guidelines regarding antenatal GBS screening and IAP administration to women who are GBS-colonized (11). During January 1, 2007–December 31, 2018, we received notification of 175 cases of LOD, of which 98 had a full assessment of maternal carriage (see definitions in Appendix). We used a special form for surveillance, designed for both EOD and LOD reporting, that included patient demographics, mode of delivery, risk factors for EOD, and IAP administration. Surveillance officers extracted all clinical information from the labor and delivery records using this standardized form; they obtained any missing data by telephone from the coordinating center. To maintain patient confidentiality, spreadsheets submitted to the principal investigator were anonymous and did not include any identifiable data of patients or caregivers. The case reporting and isolate collection were determined to be non-research public health surveillance. The local ethical committee of Azienda Ospedaliero–Universitaria of Modena approved the study (no. 423/2019). We obtained a waiver of informed consent for each of the patients included.

A full assessment of maternal carriage was available in 98 cases of LOD during 2007–2018. Most cases of LOD (89/98) came from a regional area-based surveillance in which incidence of EOD is 0.18/1,000 live births (21) and of LOD is 0.31/1,000 live births (10), and the prevalence of maternal VR colonization is 21% (22).

Eighty (81.6%) cases occurred in full-term neonates and 18 (18.4%) in preterm neonates (of which 10 were still in hospital at the time of LOD onset). Compared with full-term neonates, preterm neonates were less likely to be delivered vaginally and more likely to undergo mechanical ventilation (Table 1). Twenty mothers (3 preterm and 17 full-term) were exposed to adequate IAP; of those, 17 (85%) carried GBS at the time of LOD diagnosis.

Thirty-three (33.7%) of 98 mothers were persistently not GBS-colonized; the other 65 (66.3%) mothers were GBS-colonized, 36 (36.7%) persistently (Table 2). Maternal VR colonization was more likely to be detected at the time of onset (58/65) than at the prenatal screening (43/65; p<0.01). At the time of LOD onset, 59.2% of mothers were colonized, 18.9% had asymptomatic GBS bacteriuria, and 20.5% had positive breast milk culture.

All mothers with asymptomatic GBS bacteriuria also carried GBS at the VR site. Median urinary bacterial count was 40,000 CFU/mL (interquartile range [IQR] 10,000–100,000 CFU/mL; range 1,000–1 million CFU/mL). GBS bacteriuria was significantly more likely to be detected at the time of LOD onset (17/90 tested) rather than during pregnancy (2/92 tested; p<0.01).

Among 17 women with a positive breast milk culture, 1 mother had mastitis (1 million CFUs/mL) and 16 had no mastitis (median bacterial count 300,000 CFU/mL; IQR 100,000–725,000 CFU/mL; range 9,000–6,400,000 CFU/mL). Fourteen (82.4%) of the 17 mothers were GBS colonized at the VR site at the prenatal screening or at the time of onset, or both, but the other 3 (17.6%) were persistently not colonized.

It is crucial to understand the pathway of GBS transmission in LOD to determine the necessary interventions. We collected a large set of maternal cultures at the onset of LOD and focused on mothers with a full assessment of VR carriage. At the time of LOD onset, a substantial proportion of mothers were found to carry GBS at the VR site, although some of them were GBS-noncolonized at prenatal VR screening. Maternal GBS colonization is an important risk factor for GBS disease, and determining the extent and types of colonization is essential for the formulation of a vaccine against GBS (8,23,24).

Rates of maternal asymptomatic GBS bacteriuria were strikingly high (≈19%). GBS bacteriuria, which affects 2%–7% of pregnant women (11) (2.2% in a recent area-based study in Italy [22]), is a marker for heavy genital tract colonization. GBS bacteriuria is associated with an increased risk for EOD in neonates (11), but its role in LOD has not been previously investigated. GBS bacteriuria was present in approximately one third of the cases among mothers with VR colonization at the time of LOD onset. This observation suggests that mothers, especially those heavily colonized, may be a main source of GBS exposure for their infants. Indeed, molecular typing indicated that GBS isolates collected from mother-infant pairs were closely related. All maternal and infant bacterial strains were serotype III and possessed the surface antigen Rib, and all but 1 pair displayed the same MLST type. The common origin of the bacterial maternal–infant pairs was confirmed by PFGE analysis when performed and the comparable antimicrobial susceptibility profile. This finding is consistent with a previous longitudinal study in 160 mother–infant pairs, which demonstrated that GBS strains isolated from healthy neonates and their mothers until 8 weeks postpartum were indistinguishable (i.e., had identical band patterns) by PFGE analyses (5). Globally, serotype III strains are clinically the most important, accounting for 25% of colonizing strains and 62% of strains causing invasive disease in infants, although geographic variation exists (1).

