Spread of ceftriaxone non-susceptible pneumococci in South Korea: Long-term care facilities as a potential reservoir

Despite the availability of a pneumococcal National Immunization Program, which provides free PPSV23 vaccination for older adults aged ≥65 years in South Korea, pneumococcal pneumonia remains one of the most common respiratory infections, with increasing antimicrobial resistance. From January to December in 2015, all pneumococcal isolates were collected from a 1,050-bed teaching hospital in South Korea. All isolates were analyzed for serotype, genotype, and antimicrobial susceptibility. Demographic, clinical and microbiological data were compared between ceftriaxone susceptible and non-susceptible cases. Among 92 microbiologically identified pneumococcal isolates, ceftriaxone non-susceptible pneumococci (CNSP) accounted for 32 cases (34.8%). Some of these cases also showed levofloxacin resistance (25%, 8/32 isolates) and all CNSP cases were multidrug resistant. Compared to patients with ceftriaxone susceptible pneumococci (CSP), long-term care facility residents (odds ratio [OR] 7.0, 95% confidence interval [CI] 0.8–62.1) and patients with chronic lung (OR 4.1, 95% CI 1.1–15.0) and renal diseases (OR 9.1, 95% CI 1.2–70.5) were more common among those with CNSP on multivariate analysis. PPSV23-unique serotypes not included in PCV13 were more common in CNSP than in CSP (34.4% versus 13.3%, p = 0.02). Regarding genotypes, ST320 (10 cases), ST166 (7 cases) and ST8279 (3 cases) were dominant in CNSP, and ST8279 was only detected in previous long-term care facility residents. Clonal expansion and spread of CNSP strains should be monitored among patients with chronic lung/renal diseases and residents of long-term care facilities.


Introduction
Streptococcus pneumoniae causes a broad spectrum of infectious diseases, ranging from mild upper respiratory illnesses to severe invasive diseases [1,2]. Pneumonia and invasive pneumococcal disease (IPD) are the leading causes of morbidity and mortality after pneumococcal infection [2]. In South Korea, from 2011 to 2014, the estimated annual incidences per 100,000 persons were 44.3 and 4.8 cases for pneumococcal pneumonia and IPD, respectively [3]. The

Case definition
Ceftriaxone susceptible pneumococci (CSP) and CNSP were defined as isolates with ceftriaxone minimum inhibitory concentrations (MICs) of <2 μg/mL and �2 μg/mL, respectively. MDR was defined as non-susceptibility to three or more antimicrobial classes: β-lactam (penicillin or ceftriaxone), levofloxacin, erythromycin, linezolid, vancomycin, tetracycline, and trimethoprim-sulfamethoxazole. Extensive drug resistance (XDR) was defined as non-susceptibility to at least one agent in all antibacterial drug categories except vancomycin and linezolid. True infection was defined as any isolates from sterile fluid or respiratory isolates with clinical manifestation of pneumonia, which was diagnosed as follows: appearance of new radiological infiltrates and one or more relevant signs and symptoms: fever (temperature >38.2˚C) or hypothermia (temperature <36.0˚C); presence of purulent secretions; leukocytosis (>10,000 leukocytes/ mm 3 ) or leukopenia (<5,000 leukocytes/mm 3 ). Colonization was defined as respiratory samples without clinical manifestations of pneumonia. Five-day treatment response was determined as improvement of all baseline symptoms, signs, and radiological features within 5 days. Performance status was evaluated using the Karnofsky Performance Status Scale [19].

Serotyping, genotyping and susceptibility tests
The serotypes of all S. pneumoniae were determined with the capsular swelling (Quellung reaction) test using commercial pool and type/group-specific antisera (Statens Serum Institut, Copenhagen, Denmark) or the multiplex polymerase chain reaction assay recommended at www.cdc.gov/ncidod/biotech/strep/pcr.htm. MICs of ceftriaxone and other antimicrobial agents were determined using the broth microdilution method according to the guidelines of the Clinical and Laboratory Standard Institute (CLSI) [20]. S. pneumoniae ATCC 49619 was included as the quality control.

