Introduction

Streptococcus pyogenes or group A streptococcus (GAS) is an exclusively human pathogen that causes mild infections of the throat and skin, as well as severe infections and life threatening complications. While most common clinical manifestations are non-invasive, GAS may cause invasive infections including bacteraemia, septicaemia, pneumonia, puerperal sepsis, necrotising fasciitis (NF) and osteomyelitis. The inflammatory response of the body to the exotoxins produced by GAS may lead to streptococcal toxic shock syndrome (STSS) [1].

GAS strains have been characterized by serological identification of the surface protein M, an important GAS virulence factor. With time, serology has been replaced by emm-typing as a predictor of the M serotype [2]. To date more than 200 different emm-types have been reported and associated with various types of disease manifestations and prevalence rates worldwide. In most high-income countries emm-types 1, 3, 12 and 28 have traditionally been associated with invasive GAS (iGAS) disease [3].

Many countries including Norway witnessed a change in iGAS epidemiology in the 1980s and 1990s, with a sudden increase in both incidence and severity of infections caused by genotypes emm1 and emm3 [4, 5]. Data collected from 1998 to 2007 document a gradual change in the epidemiology of circulating iGAS genotypes in Norway, with emm1 on a steady decline and emm28 on a steady incline [6–8]. By 2006 the most frequently reported iGAS genotypes in Norway were emm28 (20 %) followed by emm1 (14 %). This change was also reported from neighbouring countries, with Denmark, Sweden and Finland reporting emm28 as the dominant genotype in their respective countries [9–11], while emm1 remained the dominant genotypes in most other high-income countries [3, 12].

A correlation between emm-types and specific clinical manifestations has been suggested. In particular, genotypes emm1, emm3 and emm18 have all been associated with outbreaks of NF and STSS in Europe and the United States [13, 14]. However, this association has not been perpetual and many interplaying and regulating factors are involved [12, 15]. Resistance to antimicrobials may increase the severity of an iGAS infection. Although resistance to penicillin (the antimicrobial of choice) is absent in naturally occurring GAS isolates, resistance to macrolides and tetracyclines have been documented in many emm-types, including emm1, emm3 and emm28 [16].

Recently, Sweden reported a shift in the iGAS epidemiology with a sudden rise of emm1 following a rapid increase in the incidence of iGAS infections. This change has not been reported from any of the neighbouring countries [17, 18]. We performed this study to provide a better insight into the current epidemiological situation of iGAS infections in Norway. Here we describe iGAS cases in terms of age, sex, demographics and clinical manifestations, and iGAS isolates in terms of emm-type and sequence type distribution in the five-year period from 2010 to 2014. Furthermore, we explore the associations between different emm-types and the most frequently reported clinical manifestations of iGAS infections.

Methods

Case definition, setting and data collection

An iGAS infection was defined as the detection of a pathogen from a normally sterile body site. Since 1993, all iGAS infections in Norway have been mandatorily notifiable to the Norwegian Surveillance System for Communicable Diseases (MSIS) and isolates have been sent to the National Reference Laboratory (NRL) at The Norwegian Institute of Public Health (NIPH). From 2010 to 2014 the NRL received a total of 808 isolates associated with iGAS infections. Isolates that were received from the same patient with an identical molecular type less than three months apart from each other (n = 17), and isolates for which molecular typing results were inconclusive (n = 3), were excluded from the dataset. The remaining isolates (n = 788) were searched against the MSIS national data registry for linked iGAS cases. A total of 756 isolates (89 % of the notified cases) were matched in MSIS, and data on demographics, clinical manifestations and outcomes were linked with the molecular data for further analyses.

Multilocus sequence typing (MLST)

MLST was performed as described by Enright et al. [19]. Briefly, primers were synthesized by Eurofins Genomics (Ebersberg, Germany), amplified DNA was purified and sequenced using an ABI 3730 DNA analyzer (Applied Biosystems, Foster City, CA) and nucleotide sequences were analysed using SeqScape v2.5 (Applied Biosystems). Allele numbers and sequence types (STs) were recovered from the MLST database (http://pubmlst.org/spyogenes), with the new alleles and STs assigned by the curator of the database.

