SDSE isolates were collected using the repository at each institution from two different southern geographical areas (Gyeongsang and Jeolla provinces) in Korea between 2012 and 2016 through two tertiary-care hospitals (Gyeongsang National University Hospital [GNUH] in Gyeongsang and Chonbuk National University Hospital in Jeolla), and from three central regions (Tokyo, Saitama, and Chiba prefectures) in Japan between 2014 and 2016 through either two regional hospitals (Kitasato Medical Center [KMS] in Saitama and Mishuku Hospital supported by Federation of National Public Service and Personnel Mutual Aid Associations in Tokyo) or one laboratory center (Sanritsu Co., Ltd. in Chiba). The isolates with the possibility of true infection were consecutively collected, and repeated isolates from the same patients were excluded. The institutions in both countries sent all the isolates to a laboratory (Gyeongsang National University College of Medicine/Kitasato Institute for Life Sciences) for further genetic analysis. Bacterial identification was conducted using an API-20 Strep or Vitek-2 system (bioMerieux Inc., Marcy l'Etoile, France) and was confirmed by 16S rRNA sequencing. This identification was confirmed if the isolate sequence had ≥98.7% similarity to the 16S rRNA sequence of SDSE type strain NCFB 1356(T). All isolates were stored at −70℃ to −80℃ before being processed for further evaluation. The clinical characteristics, including underlying diseases, infection sites, surgical intervention for treatment, and mortality were reviewed using the patients' medical charts. Infection-associated death was defined as mortality within three weeks of disease onset [11].

Institutional Review Boards at the corresponding institutions (GNUH and KMS) examined and approved this study protocol (approval number GNUH 2015-10-002 and 27-58 from each institution) before starting the investigation.

The emm genotyping and MLST were described in a previous report [15]. Bacterial DNA was extracted by the boiling method (97℃ for 10 minutes) using bacterial suspension in TE buffer. Polymerase chain reaction (PCR) was performed for the emm gene with known primers, and 180 bp DNA sequences of emm genes were submitted to http://www2a.cdc.gov/ncidod/biotech/strepblast.asp to identify the genotype. Accordingly, PCR was conducted for seven housekeeping genes (gki, gtr, murI, mutS, recP, xpt, and atoB) with known primers (https://www.cdc.gov/streplab/alt-mlst-primers.html). Each sequence was amplified from purified DNA by the dye-terminator method. DNA sequences of these genes were submitted to http://pubmlst.org/sdysgalactiae/. Allele types of seven housekeeping genes were profiled to identify sequence types (STs). The STs were grouped into clonal complexes (CCs), in which the connected STs were single-locus variants differing by only one housekeeping gene [16]. Relationships between each ST in both countries were analyzed by eBURST version 3.1 (http://eburst.mlst.net/v3/). The SDSE group G strain D166B (American Type Culture Collection [ATCC] 12394) was used as a quality control.

An antimicrobial susceptibility test was performed using 14 different antimicrobial agents, including β-lactams, tetracyclines, macrolides/lincosamides, and fluoroquinolone, to evaluate antimicrobial resistance levels by broth microdilution test using either the MICroFAST or MicroScan WalkAway System (Beckman Coulter, Inc., Brea, CA, USA) according to the CLSI guidelines [17^,18] with the same ATCC strain (Group G strain D166B) as the quality control. In Japan, tetracycline was changed to minocycline as the screening antimicrobial in 2015. Fifty-six isolates from Japan and all Korean isolates were tested with minocycline.

Antimicrobial resistance determinants, such as erm(A), erm(B), mef(A), tet(M), and tet(O), were examined by agarose gel electrophoresis after PCR amplification of these genes as previously described [19^,20].

Statistical analysis was performed using MedCalc Statistical Software version 17.6 (MedCalc Software bvba, Ostend, Belgium; http://www.medcalc.org; 2017). The Kolmogorov-Smirnov test was used to test normal distribution. To compare two non-normally distributed groups, the Mann-Whitney U test was used. Fisher's exact probability test (two-sided) was used for categorical variables. P values<0.05 were considered statistically significant.

Demographic information and clinical manifestations of SDSE infections in Korea and Japan are shown in Table 1. We observed several significant differences between the two countries. Median age was 69 years in Korea and 76 years in Japan (P=0.0002). Surgical intervention (i.e., debridement) to treat SDSE infections was used more frequently in Korea (37.7%) than in Japan (16.9%, P=0.008). SDSE was more often isolated from pus in Korea (P=0.00005), whereas it was more frequently isolated from the sputum/throat in Japan (P=0.001). SDSE often caused infections other than sepsis, cellulitis, and septic arthritis in Japan (P=0.002). Despite these differences, mortality rates were not different between the two countries (5.8% vs 9.9%, P=0.53).

