During the early 1980s, reports emerged from the then Czechoslovakia and the US describing a hitherto unrecognised complication of S. pyogenes infection, termed the ‘streptococcal toxic shock-like syndrome’ [7-10]. A review of these reports by a working group from the Centers for Disease Control and Prevention led to the establishment of a case definition for streptococcal toxic shock-like syndrome (STSS) [11]. The diverse spectrum of invasive diseases recognised as being caused by S. pyogenes included puerperal sepsis, necrotising fasciitis, septic arthritis, pneumonia, STSS and bacteraemia (without primary focus).

One of the most defining events for iGAS surveillance activity occurred in 1994 when a cluster of necrotising fasciitis cases was detected in the South West of England (Gloucestershire) [12]. This event acted as an important catalyst for a host of activity within and outside the United Kingdom (UK). Enhanced surveillance for iGAS was immediately implemented in the UK, with several European and other countries following suit. This response resulted in a small number of countries obtaining for the first time measures of disease-specific incidence, risk factors and outcome.

The impetus generated during the mid-1990s led to the establishment of an ad hoc World Health Organization (WHO) working group on S. pyogenes, comprised of representatives from streptococcal reference centres in the Czech Republic, Italy, New Zealand, UK, US and Canada. The main recommendation of the WHO consultations that followed was to support member countries in initiating comprehensive public health programmes for the control of GAS infections. The key priorities that emerged from the 1998 consultation included the urgent need to develop a mechanism to strengthen microbiological capacity and provide sustained support to an international network of laboratories, the need to evaluate the tools available for surveillance, and the need to include streptococcal infections in national public health priorities [13]. However, no definitive network across Europe was formed, and collaborations between European countries were undertaken on a largely informal basis, if at all. A formal European network was not established until 2002 [14].

Advances in microbiological characterisation of S. pyogenes
The application of molecular techniques within the last decade has contributed significantly to our understanding of the epidemiology and pathogenesis of S. pyogenes disease. The development and application of a wide range of methods for S. pyogenes characterisation led to a new era in typing and have also highlighted the importance of establishing standardised methods and agreed criteria for the interpretation and verification of types. External quality assurance (EQA) among international centres has therefore become recognised as essential for accurate epidemiological and microbiological surveillance of disease.

The emm gene which encodes the major virulence factor, the M protein, is used as the basis for the characterisation of GAS within the majority of international typing centres, although conventional serotyping is also used in some centres. A total of 93 validated M-types have been identified to date [15]. Most GAS isolates that are deemed non-typable by serological methods can be genotyped through emm sequence determination. The extensive N-terminal variability of emm genes forms the basis for the distinction of more than 170 designated emm-types described to date[16]. Marked changes in the distribution of M-types circulating in Europe were noted from the late 1980s onwards [17,18]. Observations of both severe and non-severe GAS manifestations suggest cyclical patterns of dominance between certain serotypes [18-20].

Molecular methodologies have allowed us to examine iGAS pathogenesis from a new perspective. S. pyogenes has a repertoire of pathogenic mechanisms that assure its success as a colonising and invading organism. Molecular technologies have also provided us with more subtle tools to help identify, distinguish and ultimately understand both the bacterial and the host mechanisms mediating pathogenesis.

Surveillance of iGAS disease in Europe
In the absence of an iGAS surveillance network, or informally agreed protocol, different countries within Europe have been undertaking surveillance of iGAS disease according to their own criteria. However, common methodological approaches emerged, allowing some degree of comparability of results.

