European Journal of Clinical Microbiology and Infectious Diseases

, Volume 23, Issue 1, pp 27–33

Multicentre Study of the Molecular Epidemiology, Serotypes and Antimicrobial Susceptibility Patterns of Invasive Streptococcus pneumoniae Invasive Isolated from Children in the Ille de France Area

Authors

    • Department of Hospital HealthCentre Hospitalier de Versailles
  • P. Ovetchkine
    • Paediatric DepartmentCentre Hospitalier Universitaire Jean Verdier
  • A. Collignon
    • Department of MicrobiologyCentre Hospitalier Universitaire Jean Verdier
  • C. Chaplain
    • Department of MicrobiologyCentre Hospitalier de Saint-Denis
  • E. Estrangin
    • Department of MicrobiologyCentre Hospitalier Intercommunal de Créteil
  • A. Fremaux
    • Department of MicrobiologyCentre Hospitalier Intercommunal de Créteil
  • P. Reinert
    • Paediatric DepartmentCentre Hospitalier Intercommunal de Créteil
  • P. Foucaud
    • Paediatric DepartmentCentre Hospitalier de Versailles
  • J.-C. Ghnassia
    • Department of MicrobiologyCentre Hospitalier de Versailles
  • R. Cohen
    • Department of MicrobiologyCentre Hospitalier Intercommunal de Créteil
  • J. Gaudelus
    • Paediatric DepartmentCentre Hospitalier Universitaire Jean Verdier
  • P.-Y. Allouch
    • Department of Hospital HealthCentre Hospitalier de Versailles
Article

DOI: 10.1007/s10096-003-1032-1

Cite this article as:
Decousser, J., Ovetchkine, P., Collignon, A. et al. Eur J Clin Microbiol Infect Dis (2004) 23: 27. doi:10.1007/s10096-003-1032-1

Abstract

Between January 1997 and April 2002, 73 consecutive invasive strains of Streptococcus pneumoniae were isolated from children under 16 years of age in four hospitals in suburban Paris. Their genetic diversity was investigated by serotyping and analysis of pulsed-field gel electrophoresis restriction patterns. Antibiotic susceptibility patterns were analysed by disk susceptibility testing and determination of minimal inhibitory concentrations. The genetic basis of macrolide resistance was investigated by polymerase chain reaction. Studies of penicillin and vancomycin tolerance were performed for each strain. Despite the high prevalence (45.2%) of penicillin-nonsusceptible Streptococcus pneumoniae, resistance to amoxicillin (1.4%) was rare, and no strain was resistant to cefotaxime. Overall, 4.1% of pneumococcal strains were resistant to penicillin. Penicillin or vancomycin tolerance was not detected in any of the 73 strains studied. Of the erythromycin-resistant strains (48%), all but one carried the ermB gene. No strains showing a decreased susceptibility to ciprofloxacin (MIC, >4 mg/l) or overexpressing an efflux pump inhibited by reserpine were isolated. The serotypes found, in order of frequency, were as follows: 18C, 14, 6B, 19F, 19A, 9V, 23F, 1, 7F, 9A, 38. Strains of penicillin-nonsusceptible Streptococcus pneumoniae belonged predominantly to serotypes 14, 6B, 9V, 9A, 23F, 19F and 19A. The seven-valent conjugated vaccine covered 85.5% of the serogroups isolated in children under 2 years of age and 65.6% of the serogroups identified in children over 2 years of age. The genetic analysis showed a high identity for some serotypes, such as 14/9V, 6B and 23F. The use of the seven-valent conjugated vaccine is a critical measure to prevent invasive pneumococci infections in children in the Ille de France area.

Introduction

Among bacterial species causing fatal community-acquired infections in children 10 days of age or older in France, Streptococcus pneumoniae is the second most frequent species isolated, after Neisseria meningitidis and before Bordetella pertussis [1]. The incidence of invasive pneumococcal infections among French children under 1 year of age is 42 per 100,000 and 13 per 100,000 among children between 1 and 2 years of age [2]. Moreover, France has the second highest rate of pneumococcal resistance to penicillin in Europe, just after Spain, with almost 50% of pneumococci being penicillin nonsusceptible Streptococcus pneumoniae (PNSSP) [3]. Data on serotype distribution, antibiotic resistance and epidemiological relatedness of Streptococcus pneumoniae strains from paediatric patients are sparse in France: the most recent published data are derived from strains obtained in 1997 [4, 5, 6]. This is of concern because it is possible for rapid changes to occur in the epidemiological characteristics of strains [7]. Recently, new vaccine strategies against Streptococcus pneumoniae were developed [8, 9]. The seven-valent conjugated vaccine was found to be effective in the prevention of invasive pneumococcal disease in children [10].

