Serotype distribution and incidence of invasive early onset and late onset group B streptococcal disease amongst infants in Singapore

Lancefield group B Streptococcus (GBS) or Streptococcus agalactiae invasive disease continues to cause significant mortality and morbidity in neonates and young infants. In 1996, the Centres for Disease Control and Prevention (CDC) published the consensus guideline on the prevention of perinatal GBS infection and this was subsequently updated in 2019. Based on this guideline, CDC advocated for universal screening of all pregnant women at 35 to 37 weeks of gestation for maternal vaginal and rectal GBS colonization [1, 2]. It was also recommended that intrapartum antibiotic prophylaxis should be given to the pregnant women who were screened positive for GBS colonization or had other risk factors [1, 2]. With the implementation of these guidelines, the incidence of GBS early onset disease (EOD) in the United States of America (USA) had decreased from 1.8 cases per 1000 live births in 1990 to 0.23 cases per 1000 live births in 2015 [3]. Comparatively, the incidence of GBS late onset disease (LOD) had been stable at an average of 0.31 cases per 1000 live births from 2006 to 2015. While the risk factors for GBS EOD are well studied and proven, the risk factors for LOD are less clear although preterm birth had been described to be strongly associated with LOD [2‐6].

In 2010, our institution had adopted specific steps to reduce the incidence of invasive GBS disease in our hospital based on recommendations from Royal College of Obstetricians and Gynaecologists (RCOG), CDC and National Institute for Health and Care Excellence (NICE) guidelines [1, 7, 8]. Vaginal GBS screening of pregnant women was conducted at 35 to 37 weeks if there was no planned elective Caesarean section. Indications for intrapartum GBS prophylaxis in the form of intravenous penicillin included any of the following: (i) previous infant with invasive GBS disease (ii) GBS bacteriuria during the current pregnancy (iii) positive GBS screening test during current pregnancy (iv) unknown GBS status and delivery at < 37 weeks gestation or rupture of membranes > 18 h or intrapartum temperature ≥ 38.0 °C. For patients where Caesarean section was performed before the labour onset or with intact amniotic membranes, intrapartum GBS prophylaxis was not indicated.

In a previous study done in our institution from 1996 to 1997 where the routine screening for maternal GBS status and administration of intrapartum antibiotics to high-risk pregnant women were not the standard of practice, the GBS EOD and LOD incidences were reported to be 0.265 per 1000 live births and 0.07 per 1000 live births, respectively [9]. There were no further studies that looked at the impact of the implementation of the guidelines on the EOD and LOD incidences after 2010. Data regarding the incidences of GBS EOD and LOD in Singapore and other Asia countries were also limited [10].

With a recent hexavalent capsular polysaccharide conjugate vaccine GBS6 (serotypes Ia, Ib, II, III, IV and V) that was undergoing trials [11], it has been proposed that administration of a safe and effective GBS vaccine in addition to the current GBS prevention strategies may further decrease the incidence of invasive GBS infection especially LOD [12]. We sought to understand the potential serotype coverage of this GBS conjugate vaccine in our population. In this study, we aimed to examine the causative GBS serotypes in invasive GBS disease, determine the incidences of EOD and LOD, and compare the risk factors between EOD and LOD.

This is the first study that showed the invasive GBS disease serotype distribution in Singapore. Serotype III was the most common serotype in both the EOD and LOD groups. Serotypes Ia, Ib, II, III, and V also accounted for 98.3% (n = 58) of the serotypes detected for invasive GBS diseases. We also showed that the mean incidence of invasive GBS disease was 0.42 per 1000 live births. The LOD incidence was at a mean of 2 times (range: 1.0 to 6.0) of EOD incidence per year despite the GBS preventive strategies. Known risk factors of EOD such as prematurity, peripartum maternal pyrexia and prolonged rupture of membrane were not associated with LOD. Chinese ethnicity and negative/unknown maternal GBS status were found to be associated with LOD as compared to EOD.

Worldwide, meta-analysis of serotype prevalence showed that GBS serotype III was the predominant serotype [10, 17, 18]. Around 50% of the EOD and 75% of the LOD cases were caused by serotype III. Five serotypes (Ia, Ib, II, III and V) were responsible for 97% of invasive GBS diseases around the world. Our findings mirrored this where serotype III was the most common isolate in our cohort and the serotypes Ia, Ib, II, III, and V accounted for 98.3% of the detected GBS isolates  in our cohort. Serotypes Ia, Ib and III were also frequently reported to cause invasive GBS disease around the world, and these three serotypes were seen in 89.8% of the GBS isolates that we had identified. A trivalent (serotype Ia, Ib and III) GBS conjugate vaccine or a hexavalent capsular polysaccharide conjugate vaccine GBS6 (serotypes Ia, Ib, II, III, IV and V) is likely to cover the majority of causative GBS serotypes in our population.

As compared to other Asian countries, our findings  were similar to studies from China [19, 20] and Taiwan [21] where serotype III was the most common serotype accounting for invasive GBS disease although in Taiwan, serotype VI was the leading cause of EOD from 2007 to 2010. It appears that serotypes leading to invasive GBS disease in infants can be similar although some variations can occur in different geographical locations, climate, and year of study. In our cohort, we also detected 1 EOD which was caused by serotype VI. This serotype does not lead to invasive disease commonly, although it had been reported to lead to significant proportions of maternal GBS colonization in Southeastern Asia countries previously [10]. We do not have the serotype data of maternal GBS colonization in our cohort to make this comparison.