Emm typing

The 5′-end fragment of the M protein gene was used for emm-typing, amplified as described by Beall et al. [2]. Briefly, DNA from all isolates was extracted by boiling, and an 180 bp fragment of the emm gene was amplified using primers emm1-fwd and emm2-rev as published by CDC (http://www.cdc.gov/streplab/protocol-emm-type.html), with attaching an M13-tail on the primers to facilitate sequencing [20]. Amplicons were purified using ExoSAP-IT before sequencing on ABI 3730 DNA analyzer (Applied Biosystems). Sequenced reads were inspected in Sequencer 4.8 (Gene Codes, USA) and the emm-sequence database of CDC was searched for homology. Sequences with less than 100 % similarity were re-sequenced and submitted to the curator to obtain a new emm-type or subtype.

Phylogeny

Relatedness of all isolates was investigated using the MLST data on allelic profiles. All recovered STs from the dataset were included only once in the construction of an unrooted optimized maximum parsimony tree (simulated annealing). Allele numbers were used as unordered character data values with equal cost. A phylogenetic tree was constructed with highest resampling support after 500 bootstraps in BioNumerics v.7.5 (Applied Maths, Belgium).

Antibiotic susceptibility

Antibiotic susceptibility testing for iGAS isolates was introduced in 2013, and all isolates from 2013–2014 (n = 346) were tested using E-tests according to the manufacturer’s instructions (bioMérieux SA, Marcy l’Etoile, France). Briefly, fresh overnight cultures were grown at 35 °C in 5 % CO2, suspended in Mueller-Hinton (MH) broth to a density of 0.5 McFarland. Bacteria were evenly spread on the surface of an MH-F agar plate using a cotton swab and allowed to dry for 10 min before the E-test strips were applied. Antibiotics tested were: penicillin G, erythromycin, clindamycin, tetracycline, trimethoprim-sulfamethoxazole and chloramphenicol. Interpretation of results was according to clinical breakpoints set by the European Committee for Antimicrobial Susceptibility Testing (EUCAST).

Statistics

Incidence was calculated using population data as of 1st October each year from Statistics Norway (www.ssb.no). The discriminatory power of emm-typing and MLST was estimated by the Simpsons Diversity Index (SDI) with 95 % confidence intervals (CI), and the adjusted Wallace coefficient (AW) with 95 % CI was used for measuring agreement between partitions of the two typing methodologies (http://darwin.phyloviz.net). Statistically significant differences in proportions of genders in different age groups were detected using a two-tailed Fisher’s exact test and associations between variables were estimated using prevalence ratios (PR) and 95% CI in Stata version 13.1 (StataCorpLP, USA ). Exposure variables used were age groups, sex, health regions and emm-types, with clinical manifestations (n > 10), emm-types (n > 15) or antimicrobial resistance (n > 10) as outcome measures.

Results

Descriptive epidemiology

A total of 756 cases of iGAS and corresponding bacterial isolates from Norway were analysed from a 5-year period starting in 2010 (Table 1). Isolates were submitted to the NRL from all four health regions of Norway (South-East, West, Central and North) averaging 151 per year (range, 122–173). The overall incidence of iGAS was calculated to 3.0 per 100,000 population, ranging from 2.4 to 3.4 during the study period. Incidence per 100,000 population was highest in the South-Eastern and Western health regions (3.4) compared to the Northern health region (1.7).

Table 1 Annual cases and incidence of invasive group A streptococcal infections by health regions and sex in Norway 2010 to 2014
Full size table

Cases were slightly overrepresented by males (56 %) with an incidence of 3.4 per 100,000 males compared to 2.6 in females. The median age was 59 years (range, 0–102) and the age distribution of cases displayed three peaks in the 0–9, 40–49 and 60–69 age groups. Incidence of iGAS was seen increasing with increasing age groups (Fig. 1). Males were overrepresented among the <20 year olds and between 40 and 80 year olds, most profoundly in age group 0–9 years, where 79 % were male (n = 45, p-value 0.0002).

Fig. 1
figure 1

Age and sex distribution of invasive group A streptococcal cases and age group specific incidence per 100,000 in Norway, 2010–2014. Different shaded bars represent the two sexes (see legend), distributed into age groups of 10 years from left to right on the x-axis. The absolute numbers of cases are given on the primary y-axis. The dotted line represents the age group specific incidence given per 100,000 population on the secondary y-axis

Full size image

Data on clinical manifestation were complete for 664 (88 %) cases and these were grouped into the following manifestations (Fig. 2): sepsis (n = 323), necrotizing fasciitis (n = 71), pneumonia (n = 46), erysipelas (n = 19), arthritis/osteomyelitis (n = 23), meningitis (n = 11), abscess (n = 5), pyelonephritis (n = 9), endocarditis (n = 2) and other (n = 155). No cases of STSS were reported during the study period. The clinical manifestations were evenly distributed among the different health regions and between all years. However, arthritis/osteomyelitis was more frequently described for cases in age group 40–49 years (40 %), and NF was more frequently described for cases in age group 60–69 years (30 %) compared to other age groups.

