Você está na página 1de 9


Molecular and Clinical Characteristics of Invasive Group A Streptococcal Infection in Sweden

Jessica Darenberg,1,3 Bogdan Luca-Harari,5 Aftab Jasir,5 Andreas Sandgren,1 Helena Pettersson,2 Claes Schale n,5 6 2 3 1,4 Mari Norgren, Victoria Romanus, Anna Norrby-Teglund, and Birgitta Henriques Normark
Departments of 1Bacteriology and 2Epidemiology, Swedish Institute for Infectious Disease Control, Solna, 3Center for Infectious Medicine, Department of Medicine, Karolinska Institutet, Karolinska University Hospital Huddinge, and 4Department of Microbiology, Tumor and Cell Biology, Karolinska Institutet, Stockholm, 5Department of Medical Microbiology, Institute of Laboratory Medicine, University of Lund, Lund, and 6 Department of Clinical Microbiology, Biomedical Laboratory Science, Umea University, Umea , Sweden

Background. The incidence and severity of invasive group A streptococcal infection demonstrate great variability over time, which at least, in part, seems to be related to group A streptococcal type distribution among the human population. Methods. An enhanced surveillance study of invasive group A streptococcal infection (746 isolates) was performed in Sweden from April 2002 through December 2004. Noninvasive isolates from either the throat or skin (773 isolates) were collected in parallel for comparison. Clinical and epidemiological data were obtained from 88% of patients with invasive disease and were related to isolate characteristics, including T type, emm sequence type, and the presence of 9 superantigen genes, as well as pulsed-eld gel electrophoresis pattern comparisons of selected isolates. Results. The annual incidence was 3.0 cases per 100,000 population. Among the patients with invasive disease, 11% developed streptococcal toxic shock syndrome, and 9.5% developed necrotizing fasciitis. The overall casefatality rate was 14.5%, and 39% of the patients with streptococcal toxic shock syndrome died (P ! .001 ). The T3/ 13/B3264 cluster accounted for 33% of invasive and 25% of noninvasive isolates. Among this most prevalent type cluster, emm types 89 and 81 dominated. Combined results from pulsed-eld gel electrophoresis, emm typing, and superantigen gene proling identied subgroups within specic emm types that are signicantly more prone to cause invasive disease than were other isolates of the same type. Conclusions. This study revealed a changing epidemiology of invasive group A streptococcal infection in Sweden, with emergence of new emm types that were previously not described. The results also suggest that some clones may be particularly prone to cause invasive disease. Since the mid 1980s, group A streptococci (GAS) have been identied as a reemerging cause of severe invasive infections, such as septic shock, puerperal sepsis, softtissue infections (including necrotizing fasciitis [NF]), and streptococcal toxic shock syndrome (STSS) [13]. Despite increased awareness and improved treatment of the most severe disease manifestations (i.e., NF and STSS), the mortality rate remains high, often exceeding 50% [3, 4]. The M protein is a major GAS virulence factor, evoking a type-specic host immune response. To date, 1150 different M protein sequence types (emm types) have been described [5]. Both emm and T typing are important tools in epidemiological studies. T and M/emm types correlate to each other [6], and certain type combinations have been associated with certain clinical manifestations and disease severity [711]. However, the prevalence of specic GAS types in a community shows high variability depending on geographic area and over time [1215]. The GAS superantigens (SAgs) are also important for virulence, participating in the induction of the systemic toxicity associated with severe infections [16, 17]. Currently, there are at least 12 known streptococcal SAgs [4, 9, 18]. In the present national surveillance study, performed for 2 years and 8 months, essentially all invasive GAS isolates in Sweden were collected from the microbio-

Downloaded from http://cid.oxfordjournals.org/ by guest on March 13, 2014

Received 9 January 2007; accepted 7 May 2007; electronically published 10 July 2007. Reprints or correspondence: Dr. Jessica Darenberg, Dept. of Bacteriology, Swedish Institute for Infectious Disease Control, SE 171 82 Solna, Sweden (Jessica.Darenberg@smi.ki.se). Clinical Infectious Diseases 2007; 45:4508 2007 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2007/4504-0008$15.00 DOI: 10.1086/519936

450 CID 2007:45 (15 August) Darenberg et al.

logical laboratories and characterized using molecular techniques. The results were correlated to clinical parameters. Also, a representative noninvasive material was collected. The study revealed a signicant change in GAS epidemiology in the Swedish population, with a predominance of emm types not previously shown to cause invasive disease to any large extent. Furthermore, certain clonal subgroups of isolates with higher invasiveness than others were identied. MATERIALS AND METHODS
Study Materials

the genes allelic variations (5 -CAGATATAGTAATTGATTTTA and 3 -AGCTAGAACCAGAAGAATAT). PFGE. Invasive and noninvasive isolates of emm types 89, 81, 77, 28, 12, and 1 (992 isolates in total) were subjected to PFGE analysis, using the restriction enzyme SmaI (New England Biolabs), as described elsewhere [23]. BioNumerics, version 3.5 (Applied Maths), was used for analysis; isolates with 85% similarity (3 bands difference) were considered to belong to the same clone.
Statistical Analysis

