Methicillin-Resistant Staphylococcus aureus Ocular Infection in Taiwan: Potential Role of Panton–Valentine Leukocidin Gene

Abstract

Purpose: The relationship between Panton–Valentine leucocidin (PVL), a major virulence factor of Staphylococcus aureus, and disease severity and clinical outcomes remains unclear. We investigated the molecular characteristics and role of the PVL gene in methicillin-resistant S. aureus (MRSA) ocular infection in Taiwan.

Methods: Patients with culture-proven S. aureus ocular infection in Chang Gung Memorial Hospital from 2010 to 2017 were included. The presence of the PVL gene was detected for all S. aureus isolates. MRSA isolates were characterized through pulsed-field gel electrophoresis (PFGE), staphylococcal multilocus sequence type, and staphylococcal cassette chromosome mec (SCCmec) typing. Drug susceptibility was examined using disk diffusion method and E-test. Patients’ demographics, diagnoses, and outcomes were collected.

Results: There were 112 methicillin-sensitive S. aureus and 103 MRSA isolates. Among 50 PVL(+) S. aureus isolates, 43 were MRSA. CC59/PFGE type D/SCCmec IV, V_T (38 of 43 isolates, 88%), and CC59/PFGE type C/SCCmec IV (27 of 60 isolates, 45%) were the predominant clones in the PVL(+) and PVL(−) MRSA isolates, respectively. When we compared the two CC59 strains, the patients with PVL(+)/CC59 MSRA infection were significantly younger than those with PVL(−)/CC59 MSRA (39.3 vs. 61.7 years; P = 0.001). PVL(+)/CC59 MSRA caused significantly more eyelid disorders (36.8% vs. 3.7%; P = 0.002) but less keratitis (23.7% vs. 51.9%; P = 0.034). The antibiograms of the two strains were similar.

Conclusions: PVL(+) MRSA is significantly associated with eyelid infection, especially in young patients.

Translational Relevance: PVL gene plays a role in clinical features of MRSA ocular infections.

Introduction

Staphylococcus aureus is a major bacterial pathogen affecting humans. Management of methicillin-resistant S. aureus (MRSA) that causes multidrug resistance is difficult.¹ Although healthcare-associated MRSA (HA-MRSA) was identified at the beginning of the 1960s, it was later identified even in patients without exposure to healthcare settings; thus, this infection was identified as community-associated MRSA (CA-MRSA).² The increase in CA-MRSA infections has become a global public health concern because, compared with HA-MRSA, CA-MRSA demonstrates higher ecological fitness and virulent capacity and leads to different clinical presentations. Some studies have proposed a strong correlation between highly virulent CA-MRSA and a pore-forming leukotoxin, Panton–Valentine leukocidin (PVL).³ PVL(+) MRSA isolates exhibit leucocidal activity and cause diseases with high morbidity, including necrotizing pneumonia, necrotizing fasciitis, and dermonecrosis.⁴ However, the direct association between PVL genes and their pathogenicity remains unclear. Studies analyzing the virulence determinants of CA-MRSA have indicated that PVL genes may be relevant epidemiologically but vary pathologically in different animal models.⁵ Shallcross et al.⁶ conducted a systematic review of 76 clinical studies from 31 countries to determine the role of PVL genes in disease severity and clinical outcomes. They reported that PVL genes are consistently associated with skin and soft tissue infections but are rarely associated with invasive diseases. However, the presence of PVL genes may increase the morbidity rate in pediatric patients with musculoskeletal infections. 

Although MRSA, including CA-MRSA, infections have been attracting the increasing attention of ophthalmologists in the past two decades, few studies have determined the genotypes of MRSA isolates or examined the role of PVL genes in ocular infections.^7–11 Our previous 10-year study (1999–2008) on ocular infections caused by MRSA demonstrated that CA-MRSA and HA-MRSA infections differed epidemiologically and clinically; the prevalence of CA-MRSA was higher than that of HA-MRSA; CA-MRSA usually caused mild and superficial diseases, such as eyelid and lacrimal duct disorders, whereas HA-MRSA caused severe vision-threatening diseases, such as keratitis and endopthalmitis.⁷ However, because of the retrospective nature of our previous study, we classified MRSA isolates on the basis of their epidemiological characteristics, but not molecular characteristics. Unlike HA-MRSA, the genomic characteristics of CA-MRSA vary geographically. Thus, physicians should be familiar with prevalent epidemic clones in their local areas. In this study, we prospectively collected isolates from patients with ocular S. aureus infection between 2010 and 2017 and analyzed the molecular epidemiology of MRSA isolates and determined the role of PVL genes in ocular MRSA infections in Taiwan. 

