We identified iGAS cases through ABCs, a laboratory- and population-based surveillance for severe bacterial infections currently implemented in 10 US states [13]. The ABCs case definition for iGAS disease was illness with isolation of GAS from a normally sterile site or isolation of GAS from a wound culture and accompanied by necrotizing fasciitis or streptococcal toxic shock syndrome. WGS of iGAS isolates was performed at the CDC Streptococcus Laboratory as previously described [14]. We excluded isolates with whole-genome assemblies that contained <1.5 M total bases or >150 contigs from subsequent analysis as a sequencing quality control measure. EryNS and CliNS were defined according to the Clinical and Laboratory Standards Institute MIC breakpoints [2].

Pair-wise single-nucleotide polymorphism (SNP) distances among all iGAS isolates belonging to the same emm type were calculated using the MUMmer package as previously described [10]. An isolate was defined as clustered if it differed from another isolate by ≤10 SNPs per 1.5 Mb of the 2 aligned genomes. A core-genome phylogenetic tree was constructed using Parsnp software [15] and annotated using the ggtree R package.

Proportions of EryNS or CliNS iGAS isolates were calculated. Equal group proportions were assessed using the Fisher exact test. The χ² test for trend in proportions (trend test) was used to evaluate proportion trends over time. All P values were 2-sided, and a P value <.05 was considered statistically significant. All analyses were performed using R software version 3.4.3.

ABCs case reporting and isolate collection were considered to be public health surveillance activities that were exempt from institutional review by the Centers for Disease Control and Prevention. Informed consent was not required.

A total of 4369 iGAS isolates identified in 2018–2019 and 5261 isolates identified in 2015–2017 were included in this study, accounting for approximately 85% of all iGAS infection cases reported to the ABCs during the 5-year period. Compared with years 2015–2017, the proportion of EryNS isolates increased significantly from 18.1% (953 of 5261) to 25.0% (1091 of 4369) in 2018–2019 (P < .001; Figure 1A). A similar increase was observed for the proportion of CliNS isolates (17.0% vs 24.2%; P < .001; Figure 1A). Virtually all CliNS isolates were EryNS due to shared resistance mechanisms. Among all of the 2044 EryNS isolates in 2015–2019, 2034 (99.5%) had 1 of the 4 major macrolide/lincosamide resistance mechanisms (ermB, ermT, ermTR, or mef genes). From 2015 to 2019, ermB⁺, ermT⁺, and ermTR⁺ isolates, which were both EryNS and CliNS, each showed a significant increase in proportion among all iGAS isolates (P = .01, P < .0001, and, P < .0001, respectively; Figure 1B). In contrast, the proportion of mef⁺ isolates, which were EryNS but clindamycin susceptible, did not show a significant change in the same period (P > .05; Figure 1B). The ermT⁺ isolates increased sharply in 2017–2018, mainly due to the fast expansion of emm92, which was a predominantly ermT⁺ strain.

The cluster analysis of 2015–2019 iGAS isolates (n = 9630) identified 5576 (59.9%) clustered isolates. The proportion of clustered isolates among EryNS isolates (63.6%, 1301 of 2045) was slightly but significantly higher than that among susceptible isolates (58.9%, 4465 of 7585; P < .0001). To determine the contribution of clustered isolates to the increased resistance among iGAS, we examined the proportions of EryNS isolates that were either clustered or nonclustered among all iGAS isolates (Figure 1C). The proportion of clustered EryNS isolates increased significantly from 10.5% (553 of 5261) in 2015–2017 to 17.1% (747 of 4369) in 2018–2019 (P < 0.0001; Figure 1C). In contrast, the proportion of nonclustered EryNS isolates remained comparable between the 2 periods (7.6% vs 7.8%; P = .65; Figure 1C). Consistently, the trend test for proportion of clustered EryNS isolates from 2015 to 2019 showed a significant increase (P < .0001), while no significant trend was found for proportion of nonclustered EryNS isolates (P = .60). For CliNS isolates in 2015–2019, a similar pattern of expansion in the clustered, but not the nonclustered, CliNS isolates was also observed (Figure 1D). In 2018–2019, patients who were PEH or PWID (n = 496) were significantly associated with clusters (odds ratio [OR] = 3.1, P < .001), EryNS (OR = 1.4, P < .001), and CliNS (OR = 1.5, P < 0.001), consistent with results in previous years.