In this study, strains from all but 1 case belonged to clonal complex (CC) 17, a hypervirulent clonal lineage predominantly responsible for both EOD and LOD. In animal models, GBS CC17 shows higher abilities of intestinal colonization and translocation through physiologic barriers (25,26); >80% of cases of GBS serotype III LOD worldwide are caused by the hypervirulent CC17 (6,19,26,27). The emergence of a multidrug-resistant CC17 sublineage has been increasingly reported since its identification in China, Canada, and Europe (27–29); it is identifiable by the replacement of the pilus island 1 genetic locus by mobile elements carrying both tetO and ermB genes plus aminoglycoside resistance genes. The presence of the tetO-ermB genes along with that of PI-2b alone can be considered a marker of the emerging multidrug-resistant hypervirulent CC17 clonal lineage (27–29). Although we did not perform a detailed genomic analysis for all GBS strains, the antimicrobial resistance we detected was probably due to this multidrug-resistant CC17 subclone whose dissemination is still limited among neonatal infections in Italy. GBS resistance to clindamycin is well documented, but is relevant to only a small population of women and not to infants.

Mother-to-infant postdelivery GBS transmission can be assumed in some cases. Indeed, many neonates were born to mothers who had been exposed to IAP (which interrupts maternal-to-fetal transmission). Because maternal VR carriage at the time of late onset was confirmed in 85% of mothers who received adequate IAP, a postdelivery transmission is likely. This finding is consistent with recent studies showing risks of neonatal postpartum colonization from a maternal source (5,6,25). The importance of maternal colonization is probably greater in neonates born full-term because have frequent and close contact with their mothers, whereas preterm neonates admitted to hospital have fewer chances for transmission of GBS during close contact with their mothers. Although in this study VR colonization rates at the time of LOD onset were higher in full-term mothers (64% vs. 39% in preterm mothers), the difference was not significant, perhaps because of the small sample size.

In this study, we found GBS in ≈20% of breastfeeding mothers. Mastitis in LOD was infrequent; mothers with positive breast milk culture were more often asymptomatic, although their milk bacterial counts were sometimes high. This finding suggests a silent maternal duct colonization, and it is consistent with case reports of GBS breast milk–associated LOD, in which most mothers have no sign of mastitis (13).

Furthermore, in our study most mothers with positive breast milk culture carried GBS at the VR site, which was often heavily colonized. Maternal VR carriage would appear to be associated with GBS transmission into breast milk, perhaps in some cases after translocation from the gastrointestinal tract through the lymphatic system to the mammary glands (30). In contrast, only 3 mothers who had positive breast milk culture were persistently GBS-noncolonized at the VR site. In such cases, a circular mechanism of GBS transmission to neonates could be implicated. The retrograde theory assumes that GBS, present in the infant’s throat, colonizes the mammary ducts during breast-feeding; GBS load increases in the milk, and, in turn, the infant is infected during breast-feeding. Our data do not suggest that breast milk itself is a risk factor for LOD. Breast milk is known to contain immunomodulatory and antimicrobial components (12) (i.e., sIgA and cytokines) that may protect from LOD, and the lack of these components seems to increase the risk of persistent neonatal colonization (31) and LOD (32).

Taken all together, these results show that mothers are largely the predominant source of GBS in cases of LOD, both during childbirth (especially if IAP is not given) and in the postpartum period. GBS-positive breast milk is one of the ways by which heavily colonized mothers expose their infants to GBS.

The first limitation of our investigation is that it was an observational study without a control group. Therefore, the relevance of a positive breast milk in LOD could not be clearly assessed, because we do not know how many breastfeeding GBS-colonized mothers with healthy neonates have GBS-positive breast milk. However, Berardi et al. (5) found much lower rates (≈7%) of GBS-positive breast milk in a cohort of breastfeeding mothers (GBS-colonized at the VR site) who had healthy neonates. In addition, we cannot rule out an accidental contamination of some breast milk samples during collection, although we had provided instructions for collection. Second, although we proposed doing so, we did not systematically perform full assessment of VR culture both at prenatal screening and at the time of disease onset; just over half of the mothers had the full assessment during surveillance. In fact, not all of them had prenatal screening; furthermore, the collection of cultures at the time of LOD diagnosis and then shipping the maternal strains were challenging to organize. Third, the PFGE analysis was available only in a few mother-infant pairs. However, previous studies demonstrate that the concordance of GBS strains collected from mothers and their own neonates at delivery or in the following weeks is very high, reaching ≈100% of cases (5,6,18). Finally, maternal colonization rates could be higher than we found as a result of inherent insufficient sensitivity of maternal VR cultures (11), which may lead to false-negative culture results. We did not investigate the possible role of the father in the transmission of GBS.

In conclusion, this study suggests that most cases of LOD are strictly associated with maternal VR colonization and that CC17 is the predominant clonal lineage. Rates of asymptomatic GBS bacteriuria at the time of LOD onset are strikingly high, and this finding suggests heavy maternal colonization. Positive breast milk culture is relatively common among asymptomatic breastfeeding mothers of neonates with LOD, especially if they carry GBS at the VR site. However, the causal role of breastfeeding remains uncertain, and our data do not lead to definitive conclusions. Mother-to-infant transmission may occur after delivery. Our findings call attention to maternal transmission after delivery as an underestimated source of neonatal LOD and may assist in predicting the impact of maternal GBS vaccination.