Very few countries within Europe list iGAS disease manifestations among their notifiable diseases. In Norway cases of iGAS have been notifiable since 1975, and all severe GAS (i.e., including isolates from non-sterile sites, when accompanied by severe clinical presentation) since 1995 [21]. Finland similarly made iGAS disease notifiable in 1995. Ireland added iGAS infections in their recent review of infectious disease notifications, effected in 2004 [22]. With the exception of these countries, surveillance activities are therefore predominantly reliant on voluntary reporting systems. The Table summarises national or multi-site surveillance activity results identified from WHO European Region countries. Most operate through the capture of routine local microbiological diagnoses into a central data bank [TABLE]. The quality of data available through such systems is variable, both in terms of the breadth of information collected and completeness of reporting. Many such systems do not routinely capture clinical information, which is a particular shortfall for iGAS infection given the plethora of associated conditions.

As many countries within Europe have a recognised national reference centre for microbiological identification and typing of streptococci, surveillance activities have commonly used isolate submission for surveillance purposes. This provides information on microbiological characteristics of strains circulating within these countries, such as serotype (based on T and M proteins), sequence typing of the emm gene (emmST) and antibiotic susceptibility. A potential drawback can be referral bias, depending on which criteria are applied in choosing isolates or inviting isolate submission. Isolates that are sent primarily for ’epidemiological purposes’, usually referring to the determination of strain relatedness for outbreak control purposes, will be unlikely to represent the primarily sporadic bulk of iGAS cases. Referral on the basis of atypical microbiological characteristics or clinical features would also present a biased group of isolates. However, some countries attempt to circumvent these problems of biased sampling by requesting submission of all iGAS isolates. Many countries in Europe are using isolate referral-based and laboratory report-based surveillance systems in parallel.

Regardless of the methods used to obtain cases for surveillance purposes, routine sources of information have been periodically supplemented through invoking a period of enhanced surveillance, primarily to gain additional patient, clinical, microbiological and outcome measures. In accordance with the aim of Strep-EURO, eleven countries are now undertaking enhanced surveillance (see Forming a European network - Strep-EURO). Belgium also began enhanced surveillance in 2004 following an observed sudden increase in iGAS cases detected through their laboratory surveillance system [23].

Trends in iGAS disease in Europe in the 1990s
Some interesting parallels emerge when comparing results from surveillance activities across Europe over the last decade. Surveillance data from countries who have published five or more consecutive years’ results are shown in Figure 1. Results from these primarily Northern European countries show some interesting and not entirely uniform trends, although most left the 1990s with higher rates of iGAS reports than they entered the decade. Data from the Netherlands in particular contrast with those from other countries, with rates of iGAS halving between 1995 (4.0 per 100 000) and 1999 (2.0/100 000), although an upturn was subsequently observed [17]. Most of the other countries examined showed reasonably consistent results suggestive of an overall increase in incidence during the 1990s and into the 2000s, although trend patterns varied markedly from near linear to marked peaks and troughs. Data from Scotland are among the most compelling, showing marked year-on-year rises from the mid-1990s onwards when rates of GAS bacteraemia rose from 1.49/100 000 in 1996 to 3.66/100 000 in 2002, averaging at a 41% increase per year [24;25]. A more diluted rise in reports was also seen in England and Wales throughout the 1990s[26]. Data from Finland also showed a similar pattern, rising sharply from 1996 (1.17) to 2002 (2.95) [27]. Surveillance data from neighbouring Sweden (1993-1997) showed a less clear cut pattern, rising sharply between 1993 and 1996 before dropping back again to a more stable annual rate between 2.3 - 2.9/100 000 [28,29]. Following the initial reports of a marked rise in severe clinical manifestations of GAS infection in the mid-1980s [2], rates in Norway showed an initial fall only to re-escalate around 1993 [21]. More recent web-published data indicate further sharp rises in rates of iGAS in Norway between 1996 and 1999 when rates of reports more than doubled to reach 4.9/100 000 [30]. Even more pronounced changes are apparent in Iceland, the smallest of the countries examined, where rates of iGAS have swung from lows of around 1-2 per 100 000 to peaks above 6/100 000, the highest rates observed in any European country over this period. Published data from Denmark showed further increases in the early 1990s to those identified towards the end of the previous decade [3,31], with the latest estimate from 1998 standing at 3.3/100 000 [32].

