Perinatal Streptococcus agalactiae Epidemiology and Surveillance Targets

Streptococcus agalactiae, or group B streptococcus (GBS), is a major neonatal pathogen. Recent data have elucidated the global prevalence of maternal and neonatal colonization, but gaps still remain in the epidemiology of this species.

SUMMARY

Streptococcus agalactiae, or group B streptococcus (GBS), is a major neonatal pathogen. Recent data have elucidated the global prevalence of maternal and neonatal colonization, but gaps still remain in the epidemiology of this species. A number of phenotypic and genotypic classifications can be used to identify the diversity of GBS strains, and some are more discriminatory than others. This review explores the main schemes used for GBS epidemiology and further details the targets for epidemiological surveillance. Current screening practices across the world provide a unique opportunity to gain detailed information on maternal colonizing strains and neonatal disease-causing strains, which is vital for monitoring and therapeutics, if sufficient detail can be extracted. Deciphering which isolates are circulating within specific populations and recording targets within invasive strains are crucial steps in monitoring the implementation of therapeutics, such as vaccines, as well as developing novel therapies against prevalent GBS strains. Having a detailed understanding of global GBS epidemiology will prove invaluable for understanding the pathogenesis of this organism and equipping future prevention strategies for success.

INTRODUCTION

Streptococcus agalactiae is the sole member of the Lancefield group, i.e., group B streptococcus (GBS), and therefore is commonly referred to as GBS (1). This Gram-positive coccus resides as a commensal of the gastrointestinal and genitourinary tracts of humans; however, it has the capacity to cause serious infections. These are often opportunistic in nature, affecting the elderly, the immunocompromised, and particularly neonates, for which GBS is a major cause of morbidity and mortality (2, 3). Recently, infections of healthy adults have also been reported (4, 5). In the affected populations, the spectrum of disease ranges from sepsis, pneumonia, and meningitis to endocarditis. GBS disease in neonates can be classified based on the time of onset: early-onset disease (EOD) occurs in the first week of life (6), while any disease presentation after this to up to 3 months of life is classified as late-onset disease (LOD) (7). In neonates, the majority of early-onset GBS disease occurs as a result of transmission from a colonized mother, either through ascending infection or during the process of vaginal birth (8). In contrast, late-onset sepsis is believed to be contracted via nosocomial or community-acquired means (9). Due to the potential for transmission from the mother, universal screening of pregnant women by either risk-based or culture-based methods has been implemented in most countries. In such countries, intrapartum antibiotic prophylaxis (IAP) is administered before delivery to women at risk of GBS disease (delivery at <37 weeks of gestation, intrapartum temperature of ≥38.0°C, or rupture of membranes for ≥18 h) or with GBS colonization (10). This widespread treatment strategy has been successful in reducing EOD significantly but has had no effect on LOD (10). Screening for GBS by culture currently involves vaginal and rectal swabs collected at 35 to 37 weeks of gestation and cultured in selective enrichment media (11). Additionally, diagnostic molecular techniques to detect GBS have also been described (12). The outcome of this screening is either GBS colonization or absence. However, there is a plethora of other information about these organisms that could be obtained during routine screening that may prove vital in furthering our understanding of their epidemiology; as such, we are currently underutilizing this potentially valuable surveillance source. Expanding our knowledge of current targets through the use of phenotypic and molecular assays to identify and differentiate isolates within the population will likely prove essential for future GBS interventions.

IDENTIFICATION AND CLASSIFICATION

Several methods for identifying and characterizing GBS for epidemiological and diagnostic purposes have been described (13). These include serotyping (based on the capsular polysaccharide), typing of surface proteins, multilocus sequence typing (MLST), multilocus variable-number tandem-repeat analysis (MLVA), and, more recently, clustered regularly interspaced short palindromic repeat (CRISPR)-locus assays. All have advantages and disadvantages, and all convey different levels of epidemiologic data in the context of surveillance.

Serotyping

The capsular polysaccharide (CPS) that encapsulates GBS is a major virulence factor, but it also determines the serotype. A total of 10 serotypes have been identified, with CPS IX being the most recently proposed one (14). Serotypes can be identified through latex agglutination and molecular assays, such as multiplex PCR (13, 15–17). As one of the first discriminating factors for GBS, the serotypes have been described in the literature as a way of grouping isolates in an attempt to understand their prevalence in maternal colonization and neonatal disease. Serotyping provides valuable information that allows classification into serologically similar groups, and patterns have emerged based on global prevalence. The most common global serotypes include CPS Ia, Ib, II, III, and V. In a perinatal context, however, many studies describe limited serotype testing (18–20), and others were completed prior to inclusion of CPS IX (21–23). These are major limitations in such studies, but there are still numerous studies that have included all known serotypes, and these generally support the overarching predominance of the five main CPS types in a perinatal context (24, 25).

Global serotype distribution.