Statistical analysis
We performed descriptive analyses and compared demographics and clinical characteristics between CSP and CNSP groups. To calculate ceftriaxone non-susceptibility and relate this to patient-and pathogen-specific factors, isolates were classified into two groups according to ceftriaxone susceptibility. The chi-square or Fisher's exact test was used to compare the proportions of categorical variables between the two groups, and Student's t-test was used to compare continuous variables between the groups. Clinically relevant factors and factors showing significant inter-group differences by univariate analysis were included in a multivariate logistic regression analysis. Data were analyzed using SPSS version 20.0 (SPSS Inc., Chicago, IL, USA). P values of <0.05 were considered statistically significant.

Results
A total of 92 non-duplicate S. pneumoniae were isolated. The mean patient age was 66.1 (range 15-91) years and 72 (78.3%) patients were men (Table 1). In terms of their origins, seven (7.6%) isolates were from blood or other sterile fluid, while the remaining isolates (92.4%) were from respiratory samples. Among these, CNSP constituted 32 cases (34.8%). CNSP showed high levofloxacin resistance (25%, 8/32 cases) and MDR (100%, 32/32 cases) based on the 2015 CLSI breakpoints (Table 2). MIC 50 and MIC 90 values of antimicrobial agents were higher in the CNSP group than in the CSP group.
In the ceftriaxone non-susceptible group, chronic lung disease referred to COPD (n = 7), asthma (n = 3), bronchiectasis (n = 2), and emphysema (n = 1). neither finding reached statistical significance in multivariate analysis. In the multivariate analysis, chronic renal disease, chronic pulmonary disease, and vaccine serotype pneumococcal isolates were statistically significant risk factors for CNSP acquisition.

Factors associated with 5-day treatment response among cases of true infection
Of 92 total isolates, 73 isolates were from true infection cases. Among these, four cases were excluded because treatment response could not be evaluated owing to follow-up loss (2 cases) and patient transfer (2 cases). Among 69 cases, 68 cases were pneumonia and one case was peritonitis; 48 (69.6%) of these cases showed clinical improvement within 5 days. Compared to non-responders within 5 days, responders were less severe in terms of qSOFA, CURB-65, and PSI score, although no scores reached statistical significance. Duration of antibiotic use and mortality rate were significantly higher in non-responders within 5 days (25.7 days versus 13.2 days, p = 0.003; 23.8% versus 4.2%, p = 0.024, respectively). In the univariate analysis, longterm care facility residents showed poor response within five days (p = 0.027), and duration of antibiotic use was significantly longer among poor responders (p = 0.003). In terms of antimicrobial susceptibility, penicillin resistance appeared to be associated with poor response (p = 0.065). On multivariate analysis, none of these relationships reached statistical significance, although elderly patients, long-term care facility residents, and those with nosocomial infections tended to show poor response within 5 days (Table 3).

Distribution of sequence types and serotypes
The relationships among all isolates in the MLST database are shown in Fig 1. Among 92 isolates, a total of 43 STs were identified. Seven allelic profiles (new 1 to new 7) were not recognized within the available MLST database of S. pneumoniae. eBURST analysis revealed that all 92 isolates were categorized into four major CCs (17 STs with 51 isolates) and 26 singletons (41 isolates), and each CC included two to six STs that differed at one or two alleles in the seven loci ( Fig 1A). The allelic profiles and serotypes of CNSP are shown in Table 5 and Fig  1B. Thirty-two CNSP were categorized into three major CCs (10 STs with 29 isolates) and three singletons (3 isolates). CC320 included three STs (ST320, ST2697 and ST1464) combined with three serotypes. CC166 included five STs (ST166, ST10120, ST13214, ST8279, and ST13388) associated with seven serotypes. CC81 included only two isolates, which were ST81 and ST83, respectively. Although the first two large clusters of the three above mentioned included isolates with various serotypes, one particular serotype dominated each cluster (19A in CC320, and 11A in CC166, respectively). Overall, ST320 (10 cases), ST166 (7 cases), and ST8279 (3 cases) were dominant among 32 CNSP, and ST8279 was only detected in prior long-term care facility residents. The other two isolates found among long-term care facility residents were also closely related to ST8279: one isolate differed only at ddl (profile 251-11-10-1-6-1-441), and the other was a three-locus variant (profile 15-11-10-1-315-1-113). Two ST13214 (serotypes 20 and 31/40), three ST8279 (two serotype 11A and one 9V/A), one ST13388 (serotype 11A) and one new sequence type (serotypes 20) isolates were resistant to both ceftriaxone and levofloxacin.