Invasive isolates. A national active surveillance of invasive GAS infections was conducted in Sweden from 23 April 2002 through 31 December 2004. Invasive disease was dened by isolation of GAS from blood samples or from samples obtained from other normally sterile sites. All 29 Swedish clinical microbiological laboratories, with a national catchments area of 9 million inhabitants, participated. The attending physicians completed a questionnaire regarding clinical information about the infection, such as probable port of entry, symptoms, focus, treatment, and outcome, as well as about recent use of antibiotics and underlying conditions. STSS cases were classied according to the consensus denition proposed by The Working Group on Severe Streptococcal Infections [19]. The study was approved by the ethics committee at Karolinska Institutet (Stockholm, Sweden). Noninvasive isolates. From February 2003 through June 2004, 6 of the microbiological laboratories, which were geographically spread in Sweden, submitted 10 consecutive noninvasive isolates each month, of which 5 were isolated from skin samples and 5 were isolated from throat samples. Two laboratories were located in Stockholm (Karolinska University Hospital Solna and Karolinska University Hospital Huddinge), and the others were located at the University hospitals in Lund, Go teborg, Uppsala, and Umea .
Characterization of Isolates

Data were analyzed using Stata Statistical Software, release 8.0 (Stata), and GraphPad Prism, version 4 (GraphPad Software). For nominal data, x2 test or Fishers exact test were used when appropriate. For continuous data, either Mann-Whitney U test or Kruskal-Wallis test were used. Multinomial logistic regression analysis [24] with dichotomous and continuous independent variables was tted using SAS, version 9.1.3 (SAS Institute), proc logistic. The independent variables were removed from the model in a backwards procedure using the likelihood ratio test as exclusion criteria. RESULTS Characteristics of invasive GAS infections. Seven hundred forty-six invasive cases were identied, corresponding to a mean annual incidence of 3.0 cases per 100,000 inhabitants during the study period. The incidence increased with age (gure 1), and seasonal uctuations were observed, with an increased number of cases occurring during the winter months. Questionnaires with clinical information were obtained for 88% of the patients. Among the patients with invasive cases, 11% developed STSS, and 9.5% developed NF. A majority of the patients (69%) reported underlying chronic disease (of which diabetes and cardiovascular disease were most common), predisposing factors, or drug abuse. The overall case-fatality rate was 14.5%, compared with 39% among STSS cases (P ! .001). Univariate analysis revealed that fatal outcome was signicantly associated with the presence of underlying conditions, multiorgan failure, and STSS (P ! .001) (table 1). The majority (94%) of invasive isolates were obtained from blood samples, and the remaining isolates sampled were obtained from other normally sterile sites. The T3/13/B3264 type cluster was the most common T type, accounting for one-third of the isolates. The most abundant emm types were, in descending order, emm types 89, 81, 28, 1, 12, 77, and 4, accounting for 74% of the invasive infections (gure 2 and table 2). Skin was the most common site of entry (63% of patients), followed by the respiratory tract (13%), urogenital tract (6%), and other/unknown sites (19%). Patients with skin and/or tissue involvement were older (P ! .001) and, hence, had more underlying diseases and/or conditions reported (P ! .001) (table
Invasive GAS Infection CID 2007:45 (15 August) 451

Downloaded from http://cid.oxfordjournals.org/ by guest on March 13, 2014

T typing. All invasive and noninvasive isolates were T typed using conventional poly- and monospecic T antisera (Sevapharma). emm Typing. Genomic DNA from all isolates was prepared, as described elsewhere [20]. The emm types were determined by direct sequencing of the N-terminal hypervariable portion of the emm gene, as described elsewhere [21]. The type was assigned using the emm type database [22]. SAg genotyping. All isolates were genotyped for their SAg prole using multiplex PCR with 8 specic primer pairs for the streptococcal SAg gene and the pyrogenic exotoxin genes speA, speB, speC, speF, speG, speH, and speJ, as described in detail elsewhere [14]. We used a single PCR for the streptococcal mitogenic exotoxin (sme) Z gene, with primer pairs covering

Figure 1. Age distribution, incidence, and fatal outcome for invasive group A streptococcal disease in Sweden. All 746 invasive group A streptococcus isolates were collected in Sweden from 23 April 2002 through 31 December 2004. The incidence is calculated as a yearly mean for each age group during the whole study period. Fatal cases are based on available information for 83% of all patients. F, female; M, male.