The aim of this study is to provide recent and accurate data about the antimicrobial susceptibility and genetic epidemiology of invasive Streptococcus pneumoniae strains isolated from children in four hospitals located in suburban Paris before widespread use of the seven-valent conjugated vaccine.

Materials and Methods

Strains

Strains of Streptococcus pneumoniae isolated from sterile sites (blood culture or cerebrospinal fluid [CSF]) in children over 2 months and under 16 years of age were included in the study. The strains were collected between January 1997 and April 2002 in four suburban hospitals near Paris: the Centre Hospitalo-Universitaire Jean Verdier (hospital A), the Centre Hospitalier de Saint Denis (hospital B), the Centre Hospitalier Intercommunal de Créteil (hospital C) and the Centre Hospitalier de Versailles (hospital D). The isolates were centralized in Hospital D. Only one strain per patient was studied; duplicates were discarded.

Serotyping

Strains were serotyped by means of the Quellung reaction (capsular swelling with type-specific antisera) using antisera provided by the Statens Serum Institute, Copenhagen, Denmark [11]. Cross-reactivity of serotypes within the same serogroup was assumed in the analysis of the theoretical coverage by the vaccines.

Analysis by Pulsed-Field Gel Electrophoresis

Pulsed-field gel electrophoresis (PFGE) of chromosomal DNA was performed using SmaI as described previously [12]. A contour-clamped homogeneous electric field DR III apparatus (Bio-Rad Laboratories, USA) was used for running the gels. Running conditions were 23 h at 11.3°C at 200 V ramped with an initial forward time of 5 s and a final forward time of 35 s. Gels were stained with ethidium bromide and photographed. Visual interpretation of strain relatedness on the basis of the PFGE pattern was in accordance with current consensus [13].

Antibiotic Susceptibility Testing

Determination of minimal inhibitory concentrations (MICs) of penicillin, amoxicillin, cefotaxime, erythromycin, quinupristin/dalfopristin, ciprofloxacin, levofloxacin, gatifloxacin, moxifloxacin, linezolid, teicoplanin and vancomycin was accomplished using the agar dilution method as described previously [14]. Susceptibility of the isolates to kanamycin, streptomycin, gentamicin, tetracycline, clindamycin, pristinamycin, chloramphenicol, cotrimoxazole and rifampicin was tested using the disk diffusion method as described previously [14]. Susceptibility data was interpreted according to the criteria of the Comité de l’Antibiogramme of the French Society of Microbiology [15]. Streptococcus pneumoniae BM 4203-R3 and BM 4205 from the Pasteur Institute of Paris (P. Courvalin, Unités des Agents Antimicrobiens) and ATCC 49619 were used as control strains.

Testing of Study Strains for Penicillin and Vancomycin Tolerance

The tolerance study was performed as described previously [16]. Briefly, each strain was inoculated in Mueller-Hinton broth and incubated at 37°C to reach an optical density (OD) of 0.3 at 620 nm, which corresponds with 0.5×108 colony-forming units (cfu) per milliliter. The culture was separated into three parts. Afterwards, penicillin and vancomycin were added separately to two of the three samples to reach a final concentration of 10 times the MIC; the third part was considered a control sample. After 4 h of exposure, serial dilutions were performed and inoculated onto Columbia agar supplemented with 5% sterile sheep blood. Results were interpreted for each strain as the log cfu reduction between the control and the culture in the presence of antibiotic. Each strain was tested three times in independent experiments, including the nonencapsulated laboratory control strain R6 and the tolerant, autolysin-defective laboratory mutant Lyt 4-4 (defective in the lytA gene, which encodes the major autolysin) provided by the French Pneumococci National Reference Centre (E. Varon). The limits for penicillin and vancomycin tolerance were chosen as decreases in viability of less than 2 and 2.7 log kill, respectively [16].

Identification of Macrolide Resistance Genes

The ermB and mef genes were detected after polymerase chain reaction amplification as described previously [17].