Fig. 2
figure 2

Notified clinical manifestations of invasive GAS infections by percentage of the five most common emm-types in Norway, 2010–2014. Clinical manifestations are given on the y-axis and the percentage of isolates belonging to the five most common emm-types is given on the x-axis, coloured according to legend. Clinical manifestation data were complete for 664 (88 %) isolates; the number of isolates within each manifestation is in parentheses

Full size image

Data on outcome were available for 382 (51 %) cases. Death was recorded in 19 % (n = 72) of the cases with 64 % of them attributed to sepsis (n = 46) and 21 % to NF (n = 15).

Molecular epidemiology

The 756 iGAS isolates were typed into 52 distinct emm-types (80 emm-subtypes) and 67 STs. The most frequent emm-types were emm1 (n = 208; 28 %), emm89 (n = 89; 12 %) and emm12 (n = 65; 9 %) (Fig. 3), and the most frequent STs were ST28 (n = 205; 27 %), ST101 (n = 83; 11 %) and ST36 (n = 65; 9 %).

Fig. 3
figure 3

Proportions of the most common emm-types (for n > 15) in Norway, 2010–2014 (n = 756). Each year of isolation is represented by a separate bar on the x-axis. Percentages of isolates with specific emm-types are given according to legend, with emm-types arranged from high to low bottom up. Crude numbers of isolates per year for each emm-type is in parentheses

Full size image

The SDI with 95%CI was calculated for emm-type, emm-subtype and ST, scoring the highest for emm-subtypes (0.902) compared to ST (0.880) and emm-type (0.874) with overlapping 95%CI. AW measurement of agreement between the typing methods estimated high concordance of ST partitions included within emm-types (p-value <0. 001). For the most frequent STs (n > 20) the level of concordance with emm-types was 92–100 %, e.g., ST28 and emm1 (99 %), ST101 and emm89 (99 %), ST36 and emm12 (100 %), ST15 and emm3.1 (97 %), ST52 and emm28 (98 %), ST44 and emm66 (100 %), ST39 and emm4 (92 %), ST655 and emm112.2 (100 %).

The annual distribution of the most common emm-types showed major variations in relative frequency. Noticeably, emm1 was seen as the dominant emm-type during the entire study period, although a drop in frequency was recorded in 2011 and 2012, largely attributed to the South-Eastern and Western health regions of Norway. The decrease in emm1 was mirrored by an increase of emm12 in the same health regions. For emm28 we observed an increasing trend in all regions except in the Central health region of Norway, resulting in an increase from 4 % in 2010 to 15 % in 2014. The increase of emm28 was mirrored by a decrease in emm89 in the South-Eastern and Central health regions of Norway (Fig. 4, regional level data not shown).

Fig. 4
figure 4

Trends in proportions of the five most frequently isolated emm-types in Norway, 2010–2014

Full size image

With regards to clinical manifestations, emm1 was the most frequently recovered emm-type from cases presenting with meningitis (73 %), pneumonia (46 %), erysipelas (37 %), necrotizing fasciitis (32 %) and sepsis (27 %). Also emm1 was the most frequently recovered emm-type from all age groups with the exception of 20–29, 40–49 and ≥90 age groups.

Investigations into the potential evolutionary relationship between the genotypes circulating in Norway showed the existence of five distinct clusters (resampling score 0.74) (Fig. 5). The five most frequently recovered emm-types, namely, emm1, emm89, emm12, emm3 and emm28, all seemed to originate from evolutionary distinct clusters indicating divergent evolutionary paths.

Fig. 5
figure 5

Unrooted maximum parsimony tree based on MLST profiles of estimated potential relationship between all invasive GAS isolates in Norway, 2010–2014. All nodes represent unique sequence types (n = 67), with the corresponding emm-types labelled. Nodes (most frequent) of the five major emm-types are coloured according to the legend

Full size image

Antimicrobial susceptibility

All tested isolates (n = 346) were susceptible to penicillin and chloramphenicol, 15 % were resistant to tetracycline (n = 52), 3 % were resistant to erythromycin (n = 10) and trimethoprim-sulfamethoxazole (n = 12) and 2 % were resistant to clindamycin (n = 5). Tetracycline resistance was seen in isolates of 23 different emm-types, of which 13 (25 %) were emm112.2, seven were emm22, four were emm11 and three were emm73. Tetracycline resistance among emm112.2, emm22 and emm73 isolates accounted for 87 %, 41 % and 75 %, respectively.