Surveillance data from France from the early 1990s were suggestive of a downward trend in iGAS infections, although recent reports indicate a rise between 1999 and 2002 [33]. Annual reporting rates in Belgium show marked rises between 1994 and 2000[ 34], from less than 0.5 iGAS cases per 100 000 in the mid-1990s to approaching 1/100 000 in early 2000s. The subsequent downturn after 2000 may have been short-lived: recent indications suggested a resurgence in early 2004, triggering the initiation of enhanced surveillance activities [23].

It is interesting to note the pattern of the changes in rates of iGAS reports within countries. Whereas countries such as Norway, Sweden, Iceland and the Netherlands showed quite marked up- and downswings in their rates of disease, other countries such as the UK and France have not seen such marked changes. It may be that any potential outbreaks of GAS disease are masked in national data given the larger size of these countries. Regional analyses and further microbiological characterisation of isolates would need to be undertaken to confirm if this is the case. Regardless of the patterns of iGAS disease, there is general suggestion of increasing iGAS across Europe over the past two decades. The reasons behind this change remain unclear and warrant further investigation.

In comparing estimates of the overall burden of iGAS disease between countries, it is important to take into account the different case definitions used for surveillance purposes. Whereas routinely available data from the UK, Finland and France are based on blood culture isolates only (+/- CSF), most other countries monitor all sterile site isolates. The majority of iGAS disease result in bacteraemia, however, non-disseminated invasive infections have been found to account for around 10% of cases [9,35,36], although estimates as high as 24% have been documented [37]. This would in part account for the differences in rates shown in Figure 1, with countries including all sterile sites reporting generally higher rates than others. Some countries, such as Norway, also monitor cases where the clinical presentation indicates a severe infection but without a sterile site isolate being obtained, increasing case numbers by around a quarter if included [21]. Surveillance in Belgium is somewhat unusual in including deep ear sites within its routine case definition, adding approximately two thirds more cases, although data for sterile site isolates are also available separately, as reproduced in this paper [34]. Also of note are differences in coverage within each country. Although typically all laboratories are invited to report cases or submit isolates, in many countries such as Belgium and Hungary, coverage is known to be less than complete. In the UK, coverage is very good, although under-reporting by active laboratories is also known to occur [38]. With these factors borne in mind, there is a degree of congruence between countries over this period, with rates of GAS bacteraemia between 1.7 and 2.95/100 000 in the early 2000s, and rates of iGAS (all sterile sites) between 2.7 and 3.6/100 000. Exceptions to this are the recent peaks seen in Iceland (over 5/100 000) and earlier estimates from Italy and the Czech Republic (mid-1990s), both of which measured incidence of less than 0.5 per 100 000, considerably lower than contemporaneous estimates from other countries. To what extent these lower estimates reflect true differences in incidence or other influencing factors, such as differences in common clinical practice in microbiological sampling, is unclear.

Antimicrobial resistance of iGAS isolates in Europe
Unlike other streptococci, S. pyogenes has to date remained universally susceptible to the first line treatment of choice, penicillin. Given the development of penicillin-resistance in other members of the genus, continued monitoring of penicillin susceptibility of S. pyogenes remains essential. Aside from beta-lactams, monitoring of GAS isolate susceptibility to other classes of antibiotics is also important. Latest estimates of macrolide resistance in iGAS from across Europe are given in the Table. Substantial variations in prevalence of macrolide resistance are apparent, with no clear geographical association. Results from enhanced surveillance in Italy identified 32% of iGAS isolates as exhibiting resistance to macrolides, the highest among European countries during this period [39]. France has reported a steady escalation of erythromycin resistance since the mid-1990s, reaching 23% of iGAS isolates in 2002 [40]. Analysis of small collections of invasive strains from Russia and Portugal identified 11% as resistant to macrolides [41,42]. It is conceivable that some bias in the choice of isolates reported or referred may have influenced these results, although they may equally reflect the true level of macrolide resistance among strains circulating in those countries. A further possibility is that treatment with antibiotics prior to microbiological sampling is selecting for the more resistant strains. Results from other countries during the 1990s fall between 1% and 7% macrolide resistance in iGAS.