Methods

Ethics Statement

This study was conducted in accordance with the tenets of the Declaration of Helsinki and was approved by the Institutional Review Board of Chang Gung Memorial Hospital (CGMH), Taiwan (107-2346C). Because data collected in this study were confidential and anonymized, the requirement for participants’ informed consent was waived. 

Study Population and Data Collection

We prospectively collected S. aureus isolates from patients with ocular infections from the microbiology laboratory at CGMH, Taiwan, between January 1, 2010, and December 31, 2017. We identified 215 patients with S. aureus ocular infection and retrospectively reviewed their medical records. We collected data on demographics, underlying diseases, recent medication history, ocular history, HA factors, and ocular diagnosis and management. The underlying diseases recorded included diabetes mellitus, malignancy, pulmonary disease, hypertension, liver disease, renal disease, and current non-ocular infection. Ocular diseases were categorized based on the ocular structure involved, including lid disorder, keratitis, conjunctivitis, lacrimal system disorder, endophthalmitis, and others. 

Drug Susceptibility Testing

The susceptibility of S. aureus isolates to antibiotics—namely, cefoxitin, clindamycin, fusidic acid, erythromycin, linezolid, trimethoprim/sulfamethoxazole (TMP/SMX), teicoplanin, tigecycline, and vancomycin—was examined using the disk diffusion method in accordance with the antimicrobial susceptibility testing guidelines of the Clinical and Laboratory Standards Institute.¹² The susceptibility of S. aureus isolates to fluoroquinolones was analyzed through the E-test (BioMerieux SA, Marcy-I'Etoile, France) because fluoroquinolones are not included in routine antibiotic susceptibility testing for S. aureus in our microbiology laboratory. 

Molecular Typing and Presence of PVL Genes

Polymerase chain reaction was performed to detect the presence of PVL genes in all S. aureus isolates.¹³ We determined the molecular characteristics of all MRSA isolates through pulsed-field gel electrophoresis (PFGE) and staphylococcal cassette chromosome mec (SCCmec) typing.¹⁴ A multilocus sequence typing (MLST) scheme¹⁵ was examined in selected isolates by using representative PFGE patterns. The sequence type (ST) was determined based on sequence alleles. MLST was performed and STs were interpreted following previously described procedures.^13–16 

Categorization

Patients with MRSA infection were classified into CA-MRSA and HA-MRSA groups based on their molecular findings. Patients with isolates carrying type I to III SCCmec and type IV or V SCCmec were included in the HA-MRSA and CA-MRSA groups, respectively.¹⁶ The clinical criteria for HA infection were defined by the U.S. Centers for Disease Control and Prevention Active Bacterial Core Surveillance sites¹⁷ and included, 48 hours after admission, the presence of any percutaneous device and a history of surgery, hospitalization, dialysis, or living in a nursing home within the past 1 year. 

Statistical Analysis

Continuous variables are presented as the mean ± standard deviation (SD), and nominal variables are presented as the number and percentage. The χ² test and Fisher's exact test were performed to analyze differences in variables between the groups. Differences were considered significant at a two-tailed P < 0.05. All statistical analyses were performed using SPSS Statistics 20.0 (IBM Corporation, Chicago, IL). 

Results

From 2010 to 2017, we identified 215 patients with S. aureus ocular infections, including 103 patients with MRSA and 112 patients with methicillin-sensitive S. aureus (MSSA) infections. Fifty isolates (23.3%; 43 MRSA and 7 MSSA) carried the PVL gene. The remaining 165 PVL(−) isolates included 60 MRSA and 105 MSSA isolates. The PVL gene was predominantly detected in the MRSA isolates instead of the MSSA isolates. Because the groups were independent and we wanted to ensure that the extrapolated result was applicable to the overall population, we excluded MSSA from the following analyses to prevent the small sample size from undermining the molecular impact of the PVL gene on S. aureus ocular infection. We evaluated the epidemiology of PVL(+) and PVL(−) MRSA isolates to determine the molecular impact of the PVL gene on S. aureus ocular infection. 