Isolates typed as emm11 were the largest contributors to the observed clindamycin and erythromycin resistance, accounting for 4/5 and 4/10 of the resistant isolates, respectively. Only a single emm42 isolate was identified with a MDR phenotype, which was defined as resistance to three or more classes of antibiotics. The MDR isolate was from a patient with sepsis in the Northern health region in 2014, displaying resistance to erythromycin, trimethoprim-sulfamethoxazole and tetracycline.

Associations

Exploring potential associations between age groups of iGAS cases and implicated emm-types, we found emm66 to be more frequently associated with infections in age group 40–49 years (PR 7.3, 95 % CI 4.2–12.8) compared to other age groups. In addition, emm22 had a more frequent association to age group 30–39 years (PR 6.3, 95 % CI 2.5–16.0) compared to other age groups, while emm112.2 had a more frequent association to age group 50–59 years (PR 5.5, 95 % CI 2.5–12.2).

A significant association was observed for age groups 0–9 and 10–19 years with meningitis, PR 17.7 (95 % CI 4.8–64.8). No significant association was observed between sex and clinical manifestation or emm-type. With regards to association of emm-types with clinical manifestations of iGAS, we found that emm2 had a PR of 4.5 (95 % CI 2.3–8.9) for necrotizing fasciitis. Similarly, emm12 had a PR of 3.5 (95 % CI 1.4–8.5) for arthritis/osteomyelitis. All emm22 isolates were resistant to tetracycline with a calculated PR of 7.5 (95 % CI 5.8–9.9), and a PR of 7.4 (95 % CI 5.2–10.5) was calculated for emm112.2 and tetracycline resistance.

Discussion

We present the epidemiology of iGAS infections in Norway from 2010 to 2014, exploring the isolates submitted to NRL and their associated cases. The isolate collection was recovered consistently for the entire study period from laboratories all across Norway and matched against MSIS notification data to give the best possible representation of iGAS epidemiology in Norway.

Calculated annual and overall incidences were comparable to incidence rates calculated from iGAS notification data in MSIS. For 2014 MSIS reported an iGAS incidence of 3.7 per 100,000 in Norway, a rate which has been stable since 2006 and comparable to neighbouring countries, with the exception of the sudden increase reported from Sweden in 2012 and 2013 [17]. The South-Eastern and Western health regions are the two most populated health regions in Norway covering about 77 % of the total population. Incidence of iGAS cases in these regions were relatively stable compared to the Central and Northern health regions, where there seemed to be on an upwards trend since 2012. However, the increase in the central and northern part of the country did not affect the national incidence level. The reported increase in Sweden was noticeably marked in the older population (≥ 80 years) while in Norway the incidence in all age groups has remained stable.

Typically iGAS cases follow an age and sex distribution where both sexes are equally affected and an increasing incidence rate is seen with age. During the study period we documented a slight overrepresentation of males (56 %), compared to studies performed earlier (2006–2007) [6]. Among cases <10 years of age we uncovered a significantly higher proportion of males (79 %, p-value 0.0002). The skew in this age group was recorded in all years of the study period, highest in 2013–2014 (89 %, 24/27). Cases were reported from across Norway with no apparent change in reporting patterns or identified clusters that would suggest a sudden increase in this group. Data on clinical manifestation for this age group were either missing or unspecified for 35 % of the cases; for cases with a known clinical manifestation, no significant difference was observed between the two sexes. The underlying reasons for this pronounced skew towards males in the younger age group are unknown, but warrant investigation and close monitoring in the future.

