Multistate, Population-Based Distributions of Candidate Vaccine Targets, Clonal Complexes, and Resistance Features of Invasive Group B Streptococci Within the United States, 2015–2017

Abstract

Background

Group B Streptococcus (GBS) is a leading cause of neonatal sepsis and meningitis and an important cause of invasive infections in pregnant and nonpregnant adults. Vaccines targeting capsule polysaccharides and common proteins are under development.

Methods

Using whole genome sequencing, a validated bioinformatics pipeline, and targeted antimicrobial susceptibility testing, we characterized 6340 invasive GBS isolates recovered during 2015–2017 through population-based Active Bacterial Core surveillance (ABCs) in 8 states.

Results

Six serotypes accounted for 98.4% of isolates (21.8% Ia, 17.6% V, 17.1% II, 15.6% III, 14.5% Ib, 11.8% IV). Most (94.2%) isolates were in 11 clonal complexes (CCs) comprised of multilocus sequence types identical or closely related to sequence types 1, 8, 12, 17, 19, 22, 23, 28, 88, 452, and 459. Fifty-four isolates (0.87%) had point mutations within pbp2x associated with nonsusceptibility to 1 or more β-lactam antibiotics. Genes conferring resistance to macrolides and/or lincosamides were found in 56% of isolates; 85.2% of isolates had tetracycline resistance genes. Two isolates carrying vanG were vancomycin nonsusceptible (minimum inhibitory concentration = 2 µg/mL). Nearly all isolates possessed capsule genes, 1–2 of the 3 main pilus gene clusters, and 1 of 4 homologous alpha/Rib family determinants. Presence of the hvgA virulence gene was primarily restricted to serotype III/CC17 isolates (465 isolates), but 8 exceptions (7 IV/CC452 and 1 IV/CC17) were observed.

Conclusions

This first comprehensive, population-based quantitation of strain features in the United States suggests that current vaccine candidates should have good coverage. The β-lactams remain appropriate for first-line treatment and prophylaxis, but emergence of nonsusceptibility warrants ongoing monitoring.

capsular serotypes, surface protein distributions, resistance determinants, β-lactam nonsusceptibility

(See the Editorial Commentary by Humphries on pages 1014–5.)

Lancefield group B Streptococcus (GBS) (Streptococcus agalactiae) is a major cause of sepsis and meningitis in neonates and infants [1, 2]. GBS is an increasing cause of invasive infections in adults, especially in patients aged > 65 years and those with underlying medical conditions [3]. Intrapartum antibiotic prophylaxis (IAP) to women at risk for GBS infection has dramatically lowered early-onset disease (EOD; < 7 days of age) [4], but not late-onset disease (LOD; 7–89 days of age) [5].

The β-lactam antibiotics are recommended as first-line therapy against GBS infections and for IAP because GBS is considered susceptible to these agents. However, GBS isolates nonsusceptible to β-lactams have been detected from invasive and noninvasive infections [6–9]. In the case of penicillin allergy, IAP guidelines suggest usage of cefazolin or, for women at high risk of anaphylaxis to a β-lactam antibiotic, clindamycin (if the GBS is susceptible) or vancomycin [10]. Erythromycin and clindamycin-resistant GBS is increasing in the United States (US) and elsewhere [3, 5, 11].

Concerns about emerging antimicrobial resistance and the potential for microbiome dysbiosis resulting from IAP [12], as well as a need for prevention strategies, have motivated efforts to develop vaccines against GBS [1, 3].

Ten capsular polysaccharide serotypes (Ia, Ib, II–IX) are known [13]. While GBS of all serotypes can cause invasive infections, 6 serotypes (Ia, Ib, II, III, IV, and V) account for most disease [1, 3, 14–16]. A 6-valent (Ia, Ib, II, III, IV, and V) conjugate vaccine is currently being tested [17, 18]. A number of promising protein vaccines are also under development, targeting surface adhesins and virulence factors [19–22].

Here, we use whole genome sequencing (WGS) to describe a large population-based sampling of invasive GBS (iGBS) collected in the US from 2015 to 2017. We quantitate distributions of potential vaccine targets (capsule polysaccharides, surface proteins) and their associations with major antibiotic resistance markers within clonal lineages.

MATERIALS AND METHODS

Invasive GBS Isolates

Active Bacterial Core surveillance (ABCs) conducts active, population- and laboratory-based surveillance for iGBS disease within 10 states (all or select counties) across the US, with 8 submitting isolates for characterization (see https://www.cdc.gov/abcs/methodology/surv-pop.html [23]). All available GBS isolates (n = 6340 from 7114 cases [89.1%]) during 2015–2017 were characterized. Isolates were from individuals of all ages, except for isolates from New York, which were limited to iGBS in infants < 90 days old.

