Invasive Streptococcus agalactiae infections in infants in Guangzhou, Southern China (2013–2022): molecular epidemiology and clinical management implications Authors: Xiao-lan Chen, Hua-min Zhong, Yong-qiang Xie, Wen-shan Chen, Li-Jia Wei, Kan-kan Gao, Bing-shao Liang, Su-fei Zhu, Yun-feng Liu, Fei Gao, Yan Long, Lian-fen Huang Published: 2026 ABSTRACT Streptococcus agalactiae (group B Streptococcus, GBS) is an important cause of invasive infections in infants worldwide, despite widespread implementation of intrapartum antibiotic prophylaxis (IAP). This study aimed to clarify the molecular and clinical characteristics of invasive GBS isolates from infants in Guangzhou, China, over the 10-year period 2013-2022. A total of 131 GBS strains were recovered from blood cultures (n = 98) and cerebrospinal fluid cultures (n = 33) of infants aged ≤ 90 days. All isolates were subjected to capsular serotyping, multilocus sequence typing (MLST), virulence gene profiling (including cylE, hylB, Alp family genes, and pilus island genes), and antimicrobial susceptibility testing. Of the 131 GBS isolates, 37 (28.2%) caused early-onset disease (EOD) and 94 (71.8%) caused late-onset disease (LOD). The hypervirulent III/ST17 clone was the predominant lineage (56.5%) and was associated with LOD. Serotype Ia accounted for 15.3% of all isolates and was correlated with respiratory distress in EOD cases. All isolates carried the cylE and hylB toxin genes, with rib being the most prevalent surface protein gene (71.8%). Pilus island PI-2b (67.9%) was the dominant type and was predominantly detected in serotype III/CC17 strains. Resistance rates to clindamycin, erythromycin, and tetracycline exceeded 80%, while all isolates remained susceptible to penicillin and ampicillin. This study confirms that the III/ST17 clone is the predominant lineage associated with GBS LOD and identifies an association between serotype Ia and respiratory distress in GBS EOD. These findings support ongoing molecular surveillance to guide targeted preventive strategies and improve clinical management of invasive infantile GBS disease.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12866-026-05195-1. Introduction Streptococcus agalactiae (group B Streptococcus, GBS) is a Gram-positive coccus and a common commensal organism colonizing the lower genital tract and rectum in women. It is also a major pathogen responsible for neonatal sepsis, meningitis, and pneumonia globally, often leading to severe outcomes, including neurological sequelae and neonatal death [1]. Clinically, invasive GBS infections in infants are classified into two distinct syndromes: early-onset disease (EOD), occurring at 0-6 days of life, and late-onset disease (LOD), occurring at 7–89 days of life. Multicenter clinical and molecular epidemiological analyses in China have reported an incidence of 0.31 cases of invasive GBS disease per 1000 live births, with a mortality rate of 2.3% [2]. Although intrapartum antibiotic prophylaxis (IAP) has been associated with a reduction in the incidence of EOD in many high-income countries, GBS continues to impose a substantial clinical burden, largely due to the limited efficacy of IAP for preventing LOD [3]. GBS encodes a diverse array of virulence factors [4], including pilus islands (PI), which play a pivotal role in biofilm formation [5]—a mechanism that may enable GBS to evade IAP. The Alp family of surface proteins mediates endothelial adhesion and represents a promising target for protein-based vaccines [6]. Meanwhile, the potent toxins hemolysin and hyaluronidase, encoded by the cylE and hylB genes, respectively, induce host cell damage and facilitate invasion across the placental barrier [7, 8]. However, comprehensive analyses integrating the distribution of these key virulence factors in clinical GBS isolates remain limited. Molecular characterization is useful for exploring the epidemiology and pathogenic characteristics of GBS. Traditional capsular serotyping based on capsular polysaccharide (CPS) antibodies [9] remains widely used but has limitations for isolates with low or absent capsule expression. Geno-serotyping, combining latex agglutination with PCR-based capsular gene detection, has been developed to address this shortcoming and enable more accurate characterization [10, 11]. Multilocus sequence typing (MLST) has been a standard method for defining GBS genetic lineages since 2003 [12]. Accumulating evidence indicates that distinct GBS sequence types (STs) and clonal complexes (CCs) exhibit notable differences in pathogenic potential, epidemiological patterns, and antibiotic resistance profiles [13, 14]. Our previous