Identification of Efflux System Implicated in Fluoroquinolone Resistance

The MICs of norfloxacin were determined in the presence and the absence of 10 mg/l alkaloid reserpine (Sigma, France) to investigate the possibility of efflux pump involvement [18, 19]. Overexpression of an efflux pump was concluded if, in the presence of reserpine, the MIC of norfloxacin decreased by more than two dilutions.

Results

Seventy-three strains of Streptococcus pneumoniae were collected: 62 from blood cultures and 11 from CSF samples. Hospital A supplied 18 strains, hospital B 20, hospital C 11 and hospital D 24. Forty-one strains were isolated from children younger than 2 years of age (32 from blood cultures and 9 from CSF) and 32 from children between 2 and 15 years of age (29 from blood cultures and 2 from CSF). No significant difference was established between each hospital’s contribution in terms of age of the children or origin of the strains (data not shown).

The MICs, MIC ranges, MIC50s and MIC90s for the main antibiotic compounds are listed in Table 1. Overall, 45.2% of the isolates were PNSSP (47.5% of the strains isolated from children ≤2 years of age, 40.0% of the strains from children >2 years of age). Three strains (4.1% of isolates) were resistant to penicillin, one strain (1.4%) was resistant to amoxicillin and no strain (0%) was resistant to cefotaxime. The single strain resistant to amoxicillin (MIC, 4 mg/l) was resistant to penicillin (MIC, 2 mg/l). Penicillin tolerance was not detected in any of the 73 strains studied. Among the erythromycin-resistant strains (48% of isolates), all but one carried the ermB gene; the one remaining strain carried the mef gene. Of the 40 penicillin-susceptible strains of Streptococcus pneumoniae, 10 (25%) were resistant to erythromycin. The rate of clindamycin resistance as determined by the disk diffusion method was 46.5%. All isolates were susceptible to quinupristin/dalfopristin and pristinamycin. There was no isolate for which the MIC of ciprofloxacin exceeded 4 mg/l and no isolate for which the MIC of norfloxacin in the presence of reserpine showed a significant decrease (>2 dilutions). The MIC90s of levofloxacin, gatifloxacin and moxifloxacin were low: 1 mg/l, 0.5 mg/l and 0.25 mg/l, respectively. All isolates were fully susceptible to these three fluoroquinolones. All pneumococci were susceptible to vancomycin and teicoplanin. Vancomycin tolerance was not detected in any of the 73 strains studied. No strain was resistant to linezolid. When tested by the disk diffusion method, no strain was resistant to gentamicin or rifampicin. The rate of resistance (intermediary and resistant strains) was 34.3% for tetracycline, 23.2% for chloramphenicol and 39.7% for cotrimoxazole. Multiresistance (resistance to 3 or more drugs belonging to different antibiotic classes, i.e. beta-lactam agents, macrolides, aminoglycosides) was found in 42.5% (31/73) of the strains.
Table 1

In vitro activities of eight antimicrobial agents against 73 isolates of Streptococcus pneumoniae as determined by the agar dilution method

Antimicrobial agent

No. of isolates inhibited by MIC (mg/l)

MICrange (mg/l)

MIC50 (mg/l)

MIC90 (mg/l)

0.016

0.032

0.064

0.128

0.250

0.5

1

2

4

8

16

32

64

128

Penicillin

22

11

7

3

2

5

20

3

0.016–2

0.064

1

Amoxicillin

34

4

3

1

1

4

25

0

1

0.016–4

0.032

1

Cefotaxime

30

6

5

1

3

15

13

0.016–1

0.032

1

Erythromycin

31

0

2

4

1

0

1

0

5

29

0.064–128

0.5

128

Ciprofloxacin

11

60

2

1–4

2

2

Levofloxacin

5

64

4

0.5–2

1

1

Gatifloxacin

3

40

30

0.125–0.5

0.25

0.5

Moxifloxacin

3

63

7

0.064–0.25

0.125

0.25

Twenty different serotypes were identified. Serotype distribution according to the age of the children and the susceptibility of the strains to penicillin is shown in Table 2. Three of the nine strains belonging to serotype 6B and three of the four strains belonging to serotype 23F were multiresistant (resistance to beta-lactam agents, kanamycin, erythromycin, clindamycin, chloramphenicol, cotrimoxazole). Nine of the ten serotype 14 isolates and all of the serotype 9V isolates were PNSSP and exhibited dual resistance to erythromycin and clindamycin. In terms of serogroup, the rate of coverage by the conjugated heptavalent vaccine was 85.5% for the strains isolated from children under 2 years of age and 65.6% for those isolated from the older children.
Table 2