WGS and Bioinformatics Pipeline

Chromosomal extraction, genomic libraries, and WGS were performed as previously described [9]. Genomes were sequenced on the Illumina MiSeq platform to produce 250-bp paired-end reads with average coverage depth of 60×. Following removal of adaptors, short reads were assembled de novo using VelvetOptimiser [24]. Isolate identifiers, features, genome accession numbers, and assembly metrics are listed in Supplementary Table 1. Serotypes, multilocus sequence typing (MLST), antibiotic resistance determinants, and predicted minimum inhibitory concentrations (MICs) were determined using a previously validated GBS bioinformatics pipeline [9]. The ARG-ANNOT and ResFinder databases were incorporated to detect additional resistance determinants [25, 26]. Conventional broth microdilution results to 14 antibiotics and inducible clindamycin resistance were available for all year 2015 isolates (n = 2068) [9] and years 2016–2017 isolates recovered from Minnesota (n = 1027) and were used to validate predicted resistance. Additional strain features extracted from the genomic data included presence/absence of surface protein genes encoding the hypervirulent GBS adhesin (hvga) [27], serine-rich repeat (srr) proteins [28], alpha protein family (alpha, Rib, Alp2/3, Alp1) [19], and pilus islands (PI1, PI2A, and PI2B) [29] (https://github.com/BenJamesMetcalf) (Supplementary Table 2). For each sequence query, ≥ 95% identity was required for a positive result.

MLST Data

MLST relied upon SRST2 and the database at http://pubmlst.org/sagalactiae/. Sequence types (STs) were grouped via eBURST [30] into clonal complexes (CCs) whose members shared at least 5 of 7 MLST loci; otherwise, an ST was considered an outlier. The relationships between STs and different isolate parameters were illustrated by minimum spanning tree (PHYLOViZ software version 2.0; PHYLOViZ team, Lisbon, Portugal).

RESULTS

Isolates

Among 6340 iGBS strains identified over the 3-year period, 5778 (91.1%) were from adults (≥ 18 years), with 56 isolates from individuals aged 90 days–17 years (Supplementary Figure 1) and 47 from pregnant women. There were 506 isolates from infants < 3 months of age, with 232 (3.7%) and 274 (4.3%) from EOD and LOD, respectively (Figure 1A). Most (86.8%) iGBS isolates were recovered from blood; 72 (1.1%) were from cerebrospinal fluid and the remaining 12.1% from other body fluids (data not shown). The largest number of isolates were from Maryland (1692 [26.7%]) and Minnesota (1508 [23.8%]) (Supplementary Figure 2).

Figure 1.

A, Distribution of invasive group B Streptococcus (GBS) serotypes among isolates recovered from young infants (early-onset disease [EOD] and late-onset disease [LOD]) and from all other ages. B, Distribution of invasive GBS pilus backbone protein types among isolates recovered from young infants (EOD and LOD) and from all other ages. C, Distribution of invasive GBS alpha family protein types among isolates recovered from young infants (EOD and LOD) and from all other ages. Abbreviations: EOD, early-onset disease; LOD, late-onset disease.

Vaccine Candidate Distributions

Serotypes were predicted for all 10 capsular types with few nontypeable isolates (13/6340 [0.2%]). Overall predicted serotype distribution was similar when comparing serotypes across each of the 8 states (Supplementary Figure 2). Serotypes Ia (1384 [21.8%]), V (1116 [17.6%]), II (1082 [17.1%]), III (987 [15.6%]), Ib (919 [14.5%]), and IV (748 [11.8%]) accounted for 98.4% of the isolates (Figure 1A and Supplementary Table 3). Serotypes VI, VII, VIII, and IX were rare (< 2% of isolates). The predominant serotype for both EOD (63/232 [27.2%]) and LOD (189/274 [69.0%]) isolates was serotype III; serotype III accounted for 12.6% of the 5834 remaining isolates (Figure 1A). Serotypes Ia, Ib, and III represented 138 of 232 (59.5%) EOD isolates, 242 of 274 (88.3%) LOD isolates, and 46% of remaining isolates (Figure 1A). Five serotypes (Ia, Ib, II, III, and V) accounted for 203 of 232 (87.5%) EOD isolates, 257 of 274 (93.8%) LOD isolates, and 5028 of 5834 (86.2%) remaining isolates. Serotype IV accounted for 44 of 506 (8.7%) combined isolates from young infants (27/232 [11.6%] of EOD and 17/274 [6.2%] of LOD). There were 2 serotype VI EOD isolates (not shown). In summary, the top 5 serotypes represented 91.3% (462/506) of cases in young infants, with the addition of serotype IV increasing this proportion to 99.6% (504/506).