The mean incidence of invasive GBS disease in our study was 0.42 per 1000 live births which was lower than the reported worldwide pooled incidence of 0.49 per 1000 live births in a systematic review [10]. In studies done in Asia, the overall incidence of GBS disease was 0.31 per 1000 live births and 1.1 per 1000 live births in China and Taiwan, respectively [19, 21]. Our mean EOD incidence of 0.13 per 1000 live births was also lower than the reported worldwide pooled EOD incidence of 0.41 per 1000 live births. There was a decrease in the incidence of EOD found in this study as compared to the previous finding of 0.265 per 1000 live births before the implementation of steps to reduce invasive GBS disease in our institution [9]. This shows that the burden of GBS EOD in our country had decreased over the years. However, the implementation of the current GBS preventive strategies did not have any impact on the LOD burden. Our mean LOD incidence remained high at 0.29 per 1000 live births. Similar to our findings, other studies also showed that the LOD rates were not affected by the widespread use of intrapartum antibiotic prophylaxis for high-risk pregnant women [1‐3, 12, 22]. Twenty-five percent of our infants with LOD had GBS meningitis implying significant morbidity for this group of patients. In a systematic review, 32% of GBS meningitis survivors had neurodevelopmental impairment at 18 months of follow-up, including 18% with moderate to severe neurodevelopmental impairment [23]. New interventions such as GBS vaccine are required to tackle the GBS LOD burden in our country and worldwide.

The risk factors for EOD had been well described in many studies [2, 22, 24‐29]. Prematurity, increased duration of rupture of membranes, maternal intrapartum fever, maternal vaginal GBS colonization and maternal GBS bacteriuria are proven risk factors of GBS EOD. However, the risk factors for LOD are less known. Maternal vaginal GBS colonization was reported to be a risk factor for LOD in a few studies, but these findings were inconsistent [5, 6]. On the other hand, prematurity was reported to be strongly associated with GBS LOD [3‐6, 22]. In our cohort, only the Chinese ethnicity and negative/unknown antenatal maternal GBS status were associated with LOD after multivariate analysis. This association between LOD and Chinese ethnicity was not previously reported, although the African American ethnicity was associated with GBS LOD in some studies in the United States [5, 22, 30]. This association was thought to be due to socio-economic factors and access to medical care. The association between Chinese ethnicity and LOD in our study may reflect differences in cultural practices and should warrant further investigation. We do not have the socio-economic status of the infants in our population to evaluate the impact of socio-economic factors on GBS LOD.

We also found that LOD was associated with negative/unknown antenatal maternal GBS status, and this suggests that antenatal maternal GBS status is not helpful in estimating the risk of LOD. In settings where elective Caesarean section has been arranged, vaginal GBS screening swabs will not be performed by most Obstetricians since GBS antibiotics prophylaxis will not be recommended for the clinical setting. It is known that vaginal screening swabs may yield false negative result due to the transient GBS colonization in the vaginal tract [11]. Improper sampling can also lead to false negative results as rectovaginal swab for GBS gives the best yield as compared to vaginal swabs [31]. GBS screening is recommended to be done between 35 and 37 weeks of gestation in most guidelines and GBS colonization may still occur after the initial screening tests yield negative result. Moreover, there are other potential sources of GBS transmission from the mother to baby such as contaminated breastmilk with GBS that had been reported to be associated with GBS LOD [32]. It was suggested that the breastmilk might be contaminated through the transmission of GBS from the breastfed infants who had been initially colonized with GBS during delivery or after birth, leading to multiplication of GBS in the mammary ducts and persistent exposure of the infant and mother to GBS thereafter [33, 34]. The other proposed mechanism was the translocation of GBS from the gastrointestinal tract via the lymphatics to the mammary glands [32]. High bacterial load in cases of mastitis and prematurity with less developed immune function were two other factors associated with GBS LOD after the exposure to GBS contaminated breastmilk [5, 35]. At this point, it is not proven that GBS in breastmilk is indeed the causative factor for LOD and there is not enough evidence for breastfeeding to be discontinued if the breastmilk is found to be GBS positive. There is also no consensus regarding the routine testing of breastmilk for GBS. As our study is a retrospective analysis, we do not have the data to study these potential risk factors of GBS LOD.

Another reported risk factor for LOD was HIV exposure during the antenatal period [36, 37]. None of our infants were born to HIV positive mothers. Lastly, young maternal age and maternal consumption of capsules containing dehydrated placenta were reported to be associated with LOD in the United States in some studies [4, 21, 30, 38]. However, no mothers in our study reported consumption of these capsules and we do not have the maternal ages of our cohort to make this comparison.

Our study has some limitations. A small number of GBS isolates were obtained in the 8 years of the study and the isolates in the year 2014 were not available for serotyping. However, based on the fact that the distribution of serotypes in other years of the study had remained largely similar, it is likely that the serotype distribution in 2014 will reflect the trend found in other years. This is a single-centre study although our institution is the biggest tertiary neonatal and paediatric hospital. Based on our calculations, the reported EOD and LOD incidences are estimates for the entire country during the years of study. In our cohort, all of our infants survived the invasive GBS disease and those who recovered with no outstanding concerns were not given long-term follow up. Hence, we do not have long-term developmental follow up data on these infants who were infected with GBS. Based on an earlier study conducted in our institution from Jan 1998 to May 2013, 3 out of 20 (15%) of the GBS meningitis survivors developed moderate to severe neurodevelopmental impairment [39].

Serotypes Ia, Ib, II, III, and V account for 98.3% (n = 58) of the invasive GBS diseases. The overall incidence of invasive GBS disease was 0.42 per 1000 live births, and the LOD incidence was at a mean of 2 times (range: 1.0 to 6.0) of EOD incidence per year. A GBS vaccine that covers the major causative serotypes found in our cohort can potentially reduce the overall GBS disease burden in the country.