Study Setting

The clinical laboratory at J.W. Ruby Memorial Hospital in Morgantown serves as the primary reference facility for all 19 WVUMed hospitals located throughout West Virginia, as well as facilities in western Maryland, southwestern Pennsylvania, and eastern Ohio. The WVUMed system serves an estimated patient population of 1.2 million. Most microbiological testing at J.W. Ruby Memorial Hospital and surrounding WVUMed outpatient clinics is performed by the Ruby clinical laboratory, as is referral antimicrobial susceptibility testing of many Streptococcus spp. isolates. The clinical laboratory routinely banks invasive isolates at −80^ᵒC for 1–2 years. Noninvasive isolates, including those recovered from pharyngitis cases, are not routinely held >7 days after specimen submission.

The strain collection for this study included all viable primary and referred iGAS isolates available from the freezer bank, which spanned the period January 2020–June 2021. After approval by the hospital’s Institutional Review Board (protocol no. 2202533507), we reviewed patient records to capture demographic and clinical information, such as patient age, sex, residence status, history of IVDU, intensive care unit admission requirement, number of surgical interventions, antimicrobial regimen, and clinical outcome (Appendix Table 1), for all isolates successfully retrieved. We included 2–4 replicate isolates recovered serially from 8 patients and tested them separately as a quality control measure. In all instances, intrapatient phenotypic and genotypic results were consistent, so isolates are reported per patient throughout.

Chromosomal and Plasmid DNA Isolation

We isolated genomic DNA from a 10-μL loopful of bacteria grown in Todd Hewitt broth by using the DNA extraction procedure, as described previously (14). We isolated plasmid DNA by using Gene JET Plasmid Miniprep Kit (ThermoFisher Scientific) with an additional cell-digestion step (1 mg/mL lysozyme and 0.5 U/µL mutanolysin) at 37°C for 1 hour. We analyzed plasmid DNA, uncut and digested with SwaI, on a 0.8% agarose gel for confirmation of plasmid pRW35 size in emm92 isolates (15).

Identification of Resistance Genes erm/mef and emm Typing

We used plasmid DNA as a PCR template with the ermT-specific primers, whereas we used genomic DNA as the PCR template with primers detecting ermA(TR), ermB(AM), and mefA genes (Appendix Table 2). We obtained control GAS strains harboring the corresponding erm and mef genes from the CDC Streptococcus Laboratory (https://www.cdc.gov/streplab/index.html). We used genomic DNA to determine isolate emm type by Sanger sequencing of amplicons generated with primers emm1b and emm2 (16), followed by BLAST search (https://blast.ncbi.nlm.nih.gov/Blast.cgi) on the CDC Streptococcus Laboratory.

Antimicrobial Susceptibility Testing

We performed erythromycin, clindamycin, and tetracycline susceptibility testing in the clinical microbiology laboratory by using methods described by the Clinical Laboratory Standards Institute (CLSI) (17). All automated testing used Vitek 2 ST-02 cards (bioMériuex) and was performed upon isolate recovery for clinical management purposes. All valid results were reported to and retrieved from patient electronic health records (Epic). All historic testing met ongoing quality control criteria as outlined in the laboratory Quality Management Plan and Individualized Quality Control Plan as required for laboratory accreditation. Subsequently, we performed disc diffusion and D-testing over multiple days using thawed isolates from the freezer bank. After 2 serial propagations on sheep blood agar, we inoculated swabs of 0.5 MacFarland suspensions for confluent growth on cation-adjusted Mueller Hinton agar with 5% sheep blood (BD) using discs containing conventional drug masses. Quality control organisms, including ATCC BAA-977, ATCC BAA-976, and ATCC 49619, were tested in parallel each day of use. We incubated plates at 35^ᵒC in 5% CO₂ environment for 20–24 h before measuring zones of inhibition with a manual caliper in reflected light. We interpreted zone diameters by using CLSI clinical breakpoints (17) and interpreted any degree of clindamycin zone flattening in proximity to erythromycin disc as a positive D-test result.

Susceptibility testing against aminoglycosides (gentamicin, kanamycin, and streptomycin) was performed by agar dilution on Mueller Hinton media (BD) prepared in the research laboratory. A saline suspension of each isolate at an absorbance of 1 Klett unit was prepared and a 10-µL drop (≈10⁴ CFU) was plated in singlicate onto agar medium containing arbitrarily selected concentrations ranging from 50 to 500 μg/mL, as described (5). Plates were incubated at 37°C in 5% CO₂ overnight, followed by observation of growth results.