A recent meta-analysis of maternal colonization described a global data set of serotype prevalences (24). Russell and colleagues highlighted the regional variation for the less common serotypes, such as CPS IV, VI, VII, VIII, and IX. Prior to this analysis, the general consensus was that approximately 10 to 30% of pregnant women were colonized with GBS; this analysis supports previous estimates and suggests that the global colonization rate in pregnant women is 18%, with regional estimates varying from 11 to 35%. South and East Asian countries had the lowest reported prevalences (11 to 12.5%), while the Caribbean was observed to have the highest prevalence (34.7%) (24). The subsequently estimated neonatal disease rate for GBS is 0.49 per 1,000 live births globally, with the highest incidence in Africa (1.12 per 1,000 live births) and the lowest in Asia (0.30 per 1,000 live births) (25). Five serotypes (Ia, Ib, II, III, and V) have been estimated to encompass approximately 98% and 97% of the serotypes identified during maternal colonization and neonatal disease, respectively. This leaves a small proportion of colonization/neonatal disease cases attributable to the remaining five serotypes; however, it is important to highlight these serotypes, as the dynamic has the potential to shift. Several studies have highlighted the variation in serotype prevalence globally, and the results are summarized here, along with general trends, for each serotype (Table 1).

TABLE 1.

Summary of various serotyping distribution studies that highlight serotype variation, showing the variety of patients, specimens, clinical presentations, and methods used^a

Study authors (yr)	Location	Patients	Specimen type(s)	Clinical presentation	Serotyping method(s)	Serotypes	Reference
Lin et al. (2016)	Central Taiwan	Neonates to adults	Blood	Invasive	PCR	All 10	28
		Pregnant women	Vaginal and rectal	Carriage	PCR	All 10	28
Uh et al. (2005)	South Korea	Adults	NS	Infection	Latex agglutination	Only Ia, Ib, II to VIII	29
Turner et al. (2012)	Thai-Myanmar border	Pregnant women	Vaginal and rectal	Carriage	Latex agglutination and PCR	All 10	30
Lopez et al. (2017)	Catalonia, Spain	Pregnant women and neonates	NS	Carriage	NS	NS	31
			NS	Invasive	NS	NS	31
Elikwu et al. (2016)	Nigeria	Pregnant women	Vaginal and rectal	Carriage	PCR	All 10	32
Foster-Nyarko et al. (2016)	Gambia	Infants	Nasopharyngeal	Carriage	Latex agglutination and PCR	All 10	33
Romain et al. (2017)	France	Neonates and infants	Blood and cerebrospinal fluid	Invasive	NS (French National Reference Centre for Streptococci)	Only Ia to V	35
Bjorsdottir et al. (2016)	Iceland	Adults	Blood and synovial fluid	Invasive	Latex agglutination	All 10	36
Teatero et al. (2017)	Toronto, Canada	Pregnant women	Vaginal and rectal	Carriage	PCR and latex agglutination	All 10	37
Teatero et al. (2015)	Toronto, Canada	Neonates to adults	Blood, cerebrospinal fluid, and tissue	Invasive	PCR and latex agglutination	All 10	38
Ferrieri et al. (2013)	Minnesota, USA	Infants	Blood, cerebrospinal fluid, bone, joint, and postmortem samples from liver and lung	Invasive	Immunoprecipitation	All 10	39
Blumberg et al. (1996)	Atlanta, GA, USA	Infants to adults	Blood, peritoneal fluid, and cerebrospinal fluid	Invasive	Capillary precipitin	Only Ia, Ib, II to V	41
Belard et al. (2015)	Gabon	Pregnant women	Vaginal and rectal	Carriage	Latex agglutination	All 10	42
Suara et al. (1994)	Gambia	Pregnant women	Vaginal and rectal	Carriage	Capillary precipitin	Only I to VI	43
		Neonates	Throat, rectal, and umbilical	Carriage	Capillary precipitin	Only I to VI	43
Shabayek et al. (2014)	Egypt	Pregnant and nonpregnant women	Vaginal	Carriage	PCR and latex agglutination	All 10	44
Lachenauer et al. (1999)	Japan	Pregnant women	Vaginal	Carriage	Inhibition enzyme-linked immunosorbent assay	Only Ia, Ib, II to VIII	45
Eskandrian et al. (2015)	Malaysia	Neonates and adults	Vaginal and urine	Carriage	PCR and latex agglutination	Only Ia, Ib, II to VII	18
			Blood, wound and skin ulcers, tracheal and gastric aspirates	Invasive/infection	PCR and latex agglutination	Only Ia, Ib, II to VII	18
Suhaimi et al. (2017)	Malaysia	Neonates to adults (80+ yr)	Vaginal, urine, and placental	Carriage	PCR and latex agglutination	All 10	46
			Blood and tissue	Invasive	PCR and latex agglutination	All 10	46
Dutra et al. (2014)	Brazil	Pregnant women	Vaginal and perianal	Carriage	Capillary precipitin	Only Ia, Ib, II to VIII	19
		Adults	Blood, urine, and male genital tract discharge	Invasive/infection	Capillary precipitin	Only Ia, Ib, II to VIII	19
Chaudhary et al. (2017)	India	Pregnant women	Vaginal and rectal	Carriage	Latex agglutination	All 10	49
Islam et al. (2016)	Bangladesh	Neonates	Umbilical cord	Carriage	PCR	All 10	50
Ekelund et al. (2003)	Denmark	Neonates to adults (80+ yr)	Blood	Invasive	Precipitin test	Only Ia, Ib, II to VIII	51
Paoletti et al. (1999)	Boston, MA, USA	Neonates and adults	Genitourinary tract, others NS	NS	Precipitin test	Only Ia, Ib, II, III, V, VI, VIII	52
Yan et al. (2016)	China	Pregnant women	NS	Carriage	PCR	All 10	53
Toyofuku et al. (2017)	Japan	Neonates	Nasopharyngeal and rectal	Carriage	PCR	Only Ia, Ib, II to VIII	54
Slotved et al. (2017)	Ghana	Pregnant women	Vaginal and rectal	Carriage	Latex agglutination	All 10	47
Gudjonsdottir et al. (2015)	Sweden	Neonates and adults	Blood, cerebrospinal fluid, and synovial fluid	Invasive	Latex agglutination	All 10	55
Takahara et al. (2015)	Japan	Neonate (case study)	Blood and cerebrospinal fluid	Invasive	Latex agglutination	All 10	56

^aNS, not stated.

