Necrotizing soft tissue infections caused by Streptococcus pyogenes and Streptococcus dysgalactiae subsp. equisimilis of groups C and G in western Norway
Streptococcus pyogenes (group A streptococcus, GAS) is a major cause of necrotizing soft tissue infection (NSTI). On rare occasions, other β-haemolytic streptococci may also cause NSTI, but the significance and nature of these infections has not been thoroughly investigated. In this study, clinical and molecular characteristics of NSTI caused by GAS and β-haemolytic Streptococcus dysgalactiae subsp. equisimilis of groups C and G (GCS/GGS) in western Norway during 2000–09 are presented. Clinical data were included retrospectively. The bacterial isolates were subsequently emm typed and screened for the presence of genes encoding streptococcal superantigens. Seventy cases were identified, corresponding to a mean annual incidence rate of 1.4 per 100 000. Sixty-one of the cases were associated with GAS, whereas GCS/GGS accounted for the remaining nine cases. The in-hospital case fatality rates of GAS and GCS/GGS disease were 11% and 33%, respectively. The GCS/GGS patients were older, had comorbidities more often and had anatomically more superficial disease than the GAS patients. High age and toxic shock syndrome were associated with mortality. The Laboratory Risk Indicator for Necrotizing Fasciitis laboratory score showed high values (≥6) in only 31 of 67 cases. Among the available 42 GAS isolates, the most predominant emm types were emm1, emm3 and emm4. The virulence gene profiles were strongly correlated to emm type. The number of superantigen genes was low in the four available GCS/GGS isolates. Our findings indicate a high frequency of streptococcal necrotizing fasciitis in our community. GCS/GGS infections contribute to the disease burden, but differ from GAS cases in frequency and predisposing factors.
Necrotizing soft tissue infections (NSTI) and the dominating subgroup, necrotizing fasciitis (NF), are rare and serious conditions characterized by rapid spread along tissue planes with development of necrosis [1-3]. NSTI is a heterogeneous group of infections affecting various groups of patients, caused by many microbes, leading to a wide range of clinical manifestations. Data on incidence, predisposing conditions and prognostic factors diverge, probably reflecting differences in inclusion criteria and study populations in various studies . The classification of NSTI has traditionally been based on microbiological findings: type I infections are polymicrobial, whereas type II infections are monomicrobial, and most often caused by Streptococcus pyogenes (group A streptococcus; GAS) [1-3]. To get more reproducible data it may be appropriate to investigate the different types of NSTI separately. NSTI caused by GAS seems to represent a specific disease entity, characterized by frequent development of shock and high mortality. A substantial proportion of the patients have no predisposing comorbidity. The unique pathogenesis of streptococcal fasciitis is probably dependent on many virulence factors. Although the bacterial virulence profile appears to be critical for the development of NSTI and streptococcal toxic shock syndrome (STSS), the distribution of emm types, superantigens and other virulence factors varies in streptococcal isolates associated with severe invasive disease [4-6]. Only a few previous studies have investigated the characteristics of streptococcal NSTI in detail [7, 8]. These studies have been confined to GAS infections, but recent studies have shown that S. dysgalactiae subsp. equisimilis of groups C and G (GCS and GGS) also have the ability to cause NSTI or STSS [9-13]. These two latter groups of streptococci share many virulence factors with GAS. Rarely, they even possess the group A carbohydrate antigen , which might lead to misidentification in the laboratory. To avoid this, additional tests for species identification can be used, although currently many laboratories use only serotype specificity on β-haemolytic streptococci and use the terms GAS and GCS/GGS. Streptococci within the Streptococcus anginosus group may also possess the group A, C or G antigens, but those can be easily discerned in the laboratory. In this study we describe incidence, clinical aspects and bacterial virulence gene profiles in NSTI caused by GAS and β-haemolytic GCS and GGS treated at Haukeland University Hospital, Norway during 2000–2009. The usefulness of the laboratory-based Laboratory Risk Indicator for Necrotizing Fasciitis (LRINEC) score for diagnosing streptococcal NSTI was also evaluated .