1). The most frequent emm type in this group was emm type 81, which was not the case for patients without the manifestation (P ! .05) (table 2). In contrast, emm type 1 was the most prevalent type among both patients with STSS and patients with NF (P ! .001 and P ! .05, respectively, compared with patients outside these groups). STSS was signicantly more prevalent among NF cases than among cases with other skin and/ or tissue involvement (51% vs. 7.4%; P ! .001). A signicantly higher fatality rate was noted among patients with NF who had STSS than among patients with NF who did not have STSS (28% vs. 7.7%; P p .044) (table 1). However, no type was signicantly associated with fatal outcome (table 2). Women with puerperal sepsis accounted for 3% of all patients with invasive cases, none of whom had fatal outcome. The most common emm types were emm types 28 and 89, with emm type 28 accounting for 35% of these infections, compared with 14% of other cases (P p .018) (table 2). Also, the urogenital tract as site of entry of the infection was more common among women than among men and was signicantly associated with emm type 28 (P p .037). Noninvasive GAS disease. In total, 773 isolates were collected from skin or throat specimens and were considered to represent the distribution of noninvasive isolates in the country, because the 6 laboratories involved were geographically spread. Patients with noninvasive disease were younger than those with invasive disease, and there was also a signicant age difference between patients with skin infection and patients with throat infection (table 1). The same 7 emm types that were found to be most prevalent among patients with invasive disease were also the most prevalent among patients with noninvasive disease, accounting for 72% of the isolates. However, the emm type distribution differed signicantly between invasive and noninvasive isolates (P ! .001 ) (gure 2B and table 2). The emm type distribution also differed between throat and skin isolates (P ! .001) (table 2). emm Type distribution differed with sex, geographic area,
452 CID 2007:45 (15 August) Darenberg et al.

and time. Among all invasive and noninvasive cases, 60% of emm type 81 isolates were from male patients (P p .001), whereas 59% of emm type 28 isolates were from female patients (P p .026). The difference for emm type 28 was even stronger among invasive cases alone (P p .008 ). Furthermore, the T and emm type distribution of invasive isolates varied over time (gure 2A and data not shown). In contrast, no distinct uctuations over time were found among noninvasive isolates, among which emm type 28 isolates dominated during the whole study period. Geographical differences in type distribution were also found and recognized between larger urban areas. Invasive disease clinical characteristics as predictors of emm type. To correlate clinical information to specic emm types, a multinomial regression analysis was performed. Only the 7 most prevalent emm types were considered in the analysis, and all others were grouped into a single category (others), which served as a reference in the model. The following characteristics were included as explanatory variables in the model: (1) skin and/or tissue infection, (2) puerperal sepsis, (3) NF without STSS, (4) STSS, (5) adult respiratory distress syndrome, (6) cancer and/or immunosuppressive treatment, (7) age, and (8) sex. The statistically signicant variables that we found were skin and/or tissue infection, STSS, and age (log likelihood, 2531.9; P ! .05). These variables could, to a certain degree, be used as emm type predictors. Among patients with skin and/ or tissue infection, the predicted probability of an emm type 4 infection depended on age; the highest probability of having emm type 4 was among younger patients, and the lowest probability was among elderly patients. In contrast, the probability of an emm type 89 infection was high among elderly patients. Irrespective of age, the predicted probability of an emm type 81 infection was highest among patients with skin and/or tissue infections. Also, the predicted probability was highest for emm type 1 among patients with STSS but decreased somewhat with age. Clonal analysis correlated with emm types. PFGE analysis

Downloaded from http://cid.oxfordjournals.org/ by guest on March 13, 2014

Table 1. Demographic characteristics in relation to group A streptococcal disease manifestation.

Invasive disease Puerperal sepsis NF 62 (9.5) 63.6 62 (1391) 51.6 19.0 66.7 1.77 100 0 71.0 3.1 17 (2.6) 31.7 30 (2141) 70 (1391) 0 0 20.0 0.35 0 5.9 (0) 7.7 56.7 67.6 30 (4.6) NF without STSS

Noninvasive disease

Characteristic 654 (100) 63.6 68 (099) 48.7 14.5 68.6 0.91 9.5 (18.9) 11.3 (38.9) 13.4 (19.0) 13.2 (37.3) 73.8 1.01 15.7 51.6 71 (099) 66.8 461 (70.4)


Skin and/or tissue b infections

STSS 74 (11.3) 61.9 62 (3191) 41.9 38.9 68.5 4.0

Fatal outcome 94 (14.5) 75.7 38.9 100 84.2 1.82

All 773 28.9c 47.4


Skin 384 35.6d 51.9 NA NA NA


Throat 389 22.4e 79 (3998) 27 (0102) 35 (0102) 18 (096) 43.0 NA NA NA


No. (%) of patients Age, years Mean

Median (range)

Male sex

Overall CFR

NA NA NA 100 43.2 (28.1) 51.6 (28.1) 100 11.6 29.5 NA NA NA NA NA NA

Underlying condition Mean no. of organ failuresf NF (CFR) STSS (CFR)

NOTE. Data are percentage of patients, unless otherwise indicated. CFR, case-fatality rate; NA, not applicable; NF , necrotizing fasciitis; STSS, streptococcal toxic shock syndrome.