At least 1 of the 3 distinct pilus backbone determinants were detected in almost all (98.5%) GBS isolates (Figure 1B and Supplementary Table 3). The most frequently identified “pilus type” was the combination PI-1 and PI-2a (n = 3691 [58.3%]) followed by PI-2a alone (n = 1953 [30.8%]), PI-1 + PI-2b (n = 469 [7.4%]), PI-2b (n = 203 [3.2%]), and PI-1 (n = 16 [0.3%]). Distributions of the pilus protein types among neonatal disease (EOD and LOD) isolates varied markedly from those from older ages, due to the high proportion of isolates from young infants that were positive for both PI-1 and PI-2b (186/506 [36.8%]), nearly all of which (183/186) were serotype III.

Most (n = 6289 [99.2%]) GBS isolates shared a 522-bp sequence with ≥ 95% sequence identity to 1 of 4 homologous alpha protein family gene queries (alpha, rib, alp2/3, alp1) (Supplementary Table 3) that share 68%–75% sequence identity between them (not shown). The alpha and rib queries corresponded to the 174 residue N-terminal domains used for the GBS-NN vaccine [20]. The frequencies of the alpha family protein determinants were similar: alp1 (28.1%), alpha (26.3%), alp2/3 (23.4%), and rib (22.2%) (Figure 1C); however, rib was much more common among isolates from young infants. Of the 1399 rib-positive isolates, 937 (67%) isolates were of serotype III. These rib-positive isolates included 277 of the 506 (54.7%) isolates from young infants (87 EOD, 190 LOD), of which 247 (89.2%) were serotype III. Most (1105/1766 [62.3%]) alp1-positive isolates were of serotype Ia and 864 of 1473 (58.7%) alp2/3-positive isolates were serotype V (Supplementary Table 3).

Distributions of Serotypes and Protein Vaccine Candidate Distributions Among Clonal Complexes

MLST data from 6336 GBS isolates was available and included 389 STs (Supplementary Table 4) that were grouped into 11 major CCs: CC1 (22%), CC23 (18.2%), CC19 (11%), CC459 (9.4%), CC22 (8.7%), CC17 (7.3%), CC8 (6.4%), CC12 (5.3%), CC28 (3.3%), CC88 (2.6%), and CC452 (2.3%) (Table 1 and Figure 2A and 2B). Each CC was represented by a dominant serotype: CC585/41 (V; 100%), CC17 (III; 99.8%), CC459 (IV; 99.7%), CC22 (II; 99.6%), CC28 (II; 99.5%); CC8 (Ib; 99%), CC452 (IV; 98.6%), CC23 (Ia; 98%), CC26/1087 (V; 96.7%), CC88 (Ia; 92.2%), CC12 (Ib; 69.3%), CC328 (V; 77.8%), CC19 (III; 68.8%), and CC1 (V; 62.5%) (Table 1 and Figure 3A and 3B). Twenty-eight STs (Supplementary Table 5) were represented by multiple serotypes.

Table 1.

Sixteen Clonal Complexes and Their Serotype Distributions Among the Study Isolates (2015–2017)^a

Clonal Complex	Ia	Ib	II	III	IV	V	VI	VII	VIII	IX	NT	No. of Isolates (% Total)
CC1	15	235	185	5	8	873	61	4	7	0	4	1397 (22)
CC23	1128	2	1	16	1	1	0	0	0	0	2	1151 (18.2)
CC19	15	2	6	481	0	182	0	1	10	0	2	699 (11)
CC459	0	1	0	0	591	0	0	0	0	0	1	593 (9.4)
CC22	1	0	547	0	0	0	0	0	0	0	1	549 (8.7)
CC17	0	0	0	461	1	0	0	0	0	0	0	462 (7.3)
CC8	0	403	3	0	0	0	0	0	0	0	1	407 (6.4)
CC12	2	232	89	11	0	1	0	0	0	0	0	335 (5.3)
CC28	0	0	205	1	0	0	0	0	0	0	0	206 (3.3)
CC88	154	0	12	0	0	0	0	0	0	0	1	167 (2.6)
CC452	0	0	0	1	142	0	0	0	0	0	1	144 (2.3)
CC24	62	0	4	1	1	8	0	0	0	0	0	76 (1.2)
CC10	0	39	26	1	0	6	0	0	0	0	0	72 (1.1)
CC26/1087	0	0	0	1	0	29	0	0	0	0	0	30 (0.5)
CC3	6	4	1	1	4	5	0	0	0	0	0	21 (0.3)
Singletons	0	1	2	0	0	3	0	0	0	8	0	14 (0.2)
CC328	0	0	0	7	0	2	0	0	0	0	0	9 (0.1)
CC585/41	0	0	0	0	0	4	0	0	0	0	0	4 (0.06)
Total	1383	919	1081	987	748	1114	61	5	17	8	13	6336 (100)