The dominating emm-type among iGAS isolates in Norway was emm1, which has been the most prevalent emm-type since 2010 (Table 1). The shift from emm28 to emm1 seemed to have occurred somewhere between 2007 and 2010. The top five emm-types from 2010 to 2014 accounted for 64 % of the isolates and included emm1 (27 %) followed by emm89 (12 %), emm12 (9 %), emm3 (8 %) and emm28 (8 %). Comparing emm-type distribution between 2007 and 2014, emm28 the dominating type in 2007 (20 %) had declined to 15 %, and emm1 had increased from 14 % to 33 % in 2014 (Fig. 3). Five of the six most frequent emm-types in 2007, emm1, emm28, emm12, emm4 and emm3, remained among the top six emm-types in 2010 to 2014. However, emm82, the third most frequent emm-type in 2007 (14 %) was completely absent in 2010–2014, and was replaced by emm89 which had increased from 5 % in 2007. Isolates typed as emm89 were reported consistently in all years from 2010 to 2014, and from all health regions in Norway. Also in the UK and Canada a recent upsurge of emm89 and associated iGAS disease has been reported [13, 21]. Among high-income countries emm89 is now reported among the top five emm-types [12, 13, 18, 21–24]. Whole genome sequencing data on emm89 attribute its upsurge to the development of a successful clade within emm89 genotype which has incorporated a prophage-like element similar to that of emm1 [21].

Overall emm1 was the most frequently isolated emm-type from males and females in most of the age groups and from most clinical manifestations of iGAS infections (Table 1). However, no specific associations were seen between emm1 and a particular disease manifestation, a defined age group, or a resistance phenotype. A significant association was seen for emm22 in the age group 30–39 years and for emm112.2 in the age group 50–59 years. These particular emm-types had not been associated with disease in specific age groups before in Norway. In China, emm22 has been implicated in severe iGAS disease in children [25]. Internationally, resistance to both macrolide and tetracycline has been associated with emm22 [26, 27]. In our study, emm22 also showed a positive association with tetracycline resistance (PR 7.5, 95%CI 5.8–9.9), although this association was not present in 2007 [6].

emm112.2 is an unusual emm-type. First reported in isolates obtained in the 1990s from Ethiopian children with various streptococcal diseases, it has thus far not been recorded as an important emm-type from any high-income country [28]. Interestingly, it was now among the top ten genotypes recovered in Norway (Fig. 3). It seemed to be significantly associated with the age group 50–59 years and also significantly associated with tetracycline resistance. All emm112.2 isolates were ST655 and submitted from the South-Eastern health region of Norway. Reports from India have suggested emm112 to be frequently associated with skin infections [29]. Absence of emm112.2 from the other Nordic countries, its association with a particular age group, association with tetracycline resistance and recovery only from the South-Eastern health region suggested it to be a recently imported iGAS genotype, although we did not have data on travel and the ethnicity of cases to confirm this assumption.

In our data set emm2 had a positive association with NF (PR 4.5, 95 % CI 2.3–8.9) with five NF cases out of a total of 71. These five cases represented 31 % of all infections caused by emm2. Although NF caused by emm2 has been reported earlier, no significant association has been documented [30]. For NF a strong correlation has been observed with emm18, emm3, emm5 and emm12 [12]. In our study the most frequently recovered genotypes associated with NF cases were emm1 (n = 23), emm12 (n = 8), emm89 (n = 6) and emm3 (n = 6).

Data on the trends of the most frequently recovered emm-types suggested that emm12 mirrored the development of emm1, and emm28 was mirrored by emm3 (Fig. 4). The evolutionary relationship between the circulating genotypes in Norway displayed the presence of five distinct branches (Fig. 5). Thus, the increase of emm1 and corresponding decrease of emm12 may have indicated a shift in not only a single emm-type, but an evolutionary distinct cluster. A similar shift in dominant cluster could be proposed for emm28 contra emm3. It was also interesting to note that emm89, the second most prevalent emm-type, seemed to be more closely related to emm28, which might have given way for the increasing prevalence of emm89 (Fig. 4).

When looking into associations between emm-types and clinical manifestations, it should be taken into account that data on clinical manifestations were recovered from MSIS, which primarily serves as a notification tool and thus has its limitations with regards to completeness of the reported clinical data. Also for some of the clinical manifestations the numbers of cases were low and may thus lead to an overestimation of the associations. Therefore only the most frequent manifestations and most frequent emm-types were analysed for statistical associations. Our data are strengthened by high notification completeness and isolate submitting compliance across Norway and throughout the study period. Overall we believe the data gives a good overview of the most common clinical manifestations of iGAS infections and the emm-type distribution in Norway.

We conclude that emm1 was the dominant emm-type in Norway, recovered in all years from 2010 to 2014, from all age groups and from most clinical iGAS manifestations. We documented emm89 as an emerging emm-type in Norway, along with probable import of tetracycline resistant emm112.2 type ST655, both of which will require close monitoring in the future. ST distributions of the fluctuating emm-types suggested a mutually exclusive dominance of distinct evolutionary clades.