Distribution of different serotypes among the 73 clinical isolates of Streptococcus pneumoniae according to age of the children and susceptibility of the isolates to penicillin

Serotype

Total no. of isolates

No. of penicillin-susceptible isolates

No. of penicillin-intermediate isolates

No. of penicillin-resistant isolates

Children ≤2 years

Children >2 years

4a

-

2

2

-

-

6-Ba

8

1

2

7

-

9-Aa

1

2

-

3

-

9-Na

1

-

1

-

-

9-Va

3

1

-

2

2

14a

7

3

1

8

1

18-Ca

4

7

11

-

-

19-Aa

3

2

3

2

-

19-Fa

5

2

5

2

-

23-Fa

3

1

1

3

-

1

1

2

3

-

-

7-F

1

2

3

-

-

38

2

1

3

-

-

24-F

-

2

1

1

-

3

-

1

1

-

-

5

-

1

1

-

-

13

1

-

1

-

-

31

-

1

1

-

-

15-B

-

1

-

1

-

15-C

1

-

1

-

-

aSerogroup or serotype included in the heptavalent conjugated vaccine

The study of genetic and epidemiological relatedness by PFGE showed similarities in some serotypes (Fig. 1, Table 3). According to current consensus, the 12 strains belonging to serogroup 19 were epidemiologically unrelated [13]. Of the 11 serotype 18C isolates, 3 belonged to the same pulsotype and were closely related, although they originated from two different hospitals (pulsotypes B, B1 and B2). Half of the serotype 14 strains (5 isolates from 3 different hospitals) were indistinguishable, all being PNSSP (pulsotype A). All isolates belonging to serotype 9V (4 strains, also PNSSP, from 2 hospitals) were closely related to the previous serotype 14 cluster (pulsotypes A1, A2 and A3). The three multiresistant 23F serotype isolates provided by three different centers were indistinguishable (pulsotype C). The majority of isolates belonging to the 6B serotype, also multiresistant, belonged to two different pulsotypes (D and E). Although derived from two different hospitals, the two multiresistant 9A serotype isolates were closely related (pulsotypes F and F1). Finally, four pulsotypes (A+B+C+D) included strains isolated from 18 children (Table 3).
Fig. 1

Major PFGE patterns of serogroup 19 (lanes 1–5, serotype 19A; lanes 6–11, serotype 19F); serotype 6B (lanes 12–20); serogroup 9 (lanes 21–23, serotype 9A; lane 24, serotype 9N; lanes 24–28, serotype 9V); serotype 14 (lanes 29–37); serotype 18C (lanes 38–47); and serotype 23F (lanes 48–51). Lane M, molecular weight ladder

Table 3

Different pulsotypes among the main serotypes

Serotype

No. of isolates per serotype

Pulsotypea

No. of isolates per pulsotype

No. of PNSSP isolates

Resistance pattern

14

10

A

5

5

E, CLI

9V

4

A1+A2+A3

4

4

18C

11

B+B1+B2

3

0

23F

4

C

3

3

K, S, TE, E, CLI, C, SXT

6B

9

D+D1

3

3

K, E, CLI, C, SXT

E

4

2

K, E, CLI,

9A

3

F+F1

2

2

K, TE, E, CLI, SXT

PNSSP, penicillin nonsusceptible Streptococcus pneumoniae; C, chloramphenicol; CLI, clindamycin; E, erythromycin; K, kanamycin; S, streptomycin; SXT, cotrimoxazole; TE, tetracycline

aNumbering of pulsotypes in accordance with current consensus

Discussion

In France, Streptococcus pneumoniae is among the bacteria most frequently implicated in life-threatening infections such as pneumonia or meningitis in children [1, 20, 21]. Thus, prevention of pneumococcal disease by vaccination seems to be an effective strategy [10]. Longitudinal studies are needed to evaluate the impact of the new conjugated formulation on the epidemiology of pneumococcal disease; such studies should include evaluation of morbidity, mortality, serotype distribution and antibiotic resistance [9, 22, 23]. In this work, we studied the phenotypic and genotypic properties of invasive Streptococcus pneumoniae isolated in children between 1997 and 2002 in four hospitals in suburban Paris.