Patients and Methods
Study population, case definition and clinical parameters
Haukeland University Hospital (HUH) is a teaching hospital in Bergen, Norway. Patients admitted to the neighbouring Haraldsplass Deaconal Hospital are transferred to and treated at HUH when NSTI is suspected. These two hospitals have 1100 and 170 somatic beds, respectively. Incidence was calculated on the basis of cases from the catchment area of these two hospitals. The population in this area increased from c.347 000 to 387 000 inhabitants during the study period. HUH is also a tertiary hospital for a total of c.1 000 000 people. All cases of NSTI caused by GAS, GCS and GGS treated at HUH during 2000–2009 were retrospectively registered. Only the cases meeting surgical, clinical and microbiological criteria for streptococcal NSTI were included. NSTI was defined as necrotizing cellulitis or fasciitis as described previously . Cases with surgical evidence of infectious myositis were also included, but cases of pyomyositis, defined as pus within individual muscle groups, were excluded . Evaluation was based upon macroscopic findings during surgery, histological examination or post-mortem analyses. Removal of tissue was considered mandatory for the diagnosis of NSTI, apart from cases where surgical treatment was withheld because the chances of survival were considered to be very low. STSS was defined using criteria originally meant for GAS . Streptococcal aetiology was defined by the isolation of GAS, GCS or GGS from blood or the infected soft tissue. Cases were not included if the streptococci were isolated from a mixed flora with more than two species. Relevant data were obtained from the medical records using a standard abstraction sheet. Age, gender, demographic information, comorbidities and laboratory data at the time of admission were recorded. Microbiological culture results, antibiotic and surgical treatment, per operative findings, relevant adjunctive therapy, and outcome were also recorded. Written informed consent from the patients was not needed, as judged by the Regional Committee of Medical and Health Research Ethics. The study was approved by the privacy ombudsman of HUH.
All GAS, GCS or GGS were β-haemolytic and formed large colonies on blood agar. Serogroup-specificity was determined using a rapid agglutination test (Oxoid, Basingstoke, Hampshire, UK). These methods of identification imply that rare cases of S. dysgalactiae subsp. equisimilis of serogroup A may have been misidentified as S. pyogenes (GAS).
Extraction of chromosomal DNA after mechanical lysis of the cell wall and emm typing were performed as previously described . The following primers were used for initial emm amplification and sequencing in both directions: 5′-GGG-AAT-TCT-ATT-SGC-TTA-GAA-AAT-TAA-3′ (forward primer) and 5′-GCA-AGT-TCT-TCA-GCT-TGT-TT-3′ (reverse primer) . Assignment of emm types was done using the CDC Streptococcus Laboratory emm type database (http://www.cdc.gov/ncidod/biotech/strep/strepindex.htm).
Detection of streptococcal superantigen genes and SlaA
A multiplex PCR with specific primer pairs for the 11 exotoxin genes speA, speC, speG, speH, speI, speJ, speK, speL, speM, ssa and smeZ was used as described in detail elsewhere . To include allelic variations of the smeZ gene we also used simplex PCR with a previously described primer pair . Simplex PCR amplifications of speGdys, a speG orthologue found in S. dysgalactiae subsp. equisimilis, and the gene encoding streptococcal phospholipase A2 (SlaA) were performed with primers described previously . The speGdys primers amplified segments of equal size in both speGdys and speG, whereas the speG primers amplified only speG. Hence, all the GAS and GCS/GGS were screened for the presence of the 11 GAS superantigen genes (SAg) and SlaA, whereas only the GCS/GGS were subjected to PCR with the speGdys primers.
Data were analysed using SPSS PASW Statistics, version 20.0 (SPSS Inc, Chicago, IL, USA). Categorical data were analysed using Fisher's exact test. Non-parametric data (age) were analysed using the Mann–Whitney U-test. Because multiple comparisons were performed, both unadjusted and Holm–Bonferroni-corrected p values were calculated. A two-sided p-value ≤0.05 was considered statistically significant. To identify predictors of mortality, univariate analysis of several factors previously reported to be associated with mortality was first performed. Multivariate logistic regression analysis was performed choosing factors representing both predisposing conditions and acute disease severity and with low multicollinearity and low proportion of missing values.
Seventy patients treated for NSTI caused by GAS, GCS or GGS in the period 2000–2009 were identified and included in the study. The mean annual incidence was 1.4 per 100 000 population. The annual incidence varied greatly, ranging from 0.29 to 2.84 per 100 000. Sixty-four of the infections were community-acquired, and no cases were epidemiologically linked. Sixty-one of the patients had NSTI caused by GAS, whereas GCS accounted for two and GGS for seven cases. Staphylococcus aureus was co-cultured with GAS in two samples and together with GGS in two samples. Median age at presentation was 55.5 years (range 4–87 years); and 55 and 72 years among GAS and GCS/GGS patients, respectively (p 0.11). Only one patient was below 16 years of age, and 47 of the patients were men. An overview of the clinical features, laboratory data, predisposing factors and outcome is shown in Table 1. In almost half of GAS patients both the muscle fascia (aponeurosis) and muscle tissue were affected, whereas only two of the patients with GCS/GGS infection had such signs of deep tissue involvement. A minority of the patients had only necrosis of subcutaneous tissue/superficial fascia without evidence of involvement of the muscle fascia or muscle. STSS was diagnosed more often in patients with muscle involvement than in the other patients (p 0.04). The lower limb was the site of infection in 35 of the patients. Most patients had positive tissue cultures even though most of them had received antibiotics before surgery. Only about half of the patients had an LRINEC laboratory score ≥6, which has been associated with a high probability of NSTI in a previous study on NSTI and cellulitis . Significant comorbidity was present in less than half of the GAS patients, most commonly cardiovascular disease. Surgical treatment within 24 h of admittance was performed in 45 patients, and the median number of operative procedures performed per patient was three. The amputation rate was 8.6%. Clindamycin was included in the initial antibiotic treatment for 40 patients. Only one patient was treated with intravenous immunoglobulin, and none were treated with hyperbaric oxygen. The in-hospital case fatality rates (CFR) of GAS and GGS disease were 11% and 33%, respectively, whereas the CFR after 3 months were 20% and 33%, respectively.