Patients

We included 76 GAS isolates collected during January 2020–June 2021 from 66 patients with invasive infections (Figure 1; Appendix Table 1). Median patient age was 42 (mean 45, range 23–86) years; 59% were men. On the basis of addresses listed in medical records, geographic distribution of all 56 in-state patients spanned 20 of the 55 West Virginia counties; 3 northern counties (Harrison, Marion, and Monongalia) accounted for the highest proportion (53%), likely because of larger populations and proximity to the main WVUMed campus in Morgantown. Most out-of-state patients were also within the WVUMed catchment area in neighboring counties in Maryland (n = 4) and Pennsylvania (n = 5) (Figure 1), although 1 patient was visiting from the Midwest United States. For 9 patients, details of housing status in their medical records were insufficient; among the remaining patients, 9 (16%) were reported by a case worker to be experiencing homelessness at the time of culture, despite having an address on file in their medical records. For 64 patients whose social history was sufficiently documented, 39 (61%) reported recent or remote IVDU.

Assessing infection source for all 66 patients collectively revealed 38 (59%) with skin and soft tissue infections (SSTI) of the extremities: 8 (12%) with infections of deep neck structures; 6 (9%) with endovascular sources; 4 (6%) with SSTI of the gluteal, perianal, sacral, or inguinal region; 4 (6%) with respiratory sources; 1 (2%) with ocular source; and 4 (6%) with bloodstream infections of unknown origin (Figure 2). For an additional patient with a history of IVDU who had thrombophlebitis, bacteremia, and multiple SSTIs of the extremities and gluteal region, the primary source could not be discerned. A total of 35 (53%) patients required surgical intervention: single-stage debridement/washout (n = 24), fasciotomy with 2–13 serial debridements (n = 7), below-the-knee amputation revision (n = 1), tricuspid valve replacement (n = 1), thoracotomy with pleural decortication (n=1), and vascular thrombectomy (n = 1) (Appendix Table 1). Of the remaining nonsurgical patients, 8 underwent incision-drainage procedures or chest tube placement at bedside. Overall, 18 (27%) patients required admission to the intensive care unit for >24 hours and at least 5 (7.5%) died as a result of iGAS infection, although records of follow-up care were incomplete for a substantial portion of patients.

Antimicrobial therapy varied considerably by patient acuity, duration of hospitalization, intravenous catheter availability, and degree of initial treatment response. A total of 51 patients (77%) received >1 dose of intravenous vancomycin or daptomycin during hospitalization, whereas β-lactams (amoxicillin/clavulanate, cephems, penicillin, or a combination) or clindamycin were used exclusively for 10 patients (15%). Another 3 patients who were not admitted to the hospital received no antimicrobial therapy. In total, the acuity of infection for 9 patients did not require hospital admission. Another 10 patients left the hospital against medical advice before being admitted or completing treatment, 9 of whom had histories of IVDU. All these patients received prescriptions for oral clindamycin or amoxicillin/clavulanate at discharge. Among the remaining 47 patients, median hospital length of stay was 7 (mean 14, range 1–60) days.

emm Types

Although surveillance across the United States by ABCs has demonstrated an increase in iGAS, categorization of iGAS isolates in West Virginia was lacking. The M protein type of each isolate was determined by Sanger sequencing of the 5′ end of the emm-gene PCR product (Appendix Table 2), as described (16). Analysis showed the collection was predominated by isolates of 1 emm type; of the 66 unique patient isolates, 35 (53.0%) were emm92 followed in decreasing proportion by emm types emm11 (n = 8, 12.1%) and emm89 (n = 5, 7.7%). (Table 1; Figure 3, panel A). Temporal analysis of isolate emm type recovery by 3-month periods showed an overall increase in isolates during April–June in 2020 and 2021; a substantial proportion of the isolates from all quarters were emm type 92. Although the presence of emm11 and emm89 isolates was relatively stable over time, the presence of emm92 trended upward. Of note, the collection contained only 2 emm1 isolates, 1 each of emm12 and emm28, and no emm3 isolates, which historically have been correlated with a high incidence of iGAS infections (Figure 3, panel A) (1,18,19). Although the data from this 1.5-year study period in West Virginia is less robust than ABCs national data, the findings do corroborate a continual shift in emm types responsible for iGAS disease, particularly among homeless populations and persons with a history of IVDU (11,12).