Discussion
S. pneumoniae is the most common causative pathogen of community-acquired pneumonia (CAP) and healthcare-associated pneumonia. Even in the era of PCV13, S. pneumoniae still accounts for 10-15% of CAP in developed countries [3,23]. Thus, as in the case of hospitalized patients with CAP, either a beta-lactam plus a macrolide or a quinolone alone is recommended based on the guidelines of the Infectious Diseases Society of America and the American Thoracic Society [24]. However, the prevalence of antibiotic resistance has increased worldwide, particularly among adults [7,8]. In Asian countries, more than 10% of pneumococcal isolates were resistant to ceftriaxone and levofloxacin [9,25]; 38% of non-meningeal invasive isolates were resistant to ceftriaxone in Taiwan [26], while 14.1% of respiratory isolates were non-susceptible to levofloxacin in South Korea [25]. Consistent with these reports, pneumococcal isolates showed high resistance rates to ceftriaxone (non-susceptible in 34.8%, 32/92; resistant in 17.4%, 16/92) and levofloxacin (resistant in 13.0%, 12/92) in this study. Moreover, eight cases (8.7%) were resistant to both ceftriaxone and levofloxacin at the same time. Such a high antibiotic resistance might lead to inappropriate empirical antibiotic therapy, resulting in poor clinical outcomes. In this study, we evaluated the risk factors and clinical significance of CNSP acquisition, and analyzed the serotype and genotypic characteristics of these resistant isolates.
Identifying the risk factors of CNSP acquisition is important to provide effective antibiotic treatment. Here, chronic lung and renal diseases were independent risk factors for acquisition of CNSP. Patients with end-stage renal disease and chronic lung disease are at increased risks of pneumococcal disease. Although data on carriage are lacking, these patients with chronic lung disease and chronic renal failure would have a high density of pneumococci in the nasopharynx and may have acquired MDR pneumococci with repeated exposure to antibiotics [27,28]. In patients with end-stage renal disease, monocytes, neutrophils, and dendritic cells are functionally impaired. As reported in experimental mouse studies, dysfunctional macrophages and impaired monocyte recruitment might result in decreased pneumococcal clearance in the Spread of ceftriaxone non-susceptible pneumococci nasopharynx [27]. In patients with chronic lung diseases, receptors for pneumococci, such as platelet-activating factor receptor, are overexpressed in the respiratory tract, increasing the risk of chronic colonization and invasive infections [29]. Besides these chronically ill patients, long-term care facility residents had high risks for CNSP infections. Because residents of longterm care facilities had several underlying medical conditions (diabetes mellitus and cerebrovascular disease), they were more likely to be heavily colonized with pneumococci and might be exposed to repeated antibiotic use. Thus, these individuals were more likely to be colonized with antibiotic-resistant pneumococci. In sequence, pneumococci would be transmitted to other patients by caregivers who were not strictly educated on infection control and cared for multiple patients at once. Several outbreaks of MDR pneumococcal pneumonia among longterm facilities have been reported [30,31]. Here, all five CNSP isolates from long-term care facility residents were resistant to both ceftriaxone and levofloxacin. The population aged >65 years in South Korea has been continually increasing, reaching 13.5% in 2016, and >1,500 long-term care facilities are currently operating nationwide [32,33]. Since long-term care facilities are an area of concern as potential sources of the spread of MDR pneumococci, active surveillance and interventions are required.
As antimicrobial resistance increases in pneumococci, therapeutic options become limited, and concerns regarding clinical outcomes may arise. The clinical relevance of the in vitro susceptibility of S. pneumoniae isolates has been controversial. Some studies have reported that antibiotic resistance in S. pneumoniae is not clinically relevant [34][35][36], whereas others have suggested higher mortality rate among patients infected with antibiotic-resistant S. pneumoniae [37][38][39]. In addition, data are insufficient on the clinical impact of ceftriaxone resistance, because most studies focused mainly on penicillin resistance [34,37]. Moreover, most studies focusing on