work demonstrated the dominance of the hypervirulent III/ST17 lineage in China, as well as its association with the meningeal tropism gene hvgA and major resistance genes [15]. Other studies have further linked specific STs with distinct resistance profiles: ST17 is frequently resistant to tetracycline but susceptible to levofloxacin, whereas ST19 tends to show the opposite pattern [16]; ST10 is associated with high radezolid minimum inhibitory concentrations (MICs) and fluoroquinolone resistance [17, 18]; and the III/ST19 and Ib/ST10 lineages have been identified as major multidrug-resistant clones among fluoroquinolone-resistant GBS strains in China [19]. Despite these advances, longitudinal epidemiological analyses of invasive GBS isolates from infants remain limited. In this study, we conducted a ten-year (2013–2022) retrospective molecular epidemiological investigation of invasive GBS isolates from infants aged ≤ 90 days in Guangzhou to characterize the distribution of serotypes, STs, and ten key virulence genes—including three pilus islands (PI-1, PI-2a, and PI-2b); five Alp family surface proteins (rib, alphaC, alp1, alp2/3, and alp4); and two toxin genes (cylE and hylB). This characterization may help inform more targeted GBS prevention strategies and improve the clinical management of invasive GBS disease in infants. Materials and methods Bacterial isolates and ethical approval All GBS isolates were collected from blood cultures (n = 98) and cerebrospinal fluid (CSF) cultures (n = 33) of infants aged ≤ 90 days hospitalized at Guangzhou Medical University Affiliated Women and Children’s Medical Center during 2013–2022. Species identification was confirmed using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS). Relevant clinical data were retrieved from the hospital’s electronic medical records. The study protocol was approved by the Institutional Ethics Committee of the hospital (approval no. 2025120A01). Capsular serotyping and MLST Molecular CPS serotyping was performed using a validated multiplex PCR assay and serum agglutination method [11]. Genomic DNA was extracted from GBS isolates using the SteadyPure Bacterial Genomic DNA Extraction Kit (Realy Biotechnology, China) following the manufacturer’s protocol for Gram-positive bacteria. MLST was performed by sequencing seven housekeeping genes (adhP, pheS, atr, glnA, sdhA, glcK, and tkt), and allele profiles, STs, and CCs were assigned by querying the PubMLST database (https://pubmlst.org/organisms/streptococcus-agalactiae/primers). The goeBURST algorithm was applied to assign STs to CCs and to visualize and deduce the evolutionary genetic relationship and founder genotype among all recovered GBS isolates. Virulence gene detection PCR or multiplex PCR assays were used to detect the presence of three pilus island genes (PI-1, PI-2a, and PI-2b), five Alp family surface protein genes (rib, alphaC, alp1, alp2/3, and alp4), and two toxin genes (cylE and hylB) [20]. Antimicrobial susceptibility testing The MICs of 13 antibiotics (penicillin, ampicillin, ciprofloxacin, levofloxacin, moxifloxacin, linezolid, clindamycin, quinupristin/dalfopristin, erythromycin, vancomycin, tetracycline, nitrofurantoin, and tigecycline) were determined using the VITEK 2 Compact system (BioMérieux, Marcy-l’Étoile, France). Antimicrobial susceptibility interpretations were based on the 2020 Clinical and Laboratory Standards Institute (CLSI) criteria. The double-disk agar diffusion test (D-test) was used to detect the inducible macrolide–lincosamide–streptogramin B (iMLSB) phenotype. Multidrug resistance (MDR) was defined as resistance to at least three classes of antimicrobial drug agents. Statistical analysis Categorical data were presented as numbers and percentages. Between-group comparisons of categorical variables were performed using the chi-square test or Fisher’s exact test (two-tailed). Binary logistic regression was performed to identify independent factors associated with the study periods (2018–2022 vs. 2013–2017). A P value ≤ 0.05 was considered statistically significant. All statistical analyses were conducted using SPSS software (version 26.0; IBM Corp., Armonk, NY, USA). A minimum spanning tree was generated using PHYLOVIZ version 2.0 to visualize genetic relationships between serotypes, STs, and CCs. Results Epidemiology of Invasive GBS infections Between 2013 and 2022, a total of 48,264 blood or CSF cultures were processed for hospitalized infants (≤ 90 days) at the study center, of which 137 (2.8%) were positive for GBS. Six isolates were excluded due to unsuccessful preservation, resulting in a final study cohort of 131 isolates. Clinical