General Demographics for PVL(+) MRSA and PVL(−) MRSA

As illustrated in Table 1, the patients with PVL(+) MRSA ocular infection were significantly younger than those with PVL(−) MRSA ocular infection (42.0 vs. 55.8 years; P = 0.015). The patients with PVL(−) MRSA ocular infection tended to have comorbidities compared with those with PVL(+) MRSA ocular infection (71.1% and 53.0%, respectively; P = 0.065). A higher proportion of the patients with PVL(−) MRSA infection had HA factors than did those patients with PVL(+) MRSA infection (65% and 34.9%, respectively; P = 0.003). The most common ocular disease caused by PVL(+) MRSA was lid disorder (37.2%), followed by keratitis (23.3%) and conjunctivitis (20.9%). By contrast, the most common ocular disease caused by PVL(−) MRSA was keratitis (55.0%), followed by conjunctivitis (21.6%) and lid disorder (8.3%). The patients with PVL(+) MRSA infection had a higher prevalence of lid disorder than did those with PVL(−) MRSA infection (P = 0.001); however, the prevalence of keratitis was lower in the patients with PVL(+) MRSA infection than in those with PVL(−) MRSA infection (P = 0.002). 

Table 1.

 

Comparison of Demographic and Clinical Characteristics of PVL(+) and PVL(−) MRSA

 

Antibiotic Susceptibility of PVL(+) MRSA and PVL(−) MRSA

As shown in Table 2, all MRSA isolates were susceptible to linezolid, tigecycline, teicoplanin, and vancomycin. The ratio of PVL(+) MRSA isolates susceptible to TMP/SMX and four fluoroquinolones was significantly higher than that of PVL(−) MRSA isolates (P = 0.007 and P = 0.0001–0.006, respectively). 

Table 2.

 

Comparison of Antibiotic Susceptibility Between PVL(+) and PVL(−) MRSA

 

Molecular Typing

Table 3 presents the genotypes of the 103 MRSA isolates. In the 43 PVL(+) MRSA isolates, the most dominant clone was clonal complex 59 (CC59)/PFGE type D/SCCmec IV, V_T (38 isolates, 88%). The remaining PVL(+) MRSA isolates carried CC8/PFGE type AI/SCCmec IV (four isolates) and CC30/PFGE type AG/SCCmec IV (one isolate). By contrast, higher genetic diversity was found for 60 PVL(−) MRSA isolates. The predominant clone was CC59/PFGE type C/SCCmec IV (27 isolates, 45%), followed by CC239/PFGE type A/SCCmec III, IIIA (13 isolates, 21.7%), CC45/PFGE type BM/SCCmec V (six isolates), and CC45/PFGE type AK/SCCmec IV (five isolates). 

Table 3.

 

Molecular Characteristics of 103 MRSA Ocular Isolates Stratified by PVL Gene and Pulsotype

 

Overall, all of the PVL(+) isolates carried SCCmec type IV or type V_T that represented molecular CA-MRSA. PVL(−) isolates exhibited diversity of SCCmec from type II to V, which was defined as molecular CA-MRSA or HA-MRSA. The proportions of MRSA isolates with CC59/PFGE type D/SCCmec IV, V_T/PVL(+) (38 isolates) and CC59/PFGE type C/SCCmec IV/PVL(−) (27 isolates) were the largest in their respective categories. Because both of the strains had common genetic identity (CC59) and belong to molecular CA-MRSA, we narrowed down the scope and excluded molecular HA-MRSA isolates that would interfere with the judgment required to obtain precise and in-depth knowledge regarding the role of the PVL gene in ocular infection. To juxtapose PVL(+)/CC59 CA-MSRA with PVL(−)/CC59 CA-MRSA, we investigated differences in the clinical features and antibiograms of these two strains to determine the role of the PVL gene in ocular infection. 