ceftriaxone resistance did not use the revised CLSI breakpoints of ceftriaxone for non-meningitis isolates [35,36]. In this study, ceftriaxone or other antibiotic resistance were not associated with poor clinical response. It is unclear whether pneumococcal infections with low-level ceftriaxone resistance (MIC 4 μg/mL) would respond to cephalosporin therapy or not. When comparing pneumococcal infections with high-level resistance to ceftriaxone (MIC �8 μg/mL) to the CSP group, the 5-day response tended to be better in the CSP group,  [40]. However, vaccine serotypes, especially the serotypes included only in PPSV23, were significantly higher in the CNSP group. The most common serotypes in the CNSP group were serotypes 19A, 11A, and 19F, which were known to be highly drug resistant [18,26,[40][41][42]. Serotype 19A, which was a significant cause of morbidity and mortality before the introduction of PCV13 in 2010 [43,44] was an MDR serotype [26,41], and MDR ST320 was as a major genotype of serotype 19A [41,45]. This study shows that serotype 19A/ST320 pneumococcus are still prevalent in Korea in the era of PCV13, under the influence of unrecognized factors besides vaccine selection pressure. The high level of antimicrobial resistance might have an effect on the increase in serotype 19A pneumococci [46]. In the US, there was a remarkable shift in the serotype 19A genetic structure, where the prevalence of highly resistant serotype 19A strains (CC320) continued to increase with a concurrent decrease in the prevalence of CC199 [46,47]. In agreement with previous reports [18,48], serotypes 11A and 19F were common serotypes in the present study, especially in the CNSP group. Serotype 19F is known to have a higher MDR rate [26,40,48], and an association with ST320 was observed in this study. The increase in serotype 11A prevalence may result partly from PCV introduction. The prevalence of serotype 11A in Korean hospitals was only 3.8% in 1996-2001, the period before the introduction of PCV [49]. This proportion increased to 8.0-13.2% in 2011-2013 [18], a rate similar to the findings of our study. The increase in serotype 11A isolates was also thought to be attributable to the clonal expansion of CC166 isolates, which showed very high antimicrobial resistance rates, because antimicrobial resistance has been considered to be one of the key factors influencing the survival of bacterial clones [18]. Levofloxacin resistance was also identified among serotype 11A isolates in previous Korean reports [50], and non-susceptibility to levofloxacin was associated with healthcare-associated factors and cerebrovascular disease [25]. ST 8279, a CC166 isolate known to be MDR, and even an XDR clone [42,50], was previously reported among longterm care facility residents who underwent tracheostomy [42]. Similar findings were observed in this study: three isolates among five CNSP with long-term care facility residents were ST8279, the other two isolates were also closely related to ST 8279, and all of these were XDR. It is possible that XDR ST8279 is spreading in long-term care facilities, and should be monitored closely. Pneumococcal vaccines covered the majority of the serotypes that conferred a CNSP phenotype (PCV13, 50%; PPSV23, 81.2%), emphasizing the importance of vaccination for at-risk individuals with comorbid conditions. Continuous monitoring of the dynamic nature of both ceftriaxone resistance and serotype distribution in S. pneumoniae is crucial to understanding the impact of pneumococcal vaccination in both children and adults.
In conclusion, a distinct group of patients with chronic lung/renal diseases and residents of long-term care facilities appears to be an important reservoir of CNSP in South Korea. These CNSP isolates might have emerged both de novo after repeated exposure to antibiotics and by means of person-to-person transmission in hospitals or long-term care facilities. We revealed that 81.2% of the ceftriaxone non-susceptible isolates belonged to PPSV 23 serotypes, and serotype 19A, 11A, and 19F pneumococci were predominant. Specifically, serotype 11A was associated with XDR ST8279, which is closely associated with long-term care facility residents. Our findings call for attention and collaborative strategies to monitor the spread of these strains among long-term care facilities.
Supporting information S1 File. Demographic, clinical and microbiological data of pneumococcal isolates used in this study. (XLSX) Author Contributions