and demographic characteristics of infants with EOD and LOD are summarized in Table 1; Fig. 1. Of the 131 isolates, 37 (28.2%) caused EOD and 94 (71.8%) caused LOD. The main clinical manifestations of EOD were septicemia (35.1%, 13/37) and respiratory distress (32.4%, 12/37), while LOD was predominantly characterized by septicemia (34.0%, 32/94) and meningitis (33.0%, 31/94). Other clinical manifestations included pneumonia (13.7%, 18/131), respiratory distress (9.9%, 13/131), pathological jaundice (7.6%, 10/131), and focal infections (bones, joints, soft tissues; 2.3%, 3/131). The mortality rate was higher in EOD patients (8.6%, 3/35, exclude missing values) than in LOD patients (2.8%, 2/71, exclude missing values). Notably, 80% of fatal cases (4/5, P < 0.05) involved low birthweight infants (< 2500 g). Table 1Clinical and demographic characteristics of infants with invasive GBS infection, stratified by EOD, LOD, and ST17 status (n = 131)CharacteristicEODLODST17(n = 37)NA(n = 94)NAP-value(n = 74)NAAge at onset days, median (IQR)1 (1–1)024.5 (16-45.75)0-19.5(10.25-31)0Male/female ratio23:14 (1.64)042:52 (0.81)0> 0.0536:38 (0.95)0Birthweight (g), median (IQR)3060 (2780–3445)23100 (2712.5–3475)20-3020 (2735–3470)15Low birthweight (< 2500 g), n (%)6 (17.1)212 (16.2)20> 0.059 (15.3)15Length of hospital stay, days, Median (IQR)17 (13–24)832.5 (16.75-52)30-27 (16–43)21Mortalitya n (%)3 (8.6)22 (2.8)23> 0.053 (5.1)15Preterm birth (< 37 weeks), n (%)6 (17.1)213 (17.8)21> 0.0511 (18.6)15Premature rupture of membranes, n (%)12 (34.3)26 (8.5)230.0027 (12.1)16Delivery mode (n), Vaginal/C-section ratio21:12 (1.75)448:23 (2.09)23> 0.0535:20 (1.75)19Pregnancy type, n (%), first-time15 (42.9)227 (38.0)23> 0.0522 (37.9)16GBS antenatal screening, n (%)8 (22.9)26 (8.5)230.0417 (12.1)16Positive GBS antenatal screening, n (%)4 (50.0)292 (33.3)88> 0.053 (42.8)67Maternal prenatal fever, n (%)3 (8.6)22 (2.8)23> 0.053 (5.1)15History of Adverse pregnancy outcomes, n (%)6 (17.1)25 (7.0)23> 0.055 (8.5)15EOD early-onset disease (0-6 days), LOD late-onset disease (7–89 days) , IQR interquartile range, ST sequence type, NA not availableP values were calculated by chi-square or Fisher’s exact test. Mortalitya was significantly associated with low birth weight (Fisher’s exact test, P = 0.02). Adverse pregnancy history was defined as fetal demise or spontaneous abortion Table 1Clinical and demographic characteristics of infants with invasive GBS infection, stratified by EOD, LOD, and ST17 status (n = 131)CharacteristicEODLODST17(n = 37)NA(n = 94)NAP-value(n = 74)NAAge at onset days, median (IQR)1 (1–1)024.5 (16-45.75)0-19.5(10.25-31)0Male/female ratio23:14 (1.64)042:52 (0.81)0> 0.0536:38 (0.95)0Birthweight (g), median (IQR)3060 (2780–3445)23100 (2712.5–3475)20-3020 (2735–3470)15Low birthweight (< 2500 g), n (%)6 (17.1)212 (16.2)20> 0.059 (15.3)15Length of hospital stay, days, Median (IQR)17 (13–24)832.5 (16.75-52)30-27 (16–43)21Mortalitya n (%)3 (8.6)22 (2.8)23> 0.053 (5.1)15Preterm birth (< 37 weeks), n (%)6 (17.1)213 (17.8)21> 0.0511 (18.6)15Premature rupture of membranes, n (%)12 (34.3)26 (8.5)230.0027 (12.1)16Delivery mode (n), Vaginal/C-section ratio21:12 (1.75)448:23 (2.09)23> 0.0535:20 (1.75)19Pregnancy type, n (%), first-time15 (42.9)227 (38.0)23> 0.0522 (37.9)16GBS antenatal screening, n (%)8 (22.9)26 (8.5)230.0417 (12.1)16Positive GBS antenatal screening, n (%)4 (50.0)292 (33.3)88> 0.053 (42.8)67Maternal prenatal fever, n (%)3 (8.6)22 (2.8)23> 0.053 (5.1)15History of Adverse pregnancy outcomes, n (%)6 (17.1)25 (7.0)23> 0.055 (8.5)15EOD early-onset disease (0-6 days), LOD late-onset disease (7–89 days) , IQR interquartile range, ST sequence type, NA not availableP values were calculated by chi-square or Fisher’s exact test. Mortalitya was significantly associated with low birth weight (Fisher’s exact test, P = 0.02). Adverse pregnancy history was defined as fetal demise or spontaneous abortion Clinical and demographic characteristics of infants with invasive GBS infection, stratified by EOD, LOD, and ST17 status (n = 131) EOD early-onset disease (0-6 days), LOD late-onset disease (7–89 days) , IQR interquartile range, ST sequence type, NA not available P values were calculated by chi-square or Fisher’s exact test. Mortalitya was significantly associated with low birth weight (Fisher’s exact test, P = 0.02). Adverse pregnancy history was defined as fetal demise or spontaneous abortion Fig. 1Clinical and molecular distribution of invasive GBS infections in infants. (A) Distribution of infants across hospital departments. All EOD infants and 58.5% (n = 55) of LOD infants were admitted to the intensive care unit (PICU/ICU) or newborn department (including NICU). Remaining LOD patients were from the observation unit (n = 17), emergency department (n = 14), cranial surgery (n = 5), infectious diseases department (n = 2), and orthopedics department (n = 1). (B) Serotype and disease distribution patterns between EOD and LOD groups. Abbreviations: EOD, early-onset disease (0-6 days); LOD, late-onset disease (7–89 days); PICU, pediatric intensive care unit; ICU, intensive care unit; NICU, neonatal intensive care unit Clinical and molecular distribution of invasive GBS infections in infants. (A) Distribution of infants across hospital departments. All EOD infants and 58.5% (n = 55) of LOD infants were admitted to the intensive care unit (PICU/ICU) or newborn department (including NICU). Remaining LOD patients were from the observation unit (n = 17), emergency department (n = 14), cranial surgery (n = 5), infectious diseases department (n = 2), and orthopedics department (n = 1). (B) Serotype and disease distribution patterns between EOD and LOD groups. Abbreviations: EOD, early-onset disease (0-6 days); LOD, late-onset disease (7–89 days); PICU, pediatric intensive care unit; ICU, intensive care unit; NICU, neonatal intensive care unit Approximately 86.8% of pregnant women lacked GBS screening or had no documented results, and two-thirds of available screening records were obtained after August 2021. Prenatal GBS colonization status was documented for 14 mothers: the colonization rate was higher among mothers of infants with EOD (50.0%, 4/8) than among those of infants with LOD (33.3%, 2/6), although this difference was not statistically significant, likely due to the small sample size. Among GBS-colonized mothers, only one mother of an infant with EOD received IAP. One LOD case occurred in an infant born via in vitro fertilization, whose mother had a history of recurrent miscarriage (G8P3, recurrent pregnancy loss in G2 to G6). Overall, 8.4% (11/131) of mothers had a history of adverse pregnancy outcomes (e.g., spontaneous abortion or fetal demise). Additionally, one infant with LOD sepsis had an older sibling with a history of GBS meningitis. Among LOD cases with meningitis, three infants developed long-term sequelae following treatment, including focal epilepsy, speech delay, and autism spectrum disorder. Antimicrobial susceptibility profiles All 131 isolates were fully susceptible to penicillin, ampicillin, linezolid, quinupristin/dalfopristin, and vancomycin. Tetracycline (94.1%, 95/101), erythromycin (87.8%, 79/90), and clindamycin (80.4%, 82/102) exhibited high resistance rates, each exceeding 65% annually throughout the study period. Quinolone resistance fluctuated over time: low-level resistance to levofloxacin and moxifloxacin (0-25%) was observed from 2014 to 2018 but subsequently declined (Fig. 2). Denominators for annual resistance rates varied slightly due to missing susceptibility data for some isolates in early years and differences in test panels used across periods. Resistance to ciprofloxacin, levofloxacin, and moxifloxacin was significantly associated with the ST19 lineage (P < 0.01; Table S2). Detailed annual antimicrobial susceptibility profiles are provided in Table S1. Fig. 2Temporal trends in antimicrobial resistance to seven antimicrobial agents among invasive GBS isolates, 2014-2022. Abbreviations: CIP, ciprofloxacin; LEV, levofloxacin; MXF, moxifloxacin; CLI, clindamycin; ERY, erythromycin; TET, tetracycline; NIT, nitrofurantoin Temporal trends in antimicrobial resistance to seven antimicrobial agents among invasive GBS isolates, 2014-2022. Abbreviations: CIP, ciprofloxacin; LEV, levofloxacin; MXF, moxifloxacin; CLI, clindamycin; ERY, erythromycin; TET, tetracycline; NIT, nitrofurantoin Capsular serotyping Four capsular serotypes (Ia, Ib, III, and V) were identified among the 131 isolates. In EOD cases, predominant serotypes were III (48.6%, 18/37) and Ia (32.4%, 12/37), followed by Ib (13.5%, 5/37) and V (5.4%, 2/37). In LOD cases, serotype III was predominant (80.9%, 76/94), followed by Ib (10.6%, 10/94) and Ia (8.5%, 8/94) (Fig. 1B). Serotype III was associated with LOD (80.9%, 76/94; P < 0.01), while Serotype Ia was significantly associated with EOD (60.0%, 12/20; P = 0.01). Meningitis cases were mainly caused by serotype III (74.3%, 26/35) and serotype Ib (22.9%, 8/35). A significant association was also observed between serotype Ia and respiratory distress (61.5%, 8/13, P<0.001) (Fig. 3B, Table S2). Fig. 3Minimum spanning tree, infection type, and Alp gene distribution of GBS isolates. A Minimum spanning tree based on serotypes, sequence types (STs), and clonal complexes (CCs). Circles represent STs, with circle size corresponding to the number of isolates. Colors denote serotypes; shading indicates CC assignment. Digits indicate goeBURST distance scores. B Serotype-specific distribution of clinical infection types. C Serotype-specific distribution patterns of Alp family virulence genes. Abbreviations: GBS, Group B Streptococcus; ST, sequence type; CC, clonal complex Minimum spanning tree, infection type, and Alp gene distribution of GBS isolates. A Minimum spanning tree based on serotypes, sequence types (STs), and clonal complexes (CCs). Circles represent STs, with circle size corresponding to the number of isolates. Colors denote serotypes; shading indicates CC assignment. Digits indicate goeBURST distance scores. B Serotype-specific distribution of clinical infection types. C Serotype-specific distribution patterns of Alp family virulence genes. Abbreviations: GBS, Group B Streptococcus; ST, sequence type; CC, clonal complex MLST and clonal complexes A total of 23 distinct STs were identified, with ST17 being the most prevalent (56.5%, 74/131), followed by ST12 (9.9%, 13/131) and ST19 (6.9%, 9/131) (Table 2). All STs were clustered into eight CCs, with CC17 (58.0%, 76/131), CC19 (11.5%, 15/131), and CC12 (7.6%, 10/131) being the most common (Figure S1). A minimum spanning tree illustrates relationships among serotype, STs, and CCs (Fig. 3A). The III/ST17 combination accounted for 56.5% (74/131) of all isolates and was associated with LOD (P < 0.005; Table S2). Other notable genotype-serotype associations include III/CC19 (9.9%, 13/131), Ib/ST12 (7.6%, 10/131), and Ib/CC10 (6.9%, 9/131). Table 2Distribution of serotype and sequence types (STs) among all invasive GBS isolates (n=131) and meningitis cases (n = 35)SerotypeSTs in meningitis isolates (n, %)STs in all isolates (n, %)III (n = 94)ST17(22, 62.86%), ST19(3, 8.57%), ST188(1, 2.86%)ST17(74, 56.49%), ST19(9, 6.87%), ST27(2, 1.53%), ST12(1, 0.76%), ST44(1, 0.76%), ST188(1, 0.76%), ST357(1, 0.76%), ST680(1, 0.76%), ST861(1, 0.76%), ST862(1, 0.76%)Ia (n = 20)ST12(1, 2.86%)ST23(3, 2.29%), ST12(2, 1.53%), ST24(2, 1.53%), ST485(2, 1.53%), ST529(2, 1.53%), ST890(2, 1.53%), ST27(1, 0.76%), ST28(1, 0.76%), ST399(1, 0.76%), ST7(1, 0.76%), ST1274(1, 0.76%), ST1964(1, 0.76%)Ib (n = 15)ST12(4,11.43%), ST10(2, 5.71%), ST357(1, 2.86%), ST579(1, 2.86%)ST12(10, 7.63%), ST10(3, 2.29%), ST357(1, 0.76%), ST579(1, 0.76%)V (n = 2)-ST1(2, 1.53%)For meningitis isolates, percentages are based on the total number of meningitis cases (n = 35); for all isolates, percentages are based on the total number of GBS isolates (n = 131)GBS group B streptococcus, ST sequence type Table 2Distribution of serotype and sequence types (STs) among all invasive GBS isolates (n=131) and meningitis cases (n = 35)SerotypeSTs in meningitis isolates (n, %)STs in all isolates (n, %)III (n = 94)ST17(22, 62.86%), ST19(3, 8.57%), ST188(1, 2.86%)ST17(74, 56.49%), ST19(9, 6.87%), ST27(2, 1.53%), ST12(1, 0.76%), ST44(1, 0.76%), ST188(1, 0.76%), ST357(1, 0.76%), ST680(1, 0.76%), ST861(1, 0.76%), ST862(1, 0.76%)Ia (n = 20)ST12(1, 2.86%)ST23(3, 2.29%), ST12(2, 1.53%), ST24(2, 1.53%), ST485(2, 1.53%), ST529(2, 1.53%), ST890(2, 1.53%), ST27(1, 0.76%), ST28(1, 0.76%), ST399(1, 0.76%), ST7(1, 0.76%), ST1274(1, 0.76%), ST1964(1, 0.76%)Ib (n = 15)ST12(4,11.43%), ST10(2, 5.71%), ST357(1, 2.86%), ST579(1, 2.86%)ST12(10, 7.63%), ST10(3, 2.29%), ST357(1, 0.76%), ST579(1, 0.76%)V (n = 2)-ST1(2, 1.53%)For meningitis isolates, percentages are based on the total number of meningitis cases (n = 35); for all isolates, percentages are based on the total number of GBS isolates (n = 131)GBS group B streptococcus, ST sequence type Distribution of serotype and sequence types (STs) among all invasive GBS isolates (n=131) and meningitis cases (n = 35) For meningitis isolates, percentages are based on the total number of meningitis cases (n = 35); for all isolates, percentages are based on the total number of GBS isolates (n = 131) GBS group B streptococcus, ST sequence type Temporal distribution and logistic regression analysis (2013–2017 vs. 2018–2022) Comparisons between the two study periods (2013-2017, n = 75; 2018–2022, n = 56) revealed no statistically significant differences in disease onset type or sex distribution (Table 3). Using serotype III as the reference, serotype Ia was more prevalent in the 2018–2022 period (32.1% vs. 2.7%; OR = 13.327, 95% CI: 2.753–64.518, P = 0.001). Serotypes Ib and V were detected exclusively in 2013–2017. Annual trends for serotypes and dominant STs are presented in Fig. 4. Table 3Comparison of disease onset, sex, and serotypes distribution during