Abbreviations: CC, clonal complex; NT, non-typeable.

^aThe number of isolates within the major serotype of a given CC is indicated in bold. Fourteen isolates (singletons) were unrelated to any of the 16 CCs in that their multilocus sequence types (MLSTs) did not share 5 of the 7 MLST locus sequences with members of the 16 different CCs.

Figure 2.

A, Serotype associations within individual clonal complexes (CCs). B, PHYLOViZ diagram depicting individual multilocus sequence types and associations with the 10 different serotypes. In the minimum spanning tree, each sequence type (ST) is displayed as a circle and serotypes are represented by different colors. The sizes of circles are all the same and do not represent the prevalence of each ST. Major CCs are indicated.

Figure 3.

A, Clonal complex (CC) distributions among group B Streptococcus causing early-onset disease (EOD), late-onset disease (LOD), and disease among all other ages. B, PHYLOViZ diagram depicting individual multilocus sequence types and associations with EOD, LOD, and disease among all other age groups. In the minimum spanning tree, each sequence type (ST) is displayed as a circle and serotypes are represented by different colors. The sizes of circles are all the same and do not represent the prevalence of each ST. Major CCs are indicated. Abbreviations: CC, clonal complex; EOD, early-onset disease; LOD, late-onset disease.

Distributions of serotypes and CCs among different age groups from individuals > 90 days of age were similar (not shown). Most EOD and LOD serotype III isolates corresponded to CC17 (49/63 [77.8%] and 161/186 [86.6%], respectively), which in all ages is comprised almost entirely of serotype III isolates (Figure 3A and 3B). Most serotype III isolates (444/735 [60.4%]) from other age groups were within CC19 (data not shown).

Supplementary Figures 1, 4, and 5 show distributions of 3 different vaccine candidate categories (capsular polysaccharides, pilus backbone proteins, alpha family proteins) among different CCs. Of 11 major CCs, CCs 8, 17, 22, 28, 452, and 459 were each primarily represented by a single serotype (Figure 2A and 2B), pilus type (Figure 4A and 4B), or alpha family protein (Figure 5A and 5B). Other CCs showed much less type uniformity within the 3 potential vaccine classes (Figures 2, 4, and 5).

Figure 4.

A, Pilus protein determinant associations with individual clonal complexes (CCs). B, PHYLOViZ diagram depicting individual multilocus sequence types and associations with the 5 different pilus profiles. In the minimum spanning tree, each sequence type (ST) is displayed as a circle and serotypes are represented by different colors. The sizes of circles are all the same and do not represent the prevalence of each ST. Major CCs are indicated.

Figure 5.

A, Alpha family protein associations with individual clonal complexes (CCs). B, PHYLOViZ diagram depicting individual multilocus sequence types and associations with the 4 different alpha family protein types. In the minimum spanning tree, each sequence type (ST) is displayed as a circle and serotypes are represented by different colors. The sizes of circles are all the same and do not represent the prevalence of each ST. Major CCs are indicated.

HVGA Virulence Protein

All 465 serotype III/CC17 GBS isolates were positive for hvgA encoding the major GBS adhesin/invasion, including 161 LOD and 49 EOD isolates. The remaining 8 hvgA-positive isolates were serotype IV (CC17, 1; CC452, 7) from adults ranging in age from their early 40s to 90 years (Supplementary Table 3).

srr1 and srr2

Serine-rich repeat glycoprotein determinants (srr1 or srr2) were identified in most isolates (n = 6113 [96.4%]) (Supplementary Table 3), with srr1 predominant (n = 5507 [90.1%]). srr2 was associated primarily with 2 CCs: CC17 (n = 460 [76%]) and CC452 (n = 142 [23.4%]). Isolates negative for srr genes were mostly of ST28 lineage (80%).