Prevention and control of iGAS disease
The overwhelming majority of iGAS cases arise in the community, 4%-13% of infections being acquired in hospital [5,37,43]. Of community-acquired cases, the vast majority are thought to arise sporadically [43,44], limiting opportunities for implementation of control measures. Although many clusters linked in time and place have been reported, many involve strains belonging to different M-types and/or cases with no apparent epidemiological link [12,45,46]. Where household clusters occur, their presentation can often be almost simultaneous, limiting the potential for effective intervention [47,48]. Attempts have been made in the US and Canada to quantify the risk of secondary or ‘subsequent’ cases, a third party often being the source of infection [43,44]. No European country has yet to publish any full guidance on the management of iGAS in the community, although France and the UK have begun this process [49].

Of the cases that arise in hospitals, clusters are not uncommon [50,51]. Cases of puerperal fever still occur, occasionally resulting in the death of otherwise healthy young women [21,29,52]. Reviews of maternal death both in the UK and the Netherlands identified increases in maternal sepsis involving S. pyogenes over the past decade [52,53].

Few countries within Europe undertake continuous surveillance in sufficient depth to be able to monitor trends according to specific risk factors or modes of acquisition. Enhanced surveillance data obtained during the 1990s in European countries confirmed the following as risk factors for iGAS disease: alcoholism, malignancy, diabetes, skin lesions, recent childbirth, steroid use and chickenpox [5,54,55]. However, in a substantial proportion of cases there is no evidence of any particular risk or predisposing factors, between 17-31% [3,56,57]. Risk factor information accompanying isolates referred to the national reference laboratory in the UK has yielded some startling trends in iGAS associated with injecting drug use (IDU). Of the isolates referred to the reference laboratory between 1995 and 2002, the proportion emanating from IDUs has risen dramatically from <5% to 15% [58]. Early results from the Strep-EURO programme in the UK have substantiated the importance of IDUs as a risk group for iGAS disease [35]. Analysis of S. pyogenes strains circulating in Switzerland between 1993 and 1997 identified the spread of distinct clones among IDUs [59], further supported by the results of a case-control study which identified a common place of drug purchase as strongly associated with iGAS disease [60].

Some countries in Europe have reported higher incidence of iGAS disease in particular ethnic groups. A study from southern Israel found a significantly higher risk of iGAS disease in its Bedouin population compared to the Jewish population [61]. Without adjustment for key variables that could explain these differences, it remains unclear to what extent this is attributable to genetic factors, frequency of underlying illnesses or living arrangements. Interestingly, a study from north London found higher rates of pharyngeal carriage of GAS in orthodox Jewish children and adults attending primary care services than other attendees, regardless of sore throat symptoms [62].

That iGAS disease occurs more commonly in late winter-early spring is a fairly ubiquitous finding across Europe [2,3,17]. Similar patterns are seen for scarlet fever and streptococcal pharyngitis [63,64]. The degree to which these seasonal patterns in GAS disease manifestations reflect climate-induced vulnerability to respiratory pathogens, or a sequel to viral respiratory infections remains unclear.