Based on information from returned questionnaires for 654 of 746 patients with invasive disease. The patient groups for invasive disease are overlapping, which has been taken into consideration during statistical analysis (results not shown in the table). b Patients with symptoms from skin and/or soft-tissue infection or skin stated as the probable entry of infection. c Information available for 753 patients. d Information available for 374 patients. e Information available for 379 patients. f Per case presenting with a specic disease manifestation.

Downloaded from http://cid.oxfordjournals.org/ by guest on March 13, 2014

Downloaded from http://cid.oxfordjournals.org/ by guest on March 13, 2014

Figure 2. Major emm and T types for invasive and noninvasive isolates. A, The emm type distribution for invasive isolates in 4-month intervals. B, The proportions and combinations of emm and T types for the 746 invasive isolates and 773 noninvasive isolates collected during the study period. Types including !5% of either invasive or noninvasive isolates are not specied in detail but are grouped in the other category. T types are shown as the percentage of combined invasive and noninvasive isolates.

of all invasive and noninvasive isolates of emm types 89, 81, 77, 28, 12, and 1 (992 isolates total) revealed that most of the isolates tested (97%) grouped in concordance with their emm type, and no discrimination between PFGE patterns of invasive and noninvasive isolates could be noted. Strikingly, emm types 1 and 81 isolates were highly homogenous, clustering into 1 PFGE group each. In contrast, emm types 89 and 28 formed 4 groups each, and 2 groups were noted for both emm type 12 and 77. No specic disease manifestation or outcome correlated to any specic PFGE group. SAg gene proles as markers for invasiveness. The mean number of SAg genes present in invasive and noninvasive isolates ranged from 4.0 to 6.6 (table 3), and differences in SAg gene proles were noted between emm types and their PFGE determined subgroups (table 4). The speA gene was commonly detected among both invasive and noninvasive emm type 1 isolates (83% and 79%, respectively) but was rarely found among other types (P ! .001). speH was particularly linked to emm type 12, speJ was linked to emm types 1 and 28, and the gene for the streptococcal SAg was most common among emm type 4 isolates. The speC gene was less common among emm types 89, 81, and 1 than among other types (43% total positive isolates among emm types 89, 81, and 1 vs. 73% among the
454 CID 2007:45 (15 August) Darenberg et al.

other types; P ! .001). The speB, speF, speG, and smeZ genes were found in the majority of isolates. Thus, some invasive emm type 81 isolates lacked the same smeZ gene or had an allelic variant not detected by our primers. These isolates showed similar PFGE banding patterns as isolates with the gene present (table 4). Similar results were found for speG among emm type 77 and emm type 4 isolates (table 3); however, the emm type 77 negative isolates did not belong to the same PFGE subgroup. The emm type subgroups, as identied by PFGE, could be divided by SAg genotyping. As shown in table 4, 2 specic SAg proles within groups 89A and 81A were signicantly more prevalent among invasive than noninvasive isolates. No specic SAg prole within any of the 14 PFGE subgroups, was statistically more prevalent among noninvasive isolates. DISCUSSION The present nationwide active surveillance study of invasive GAS infections in Sweden includes 11500 isolates from patients with invasive and noninvasive GAS disease. Clinical information was related to molecular characteristics of all isolates, aiming at identifying GAS virulence markers and other attributes

Table 2. emm type distribution in relation to invasive and noninvasive disease, invasive disease manifestation, and origin of noninvasive infection.
Origin of noninvasive infection

Invasive disease manifestation Skin and/or tissue infection 461 15.2 17.1 11.3 12.8 6.3 7.0 6.5 23.9 .029