In 2018–2019, 65% EryNS isolates (711 of 1091) belonged to the 6 emm types that already had >50% EryNS isolates within the emm type in 2015–2017, including emm types 92, 11, 83, 169, 58, and 94. A notable exception was observed in emm49. In 2015–2017, all emm49 isolates (368 of 368) were susceptible to both erythromycin and clindamycin (Supplementary Table 1). However, the proportion of EryNS and CliNS isolates within emm49 increased to 8.3% (12 of 144) in 2018 and further increased to 31.7% (32 of 101) in 2019 due to the emergence of ermTR⁺ (n = 43) and ermB⁺ (n = 1) isolates (Supplementary Figure 1A). Phylogenetic analysis of all emm49 isolates indicated that 38 of the 44 EryNS and CliNS isolates belonged to a single ermTR⁺ sublineage detected exclusively in Maryland during 2018–2019 (designated M49_MD; Supplementary Figure 1B). Within the M49_MD sublineage, median pairwise distance among the 38 isolates was 3 SNPs (range, 0–7), suggesting a single genomically closely related cluster that emerged recently. In Maryland, the M49_MD sublineage was first identified in August 2018, with 37 additional invasive infections by the same subclone occurring within the next 15 months. By the fourth quarter of 2019, M49_MD had become the dominant emm49 sublineage in Maryland, accounting for 71% (10 of 14) of emm49 iGAS isolates identified in that state.

In this study, we found that combined macrolide and clindamycin resistance among iGAS isolates remained substantial in 2018–2019 after a continued increase since 2017. The high prevalence of resistance appeared to result from both expansion of predominantly resistant emm types (eg, emm types 92, 11, and 83) and emergence of new resistant sublineages among previously susceptible emm types (eg, emm49). Importantly, nearly all increases of EryNS and CliNS could be accounted for by genomic clusters of resistant isolates, suggesting a prominent role of temporally and genomically related iGAS infections in facilitating the spread of antimicrobial resistance. Genomic clusters were more frequently observed in certain populations including PWH and PWID [10, 11]. These disadvantaged populations also showed a higher risk of EryNS and CliNS infections [4]. Thus, targeted control of clustered iGAS infections, particularly among disadvantaged populations, could help in containing antimicrobial resistance.

The study results also highlight the impact of the fast evolutionary potential of GAS. In this case, the acquisition of ermTR genes resulted in the emergence of an EryNS and CliNS emm49 sublineage that differed from the susceptible emm49 ancestor by only a few core-genome SNPs. The acquisition event might have increased the fitness advantage of this GAS lineage in the presence of antibiotics, which could in turn facilitate the evolution of further potentially advantageous mutations by allowing continued bacterial replication in antibiotic-treated host individuals. On a more immediate note, the study findings suggest that caution is warranted when using macrolides and clindamycin to treat GAS infections in the United States, given the continued increase of EryNS and CliNS among clinical isolates.

The study had several limitations. The focus on invasive GAS alone provides limited information on the resistance trend in noninvasive infections that could have significant community impacts with respect to antimicrobial selection and to sequelae of incompletely treated streptococcal pharyngitis. The retrospective nature of the study could also limit the data available for variables such as clinical outcomes for those with resistant isolates. Future studies are needed to evaluate interventions that interrupt the development of resistant clusters and to inform guidance development regarding treatment of streptococcal infections in penicillin-allergic patients.