Table 1. Clinical characteristics and outcome
|Subcutaneous involvement onlyb,c||5||59||3||7||8||66||0.033 (0.363)|
|Blood culture positive||27||61||2||9||29||70||0.289|
|Tissue culture positive||44||60||8||9||52||69||0.435|
|LRINEC score ≥6d||29||58||2||9||31||67||0.161|
|Age ≥65 years||15||61||5||9||20||70||0.107|
|Died in hospital|| 7||61||3||9||10||70||0.112|
|Died before 90 days||12||61||3||9||15||70||0.392|
The impact of various factors on in-hospital mortality is presented in Table 2. Premorbid factors (age ≥65 years and comorbid heart disease) as well as disease manifestations (STSS, muscle involvement) and treatment factors (fewer than three surgical debridements) were associated with mortality in the univariate analysis. When corrected for multiple comparisons, muscle involvement was not a significant factor. The association between mortality and few debridements could be explained by the fact that in five of the six patients with fewer than three debridements who died, surgical treatment was withheld because of a moribund state. In a logistic regression analysis age, STSS and comorbidity were chosen as predictor variables. Independent predictors of mortality in this model were increased age (OR per year 1.095, 95% CI 1.023–1.172; p 0.009) and STSS (OR 9.596; 95% CI 1.016–90.640; p 0.048).
Table 2. Predictors of in-hospital mortality
|STSS||24||60||9||10||13.50 (1.61–113.55)||0.005 (0.045)|
|Muscle involvementc||20||55||8|| 9||14.00 (1.63–120.22)||0.008 (0.056)|
|Surgery >24 hd||22||60||3||10||0.74 (0.17–3.16)||1.000|
|Surgical revisions <3e|| 8||60||6||10||9.75 (2.25–42.32)||0.003 (0.033)|
|Not clindamycine||27||60||3||10||0.52 (0.12–2.22)||0.498|
|Age ≥65 years||13||60||7||10||8.44 (1.91–37.26)||0.004 (0.040)|
|Heart disease|| 6||60||5||10||9.00 (2.01–40.30)||0.006 (0.048)|
|LRINEC score ≥6f||26||58||5|| 9||1.54 (0.37–6.32)||0.723|
|emm type 1/3/5/12/18/87/89g||19||36||4|| 6||1.79 (0.29–11.04)||0.673|
Forty-two GAS isolates were available for molecular analysis. The emm type distribution in relation to SAg gene profile of the GAS isolates is shown in Table 3. Thirteen isolates possessed emm1. All isolates except those possessing emm4.0 and emm77.0 contained speG, and all but three isolates, belonging to emm2, emm49 and emm96 contained smeZ. All the 13 isolates of emm1 showed identical SAg profiles and contained the phage-mediated superantigen speA and the chromosomally encoded speG, speJ and smeZ. For other emm types found in more than one case, the isolates sharing the same emm type also had the same virulence gene profile in all but four isolates. The four available GCS/GGS isolates belonged to four different emm types (stC74a.0, stG10.0, stG2574.0, stG6.3). The speGdys gene was found in isolates of the former three of these emm types. GAS SAg or SlaA genes were not identified in GCS/GGS.
Table 3. Virulence genes of 42 group A streptococcus isolates
There was no significant correlation between fatal outcome and specific virulence gene profiles, although three of the six genotyped GAS isolates associated with fatal outcome had emm types previously reported to give a CFR above 30% when associated with NF .