the period of 2013–2017 and 2018–2022Characteristic2013-2017(n = 75)n (%)2018-2022(n = 56)n (%)P-valueOR(95% CI)Disease onsetEOD19 (25.3)18 (33.3)0.392-LOD56 (74.7)38 (66.7)SexMale37 (49.3)28 (50.0)0.940-Female38 (50.7)28 (50.0)SerotypeIa2 (2.7)18 (32.1)0.00113.327(2.753–64.518)Ib15 (20.0)0 (0)0.998*-V2 (2.7)0 (0)0.998*III56 (74.6)38 (67.9)Ref.EOD early-onset disease, LOD late-onset disease, OR odds ratio, CI confidence interval, Ref. reference*Due to complete separation, Fisher’s exact test was used for serotypes Ib and V Table 3Comparison of disease onset, sex, and serotypes distribution during the period of 2013–2017 and 2018–2022Characteristic2013-2017(n = 75)n (%)2018-2022(n = 56)n (%)P-valueOR(95% CI)Disease onsetEOD19 (25.3)18 (33.3)0.392-LOD56 (74.7)38 (66.7)SexMale37 (49.3)28 (50.0)0.940-Female38 (50.7)28 (50.0)SerotypeIa2 (2.7)18 (32.1)0.00113.327(2.753–64.518)Ib15 (20.0)0 (0)0.998*-V2 (2.7)0 (0)0.998*III56 (74.6)38 (67.9)Ref.EOD early-onset disease, LOD late-onset disease, OR odds ratio, CI confidence interval, Ref. reference*Due to complete separation, Fisher’s exact test was used for serotypes Ib and V Comparison of disease onset, sex, and serotypes distribution during the period of 2013–2017 and 2018–2022 13.327 (2.753–64.518) EOD early-onset disease, LOD late-onset disease, OR odds ratio, CI confidence interval, Ref. reference *Due to complete separation, Fisher’s exact test was used for serotypes Ib and V Fig. 4Annual distribution of capsular serotypes and major sequence types among invasive GBS isolates, 2013-2022. (A) Annual distribution of capsular serotypes. (B) Annual distribution of predominant sequence types (ST17, ST12, ST19, and others). Abbreviations: GBS, group B streptococcus; ST, sequence type Annual distribution of capsular serotypes and major sequence types among invasive GBS isolates, 2013-2022. (A) Annual distribution of capsular serotypes. (B) Annual distribution of predominant sequence types (ST17, ST12, ST19, and others). Abbreviations: GBS, group B streptococcus; ST, sequence type Virulence gene profiles All 131 isolates were screened for the presence of cylE, hylB, Alp family genes, and pilus island genes (Figure S1). Among Alp family genes, rib was the most prevalent (71.8%, 94/131), followed by alphaC (19.8%, 26/131), alp1 (6.9%, 9/131), and alp2/3 (0.8%, 1/131); alp4 was not detected. Three isolates carried no Alp family genes, and two isolates harbored two Alp genes simultaneously. The rib gene was significantly associated with the III/ST17 clone (74.5%, 70/94; P < 0.001, Table.S2). The alphaC was mainly detected in serotype Ia (42.3%, 11/26) and Ib (42.3%, 11/26) isolates, and all serotype Ib/CC10 isolates (n = 9) exclusively carried alphaC (Fig. 3C, S1). All isolates harbored at least one pilus island: PI-2a (32.1%, 42/131) or PI-2b (67.9%, 89/131). PI-1 was detected in 49.6% (n = 65) of isolates (Table 4). Pilus islands’ distribution varied significantly across serotypes and was strongly associated with specific CCs. Specifically, III/CC17 strains exclusively carried PI-1 + PI-2b or PI-2b alone, whereas 76.9% (10/13) of III/CC19 strains carried PI-1 + PI-2a. The PI-1 + PI-2a or PI-2a genotype was present in 80% of serotype Ia (16/20) and Ib (12/15) isolates, as well as both serotype V isolates (n = 2). The most prevalent virulence gene combination was PI-2b-cylE-hylB-rib (35.1%, 46/131) (Figure S1). Table 4Distribution of Pilus islands (Pls)stratified by capsular serotype (n = 131)SerotypesPilus Island (n, %)Total (n)PI-2aPI-2bPI-1 + PI-2aPI-1 + PI-2bIa7, 5.33, 2.39, 6.91, 0.820, 15.3Ib3, 2.31, 0.89, 6.92, 1.515, 11.5III1, 0.851, 38.911, 8.431, 23.794, 71.8V002, 1.502, 1.5Total (n, %)11, 8.455, 4231, 23.734, 26131P*< 0.001< 0.001< 0.001< 0.001Data are presented as number and percentage (%)PI pilus islandP* values were calculated by Fisher’s exact test Table 4Distribution of Pilus islands (Pls)stratified by capsular serotype (n = 131)SerotypesPilus Island (n, %)Total (n)PI-2aPI-2bPI-1 + PI-2aPI-1 + PI-2bIa7, 5.33, 2.39, 6.91, 0.820, 15.3Ib3, 2.31, 0.89, 6.92, 1.515, 11.5III1, 0.851, 38.911, 8.431, 23.794, 71.8V002, 1.502, 1.5Total (n, %)11, 8.455, 4231, 23.734, 26131P*< 0.001< 0.001< 0.001< 0.001Data are presented as number and percentage (%)PI pilus islandP* values were calculated by Fisher’s exact test Distribution of Pilus islands (Pls)stratified by capsular serotype (n = 131) Data are presented as number and percentage (%) PI pilus island P* values were calculated by Fisher’s exact test Discussion GBS is an important cause of sepsis and meningitis in newborns worldwide, with preterm infants at particularly high risk. While IAP effectively reduces