Demographics and Clinical Features of CC59/PVL(+) CA-MSRA and CC59/PVL(−) CA-MRSA

As indicated in Table 4, patients with PVL(+)/CC59 CA-MSRA infection were significantly younger than those with PVL(−)/CC59 CA-MSRA infection (39.3 vs. 61.7 years; P = 0.001). No differences in comorbidities and HA factors were noted between these groups. PVL(+)/CC59 CA-MSRA caused a higher prevalence of eye lid disorder (36/8% vs. 3.7%; P = 0.002) but a lower prevalence of keratitis (23.7% vs. 51.9%; P = 0.034). 

Table 4.

 

Comparison of Demographic and Clinical Characteristics of PVL(+) and PVL(−)/CC59 CA-MRSA

 

Antibiotic Susceptibility of PVL(+)/CC59 CA-MSRA and PVL(−)/CC59 CA-MSRA

Table 5 illustrates the percentage of antibiotic susceptibility between PVL(+)/CC59 CA-MRSA and PVL(−)/CC59 CA-MRSA. Both isolates displayed high susceptibility toward all tested antibiotics, except for erythromycin and clindamycin. 

Table 5.

 

Comparison of Antibiotic Susceptibility of PVL(+) and PVL(−)/CC59 CA-MRSA

 

Discussion

This study readdressed the inadequacy of molecular typing in our previous study on MRSA ocular infection in Taiwan⁷ and investigated the role of the PVL gene, a common feature of CA-MRSA infection that has been reported to be associated with poor prognosis.¹⁸ We observed that PVL(+) isolates accounted for 23.3% of S. aureus infections (50/215), and they were mainly associated with MRSA infection (43/50). The patients with PVL(+) MRSA infection were younger and had fewer HA factors than did those with PVL(−) MRSA infection. Moreover, a higher proportion of the patients with PVL(+) MRSA infection had eyelid disorders than did those with PVL(−) MRSA infection. However, a lower proportion of the patients with PVL(+) MRSA infection had keratitis than did those with PVL(−) MRSA infection. Although both of the isolates were resistant to multiple drugs, PVL(+) MRSA isolates were more susceptible to TMP/SMX and fluoroquinolones. These findings are in agreement with those of our previous study examining differences between CA-MRSA and HA-MRSA.^7,9 The predominant strains of PVL(+) and PVL(−) MRSA isolates are CC59/PFGE type D/SCCmec IV, V_T and CC59/PFGE type C/SCCmec IV isolates, respectively. Both of these strains belong to the CC59 molecular lineage and are prevalent CA-MRSA strains in Taiwan. Therefore, we compared clinical features between patients infected with CC59/PVL(+) CA-MRSA and those infected with CC59/PVL(−) CA-MRSA, and the results demonstrated that the patients infected with CC59/PVL(+) CA-MRSA were younger and had a higher prevalence of eyelid disorders than did those infected with CC59/PVL(−) CA-MRSA. Thus, the differences in demographics and clinical presentations between patients with CC59/PVL(+) CA-MRSA and CC59/PVL(−) CA-MRSA may be attributable to the presence of the PVL gene because both of the strains exhibited a similar molecular lineage and antibiograms. 

CA-MRSA isolates vary genetically across different geographic regions. However, the strong association of the PVL gene with the small SCCmec element is commonly observed in major clones such as ST8 (USA 300) in the United States and Canada, ST80 in Europe, and ST59 in Asia.^16,19,20 In our study, the most common clone of PVL(+) MRSA isolates was CC59/PFGE type D/SCCmec IV, V_T (88.0%), which is a common endemic CA clone in Taiwan. By contrast, the most common clone of PVL(−) MRSA isolates was CC59/PFGE type C/SCCmec IV (45.0%), another endemic CA clone in Taiwan, followed by CC239/PFGE type A/SCCmec III, IIIA (21.7%), an endemic HA clone in Taiwan, and CC45 with PFGE type BM/SCCmec V and PFGE type AK/ SCCmec IV (18.3%). In terms of genotypes, PVL(−) MRSA isolates consisted of both molecular CA- and HA-MRSA isolates; this explains the significant difference in HA factors between PVL(−) MRSA and PVL(+) MRSA isolates in our study. 