Variable Total no. of isolates emm Type 89 81 28 1 12 77 4 Other P

Invasive disease 746 15.7 14.5 13.9 11.9 6.3 5.9 5.9 25.9

Noninvasive disease 773 8.1 10.6 16.2 7.4 14.2 6.1 9.1 28.3

Total 654

Puerperal sepsis 17 23.5 0 35.3 11.8 11.8 0 0 17.6


NF without STSS 30 6.7 23.3 20.0 10.0 10.0 6.7 13.3 10.0 .090

NF 62 8.1 14.5 12.9 22.6 6.5 4.8 9.7 21.0 .128


STSS 74 13.5 12.2 10.8 28.4 6.8 5.4 4.1 18.9 .003

Fatal outcome 95 19.0 11.6 10.5 14.7 8.4 8.4 3.2 24.2 .537

Skin 384 8.6 18.5 15.6 5.5 8.3 7.0 8.1 28.4

Throat 389 7.5 3.1 16.7 9.3 20.1 5.1 10.0 28.3

16.4 15.3 13.5 12.1 6.4 5.8 6.0 24.6

Downloaded from http://cid.oxfordjournals.org/ by guest on March 13, 2014

NOTE. Data are percentage of isolates, unless otherwise indicated. NF , necrotizing fasciitis; STSS, streptococcal toxic shock syndrome. Based on information from returned questionnaires for 654 of 746 patients with invasive disease. The type distribution within each patient group is compared with the distribution among all patients who did not present with the specic disease manifestation. c Determined using x2 test. d Comparison of invasive and noninvasive isolates. e Determined using Fishers exact test. f Comparison of skin and throat isolates.
b a

that are important for disease severity and outcome. Although several clinical studies involving invasive GAS infections have been conducted around the world, few reports have included noninvasive isolates (e.g., from tonsillitis, impetigo, or asymptomatic carriage) [2528]. The present study compares isolates from invasive and noninvasive GAS disease, collected during the same time period and from the same geographic area, to correlate molecular characteristics to disease severity and to estimate the invasiveness of different GAS isolates and/or clones. Previous national epidemiological studies of invasive GAS infections in Sweden revealed an increasing incidence, from 1.8 cases per 100,000 population in 1987 [29] to 12.9 cases per 100,000 population during 19941995 and 19961997 [27, 30]. This is in accordance with the increase noted in several European countries during the 1990s [31]. In our study, the mean annual incidence was 3.0 cases per 100,000 population. Skin was found to be the major port of entry of infection, and skin infection was the major clinical feature among patients with invasive disease. The majority (69%) of all patients had underlying medical conditions, except for those with puerperal sepsis. The overall fatality rate was 14.5%, but the rate was higher (39%) among the 11% of patients who developed STSS. These ndings are consistent with previous ndings from the period 19961997 [27]. However, patients with STSS did not differ with regard to age or prevalence of underlying disease,

compared with all patients with invasive disease. This suggests that host susceptibility plays an important role in determining severity of disease. Kotb et al. [32] reported an association between outcome of invasive GAS infection and HLA class II haplotype, which was linked to the inuence of HLA class II and the magnitude of SAg-mediated inammatory responses. In contrast to the Swedish surveillance studies performed during 19941995 and 19961997, in which T types 1 and 28, respectively, dominated the T type cluster T3/13/B3264 dominated among both invasive and noninvasive cases in our study. This cluster was mainly associated with the recently described emm types 89 and 81, which is a new nding, because emm type 89 has previously been most commonly associated with T11 [6]. Thus, emm types 89 and 81 accounted for 30% of invasive isolates, and to our knowledge, it is the rst time in western countries that 2 emm types of high numbers (180) have dominated a national surveillance study. Furthermore, the type distribution also uctuated over time and with geographic area. In agreement with previous reports, associations between emm types and particular disease manifestations were observed [711]. emm type 28 was associated with puerperal sepsis and, also, was more prevalent among female patients. As expected, emm type 1 was strongly associated with STSS. Furthermore, emm type 81 was more prevalent among male patients and was signicantly associated with skin or tissue involvement and/or
Invasive GAS Infection CID 2007:45 (15 August) 455

Table 3. Superantigen (SAg) gene distribution within major emm types of invasive and noninvasive isolates.
Percentage of positive isolates Mean no. a of SAgs I 4.5 4.0 5.9 5.9 5.8 4.2 5.4 5.1 NI 4.8 4.6 5.9 6.1 5.6 4.2 5.4 5.2 I 1 0 0 83 0 0 0 1

No. (%) of isolates emm Type 89 81 28 1 12 77 4 Other I 117 (16) 108 (14) 104 (14) 89 (12) 47 (6) 44 (6) 44 (6) 297 (26) NI 62 (8) 83 (11) 125 (16) 57 (7) 110 (14) 47 (7) 70 (9) 219 (28)

speA NI 6 2 5 79 7 0 0 1 I

speC NI 63 41 92 54 66 77 97 68 I

speG NI 100 100 99 100 100 30 36 99 I 3 11 15 5 94 9 0 16

speH NI 11 11 20 5 91 2 3 17 I 2 1 76 63 9 0 0 13

speJ NI 3 11 73 70 2 6 16 15 I 1 0 0 5 6 5 98 30

ssa NI 0 5 5 5 2 6 90 24 I

smeZ NI 98 94 99 98 100 100 100 96

47 29 97 36 75 80 100 52

99 100 100 100 100 30 43 98

98 59 100 100 100 98 100 95

NOTE. The vast majority of the isolates harboured the genes for speB (195%) and speF (198%). I, invasive isolates; NI, noninvasive isolates; smeZ, streptococcal mitogenic exotoxin Z; spe, streptococcal pyogenic exotoxin; ssa, streptococcal SAg.