Resistance Determinants

Of 6340 isolates, 87 (1.4%) did not yield PBP2X types due to assembly errors. These 87 isolates were found to be susceptible to β-lactam antibiotics by MIC testing. There were 79 PBP2x types (Supplementary Table 1), defined as any amino acid sequence difference from the major allele (3615/6235 [58%]). There were 54 isolates (0.87%) among the 6235 examined with 1 of 31 point mutant alleles of pbp2x that were determined through phenotypic testing to have decreased susceptibility to 1 or more β-lactam antibiotics based on previously described cutoffs [9] (Table 2 and Figure 6). Of these 54 isolates, 37 (68.5%) were also macrolide-resistant (compared to 55% for all isolates combined). These pbp2x point mutants were distributed over the 3 years, and 8 alleles (PBP2X types 17, 24, 26, 32, 42, 47, 59, and 68) accounted for 2 or more of the 54 isolates recovered during at least 2 of the 3 years. Six of these 8 types were overrepresented by a major strain complex (PBP2x-17, 3/6 Ib/CC12; PBP2x-24, 4/4 Ia/ST23; PBP2x-26, 5/5 Ia/ST26; PBP2x-42, 2/2 Ia/ST23; PBP2x-47, 3/3 V/ST1; PBP2x-59, 2/2 V/ST1).

Table 2.

PBP2x Substitutions Associated With Reduced Susceptibility to β-Lactam Antibiotics Among Group B Streptococci From Invasive Disease, 2015–2017

PBP2X Type	PBP2x Substitutionsa	No. of Isolates	Isolate (Serotype, ST)	Ampicillinb	Cefotaximeb	Cefoxitinb	Cefazolinb	Penicillinb	Ceftizoximeb
10	L534S	1	20155117 (V, 1)	0.12	0.12	16	0.5	0.12	4
17	A400V	6	20154959 (Ib, 8)	0.25	0.12	16	0.5	0.12	1
			20153631 (Ib, 1)	0.25	0.12	16	0.5	0.12	1
			20165790 (Ib, 10)	0.25	0.12	16	1	0.12	1
			20173788 (Ib, 8)	0.25	0.12	16	0.5	0.12	1
			20175374 (V, 1)	0.25	0.12	16	1	0.12	1
			20175554 (V, 1)	0.25	0.12	16	0.5	0.12	1
24	I377V, V510I, V525I	4	20155719 (Ia, 23)	0.25	0.12	16	0.5	0.12	1
			20165538 (Ia, 88)	0.25	0.06	8	0.5	0.12	1
			20166195 (Ia, 23)	0.25	0.12	16	0.5	0.12	1
			20175104 (Ia, 23)	0.25	0.12	16	0.5	0.12	1
26	I377V, G415E, V510I	10	20156531 (Ia, 23)	0.25	0.12	16	0.5	0.12	0.5
			20162445 (Ia, 23)	0.25	0.12	16	0.5	0.12	0.5
			20164013 (Ia, 23)	0.25	0.12	16	0.5	0.12	0.5
			20172717 (Ia, 23)	0.25	0.12	16	0.5	0.12	0.5
			20174129 (Ia, 23)	0.25	0.12	16	0.5	0.12	1
			20164903 (Ia, 1031)c	0.12	0.06	16	0.25	0.06	0.5
			20166529 (Ia, 23)	0.25	0.12	16	0.5	0.12	0.5
			20171112 (Ia, 23)	0.25	0.12	16	0.5	0.12	0.5
			20171215 (Ia, 23)c	0.25	0.12	16	0.5	0.12	0.25
			20176075 (Ia, 23)	0.25	0.12	16	0.5	0.12	0.5
29	521-Y-522 insertion	1	20160409 (V, 1)	0.25	0.12	16	0.25	0.12	2
32	T394I	2	20161556 (IV, 459)	0.25	0.12	8	0.25	0.12	2
			20162626 (V, 1)	0.25	0.25	16	0.5	0.12	4
33	R433C	1	20161558 (III, 19)	0.12	0.25	8	0.5	0.12	0.5
34	I377V, G406D	1	20161102 (II, 22)	0.5	0.25	16	1	0.25	16
35	I377V, A400V, V510I	1	20161114 (IV, 452)	0.25	0.12	16	0.5	0.12	1
37	I377V, A400T	1	20162216 (II, 22)	0.25	0.12	16	0.5	0.12	1
39	I377V, G398A, G627V	1	20175390 (III, 109)c	0.5	0.25	16	0.5	0.25	1
42	I377V, A400T, V510I	2	20162746 (Ia, 23)	0.25	0.12	8	0.25	0.12	1
			20164794 (Ia, 23)	0.25	0.12	16	0.5	0.12	1
44	A400T, Q412K	1	20165260 (V, 1)	0.25	0.25	16	0.5	0.12	4
47	F399V	3	20166135 (V, 1)	0.25	0.12	8	0.5	0.12	4
			20176687 (V, 1)	0.5	0.12	16	0.5	0.12	4
			20163262 (V, 1)	0.25	0.12	8	0.5	0.12	2
48	I378V, V510I, A514G, N575K	1	20166262 (Ia, 23)	0.5	0.12	4	0.25	0.12	0.25
55	I377V, T394I	1	20164062 (II, 22)	0.25	0.12	16	0.5	0.12	2
58	I377V, Q412R	1	20166357 (Ia, 22)	0.25	0.12	16	0.25	0.12	0.5
59	G406D	2	20170772 (V, 1)	0.5	0.25	16	1	0.25	32
			20183967 (V, 1)	0.25	0.12	8	1	0.12	16
60	Del K358	1	20171010 (III, 19)	0.12	0.06	16	0.25	0.06	2
63	I377V, V510I, Q557E	1	20172108 (Ia, 23)	0.25	0.25	16	1	0.25	2
66	I377V, G406D, I510V	1	20172365 (Ia, 23)	0.5	0.12	16	1	0.25	16
68	T394A	2	20172646 (Ib, 8)	0.12	0.12	8	0.25	0.06	2
			20181729 (IV, 459)	0.12	0.12	8	0.25	0.06	2
70	I377V, A400V	1	20173768 (III, 17)	0.25	0.12	16	0.5	0.12	1
80	I377V, V510I, A514T	1	20174532 (Ia, 23)	0.25	0.12	32	0.5	0.12	0.5
81	A400V, V525I	1	20176362 (V, 1)	0.25	0.12	16	0.5	0.12	8
83	A550V	1	20175389 (IV, 821)	0.12	0.12	16	0.5	0.06	2
85	I377V, V510I, A616V	1	20177021 (Ia, 23)	0.25	0.12	16	0.5	0.12	0.5
87	T432I	1	20180100 (III, 19)	0.12	0.12	16	0.5	0.12	2
91	F395C	1	20181122 (III, 19)	0.12	0.12	4	0.5	0.12	1
94	I377V, L534S, G627V	1	20182381 (III, 828)	0.25	0.25	16	0.5	0.25	8
95	F486L,V510L	1	20183394 (III, 23)	0.25	0.12	16	0.5	0.12	0.5