Ia.

The frequency of different serotypes as colonizers makes determination of disease development an extremely difficult task. The asymptomatic colonization in some yet invasive disease progression in others emphasizes the complexity of the host-GBS interaction. Maternal colonization by serotype Ia is seen globally (24), and it is the most prevalent serotype in maternal disease, according to a limited meta-analysis (including only the United States, United Kingdom, and France) (26). Regarding neonatal disease, a previous meta-analysis found Ia to be the most frequent serotype contributing to EOD; however, this global estimate lacked data from Asia (27). The latest analysis includes Asia, and while overall trends of prevalence remain similar, South America is the only region in which CPS Ia predominates over CPS III in cases of EOD (25). In contrast, in East Asia, unlike other regions, the prevalence of Ib surpasses that of Ia (25).

Ib.

East Asia has the largest proportion of neonatal disease attributable to CPS Ib, and within the Asian region, CPS Ib is second only to CPS III in predominance (25). In Taiwan, Lin and colleagues found CPS Ib to account for 21.6% of the isolates collected from a mixed population of pregnant women, neonates, and nonpregnant adults, representing the most common serotype in the study (28). This frequency, however, was largely accounted for by nonpregnant adult invasive disease isolates, which represented 26/29 CPS Ib isolates. Similarly, in South Korea, 22% of GBS serotypes causing infection in adults were identified as Ib (29). Differences in distribution have been observed within different cohorts, such as pregnant, neonatal, and adult populations. Maternal colonization with CPS Ib in Mexico and South America has been reported to be considerably higher than that in other countries (24). On the opposite end of the spectrum, CPS Ib was the least prevalent serotype, after nontypeable (NT) isolates, to account for maternal disease (in the United States, United Kingdom, and France) (26).

II.

Serotype II is the fourth most prevalent serotype in maternal disease, according to a study including the United States, United Kingdom, and France (26), and is the fifth most prevalent in neonatal disease worldwide (25). The predominance of this serotype was observed in pregnant women at the Thai-Myanmar border, where 24% of isolates recovered were CPS II, as reported by Turner and others (30). A study in Catalonia, Spain, assessed GBS isolates from pregnant women and pathogenic strains isolated from neonates with sepsis and found that serotypes Ia and II were significantly more frequent than others as colonizing strains (31). Colonization of Nigerian pregnant women also showed a predominance of serotype Ia, while serotype II accounted for 71.4% of the neonatal colonizing isolates (32). Likewise, GBS isolated from the nasopharynxes of Gambian neonatal carriers found the highest prevalences of CPS V and II (33).

III.

CPS III is well known for its association with disease, in particular EOD, and it has been reported thoroughly worldwide and remains the leading cause of neonatal disease (25). Maternal prevalences differ by continent or region, as Africa, Australia, and New Zealand have much higher prevalences, while the Middle East, Eastern Europe, South Asia (including India and Bangladesh), and Central America have lower rates (24). In coastal Kenya, CPS III accounted for the majority of maternal colonizing isolates and 70% of the neonatal disease strains (34). In France, serotype III accounted for 83.9% of infant meningitis cases caused by GBS infection (35). The temporal dynamics of this serotype were reported for adult invasive infections in Iceland, where the only significant decline in serotype prevalence over the 39-year period was for CPS III (36). This observation contrasts with those for other populations, such as neonates, in which it is attributed to the majority of invasive disease cases (25).

IV.

Serotype IV is less prevalent globally than the other main serotypes (Ia, Ib, II, III, and V). Recent studies, however, have noticed the emergence of this serotype, particularly in Canada and the United States (37–39). These studies also reported evidence of capsular switching relevant to this serotype, a concern for targeted therapies. The presence of serotype IV within a predominantly CPS III cluster of isolates from France was suggested to have occurred through capsular switching (40), and capsular switching to serotype IV within clonal complex 17 was also reported in Kenya (34), which further supports this as a cause for concern.

V.

Early studies reporting the emergence of CPS V noted a close genetic relationship among the strains of this serotype and its recovery from all populations, but more so from nonpregnant adults (41). It is recognized as the third most prevalent serotype associated with maternal disease in the United States, United Kingdom, and France (26). Africa, more specifically West Africa, has a higher frequency of CPS V maternal colonization than those of other regions, such as North America, Australia, and New Zealand, which have much lower prevalences (24). In Gabon, approximately one-third of strains isolated from pregnant women were CPS V (42). This supports previous observations of CPS V predominance by other studies of Gambian mothers and their infants and of pregnant women in Egypt (43, 44).

VI.

Considered one of the rarer serotypes (with VII, VIII, and IX), CPS VI is often reported in Asian countries, in particular Japan (45) and Malaysia (18, 46). While global colonization estimates place CPS VI in the minority, this serotype is thought to account for as much as 20% of serotypes observed in Southeast Asian pregnant women (24).