The present study is among the largest series of streptococcal NSTI reported to date. Several large series of NSTI have been published during the past few years, but none of these presented more than a few clinical parameters in relation to bacterial aetiology [24-27]. Detailed and stringent clinical and microbiological information from large series on streptococcal NSTI is lacking, apart from prospective studies from Ontario, Canada performed in the 1990s [7, 8]. Compared with the incidence of GAS NF in prospective studies from Canada (0.08–0.49/100 000/year) and Europe (0.22–0.6/100.000/year) a relatively high frequency of streptococcal NSTI was observed in our study, and GCS/GGS infections contribute to the disease burden [8, 23, 28]. Differences in reported incidence rates may also be explained by the use of imprecise inclusion criteria. In studies on NF, cases with engagement of the muscle fascia are probably often the only ones included. Excluding cases without such involvement might be inappropriate, because streptococcal disease with rapid spread along the subcutaneous tissue planes probably represents the same disease as the slightly more deeply situated infections. To improve comparability of studies, it may be more feasible to use the term NSTI rather than NF, as suggested by several authors, implying that cases are included irrespective of soft tissue level [2, 3, 29, 30]. Almost half of our patients had involvement of the muscle fascia and muscle tissue. The incidence in our study is therefore high even when comparing only definite NF cases.
The high proportion of patients with intraoperative signs of muscle tissue involvement shows that streptococcal infections not only spread along tissue planes; they often disrespect borders between tissue compartments. GAS-infected patients had muscle tissue involvement more often than GCS/GGS-infected patients. However, the difference was not statistically significant, and the number of GCS/GGS cases was low.
During the study period, <5% of all NSTI (type I and type II infections) at our hospital were culture negative (data not shown). This leads us to believe that almost all streptococcal cases treated in our institution were included in this study. High bacterial load and reduced bioavailability of antibiotics are known to reduce the effect of antibiotics in NSTI. These observations are supported by the high number of positive tissue cultures among the included patients, although most patients received antibiotic treatment before surgical intervention.
The LRINEC score was an insensitive tool for the diagnosis of streptococcal NSTI in our patients. A possible explanation might be the high awareness of NSTI in our region, contributing to early hospitalization and thus a low LRINEC score at admission. The usefulness of this score has been questioned also in several other recent studies [2, 3]. Early recognition of typical symptoms and signs that requires surgical exploration therefore remains the major tool in the diagnosis of streptococcal NSTI . High-quality education and training of the physicians involved is therefore essential.
The significant proportion of streptococcal NSTI caused by GCS/GGS in our study confirms the ability of these streptococci to cause severe soft tissue infections. In accordance with other studies of invasive GCS and GGS disease, our GCS/GGS-infected patients tended to be older and more often had underlying disease than GAS patients [9-12]. Among the 61 GAS cases 27 neither had comorbidity nor age ≥65 years, supporting the impression that GAS are highly virulent bacteria able to cause particularly severe soft tissue infections even in patients without predisposing conditions .
Although the CFR of GAS infections in the present study was relatively low, several factors were found to be associated with mortality. A high CFR in those with muscle tissue involvement found here is consistent with the high mortality previously reported in streptococcal myositis, a particularly severe form of NSTI . Our findings differ from observations reported in a study of NSTI where the presence of myonecrosis did not influence mortality. This is possibly explained by a low proportion of streptococcal disease . We found that STSS and age are independent predictors of mortality. These results are consistent with those from other studies [7, 8, 25, 26]. Even though early surgery or clindamycin treatment was not significantly associated with favourable outcome in this material, these factors were protective in other series .
Certain emm types, like emm1, emm3, emm4 and emm28, dominated our GAS samples. Particularly emm1 and emm3 have previously been linked to NSTI, STSS and mortality, probably indicating that isolates within certain emm types are genetically fit to produce more severe disease manifestations [4, 23]. Nevertheless, the overall emm type diversity in our material was substantial. This, along with a lack of correlation between emm type/superantigen gene profile and fatal outcome, indicates that host susceptibility is a key factor in the development and course of streptococcal NF. The low content of superantigen genes detected in GCS/GGS probably reflects the fact that these virulence factors are not a prerequisite for the development of NSTI caused by these bacteria [9, 10, 12].
The major limitations of this study arise from the retrospective inclusion of data. Such a design makes it difficult to perform a detailed and standardized characterization of all the patients, including preoperative clinical findings, macroscopic findings upon surgery and comprehensive laboratory analyses.
In conclusion, we report a relatively high incidence of streptococcal NSTI in our region of western Norway. GCS and GGS infections contributed to the disease burden, but they differ from GAS cases in frequency and predisposing factors. The overall mortality rates were relatively low, probably related to well-implemented routines of early surgery, appropriate antibiotic therapy and adequate supportive measures.
We thank M. Gjerde Tellevik (Centre for Tropical and Imported Infectious Diseases, Haukeland University Hospital) for excellent technical support. Preliminary results for some of the patients were presented at the 18th Lancefield International Symposium in Italy, September 2011 (abstract book p.41). The study was supported by the Department of Clinical Science, University of Bergen. The results form part of the clinical background material for the EU FP7-funded INFECT study (www.fp7infect.eu).
The authors declare that there are no potential conflicts of interests.