Abbreviation: ST, sequence type.

^aResidues in bold appear to be associated with decreased β-lactam susceptibility.

^bValues in bold indicate elevated minimum inhibitory concentrations.

^cIsolates 20164903 (early-onset disease [EOD]), 20171215 (late-onset disease [LOD]), and 20175390 (EOD) were associated with EOD and LOD cases. All other isolates were from adult disease cases.

Figure 6.

Serotype distribution of the 54 group B Streptococcus isolates from this study with reduced susceptibility or low-level nonsusceptibility of up to 6 different β-lactam (BL) antibiotics (also shown in Table 2), including data for reduced or low-level susceptibility for each of the 6 β-lactams used. The number of isolates are shown above each of the 10 columns. The total number of isolates represented by each serotype is the sum of the first 6 columns. Abbreviation: BL, β-lactam.

Only 9 isolates within 8 PBP2X types (34, 39, 47, 48, 59, 63, 66, and 94) had MICs above the susceptible breakpoints for ampicillin and/or penicillin (≥ 0.25 mg/L and ≥ 0.12 mg/L, respectively) (Table 2). Of these 8, types 34, 39, 59, and 94 also revealed elevated MICs for the third-generation cephalosporins cefotaxime (0.25 mg/L) and ceftizoxime (1–32 mg/L). Relevant substitutions within these PBP2X types included G406D (n = 4), G398A (n = 1), F399V (n = 1), A514G + N575K (n = 1), L354S (n = 1), and Q557E (n = 1), which generally map close to 3 conserved catalytic motifs within PBP2X: 344-STMK-347, 402-SSN-404, and 552-KSG-554. PBP2X type 59 (G406D), found within 2 serotype V/ST1 isolates, was unique in that it was observed in > 1 isolate (once in each of years 2016 and 2017) and was also associated with above-normal MICs for all 6 of the β-lactam antibiotics tested. In total, there were 14 isolates of 1 of 12 different PBP2X types with elevated MICs for 1–4 of the clinically relevant β-lactams (ampicillin, penicillin, cefotaxime, cefazolin) used in this study. Thirteen isolates contained a substitution (V or T) at the position A400 (PBP2X types 17, 35, 37, 42, 44, 70, and 81). The PBP2X G415E substitution was found in 10 serotype Ia/CC23 isolates. There was no evidence of contributing mutations within PBP genes pbp1a (Supplementary Table 1) and pbp2b (data not shown). The exception was PBP1a-62 in association with PBP2X-83, with a D370N substitution. Elevated MICs for cefoxitin and/or ceftizoxime were most common, with 78 instances compared to 26 instances for the 4 other β-lactams combined (Figure 6). There were only 12 isolates with reduced susceptibility to 3 or more of the 6 β-lactams (Figure 6).