VII.

Similarly, CPS VII is less well represented across the globe (24). Interestingly, it was found to be the most common serotype in southern Ghana, in the first study to report such serotype prevalence in pregnant women from West Africa (47). A number of Malaysian studies observed serotype VII isolates in their cohorts, ranging from 1.7% to 21.4% of serotypes (18, 46, 48). This supports the observation of intracountry variation that was similarly reported (although not for CPS VII) for serotypes in different regions of Brazil (19). A study of maternal colonization in India observed CPS VII accounting for 6.7% of the serotypes and having a higher prevalence than that of CPS Ib, which is indicative of Asian countries yet shows a unique distribution (49). More recently, Islam et al. observed that 37.1% of the GBS specimens isolated in Bangladesh from newborn umbilical cord areas (colonizing isolates) belonged to CPS VII (50).

VIII.

Serotype VIII is strongly associated with Japan, with an early study reporting a high frequency in pregnant women (45). That study found a 35.6% prevalence of CPS VIII among serotypes, followed by CPS VI (also considered rare globally), among colonizing isolates from Japanese pregnant women (45). Thereafter, incidences of its presence began to be described outside Japan. In Denmark, seven instances of invasive disease isolates belonging to CPS VIII among nonpregnant adults were reported from 1999 to 2002 (51). Also, one isolate colonizing a pregnant woman in Maryland was recognized as CPS VIII (52). Lately, the frequency of CPS VIII is generally low outside Japan, and studies often describe few isolates (53, 54).

IX.

The newest serotype, described in 2007, has had fewer chances for description in the literature; however, in the last decade, studies have adopted new techniques, and now most test for all serotypes. As mentioned above, a study in southern Ghana observed CPS IX as having the second highest prevalence, behind CPS VII (47). A Swedish study assessed changes in serotype distribution over time for invasive disease (neonatal and adult populations) and suggested that circulating serotypes remained stable from 1990 onwards; however, the study noted an increase in CPS VII to IX (55). Additionally, a case study in Japan described an incidence of ultra-late-onset sepsis caused by CPS IX, and this demonstrates its capacity to play a role in disease (56).

Nontypeable isolates.

Numerous serotyping studies result in isolates that do not represent any of the defined types and are thus considered nontypeable (NT). It is not known whether this is a result of technique specificity or whether such isolates may represent novel serotypes; either way, NT isolates may potentially skew the available data. Capsule loss and lack of expression are both potential contributors to this NT category (57, 58). One study assessed a collection of GBS isolates classified as NT by serological methods and went on to characterize them by a number of different techniques (immunodiffusion, pulsed-field gel electrophoresis [PFGE], and multilocus sequence typing) (59). This resulted in all isolates being serotyped and found to represent all types, excluding IV and VII to IX, with instances of multiple serotypes per isolate, which has been described previously (52). This may represent an incidence of capsular switching, a phenomenon that involves the genetic exchange of capsular genes and thus enables expression of alternative capsule types, and it is proposed that such horizontal transfer may be responsible for the capsular locus diversity (60–62). In a former study by Ramaswamy and colleagues, 80% of isolates containing multiple serotypes contained a type V CPS-specific gene, and the authors proposed that this serotype may have an association with increased competency and ability to acquire DNA (59). This is an interesting notion, but given that GBS is not thought to be naturally transformable, the idea that capsular switching may result from increased competence of CPS V seems unlikely; however, capsular switching has been reported by others (37, 40, 63, 64). This is a concern for development of vaccines targeting the capsule, as this has been an issue for the naturally transformable organism Streptococcus pneumoniae due to capsular switching (65, 66).

One of the major limiting factors in discussing GBS serotype distribution is that a direct comparison between studies can be difficult due to the number of different study parameters assessed, such as specimen type and study population, and the added complexity of various serotyping methodologies. For optimal surveillance outcomes, data on all types of patient cohorts are vital to completing the broader picture of GBS epidemiology. Increasing our understanding of prevalent isolates associated with specific body sites and their potential to cause disease by monitoring the serotypes is the first step toward narrowing down invasive strain targets and surveying the serotype distribution dynamics.

Surface Proteins

In addition to the CPS, proteins are also present on the surfaces of GBS cells, and studies have shown that antibodies against both CPS and surface proteins are protective against GBS infection (67–69). The alpha-like protein (Alp) family proteins are among the well-characterized GBS surface proteins and are significant virulence factors. These include alpha-C protein, Rib, Alp2, Alp3, Alp4, and epsilon (Alp1) and are encoded by the bca, rib, alp2, alp3, alp4, and epsilon/alp1 genes, respectively (70). Detection of the expressed surface proteins by serology or the presence of their genes by molecular techniques, such as PCR or sequencing, can be used as another grouping tool. The C antigen was the first identified surface protein antigen and has been found in numerous GBS strains, except for serotype III strains (71). Alps have been present in the majority of GBS isolates, but non-Alp isolates do exist. A total of 1.1% of isolates in one study were non-Alp isolates, and all were associated with invasive bloodstream infections (72). Smith and others examined associations between the common serotypes (Ia, Ib, II, III, and V) and specific genes. They found that most genes tested (sbp1, bca, bac, rib, brp, pag, and psp) were common in only one or two serotypes, except for bca, which was present in >50% of all serotype Ib, II, and V strains (73). A review of several surface proteins reported correlations for alpha protein presence in CPS Ia, Ib, and II strains, yet it was found almost never in CPS III and only rarely in CPS V strains. Rib is expressed by the large majority of CPS III, many CPS II, and a few CPS V strains. Alp3 is expressed by CPS V and VIII strains and Alp2, more rarely, by CPS Ia, III, and V strains (70).