MLS_B Susceptibility and Resistance Profiles

Next, we assessed antimicrobial resistance among isolates in our iGAS collection. In aggregate, 76% (50/66) of isolates were resistant to erythromycin (Table 1), which is considerably higher than the percentage reported in a larger collection (11). Aside from emm77, which included 1 erythromycin-resistant isolate and 1 erythromycin-susceptible isolate, all other emm types exclusively harbored either erythromycin-resistant or erythromycin-susceptible phenotypes (Table 1; Figure 3, panel B). We used disc diffusion and D-testing to assess clindamycin susceptibility and to determine whether resistance was constitutive or inducible (Figure 3, panel C). Clindamycin susceptibility mirrored that of erythromycin; 16 erythromycin-susceptible isolates also demonstrated clindamycin susceptibility. Of the 50 erythromycin-resistant isolates, 40 exhibited inducible clindamycin resistance (i.e., not detectable without erythromycin induction), 9 demonstrated constitutive clindamycin resistance, and 1 isolate (emm22) was clindamycin susceptible without evidence of inhibition zone flattening. Similar to erythromycin, most emm types were uniformly susceptible or resistant to clindamycin except for 2 emm77 (1 susceptible and 1 inducible-resistant isolate) (Table 1; Figure 3, panel C). Phenotypic heterogeneity was also noted among emm92 isolates (of which 4 exhibited constitutive clindamycin resistance and 31 exhibited inducible clindamycin resistance), as well as among emm11 isolates (of which 5 isolates produced inducible and 3 produced constitutive phenotypes) (Table 1).

Detection of Erythromycin-Resistance Determinants

We tested isolates for common erythromycin resistance genes by PCR amplification to detect the presence of the methyl transferase genes ermA, ermB, and ermT, as well as mefA, a gene-encoding protein associated with an efflux pump (Appendix Table 2). On the basis of a previous ABCs report that ermT in emm92 GAS was carried on the pRW35 plasmid (15), extrachromosomal DNA was isolated from resistant isolates of various emm types, yet only emm92 isolates harbored plasmid DNA (Figure 4, panel A). Restriction digestion targeting a conserved SwaI site confirmed a ≈4.9-kb size of this pRW35-like plasmid (data not shown). All 35 emm92 isolates were resistant to erythromycin (Table 1) and contained ermT detected by PCR (Figure 4, panel B).

Chromosomal DNA was used for the detection of ermA/B (Figure 4, panels C, D) and mef (data not shown) genes. Erythromycin-resistance genes identified in emm11 isolates varied; 6 carried ermB and 2 harbored ermA (Table 1). Of the remaining 7 erythromycin-resistant isolates of various emm types, ermA was detected in 5, ermB in 1, and mefA in 1. Collectively, 86% of ermA-containing isolates (6 of 7) showed inducible clindamycin resistance, whereas isolates containing ermB had a more evenly split phenotype for clindamycin resistance (3 iMLS_B vs. 4 cMLS_B) (Table 1). The mefA gene, which encodes a component of the mefA-msrD efflux pump, was detected in a single emm22 isolate and corresponded to the erythromycin-resistant, clindamycin-susceptible phenotype referred to as the M phenotype (20).

Additional Susceptibility and Proposed Resistance Determinants

Isolates also underwent susceptibility testing for tetracycline by disc diffusion, as well as for the aminoglycosides (gentamicin, streptomycin, and kanamycin) by agar dilution method using concentration ranges as previously described (5). In aggregate, 76% of isolates were resistant to both tetracycline and erythromycin, whereas the single emm87 isolate was erythromycin sensitive but tetracycline resistant. For aminoglycosides, all 66 isolates were susceptible to gentamicin using CLSI Staphylococcus aureus breakpoints, whereas we observed presumed resistance to kanamycin (MIC >500 µg/mL) in 89% of tested isolates and resistance to streptomycin (MIC >500 µg/mL) in 61% of isolates. In addition to their universal plasmid-encoded MLS_B phenotype, all emm92 strains in this collection were uniformly resistant to tetracycline, kanamycin, and streptomycin, presumably encoded by the ICESpyM92 mobile element (Table 2) (5). The remaining isolates of various emm types with MLS_B phenotype demonstrated resistance to either kanamycin or tetracycline.