Erythromycin resistance (with clindamycin resistance or alone) within the 6 major serotypes ranged from 39.2% (serotype III) to 77.7% (serotype IV) (Figure 7). Several different genes conferred resistance to 1 or more of the antibiotic classes: macrolides, lincosamides, and streptogramins (Supplementary Table 6). Overall, resistance to erythromycin and clindamycin was predicted to be 55.2% (n = 3497) and 43.9% (n = 2783), respectively. The erm determinants (ermB, ermTR/A, ermT) that confer resistance to macrolides, lincosamide, and streptogramin B (MLS_B) antibiotics were found alone or in combination with other determinants in 2690 strains and represented most (99.3%) of the combined resistance to erythromycin and clindamycin. The remaining isolates (18/2710 [0.7%]) contained mef/msrD (conferring erythromycin resistance) and either the lnu (lnuB or lnuC) and/or lsa (lsaA, lsaC, or lsaE) determinants known to confer resistance to lincosamides [9, 31]. Erythromycin resistance and clindamycin susceptibility was predicted in 779 mef/msrD–positive isolates. One isolate (20161487) contained an apparent mef homologue (designated mefSL1) (Supplementary Table 6). Sixty-seven isolates had predicted resistance to clindamycin only; 59 isolates contained lsaC, 7 with both lnuB and lsaE, and 1 with lnuC alone. Tetracycline resistance was predicted in 85.2% (n = 5399) of isolates, primarily conferred by tetM (n = 5011 [92.8%]) and tetO (4.6%) (Supplementary Table 7). There were 103 (1.62%) isolates with predicted MICs of ≥ 4 mg/L for levofloxacin (Supplementary Table 8). GyrA S81L combined with ParC S79F/Y was the most common substitution (n = 69 [67%]) conferring fluoroquinolone resistance. Two isolates (20166174 and 20170296), both serotype V/ST1 from adult patients in Maryland, contained vanG elements conferring vancomycin MIC of 2 mg/L (Supplementary Table 9). Chloramphenicol resistance conferred by catQ was detected in 10 isolates. Eighteen isolates had substitutions within RpoB that conferred rifampin resistance. Two isolates carried dfrG that confers high-level trimethoprim resistance. Determinants found to be associated with gentamicin (aac6-aph2) and other aminoglycosides (aph3-III or ant6-Ia) were detected in 17 and 238 isolates, respectively.

Figure 7.

Proportions of each group B Streptococcus serotype resistant to the individual antimicrobial classes shown and the proportion susceptible to all the antimicrobials assessed in this study (includes those susceptible to the drugs shown and additionally susceptible to clindamycin alone, vancomycin, rifampin, chloramphenicol, and aminoglycosides). Abbreviation: NT, non-typeable.

DISCUSSION

During the past 70 years, GBS has emerged as the most common cause of newborn sepsis. Efforts are ongoing toward developing effective vaccines to be implemented in pregnancy, and there is more recognition of the increasing iGBS disease burden within adults.

GBS has an extensive and variable array of surface structures, some of which are promising vaccine candidates [19]. A 6-valent polysaccharide formulation would target nearly all combined EOD and LOD cases. Even so, the concept of universal protein-based vaccines that could target all GBS is attractive. Our data are consistent with past work indicating that a vaccine containing 3 pilus protein constituents could conceivably target virtually all GBS [32, 33]. The GBS-NN vaccine, consisting of a protein that correlates to the rib and alpha queries used for this survey, could also possibly protect against GBS strains, since nearly all isolates potentially express highly homologous alpha family proteins [21]. There is uncertainty in this prediction, as published data indicating cross-protection of GBS-NN against strains expressing Alp1 and Alp2/3 are lacking [20, 34]. Strains within the serotype III MLST type 17 (ST17) lineage, which uniformly express the hypervirulent GBS adhesin HvgA, are disproportionally associated with neonatal meningitis and sepsis [27] and, as shown in this work, are uniformly positive for the Rib protein that comprises half of the GBS-NN vaccine.