The STs and emm genotypes of SDSE isolates from Korea and Japan are shown in Tables 2 and 3, respectively. Among the 69 Korean isolates, there were 20 different combinations of ST-emm types. ST127-stG245 (N=16), ST128-stG485 (N=10), ST138-stG652 (N=8), and ST146-stG6 (N=7) were the most common types. Among the 25 blood-derived isolates, six, six, and four isolates were ST127-stG245, ST128-stG485, and ST146-stG6, respectively. A novel ST295 (allele profile, 3-3-2-2-9-22-2) was identified. The ST-emm types for four mortality cases were ST29-stC74a, ST148-stG166b, ST128-stG485, and ST138-stG652. All four strains were isolated from blood.

Among the 71 Japanese isolates, there were 29 different combinations of ST-emm types. ST17-stG6792 (N=11), ST29-stG485 (N=7), and ST205-stG6792 (N=6) were the most frequent types. Among the 36 blood-derived isolates, ST17-stG6792 and ST127-stG245 represented nine and three isolates, respectively. Combinations of ST-emm were more diverse for Japanese blood isolates. Seven mortality cases possessed three ST17-stG6792 and each of ST205-stG6792, ST29-stG485, ST37-stG485, and ST25-stG5420. Three strains were isolated from blood, two from open pus, one from cerebrospinal fluid, and one from sputum.

The distribution of the common emm types was quite different between the two countries. The stG245 (24.6%), stG6 (15.9%), and stG485 (14.5%) were most common in Korea, while stG6792 (29.6%), stG485 (14.1%), and stG245 (12.7%) were most frequent in Japan. The ST-relationships consisting of several CCs were observed among the Japanese isolates only. Among the Korean isolates, only CC25 was observed, whereas the Japanese isolates indicated a variety of CCs (i.e., 17, 37, 25, 15, and 8) (Figs. 1, 2).

There was no statistically significant difference in antimicrobial resistance rates between the two countries (P>0.05).

Resistance rates to erythromycin, azithromycin, clindamycin, and minocycline were 34.8%, 34.8%, 17.4%, and 30.4%, respectively, for Korean isolates (Table 2). Several ST-emm types were associated with macrolide/lincosamide and/or tetracycline resistance. Among the 24 erythromycin-resistant strains (MIC≥1 µg/mL), 12 were ST127-stG245 and six were ST128-stG485 (Table 2). Antimicrobial resistance determinants of erm(A), erm(B), and mef(A), comprised three, 14, and six, respectively. All the 12 isolates resistant to clindamycin (MIC≥1 µg/mL) contained the erm(B) gene; 10 were ST127-stG245. Among the 21 minocycline-resistant strains (MIC>4 µg/mL), seven were ST138-stG652, six were ST127-stG245, and six were ST128-stG485. Antimicrobial resistance genes of tet(M) and tet(O) comprised 20 and one, respectively. Nine strains (42.9%) resistant to tetracycline also possessed macrolide resistance determinants.

Resistance rates to erythromycin, azithromycin, clarithromycin, clindamycin, tetracycline, and minocycline were 28.2% (No. of examined isolates=71), 32.1% (N=56), 13.3% (N=15), 14.1% (N=71), 26.7% (N=15), and 21.4% (N=56), respectively, for Japanese isolates (Table 3).

Among the 20 erythromycin-resistant strains, stC36 and stG485 comprised four each (Table 3). Antimicrobial determinants of erm(A), erm(B), and mef(A) comprised 11, eight, and two isolates, respectively. There was a significant difference in the frequency of erm(A) (P=0.009) between the two countries, whereas other genes showed no significant frequency differences.

Among the 10 isolates resistant to clindamycin, erm(B) and erm(A) comprised eight and two respectively; there were three of stC36, two of stG245, two of stG840, and two of stG653. Among the 12 minocycline-resistant strains out of 56 tested, five were stG10 and three were stG245. All strains contained tet(M) except two strains with tet(O). Five strains (41.7%) resistant to tetracycline also possessed macrolide resistance determinants. One strain (ST274-stG245) isolated from blood concurrently contained erm(B), tet(M), and tet(O) genes.

The 50% minimum inhibitory concentrations (MIC₅₀) of erythromycin, clindamycin, and minocycline were ≤0.12 µg/mL, ≤0.12 µg/mL, and 0.5 µg/mL, and the MIC₉₀ of these antimicrobial agents were >2 µg/mL, >1 µg/mL, and >4 µg/mL, respectively, for both Korean and Japanese isolates. Fluoroquinolone-resistant isolates (N=5) were identified only in Japan. None of the tested SDSE isolates from either country were resistant to β-lactams.