Mean number of SAg genes for all invasive isolates and noninvasive isolates was 5.1 and 5.4, respectively.

Downloaded from http://cid.oxfordjournals.org/ by guest on March 13, 2014

originthe latter in line with studies performed in the United Kingdom that also reported emm type 81 as an important cause of skin diseases [33, 34]. However, when comparing the present study with the previous Swedish study performed during 1996 1997, we found that the spectrum of clinical manifestations and their severity did not change markedly, despite changes in type distribution [27]. Interestingly, we found a correlation between patient age and emm type distribution. For example, emm type 4 was more common among younger patients with skin and/or tissue infections; in patients with STSS, the prevalence of emm type 1 decreased with age, and the prevalence of emm type 89 increased with age. Also, we found differences in type prevalence between

invasive and supercial diseases. Among invasive cases, emm type 89 was the most prominent type, followed by 81, 28, 1, and 12. In contrast, emm type 28, followed by types 12, 81, 4, and 89, was the leading type among noninvasive cases. Finally, emm type 12 was the most prevalent type among noninvasive throat isolates, which is in agreement with previous ndings [35]. However, we could not pinpoint any specic emm types that were responsible for invasive or noninvasive disease, because all major types could cause a wide spectrum of disease manifestations. Several GAS genomes, of 9 different M types, have been sequenced [36]. A major part of genetic differences observed are located within 35- to 45-kb insertions corresponding

Table 4. Superantigen (SAg) gene proles among emm81 and 89 PFGE groups.
Major SAg gene proles Spe PFGE group, prole 81A Prole 1 Prole 2 Prole 3 89A Prole 1 Prole 2 89B 89C 89D

No. (%) of isolates G + + + + + + + + H J ssa smeZ + + + + + + + All


B + + + + + + + +

C + + + + +

F + + + + + + + +


Noninvasive 32 (42.1) 20 (26.3) 2 (2.6) 18 (45.0) 11 (27.5) 5 (71.4) 9 (100) 4 (100)

67/180 (37.2) 39/180 (21.7) 38/180 (21.1) 73/109 (67.0) 21/109 (19.2) 33/40 (82.5) 15/17 (88.2) 5/6 (83.3)

35 (33.6) 19 (18.2) 36 (34.6)


55 (79.7)d 10 (14.4) 28 (84.4) 6 (75.0) 1 (50.0)

NOTE. smeZ, streptococcal mitogenic exotoxin Z; spe, streptococcal pyogenic exotoxin; ssa, streptococcal SAg. All isolates within the PFGE group with identical SAg prole. Invasive isolates within the PFGE group with identical SAg prole. Noninvasive isolates within the PFGE group with identical SAg prole. d Invasiveness of a PFGE group (i.e., difference in prevalence of invasive and noninvasive isolates with identical SAg proles and belonging to the same PFGE group) were assessed by Fishers exact test, 2-sided P ! .001.
b c

456 CID 2007:45 (15 August) Darenberg et al.

to prophages, prophage-like elements, and other exogenous sequences [37, 38]. Several of these prophages encode extracellular virulence factors, such as SAgs (SpeA, SpeC, SpeH, SpeI, SpeK, SpeL, SpeM, and streptococcal SAg), whereas other SAgs (SpeB, SpeF, SpeG, and SmeZ) are chromosomally encoded. We characterized the clinical isolates with respect to 9 SAg genes and found associations between emm/PFGE subgroups and SAg proles. Importantly, some of these proles, in combination with the PFGE subgroups, differed depending on invasiveness, which argues that the mechanisms for differences in invasive disease potential among different GAS isolates may, at least in part, be attributed to horizontally acquired prophages and other mobile DNA elements. However, the mechanisms for the increased invasiveness of a particular PFGE subgroup with a certain SAg prole may not be because of SAg expression per se, because SAg proles that were overrepresented among invasive isolates of PFGE groups 89A and 81A harbored fewer SAg genes than did other proles within the same groups (table 4). Overall, noninvasive isolates harbored slightly more SAg genes than did invasive isolates. The gene proles were, to a large extent, also linked to emm types of the isolates (table 3). Importantly, the use of these combined typing methods allowed identication of particular clones with higher invasive disease potential. It will be interesting to make a comparative genomic analysis between the clones identied, to possibly identify genes of direct relevance for invasive disease. In conclusion, the present surveillance study reveals a shift in the epidemiology of invasive GAS infection in Sweden, with a dominance of emm types that, previously, were not observed to have a high incidence. No correlation was found between invasiveness and the presence of single SAg genes. However, certain SAg proles within clonal clusters, as determined by PFGE and emm type, were found to be more prominent among invasive isolates, suggesting an increased invasive disease potential. If we can identify clones that are particularly invasive and, in the future, also understand which properties affect virulence, we will enable better and more focused strategies for treatment and intervention, as well as for prevention (by creating a basis for vaccine development). Both the M protein and the recently described pilus-like structure [39], which has been associated with T types, constitute potential vaccine candidates, emphasizing the need for knowledge of emm and T type distributions among GAS isolates. In addition, studies of GAS isolate SAg proles may represent a useful strategy to identify subclones that are particularly prone to cause invasive disease, to the benet of surveillance and prevention of these important infections.
We thank all clinical microbiology laboratory workers in Sweden, for sending the isolates to the Swedish Institute for Infectious Disease Control,