Pulsed-Field Gel Electrophoresis

Pulsed-field gel electrophoresis (PFGE) is another molecular method that distinguishes isolates with greater discrimination than that of serotyping for evaluation of genetic relatedness. It involves nucleic acid digestion with one or more restriction enzymes (often SmaI in the case of GBS) to produce fragment profiles on an agarose gel, after which the profiles are then compared for similarity and grouped accordingly (74). Fasola and colleagues described this method for use in GBS in 1993, and Benson and Ferrieri later reported an updated and more rapid PFGE technique for GBS, reducing the workflow from 6 to 8 days to 3 days total (75, 76). PFGE profile correlations with serotypes have shown variation; Skjaervold et al. observed homogeneity for Ib, III, and V strains, while IV, Ia, and II strains were heterogenous (77). This suggests that serotype grouping does not permit an understanding of close genetic relationships. While a greater depth of genetic relationship can be observed using PFGE than using serological typing, there are limitations with this method. Due to the lack of a uniform typing system with respect to naming, PFGE analyses can be difficult to compare between studies: what may be PFGE clone A in one study could be clone 37 in another (78, 79). Therefore, pattern and band sizes must be used for comparison between studies. In this respect, it is mainly useful for comparison within a study to see diversity among isolates and modes of transmission. For example, mother and infant paired GBS specimen fragment profiles were observed to have common patterns, highlighting vertical transmission (80). In the past, identifying fragment banding on gels was subjective and lacked accuracy in the reporting of band sizes, making comparison across studies difficult. The idea of using restriction enzymes with minimal cutting sites aims to reduce this by producing fewer fragments for visualization on the gel (74). Regardless of any limitations, PFGE has identified variation within and across serotypes and has led to the progression of the molecular epidemiology of GBS.

Multilocus Sequence Typing

Jones and colleagues developed the multilocus sequence typing (MLST) method for GBS by using seven defined housekeeping genes (adhP, atr, glcK, glnA, pheS, sdhA, and tkt) (81). The MLST database was produced by Jolley and colleagues and currently includes over 4,000 GBS isolates (82). Allelic variations among housekeeping genes allow the determination of different sequence types (STs). These STs can then be clustered into clonal complexes (CCs), which are conservatively defined by members having no more than one allelic difference from other members of the cluster.

The MLST database is a valuable resource that enables sharing of data; the publicly available information for GBS includes a range of criteria to assess, including the country of origin, serotype, and specimen type. Although there remains a reporting bias, in that only submitted information is available, the data represent a mix of global information which is very useful. An audit of the GBS MLST database (https://pubmlst.org/sagalactiae/) found a total of 4,131 isolates from 38 different countries. Of these, 3,949 isolates had been assigned an ST (n = 635). Twelve STs account for 73% of the GBS isolates with available MLST data (Fig. 1). The remaining STs (27%) each account for less than 1% of all isolates, and 525 STs have a single isolate submitted to the database.

FIG 1.

Frequencies of the 12 most prevalent GBS sequence types available from the MLST database, which includes a total of 635 sequence types (remaining STs account for <1% each).

Having access to the location of each isolate's origin enabled categorization of STs by continent (Fig. 2). This highlighted gaps in the global data, with no isolates from Australia and New Zealand available, unless they are not stated (categorized as “unknown”). Examining the main contributors within each continent, Netherlands (n = 1,230/1,610 isolates), Kenya (n = 1,133/1,172 isolates), Japan (n = 490/550 isolates), United States (n = 325/491 isolates), and Brazil (n = 3/5 isolates) have submitted the most isolates within Europe, Africa, Asia, North America, and South America, respectively.

FIG 2.

Contributions by continent of GBS isolates available in the MLST database, as percentages of all isolates (South America represents 0.1% and is therefore not visible).

Associations between ST and serotype have been reported in the literature, with some showing a strong correlation and others very little; this is also observed in the MLST database (Fig. 3). Ramaswamy et al. observed a correlation between CPS III and ST-17 and between CPS Ib and ST-12, and they found that CPS V was represented in all STs except for ST-17 (59). Similar trends are observed within the MLST database, where each of the most prevalent STs appears to be dominated by a particular serotype (Fig. 3). The serotype variability in some instances, for example, ST-1, may be suggestive of discordance with serotype profiling. Trends observed by various studies reveal a variable serotype presence within ST-1, variation between CPS Ib and II within ST-10, and a predominance of CPS III within ST-17 and ST-19 and of CPS Ia within ST-23 (83, 84). Within the clonal complexes, homogeneity of serotypes is evident in the CC-17 group, with all isolates representing CPS III, except for a small proportion with unspecified serotypes. Similarly, within CC-19, ST-182 includes CPS III only, but other STs in this complex show variation, especially ST-28, which is predominantly CPS V (Fig. 3). From an evolutionary perspective, caution must be taken when inferring phylogenetic relationships, as suggested by Sorensen et al., due to the indication that different segments of the GBS genome evolved independently (85). With this in mind, MLST is perhaps not the best method for describing evolutionary relationships, but for the purposes of epidemiology at the current scale, it should be acceptable now that we are aware of these factors.