the incidence of EOD, it is not fully protective, and rare cases of EOD still occur even among infants born to mothers who received guideline-recommended IAP [21]. Beyond antibiotic resistance, the biofilm-forming capacity of GBS has been reported to be associated with IAP failure and persistent maternal colonization [22]. Invasive GBS infections continue to pose a substantial clinical burden for infants at the study center, with a median of 12.5 cases annually (IQR: 6.5–19.25) during 2013–2022. Consistent with data from Singapore [23], we observed a higher incidence of LOD than EOD (Table 1), a pattern that differs from that reported in South Africa [24], likely reflecting regional variations in IAP implementation and clinical practice [25]. The median age at EOD onset was 1 day (IQR: 1–1), consistent with vertical transmission from colonized mothers and known obstetric risk factors, such as premature rupture of membranes and maternal fever [26]. In contrast, LOD typically has a subacute presentation, with potential transmission routes including postnatal maternal contact (e.g., breast milk), horizontal transmission from household contacts, community acquisition, or nosocomial spread [27]. The Chinese Experts’ Consensus on Prevention of Perinatal GBS Disease was issued in August 2021 [28]. In this study, approximately 86.8% of pregnant women had no GBS screening or no documented results, and two-thirds of available screening records were obtained after August 2021. These findings highlight the necessity to promote standardized prenatal GBS screening and consistent IAP implementation across healthcare facilities in China. The distribution of GBS serotypes and STs varies geographically [29–31]. In this study, the III/ST17 clone (56.5%) was the predominant lineage among infantile GBS isolates, a prevalence higher than the Chinese national average of 40.6% [28]. We confirmed a significant association between LOD and III/ST17 (P < 0.005), consistent with European studies linking CC17 to hypervirulence, neonatal meningitis, enhanced blood-brain barrier penetration, and immune evasion [32–34]. A notable epidemiological shift was observed between 2013–2017 and 2018–2022; the prevalence of serotype Ia increased from 2.7% to 33.3% (P<0.05, OR = 13.327), while serotype Ib decreased from 20.0% to 0% (Table 3). This shift may be related to clonal expansion of fitter serotype Ia lineages, potentially driven by capsular switching [35]. We also identified a significant association between serotype Ia and respiratory distress/failure, consistent with previous studies in southern China linking serotype Ia to pneumonia [17], suggesting the clinical value of molecular serotyping for early risk stratification and targeted management of invasive GBS disease. All GBS isolates in this study remained susceptible to penicillin, ampicillin, and vancomycin, while resistance rates to erythromycin, clindamycin, and tetracycline each exceeded 80%. This susceptibility profile aligns with recent epidemiological data from mainland China, confirming penicillin as the first-line agent and reporting a pooled erythromycin resistance rate of 65.4%, with geographically variable clindamycin resistance (exceeding 70% in northeast China) [36]. High macrolide and lincosamide resistance is a well-documented global trend [37], driven mainly by erm genes (ribosomal target modification) and mefA/E genes (efflux-mediated resistance) [36, 37]. Tetracycline resistance (> 80% nationally) is mainly attributed to tetM and tetO [36, 37]. Multidrug resistance was detected in 66.7% (10/15) of fluoroquinolone-resistant isolates, particularly in III/ST19 (50%, 5/10); 80% (4/5) of these isolates were coresistant to fluoroquinolones and lincosamides, macrolides, or tetracyclines. The underlying mechanisms may involve fluoroquinolone target gene mutations and resistance clusters such as lnu(B), disseminated via integrative conjugative elements or transposons [19, 38]. Vertical transmission is a primary route and an important reservoir for LOD [39, 40]. The hylB (hyaluronidase) and cylE (hemolysin) genes are key virulence determinants that facilitate GBS across of the placental barrier and establishment of neonatal infection [41, 42]. The cylE gene, located within the cyl operon, was originally thought to encode the β-hemolysin/cytolysin toxin but is now recognized as an acyl‑CoA acyltransferase that catalyzes the biosynthesis of the hemolytic glycolipid ornithine rhamnolipid; deletion of cylE abrogates toxin production and results in a non‑hemolytic phenotype [43, 44]. The hylB gene encodes a secreted hyaluronate lyase that degrades hyaluronic acid