and especially those who additionally supplied noninvasive isolates; clinicians, for completed patient questionnaires; C. Johansson, G. Mo llerberg, I. Andersson, M. Kais, and S. Melin, for skillful technical assistance; and S. Ku hlman Berenzon, for expert statistical advice. Financial support. The European Unions 5th Framework Program (Strep-EURO,QLK2.CT.2002.01398). Potential conicts of interest. All authors: no conicts.

1. Vlaminckx B, van Pelt W, Schouls L, et al. Epidemiological features of invasive and noninvasive group A streptococcal disease in The Netherlands, 19921996. Eur J Clin Microbiol Infect Dis 2004; 23:43444. 2. Efstratiou A. Group A streptococci in the 1990s. J Antimicrob Chemother 2000; 45:312. 3. Davies HD, McGeer A, Schwartz B, et al. Invasive group A streptococcal infections in Ontario, Canada. Ontario Group A Streptococcal Study Group. N Engl J Med 1996; 135:54754. 4. McCormick JK, Yarwood JM, Schlievert PM. Toxic shock syndrome and bacterial superantigens: an update. Annu Rev Microbiol 2001; 55: 77104. 5. Li Z, Sakota V, Jackson D, Franklin AR, Beall B. Array of M protein gene subtypes in 1064 recent invasive group A streptococcus isolates recovered from the active bacterial core surveillance. J Infect Dis 2003; 188:158792. 6. Johnson DR, Kaplan EL, VanGheem A, Facklam RR, Beall B. Characterization of group A streptococci (Streptococcus pyogenes): correlation of M-protein and emm-gene type with T-protein agglutination pattern and serum opacity factor. J Med Microbiol 2006; 55:15764. 7. Talkington DF, Schwartz B, Black CM, et al. Association of phenotypic and genotypic characteristics of invasive Streptococcus pyogenes isolates with clinical components of streptococcal toxic shock syndrome. Infect Immun 1993; 61:336974. 8. Beres SB, Sylva GL, Barbian KD, et al. Genome sequence of a serotype M3 strain of group A streptococcus: phage-encoded toxins, the highvirulence phenotype, and clone emergence. Proc Natl Acad Sci U S A 2002; 99:1007883. 9. Bisno AL, Brito MO, Collins CM. Molecular basis of group A streptococcal virulence. Lancet Infect Dis 2003; 3:191200. 10. Areschoug T, Carlsson F, Stalhammar-Carlemalm M, Lindahl G. Hostpathogen interactions in Streptococcus pyogenes infections, with special reference to puerperal fever and a comment on vaccine development. Vaccine 2004; 22(Suppl 1):S914. 11. Green NM, Zhang S, Porcella SF, et al. Genome sequence of a serotype M28 strain of group a streptococcus: potential new insights into puerperal sepsis and bacterial disease specicity. J Infect Dis 2005; 192: 76070. 12. Kaplan EL, Wotton JT, Johnson DR. Dynamic epidemiology of group A streptococcal serotypes associated with pharyngitis. Lancet 2001; 358: 13347. 13. OBrien KL, Beall B, Barrett NL, et al. Epidemiology of invasive group a streptococcus disease in the United States, 19951999. Clin Infect Dis 2002; 35:26876. 14. Schmitz FJ, Beyer A, Charpentier E, et al. Toxin-gene prole heterogeneity among endemic invasive European group A streptococcal isolates. J Infect Dis 2003; 188:157886. 15. Martin JM, Green M, Barbadora KA, Wald ER. Group A streptococci among school-aged children: clinical characteristics and the carrier state. Pediatrics 2004; 114:12129. 16. Kotb M. Bacterial pyrogenic exotoxins as superantigens. Clin Microbiol Rev 1995; 8:41126. 17. Norrby-Teglund A, Kotb M. Host-microbe interactions in the pathogenesis of invasive group A streptococcal infections. J Med Microbiol 2000; 49:84952. 18. Fraser J, Arcus V, Kong P, Baker E, Proft T. Superantigens: powerful modiers of the immune system. Mol Med Today 2000; 6:12532.

