GBS isolates were sent to the National Reference Center by correspondents located throughout the national territory on a voluntary basis. Only invasive infections, such as GBS isolated from a normally sterile site, were considered for this study. A total of 1,262 neonatal invasive infections (EOD, n = 394, 31%; LOD, n = 868, 69%) were reported during 2007–2019. The annual number of cases increased significantly over time as a result of a marked rise in LOD cases since 2013 (Appendix Figure 1). Bacteremia without focus was the main clinical presentation during both EOD and LOD (Table 1). Meningitis represented a frequent complication and was more common in LOD, in which it affected nearly half of infants (p<0.0001; Table 1). The proportion of meningitis during LOD dropped significantly over time, from 69% (95% CI 51%–83%) in 2007 to 33% (95% CI 25%–43%) in 2019 (p for trend = 0.008; Appendix Figure 2). The French recommendations for lumbar puncture indication in neonates did not change during the study period. This observation, together with the increased annual number of cases, suggests a better reporting of bacteremia and a better representativeness of our collection over time.

Molecular capsular typing of the 1,262 GBS isolates was performed (11) (Table 1). Serotype III was overrepresented, especially in LOD, accounting for 57% (95% CI 52%–62%; n = 223/394) of EOD cases and 82% (95% CI 79%–84%; n = 712/868) of LOD cases. Identification of the hypervirulent CC17 GBS, a highly homogenous CC that includes the sequence type (ST) 17, was performed using a specific PCR (12) and showed that it caused 66% (95% CI 63%–68%; n = 827/1,262) of GBS neonatal invasive disease. CC17 GBS prevalence was particularly overwhelming in LOD (74%, 95% CI 71%–77%) compared with EOD (48%, 95% CI 43%–53%; p<0.0001) and, during EOD, in cases of meningitis compared with bacteremia (68%, 95% CI 59%–77% vs. 41%, 95% CI 36%–47%; p<0.0001). Furthermore, CC17 GBS prevalence increased by »50% over the study period, rising from 53% (95% CI 40%–65%) in 2007 to 76% (95% CI 68%–82%) in 2019 (p for trend = 0.0001; Figure 1). This evolution was linked with its prevalence in LOD, which gradually increased from 59% (95% CI 41%–75%) to 85% (95% CI 77%–91%) of the cases during 2007–2019 (p for trend = 0.025).

We determined the susceptibility of the 1,262 GBS isolates to antimicrobial drugs and performed the detection of resistance genes as previously described (13). All isolates were susceptible to penicillin, amoxicillin, and vancomycin. Resistance to tetracyclines did not vary through the study period and concerned 91% (95% CI 89%–92%) of the strains, owing to the genetic determinant tet(M) in 92% of the cases (data not shown). Only 3 isolates (0.2%, 95% CI 0.1%–0.7%) showed high-level resistance to gentamicin, but high-level resistance to amikacin increased from 0% (95% CI 0%–7%) in 2007 to 18% (95% CI 12%–26%) in 2019 (p for trend <0.0001; Table 2). Similarly, resistance to erythromycin increased from 22% (95% CI 13%–34%) to 30% (95% CI 23%–38%; p for trend = 0.019). Resistance to erythromycin was mostly the result of modifications of the ribosomes that confer cross resistance to lincosamides and are encoded by the genetic determinants erm(B) (64%), erm(A/TR) (13%), or erm(T) (1%), and in 22% of the cases were the result of an efflux mechanism encoded by the genetic determinant mef.

Next, we specifically investigated CC17 GBS resistance to erythromycin and amikacin and found an increase over the study period from 17% (95% CI 8%–35%) to 29% (95% CI 21%–38%; p for trend = 0.0042) for erythromycin resistance and from 0% (95% CI 0%–11%) to 14% (95% CI 9%–21%; p for trend <0.0001) for amikacin resistance (Table 2). We postulated that these evolutionary trends were attributable to the emergence of the MDR CC17 GBS sublineage, which exhibits resistance to tetracyclines, macrolides, lincosamides, and amikacin as a result of the replacement of the pilus island 1 genetic locus by mobile genetic elements carrying the resistance determinants tet(O), erm(B), and aphA-3 (8,9). The proportion of CC17 GBS harboring tet(O), erm(B), and aphA-3 among neonatal GBS isolates increased from 0% (95% CI 0%–6%) in 2007 to 14% (95% CI 9%–21%) in 2019 (p for trend <0.0001; Figure 2). Whole-genome sequencing of 8 of these MDR CC17 GBS (Appendix Table) confirmed the replacement of the pilus island 1 locus by large integrative and conjugative elements (ICEs) similar to those previously described in China and Canada (8,9). Interrogation of the ICEberg database (http://db-mml.sjtu.edu.cn/ICEberg/) showed that these ICEs displayed the highest sequence similarity (92%–98%; Appendix Figure 3), with the GBS ICESag37 described in a CC10 isolate responsible for a neonatal bacteremia in China (14).

We analyzed a total of 1,262 neonatal invasive infections over a 13-year study period in France, which represents »30% of the total national estimated cases (4). A selection bias toward the more severe cases cannot be excluded. However, the proportions of EOD and LOD and the associated clinical manifestations described here are very close to the national estimations. Thus, we can assume that our study reflects the national epidemiology without major discrepancies.

We observed a higher reporting of LOD in contrast to EOD over the 13-year study period. This trend mirrors the data from the surveillance network in France, which show a continuous increase in LOD incidence with an overall 65% rise over the past 20 years (4). We describe a growing prevalence of the hypervirulent CC17 GBS and of its MDR sublineage in LOD, which might account for the increasing incidence of this syndrome. Whether these trends are the result of a higher tropism of the MDR sublineage for neonatal infections or merely of its selection and clonal expansion as a result of antibiotic selection pressure requires further investigation. Given the worldwide expanding burden of GBS LOD, the adaptability of GBS to its environment through horizontal gene transfer (15), and the resulting potential reduction of the therapeutic arsenal against this major neonatal pathogen, our results reinforce the need for a continued surveillance of GBS diseases and for the development of alternative preventive strategies.

Additional information about multidrug-resistant hypervirulent group B Streptococcus in neonatal invasive infections in France.