CC59 is the dominant epidemic lineage of CA-MRSA in Taiwan, and it can be classified into at least two major clones: the Taiwan clone, which is characterized as PFGE type D/SCCmec V_T/PVL(+), and the Asia-Pacific clone, which is characterized as PFGE type C/SCCmec IV/PVL(−).²¹ These clones differ particularly in terms of the presence of the PVL gene, which provides us an opportunity to determine the role of PVL in the pathogenesis of CA-MRSA infections. Several studies on Taiwanese children have reported that the Taiwan clone, PVL(+), was prevalent in patients with severe infection, whereas the Asia-Pacific clone, PVL(−), was a colonizer often seen in healthy children.^22,23 Chen et al.²⁴ investigated the characteristics of these two clones in cultured cells and a murine model and determined that the Taiwan clone had a higher potential of causing neutrophil lysis and sepsis with a higher mortality rate in vivo than did the Asia–Pacific clone. Epidemiological and experimental studies have demonstrated the increased pathogenic potential of the Taiwan clone, which is likely due to the presence of the PVL gene. 

In this study, we determined that the presence of the PVL gene results in differences in epidemiological and clinical features among S. aureus ocular infections. We observed that the patients with PVL(+) MRSA or CC59/PVL(+) CA-MRSA infection were younger and had a higher prevalence of eyelid disorders. This finding is in agreement with those of previous studies in other fields indicating that PVL(+) isolates are mainly associated with skin and soft tissue infections in young individuals in Taiwan²⁵ and other regions.⁶ By contrast, PVL(−) MRSA or CC59/PVL(−) CA-MRSA results in a higher prevalence of keratitis, a vision-threatening disease, in older patients. Bispo et al.²⁶ examined the population structure of 68 ocular MRSA isolates collected at Massachusetts Eye and Ear in the United States and observed the dominance of two major CC clones: CC8/SCCmec IV/PVL(+) and CC5/SCCmec II or V/PVL(−). CC8/PVL(+) isolates mainly infected younger people, and the majority of the patients had preseptal or orbital cellulitis. By contrast, CC5/PVL(−) isolates frequently caused keratitis in older people. Tissue tropism can be the driving force for the differences in clinical presentations; that is, we observed that the CC8/PVL(+) strain mainly causes infection of the keratinized epithelium (lids) and soft tissue, whereas the CC5/PVL(−) strain mainly causes infection of the wet epithelial ocular surface (cornea). Despite differences in the dominant clones, epidemiological and clinical characteristics were similar between PVL(+) and PVL(−) MRSA isolates in our and the study conducted by Bispo et al.²⁶ These findings indicate that the PVL gene affects the clinical presentation of MRSA infection. 

PVL(+) MRSA is known to be associated with severe CA-MRSA infection and poor prognosis.¹⁸ In our study, no differences were noted for the need of surgery or hospital admission between patients with PVL(+) or PVL(−) MRSA/CA-MRSA infection despite differences in clinical presentations. In 2006, Rutar et al.⁸ first reported the details of nine patients who had severe ocular infections, including orbital cellulitis, endophthalmitis, and venous thrombosis, caused by USA300, a PVL(+) strain. However, Foster et al.¹⁰ studied the molecular characteristics of 85 S. aureus isolates from children with periorbital and orbital cellulitis and reported no differences in clinical characteristics between MRSA and MSSA or USA300 and non-USA300 infections. Sukek et al.²⁷ compared the clinical outcomes of 95 cases of keratitis caused by PVL(+) and PVL(−) S. aureus in the United Kingdom and reported that PVL(+) isolates were associated with poorer clinical outcomes and more surgical interventions. Notably, in their study, eight of nine PVL(+) isolates were identified as MSSA, which is common in the United Kingdom. The discrepancy in the results of these studies may be due to differences in the genotypes of PVL(+) isolates; the mixture of MSSA, HA-MRSA, and CA-MRSA for PVL(−) isolates; and differences in ocular diseases. However, we could not analyze the outcomes of individual ocular diseases between PVL(+) or PVL(−) MRSA/CA-MRSA due to the small sample size. Additional studies on different PVL stains in different countries are warranted to determine the role of the PVL gene in different S. aureus ocular infections. 