FIG 3.

Serotype distribution within each sequence type, using the public MLST database for GBS, for the 12 most represented sequence types within the database, as well as their corresponding clonal complexes (CCs), where appropriate.

Recurring themes in the literature include the numerous studies that have described the association between ST-17 and invasive neonatal disease (83, 86–90) and, further, the relationship between this lineage and meningitis (91). Neonatal invasive isolates collected over a 10-year period in Sweden showed variation in genotype prevalence over time and in different regions, in particular the emergence of CC-1 and subsequent disappearance of CC-23 (88). These temporal changes emphasize the importance of understanding GBS epidemiology and tracking changes in prevalence worldwide: ST-17 may expand and/or other STs may become hypervirulent over time.

The MLST profiling scheme is publicly available and curated, making it an invaluable epidemiology resource. If the database were treated like a repository, the epidemiological surveillance would be greatly enhanced, which would be desirable for targeted therapeutics assessment.

Random Amplified Polymorphic DNA and Repetitive-Sequence-Based PCRs

Random amplified polymorphic DNA PCR (RAPD-PCR) and repetitive-sequence-based PCR (rep-PCR) are techniques that rely on the amplification of DNA via random primers (RAPD-PCR) (92) or repetitive sequence target primers (rep-PCR) and result in genomic fingerprinting profiles, as the generated amplicons lead to strain-specific band patterns on an electrophoretic gel.

The RAPD assay has been used in a number of genotyping contexts, such as evolution and genetic relatedness (93), epidemiology (46, 94), transmission (95), and isolate variation within individual patients (96–98). One study examined RAPD profiles of GBS isolates from a number of sites, such as blood, cerebrospinal fluid, and maternal breast milk, in cases of LOD in neonates (n = 4) (96). It found identical banding profiles for all sample types from individual patients and differing types between patients, but this was representative of a small sample size. Likewise, Brzychczy-Wloch et al. used RAPD-PCR to suggest that GBS isolates from colonized neonates originated from the colonized mothers, regardless of mode of delivery, as they had identical genotypic profiles (99). While detection of identical clones via RAPD profiles can suggest vertical transmission, other studies have used this to demonstrate horizontal transmission in neonates born to noncolonized mothers that likely acquired GBS from the hospital environment, as identical profiles were observed in other hospital wards (100).

Toresani et al. found 16 profiles for 21 GBS isolates tested by RAPD assay, and furthermore, successive samples from two patients showed different genotype profiles, suggesting reinfection (97). In contrast, El Aila et al. compared culture methods and genotyped multiple isolates collected from individual patients. RAPD testing of 118 isolates from 32 pregnant women resulted in 66 genotype profiles, including cases of multiple different genotypes at each site (vagina and rectum), demonstrating that assumed reinfection may be due to a lack of initial identification (98). This highlights the use of RAPD assay for single-center studies and its integration with high-resolution capillary electrophoresis, which enhances accuracy and provides digital fingerprints (98).

Numerous studies have observed correlations between RAPD profiles and clustering of strains with resistant phenotypes (101, 102). The results from a study of macrolide-resistant GBS isolates suggested that PFGE was a more precise tool than RAPD-PCR for analyzing molecular epidemiology (95). In contrast, Zhang and colleagues (103) found that the clustering of the isolates by RAPD-PCR was in concordance with previous PFGE findings suggesting that clusters could be serotype specific (80).

The great advantages of RAPD assay are a relatively short procedure time and a simple methodology, which make it a suitable tool for suggesting clonal relatedness in clinical settings and for monitoring transmission. Comparison between studies can be difficult, however, particularly if different primers are used. The random nature of this profiling technique provides variation and enables discrimination between clones, but similar to the PFGE technique, due to the lack of specific targets, the information this provides beyond strain similarity is somewhat limited unless further genetic analysis is undertaken.

The literature assessing rep-PCR for GBS is limited, with one study comparing the rep-PCR Diversilab system with MLST and PFGE by using clinical adult and neonatal GBS isolates (n = 179) (104). The clustering concordances of PFGE and rep-PCR were similarly low, but PFGE had a greater discriminatory power than that of rep-PCR (104). One study compared both RAPD-PCR and rep-PCR techniques to genotype GBS isolates from fish and observed RAPD-PCR to have a greater level of variability than that of rep-PCR between strains, with 13 and 9 profiles, respectively (105). This is not supportive of the use of rep-PCR over other methods tested, including MLST and PFGE. rep-PCR typing is notorious for its susceptibility to minor variations in experimental conditions and reagents, resulting in poor reproducibility (106, 107).