Within the major CCs, it appears that there has been frequent recombinational switching of genetic determinants encoding capsular serotype, pilus type, and alpha family protein type. It is theoretically possible that 1 or more of the 3 rarely occurring serotypes (1.64%) could emerge as a major cause of iGBS disease, possibly facilitated through expansion of a serotype-switch variant within a major virulent clonal complex. The primary CCs of each of these 3 rarely occurring serotypes in invasive disease are highly associated with an alpha family determinant type and 1 or more pilus types.

Although there have been reports of decreased susceptibility to β-lactam antibiotics during the past 11 years [6–8, 35], our data from 3 years of surveillance indicate that this phenotype remains rare within the US and that impactful emergence of such strains has not occurred. Nonetheless, the small number of these rare mutant alleles within members of the same major strain complexes within multiple years suggests that in each of these strains a level of fitness has been attained that compensates for an altered Ppb2X protein (an essential component of the peptidoglycan synthetic apparatus).

We performed conventional testing for > 2000 of the isolates described here that were recovered during 2015 and, as other studies have described [8, 36, 37], we found that the detection of unusual PBP2X substitutions within its transpeptidase domain effectively flagged all GBS isolates with unusually elevated β-lactam MICs.

Recently we identified 2 independent vancomycin-nonsusceptible serotype II/ST22 iGBS strains (each carrying a distinct vanG element) [38]. In the current study, we identified and phenotypically confirmed 2 vanG-positive, serotype V/ST1 isolates. The vanG elements detected in GBS to date are nearly identical to elements found within Enterococcus, suggestive of a common gastrointestinal interspecies reservoir for accessory element resistance determinants. Although GBS nonsusceptibility to clinically relevant antibiotics such as β-lactams and vancomycin remains very rare, continued vigilance is warranted. Continued monitoring of diverse resistance features is essential in view of the rapidly growing proportion of iGBS that is resistant to macrolides and lincosamides [3, 5] conferred by diverse genetic determinants [9, 31].

CONCLUSIONS

We present population-based analysis of > 6300 iGBS isolates from the US. Annotated features from each isolate are linked to a publicly available genomic sequence. These data establish an important baseline for monitoring this rapidly evolving pathogen and evaluating vaccine candidates. Current capsular serotype and surface protein distributions indicate that these potential vaccine targets would provide excellent coverage, especially in young infants.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. The authors thank all persons in the Active Bacterial Core Surveillance (ABCs) areas who are involved with surveillance and maintenance of the system; the laboratory staff at the 10 sites who isolate ABCs pathogens and make it possible to track these infections; the Centers for Disease Control and Prevention (CDC) ABCs Team and the Streptococcus Laboratory for isolate characterization; and the Minnesota Department of Public Health laboratory for performing phenotypic antimicrobial susceptibility testing of all isolates from Minnesota. The authors also acknowledge the following members of the ABCs team at the study sites: California (Art Reingold, Mirasol Apostol, and Herschel Kirk), Colorado (Rachel Herlihy and Shelli Marks), Georgia (Amy Tunali, Stephanie Thomas, Melissa Tobin-D’Angelo, and Ashley Moore), Minnesota (Corinne Holtzman, Kathy Como-Sabetti, and Anita Glennan), New Mexico (Lisa Onischuk and Nicole Espinoza), Oregon (Tasha Poissant and Heather Jamieson), and New York (Kari Burzlaff, Glenda Smith, Nancy Spina, and Rachel Wester). This publication made use of the Streptococcus agalactiae multilocus sequencing typing website (https://pubmlst.org/sagalactiae/) sited at the University of Oxford [39]; development of this site has been funded by the Wellcome Trust.

Disclaimer. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the CDC.

Financial support. This work was supported by CDC funding for the ABCs program through cooperative agreements with the ABCs sites. The whole genome sequencing of group B Streptococcus isolates was supported in part by the CDC’s Advanced Molecular Detection program.

Potential conflicts of interest. L. H. H. has received personal fees from GSK, Merck, Pfizer, and Sanofi Pasteur, outside the submitted work. N. A., M. M. F., and P. S. V. report grants from the CDC during the conduct of the study but not related to the study. R. L. reports grants from the CDC during the conduct of the study; was a coeditor of a book on infectious disease surveillance; received royalties donated to the Minnesota Department of Health; and and has served as associate editor of the Red Book Report of the American Academy of Pediatrics Committee on Infectious Diseases, with compensation donated to the Minnesota Department of Health. All other authors report no potential conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.