in the extracellular matrix, thereby disrupting tissue integrity and facilitating bacterial dissemination implicated in ascending uterine infection and adverse pregnancy outcomes, including preterm birth, stillbirth, and fetal injury [41].Both the cylE and hylB genes were detected in all isolates (100%), consistent with reports from Fuzhou, China [45], and the Argentine Pampa region [46]. Given their universal presence and important role in GBS pathogenesis, inhibitors targeting cylE or hylB may represent potential adjunct strategies to conventional IAP for reducing EOD incidence. Despite the implementation of IAP, EOD cases still occur worldwide [47], potentially attributed to biofilm formation in GBS-colonized mothers. Pilus islands (PI-1, PI-2a, and PI-2b) are essential for biofilm formation and GBS virulence [48]. Consistent with previous Chinese reports [49, 50], PI-2b was universally detected in III/ST17 isolates in our study and has been linked to enhanced macrophage internalization and intracellular survival [51]. PI-2a, associated with biofilm formation in chronic infections and antibiotic resistance [52], was carried by > 75% of non-III/ST17 isolates (Table 4 and Figure S1). The III/ST17 lineage exhibit biofilm formation independent of pilus genes, relying instead on capsular polysaccharides, biofilm regulatory proteins, and CsrR-mediated regulation of the adhesin BapB [53, 54]. One EOD case occurred despite maternal IAP; the causative isolate was the serotype III/CC19 strain fully susceptible to penicillin and clindamycin and harboring PI-2a and PI-1, indicating that IAP failure may be associated with biofilm formation in this case. Vaccine development is important for LOD prevention, although no licensed GBS vaccine is currently available. Surface protein genes (Alp family) are important targets for vaccine design. In this study, rib was the most prevalent Alp family gene (71.8%, 94/131) and strongly associated with serotype III, suggesting a potential neurotropic role [55]. A hexavalent capsular polysaccharide-conjugate vaccine targeting serotypes Ia, Ib, and Ⅱ-V would cover all isolates in this study [56]. Alp family protein-based vaccines (e.g., alphaC, rib, alp1, and alp2/3) could offer broader protection [57] and would cover 97.7% of our isolates. Additional strategies, including predictive models integrating clinical and molecular data for highest-risk neonates and biofilm-inhibiting agents, may further reduce the clinical burden of GBS infections, especially LOD. With a median of 12.5 annual cases at our center, LOD complicated by meningitis and neurological sequelae (including epilepsy, speech delay, and autism spectrum disorder in three of our patients) constitutes a substantial clinical burden. Therefore, standardized long-term multidisciplinary neurodevelopmental follow-up and timely rehabilitation are important for infants surviving GBS meningitis. Early identification of subtle developmental disorders, prompt targeted rehabilitation, and individualized early intervention are necessary to minimize long-term neurological sequelae, maximize developmental potential, and improve long-term quality of life. These findings further support the importance of effective prevention, optimized clinical management, and refined risk stratification for neonatal invasive GBS disease. Several limitations should be addressed in this study. First, this was a single-center retrospective study, which may restrict the external validity and generalizability of our conclusions. Second, the limited follow-up duration resulted in insufficient data on long-term neurodevelopmental outcomes of survivors. Third, the exclusion of mild and outpatient GBS cases may introduce selection bias. Finally, although multivariate analysis was performed, unmeasured or residual confounding factors could not be entirely excluded. Conclusion Invasive GBS infections in infants aged ≤ 90 days in Guangzhou were associated with the hypervirulent III/ST17 clone, and serotype Ia was associated with respiratory distress in EOD cases. No substantial shifts in antimicrobial resistance patterns were observed over the 10-year study period, indicating that penicillin and ampicillin were still the first-line agents for GBS prophylaxis and treatment. Our findings highlight the necessity to strengthen prenatal GBS screening and improve IAP implementation in China. Given that cases of IAP failure may be related to biofilm, vaccine development and nonvaccine approaches (e.g., anti-biofilm agents or monoclonal antibodies targeting key virulence factors) should be considered to help reduce the burden of infantile invasive GBS disease.