Multilocus Variant-Number Tandem-Repeat Assay

Various bacterial species have been typed through the detection of variable-number tandem repeats (VNTRs), which is based on the repeated sequences that make up DNA at different loci and that are found throughout bacterial genomes (108). As the name suggests, these VNTR loci vary in repeat number, allowing for an assay of a combination of several VNTR loci to generate different strain-specific profiles. This is known as a multilocus variant-number tandem-repeat analysis (MLVA) and has a principle similar to that of MLST; however, rather than housekeeping genes, MLVA uses VNTR loci (109). This typing scheme was developed for GBS by Radtke and colleagues and includes five of the most diverse loci (SATR1 [110], SATR2, SATR3 [111], SATR4 [112], and SATR5 [113]) (109). The authors found clustering by CPS/surface protein typing or MLST to generally correspond with the MLVA profiles, but the latter was more discriminatory. A total of 126 GBS isolates tested were profiled as having 70 different MLVA types (MTs), compared to 36 STs by MLST and 19 CPS/protein types. A specific number of 15 repeats within SATR2 was observed to correspond with serotype IX. Additionally, for one of the three described SATR3 alleles, all isolates of that allele (except one) belonged to ST-17, known to account for a large proportion of neonatal infections (81, 88). Other MLST groups were more dispersed, including CPS V/ST-1, for which SATR5 resolved into a number of profiles. This is interesting due to the often homogeneous reports of this subgroup in other studies (28, 77). The SATR1 locus is considered to be composed of clustered regularly interspaced short palindromic repeats (CRISPR), and this worked efficiently as part of the MLVA scheme.

Additional MLVA schemes have been reported, one of which used six VNTR loci, including three described by Radtke et al. (109, 114). Similar results were observed in that study, with greater isolate discrimination reported with the MLVA technique (98 types) than with MLST (51 types). The clustering patterns observed for MLVA types showed similarity to MLST CCs; in particular, all human strains of MLST CC-17 appeared in MLVA cluster 9 (114). This scheme was able to categorize MLST CC-23 into two clusters, one accounting for CPS III and the other for CPS Ia, which highlights this scheme's ability to distinguish strain variability within lineages (114).

A study of commensal GBS colonization of nonpregnant women of reproductive age utilized MLVA for genotyping and observed 15 types among the 86 GBS isolates (115). For pregnant women, 30 MTs from a total of 41 GBS isolates were observed, which suggested a highly diverse population structure (116). The MTs are difficult to compare between studies, though, as there is no mention of a database in which naming conventions are uniformly enforced; however, a database has been created for this purpose (http://www.mlva.net) but unfortunately does not include GBS.

CRISPR Locus 1

Lastly, the molecular target which was mentioned above as one of the loci of the MLVA scheme is CRISPR as well as Cas (CRISPR-associated proteins). These are known for their role in bacterial host defense and adaptive immunity against foreign invading DNA, such as bacteriophages and mobile genetic elements (117). The genetic arrangement that this system follows results in a polarized orientation, which allows tracking of ancestral invasions over time (118). In GBS, two CRISPR-Cas systems have been identified, including a type II-A system (ubiquitous and functional) and a type I-C system (rare and often incomplete), associated with CRIPSR locus 1 (CRISPR1) and CRISPR2, respectively (119, 120). The ubiquitous nature of CRISPR1 makes it an attractive molecular marker with which to track changes in the genome. The highly conserved direct repeats form the spacers between introduced genetic materials and can therefore be monitored for changes. Using this method, Beauruelle and others were able to monitor the vaginal GBS colonization of 100 women over an 11-year period (121). This extensive study highlights the influence of genetic variables that alter these organisms over time, for example, bacteriophage infection. This is an example of new epidemiological markers that may assist in understanding these bacteria, their colonization, and their pathogenic tendencies.

FROM TARGETS TO TREATMENT

The purpose of the targets and molecular assays presented above is to understand the population of GBS with respect to colonization and disease; by distinguishing differences between strains, we can begin to associate specific features with clinical outcomes. Equally, by identifying similarities that are represented across the majority or the entire population of GBS strains, we can target these bacteria more broadly. Within antenatal settings, the current preventative measures for GBS include antenatal GBS screening (risk or culture based) and subsequent intrapartum antibiotic prophylaxis prior to delivery. This has resulted in reductions in EOD rates as reported by the Centers for Disease Control and Prevention and is also recommended in other countries (10, 11, 122, 123). Unfortunately, no change in LOD has been observed with such preventative strategies; however, this is likely due to the nature of transmission (nosocomial or community acquired) (124).

Due to the recommended IAP, antibiotic exposure has become widespread, but the potential for resulting adverse events is unclear, as highlighted in a recent systematic review (125). While consistency is lacking for data on development of antimicrobial resistance and long-term adverse events as a result of IAP for GBS, infant microbiome disruption has been observed by numerous studies (126–132). However, due to the lack of high-level (species) taxonomic data and the myriad other methodological problems that are still rampant in microbiome studies (lack of negative controls and proper selection of primer sets) (133), with the current data it remains unclear what impact this may have on neonates. With the potential implications in mind, however, it is important to examine alternative preventative and therapeutic strategies.

Maternal GBS vaccination has been proposed as an alternative preventative strategy due to the ability of maternal IgG to cross the placenta and provide immunity for the fetus (134). Baker and Kasper identified an association between deficient maternal antibody levels and the occurrence of neonatal GBS infection. That study also detected GBS antibody titers in the umbilical cord sera of three healthy neonates born to mothers with demonstrable antibody levels (60 to 80% of the maternal levels were detected), suggesting transplacental transfer and illustrating the potential for a maternal vaccine (135). The benefit of maternal vaccination has since been described, and a number of GBS vaccines have been developed (134, 136–138).