Forming a European network - Strep-EURO
In light of recent reports suggesting global changes in the epidemiology of severe clinical manifestations of GAS infection, and given the disparate and disconnected surveillance activities across Europe, leading microbiologists from across Europe have formed a unified network to take forward a programme of work relating to iGAS. Funded by the Fifth Framework Programme of the European Commission’s Directorate-General for Research, the Strep-EURO network was launched in September 2002 (www.strep-euro.lu.se). Eleven countries are participating in Strep-EURO [FIGURE 2], with overall coordination coming from the University of Lund in Sweden. The programme of work is divided into seven work packages. Central to these work packages is the collection of enhanced surveillance data, which each country undertook for a period of two years (1 January 2003 – 31 December 2004) using a standardised questionnaire for the capture of patient, clinical, risk factor and outcome data. Cases were defined through isolation of S. pyogenes from a site that is normally sterile or from a non-sterile site where accompanied by clinical symptoms indicative of STSS[11].

Each country undertook to capture information on all cases arising within their country, or within a geographical area with a definable catchment population. Some countries, such as the UK and Finland, maximised case ascertainment through linkage of microbiological reports and isolates referred to the national reference centre [65]. Isolates were sought from each case and sent to a central laboratory, usually the national reference laboratory. These strains will be subject to the same microbiological characterisation, primarily M serotyping and/or emm sequence typing (emmST). Detection of toxin and antibiotic resistance genes will also be performed on a subset of strains. A selection of strains will be characterised further by multilocus sequence typing (MLST) and pulsed-field gel electrophoresis (PFGE).

Three external quality assessment studies are included within the remit of the Strep-EURO programme, for serological and molecular (emm) typing, PFGE subtyping and antimicrobial susceptibility determination using phenotypic and genotypic methods.

Collection of these data should allow some meaningful and robust comparisons of cases arising in each country, in particular clinical manifestations, risk factors and strain characteristics. Results are also being pooled into a single Strep-EURO database held in Helsinki. Amalgamation of these diverse clinical, epidemiological and microbiological data promises to provide a powerful tool for examining the interrelation between these factors. Given the number of different M-types, and the diverse range of clinical manifestations associated with invasive GAS disease, pooling of data is essential to gain statistical power for the identification of significant associations, for example between virulence markers and pathogenicity or mortality.

It is hoped that results from Strep-EURO will substantially improve our understanding of the epidemiology of iGAS disease in Europe. This should in turn yield findings of public health value. Identification and quantification of clusters of iGAS will help direct prevention guidance, with early results already showing success in using Strep-EURO data for such purposes [49,66]. Efforts to enhance or initiate a system for the capture of iGAS cases are also proving successful in many countries. Evaluation of serotype distribution could be of benefit in the evaluation of the possible impact of polyvalent vaccines.

This article was written with contributions from the following on behalf of the Strep-EURO group: A Bouvet (National Reference Centre for Streptococci, France), R Creti (Istituto Superiore di Sanità, Italy), K Ekelund (Statens Serum Institut, Denmark), B Henriques-Normark (Swedish Institute for Infectious Disease Control, Sweden), M Koliou (Archbishop Makarios Hospital, Cyprus), P Kriz (National Institute of Public Health, Czech Republic), P Tassios (University of Athens Medical School, Greece), V Ungureanu (Cantacuzino Institute, Romania).

Acknowledgements
We acknowledge with thanks the following for supplying data for their country:

M Coyne (Scottish Centre for Infection and Environmental Health, Scotland) G Hanquet (Institut Scientifique de Santé Publique, Belgium), A Høiby (Statens Institutt for Folkhelse, Norway), KG Kristinsson (Landspítali Háskólasjúkrahús, Iceland), D Lévy-Bruhl (Institut de Veille Sanitaire, France), B Libisch (Országos Epidemiológiai Központ, Hungary), and B Vlamincx (Rijksinstituut voor Volksgezondheid en Milieu, Netherlands).

We would also like to extend our thanks to the following for their help with sourcing local information: J Begovac, G El Belazi, R Cano Portero, A Detcheva, C Furtado, J Granerød, I Lucenko, J McCarroll, J Paciorek, J Papaparaskevas, and M Straut.

The Strep-EURO project is funded by the European Commission’s Directorate-General for Research’s Fifth Framework Programme (QLK2.CT.2002.01398)