Downloaded from http://cid.oxfordjournals.org/ by guest on March 13, 2014

Invasive GAS Infection CID 2007:45 (15 August) 457

19. Working Group on Severe Streptococcal Infections. Dening the group A streptococcal toxic shock syndrome: rationale and consensus denition. The Working Group on Severe Streptococcal Infections [comment]. JAMA 1993; 269:3901. 20. Jasir A, Tanna A, Efstratiou A, Schalen C. Unusual occurrence of M type 77, antibiotic-resistant group A streptococci in southern Sweden. J Clin Microbiol 2001; 39:58690. 21. Beall B, Facklam R, Thompson T. Sequencing emm-specic PCR products for routine and accurate typing of group A streptococci. J Clin Microbiol 1996; 34:9538. 22. Centers for Disaease Control and Prevention. Streptococcus pyogenes emm sequence database. 2006. Available at: http://www.cdc.gov/ncidod/biotech/strep/strepblast.htm. Accessed 29 June 2007. 23. Stanley J, Desai M, Xerry J, Tanna A, Efstratiou A, George R. Highresolution genotyping elucidates the epidemiology of group A streptococcus outbreaks. J Infect Dis 1996; 174:5006. 24. Agresto A. Categorical data analysis. 2nd edit. Hoboken, NJ: Wiley, 2002. 25. Johnson DR, Wotton JT, Shet A, Kaplan EL. A comparison of group A streptococci from invasive and uncomplicated infections: are virulent clones responsible for serious streptococcal infections? J Infect Dis 2002; 185:158695. 26. Johnson D, Stevens D, Kaplan E. Epidemiologic analysis of group A streptococcal serotypes associated with severe systemic infections, rheumatic fever, or uncomplicated infections. J Infect Dis 1992; 166:37482. 27. Eriksson BK, Norgren M, McGregor K, Spratt BG, Normark BH. Group A streptococcal infections in Sweden: a comparative study of invasive and noninvasive infections and analysis of dominant T28 emm28 isolates. Clin Infect Dis 2003; 37:118993. 28. Ekelund K, Darenberg J, Norrby-Teglund A, et al. Variations in emm type among group A streptococcal isolates causing invasive or noninvasive infections in a nationwide study. J Clin Microbiol 2005; 43: 31019. 29. Stro mberg A, Romanus V, Burman L. Outbreak of group A strepto-











coccal bacteremia in Sweden: an epidemiological and clinical study. J Infect Dis 1991; 164:5958. Svensson N, Oberg S, Henriques B, et al. Invasive group A streptococcal infections in Sweden in 1994 and 1995: epidemiology and clinical spectrum. Scand J Infect Dis 2000; 32:60914. Lamagni T, Efstratiou A, Vuopio-Varkila J, Jasir A, Schalen C. The epidemiology of severe Streptococcus pyogenes associated disease in Europe. Euro Surveill 2005; 10:17984. Kotb M, Norrby-Teglund A, McGeer A, et al. An immunogenetic and molecular basis for differences in outcomes of invasive group A streptococcal infections. Nat Med 2002; 8:1398404. Colman G, Tanna A, Efstratiou A, Gaworzewska ET. The serotypes of Streptococcus pyogenes present in Britain during 19801990 and their association with disease. J Med Microbiol 1993; 39:16578. McGregor KF, Spratt BG. Identity and prevalence of multilocus sequence typing-dened clones of group A streptococci within a hospital setting. J Clin Microbiol 2005; 43:19637. Espinosa LE, Li Z, Gomez Barreto D, et al. M protein gene type distribution among group A streptococcal clinical isolates recovered in Mexico City, Mexico, from 1991 to 2000, and Durango, Mexico, from 1998 to 1999: overlap with type distribution within the United States. J Clin Microbiol 2003; 41:3738. Holden MT, Scott A, Cherevach I, et al. Complete genome of acute rheumatic fever-associated serotype M5 Streptococcus pyogenes strain manfredo. J Bacteriol 2007; 189:14737. Banks DJ, Beres SB, Musser JM. The fundamental contribution of phages to GAS evolution, genome diversication and strain emergence. Trends Microbiol 2002; 10:51521. Green NM, Beres SB, Graviss EA, et al. Genetic diversity among type emm28 group A streptococcus strains causing invasive infections and pharyngitis. J Clin Microbiol 2005; 43:408391. Mora M, Bensi G, Capo S, et al. Group A streptococcus produce piluslike structures containing protective antigens and Lanceeld T antigens. Proc Natl Acad Sci U S A 2005; 102:156416.

Downloaded from http://cid.oxfordjournals.org/ by guest on March 13, 2014

458 CID 2007:45 (15 August) Darenberg et al.