While no vaccines against GBS are currently licensed and available, several polysaccharide conjugate and surface protein vaccines are under trial (137, 138). Through comparison of the vaccine targets with the global GBS epidemiology, we can identify areas of vaccine coverage. Numerous studies that have identified the less commonly recognized serotypes within their populations have raised concern about vaccinations targeting specific serotypes, such as Ia, Ib, and III, all of which are included in the trivalent vaccine currently furthest along in progress, with completion of a phase Ib/2 clinical trial (ClinicalTrials.gov number NCT01193920) (139, 140). A review by Lin et al. describes such challenges faced by capsule-based conjugate vaccines against GBS, in addition to dealing with the potential for serotype replacement and capsular switching (138). More recently, a pentavalent vaccine targeting serotypes Ia, Ib, II, III, and V commenced a phase I trial on healthy volunteers (ClinicalTrials.gov number NCT03170609); if successful, this may be an effective GBS vaccine outside Asia, as it covers the five major circulating serotypes in the United States, Europe, and Australia (24).

Alternative vaccination strategies have focused on proteins. Surface proteins have been included as vaccine targets in addition to capsule types, and given the variety of circulating serotypes described, these may provide a broader, more universal target to exploit, which would be ideal for vaccine development (137). Although few surface proteins are conserved across all strains, Maione and colleagues provided evidence of protein vaccination in female mice with the ability to elicit protection in 43 to 80% of cases of GBS challenge to pups born to the vaccinated mother (141). Four proteins, including a conserved Sip protein (SAG0032) encoded within the core genome of strains, were included in this study, in addition to three other antigens (SAG1404, SAG0645, and SAG0649) which had a variable presence in the genome (141). Studies in humans have been described by Heath (137) and include multiple studies of monovalent, one bivalent, and several trivalent CPS-based vaccines, all in the proof-of-concept stage involving trials in pregnant women. A bivalent vaccine including the N-terminal domains of the surface proteins alpha-C and Rib is currently in development by MinervaX and has completed phase I clinical trials (ClinicalTrials.gov number NCT02459262). Larsson et al. reported the association between low protein alpha and Rib antibodies and the occurrence of neonatal invasive infection (142). While the type of vaccine is important in terms of coverage, the implementation is equally important in considering access, especially in low-income settings and developing countries, where screening and widespread prophylaxis are often not practical (143). Based on the World Health Organization guidelines, GBS vaccination, particularly in South Africa, would be very cost-effective (144). In developed countries, such as the United States, Kim and others evaluated the cost-effectiveness of a maternal vaccine compared to screening/IAP (145). They concluded that ≥70% vaccine efficacy in addition to IAP for unvaccinated women would prevent more disease than current strategies, with a similar economic burden (145). Epidemiological data about GBS are invaluable not only for vaccine development and success but also for other future targeted therapeutic agents, such as bacteriophage therapy (146).

CONCLUSIONS

This review describes different commonly used and emerging techniques to define GBS epidemiology and phylogeny. Some traditional targets, such as serotypes, are widely reported but lack the level of definition that other molecular targets can provide with respect to genetic relatedness and pathogenic features. The difficulty lies in recognizing these targets and widely reporting them to provide a comparable basis for global surveillance. It is likely that future epidemiological efforts to monitor GBS will include the reporting of a combination of these targets and will in turn refine the association between these different components. The epidemiology of GBS provides insight into the adaptation and evolution of these organisms through examination of their genetic relatedness and population structure. Understanding GBS as an opportunistic pathogen is one aspect to arise from the proposed surveillance efforts; another is guidance for development of future therapeutics. Vaccines based on serotypes commonly found in one area may not include prevalent serotypes of another due to the geographic variation observed even at the regional level. To vet these coverage issues, surveillance of different targets within GBS is required to monitor prevalent isolates in all populations. Additionally, following vaccine implementation, this will provide a platform to monitor shifts in GBS dynamics and to identify emerging pathogen types before outbreaks occur. Broader coverage may result from protein vaccine development, which further emphasizes the importance of characterization of a range of targets. The targets outlined here each have individual relevance, although uniformity in reporting profiles would be ideal for global comparisons. Currently, MLST, while less discriminatory than MLVA, comprises the only method with an established database to enable this comparison and provide the added confidence of curation.

Biographies

Lucy L. Furfaro is currently completing her Ph.D. at the University of Western Australia and previously completed her B.S. and Honors programs at the University of Western Australia. Her research interests include perinatal microbiology, with a focus on group B streptococcus and treatment alternatives to antibiotics, such as bacteriophage therapy. This area of research has been pursued over the last 4 years as part of Honors and postgraduate studies.

Barbara J. Chang received her Ph.D. from Monash University, Australia, followed by postdoctoral positions at the University of Calgary, Canada. Currently, she is the Director of Postgraduate Programs in Infectious Diseases at the School of Biomedical Sciences at the University of Western Australia. Her research interests include the biology and genetics of bacteriophages, anti-quorum-sensing agents, and bacterial pathogens, including Clostridium difficile, Moraxella catarrhalis, and group B streptococcus.

Matthew S. Payne was awarded his Ph.D. in molecular microbiology from the University of Queensland, Australia, and has since conducted postdoctoral research at Kings College London, United Kingdom, and La Trobe University, Australia. Dr. Payne is currently at The University of Western Australia, where he is focused on preventing infection-mediated preterm birth. His research interests include the microbiology of preterm birth, microbial metagenomics, Ureaplasma and Mycoplasma spp., and antibiotic and nonantibiotic treatment regimens for prevention of preterm birth and neonatal sepsis.