In this study, we examined the susceptibility of the isolates to fluoroquinolones, the most widely used empirical antibiotic for treating ocular infections, which was not mentioned in our previous study.⁷ In our study, PVL(+) MRSA isolates exhibited significantly higher susceptibility to four tested fluoroquinolones than did PVL(−) MRSA isolates (93.0% vs. 58.3%–70.0%); PVL(+) and PVL(−) CC59 CA-MRSA isolates were susceptible to levofloxacin, gatifloxacin, or moxifloxacin. The relatively lower susceptibility to fluoroquinolones of PVL(−) MRSA isolates was due to the mixture of different strains, because the antibiotic susceptibility of MRSA is associated with the strains instead of the PVL gene. In contrast to our findings, several national surveys of ocular isolates conducted in the United States have reported that the rate of susceptibility to fluoroquinolones for MRSA was low (only 15%–25%).^28–30 The present study, however, indicates that fluoroquinolones are highly effective for treating CA-MRSA ocular infection in Taiwan, which might explain why some surrogate outcomes, such as the rates of surgery and admission, were similar in the ocular infections caused by PVL(+) and PVL(−) CC59/CA-MRSA isolates in our study. 

The necessity to decolonize the PVL gene in a certain age group in the community has been suggested in some countries. For example, the infection control policy in the United Kingdom advocates screening and/or treatment of PVL(+) S. aureus cases and their contacts.³¹ However, the screening should only be advocated on the assumption that PVL(+) infection is rare and severe or it would be economically infeasible to decolonize the PVL gene in the community only to minimize the future risk of infection. In addition, patients may receive unnecessary antibiotic treatment, which would increase their vulnerability to the risk of recolonization by another more virulent pathogen if the PVL gene that may not be fully accounted for in invasive diseases is eradicated. Because we found that PVL(+) isolates caused relatively mild ocular disease and were highly susceptible to fluoroquinolones, we do not think PVL screening and consecutive decolonization are necessary for patients with ocular infection, at least in Taiwan. 

This study has some limitations. First, although the specimens were collected prospectively, we analyzed clinical data retrospectively; thus, data on some risk factors might be incomplete. Second, a small sample size might affect the statistical significance of analysis results, but a few differences between groups were observed. Third, our findings were geographically distinct and may not be generalized to other regions or populations. Finally, although we used two related clones (the Taiwan and Asia–Pacific clones) to determine the potential role of the PVL gene in MRSA ocular infections based on epidemiological observations, we cannot rule out the effect of another prophage, such as SA3, which is truncated in the Taiwan clone.²⁵ Additional experimental studies examining proteomic, transcriptomic, and gene expressions are warranted to identify virulence markers specific to each endemic CA-MRSA lineage. 

In conclusion, the PVL gene was present in approximately one-fourth of ocular S. aureus isolates and was mainly present in MRSA isolates in Taiwan. All PVL(+) MRSA isolates were molecularly identified as CA-MRSA, mainly composed of CC59/PFGE type D/SCCmec IV, V_T strains. By contrast, PVL(−) MRSA isolates were both HA-MRSA and CA-MRSA; however, approximately half of the isolates were CC59/PFGE type C/SCCmec IV strains. The two predominant CA-MRSA strains, CC59/PVL(+) and CC59/PVL(−), had different clinical features; CC59/PVL(+) resulted in a higher prevalence of eyelid disorders in younger individuals, suggesting that the PVL gene plays a role in S. aureus ocular infections. A larger study with various PVL stains from different countries should be conducted to verify the role of the PVL gene in S. aureus ocular infections. 

Acknowledgments

The authors thank Wen-Hsuan Chen for technical assistance. 

Supported by grants from the Ministry of Science and Technology (MOST 109-2635-B-182A-003, NMRPG3K0461) and Chang Gung Memorial Hospital, Taiwan (CMRPG1I0021). The funding organizations had no role in the design or conduct of this research. 

Disclosure: Y.-T. Liu, None; E.Y.-C. Kang, None; Y.-L. Chen, None; L.-K. Yeh, None; D.H.K. Ma, None; H.-C. Chen, None; K.-H. Hung, None; Y.-C. Huang, None; C.-H. Hsiao, None