The proportion of PNSSP among invasive strains isolated from children reached 45.2%, as high as in previous studies [3, 24]. The rate of PNSSP was higher among younger children (≤2 years of age, 47.5%) than among older children (>2 years of age, 40%), as described previously [24]. In comparison to currently available French data, the number of strains highly resistant to penicillin was low (4.1% of isolates vs. 19%) [24]. The rate of amoxicillin resistance (1.4%) was consistent with the data reported in the literature [24]. The MIC of penicillin for the single strain resistant to amoxicillin was lower than the MIC of amoxicillin for the same strain (2 mg/l vs. 4 mg/l). This phenomenon was described previously in France [25]. The absence of cefotaxime-resistant strains also was reported previously [24]. In a recent multicentre study, no Streptococcus pneumoniae strains highly resistant to third-generation cephalosporins were identified among 315 blood isolates obtained from adults [14].

The comparison of these results with the situation in neighbouring countries emphasized two aspects particular to the French setting. First, the prevalence of PNSSP in France (43%) was high compared to northern European countries such as Germany (7%) [3, 26]. This difference may be explained at least in part by different antibiotic prescribing practices as well as different cultural and social factors [26]. Second, the number of strains highly resistant to amoxicillin or cefotaxime was significantly higher in Spain (5 and 7%, respectively), even though the rates of PNSSP in these two countries were similar [3, 27]. We found no explanation for this dramatic difference.

Concerning the non-beta-lactam antibiotics, it has been now clearly established that macrolide resistance is clinically relevant [28]. In our study, 48% of the Streptococcus pneumoniae isolates were resistant to erythromycin. This high rate of erythromycin resistance was in accordance with the well-established dual resistance to beta-lactam antibiotics [3]. The level of macrolide resistance was so high that 25% of penicillin-susceptible Streptococcus pneumoniae isolates were resistant to erythromycin. This is consistent with the data published previously in France [3, 14, 24]. These results still place France in the group of countries with the highest rates of macrolides resistance; at the same time, they confirmed the susceptibility of French isolates to synergistin, pristinamycin and dalfopristin/quinupristin [3, 29, 30]. Among erythromycin-resistant isolates, all but one were found to carry the methylase ermB gene. These data confirm previous reports of the distribution of the macrolide resistance gene in France [29, 30]. In France and globally in Europe, the high rate of erythromycin resistance in pneumococcal paediatric isolates was not due to the spread of mef-carrying strains, as in other countries [27, 31].

Some authors have examined the use of fluoroquinolones in children for the therapy of life-threatening infections caused by multiresistant strains [32, 33]. In our study, there was no Streptococcus pneumoniae isolate for which the MIC of ciprofloxacin exceeded 4 mg/l, which represented the threshold for low-level resistance. No isolate showed a significant decrease in norfloxacin susceptibility in the presence of reserpine, which means that no efflux susceptible to this alkaloid was overexpressed. This is of interest since this mechanism is probably implicated in the preliminary stages of development of fluoroquinolone resistance [18, 19].

Antibiotic-tolerant strains have been studied in experimental settings and in cases of clinical failure [34, 35, 36, 37]. The phenomenon of antibiotic tolerance may also represent a precursory stage leading up to antibiotic resistance [34]. Tolerance of some Streptococcus pneumoniae strains to the lytic action of antibiotic compounds such as beta-lactam agents has been described since the mid-1980s [38]. Vancomycin tolerance in Streptococcus pneumoniae has been demonstrated more recently [39]. To the best of our knowledge, the present work is the first large study investigating the prevalence of the penicillin and vancomycin tolerance among pneumococci isolated from children in France, although these two groups of antimicrobial compounds are indicated as first-line therapy for meningitis by national guidelines [40]. No strains showing penicillin or vancomycin tolerance were detected in any of the 73 strains studied. In previously published data, 8–30% of clinical isolates of pneumococci showed reduced lysis by benzylpenicillin and 3% exhibited a phenotype for tolerance to vancomycin [16, 34]. Henriques Normark et al. [16] have described a putative link between penicillin resistance and tolerance. Despite a high rate of PNSSP, no tolerant strains were identified in our work. In a recent study, a similar finding was established in Spain, a country with a high rate of intermediary and highly resistant pneumococci [41].

The main serotypes identified in this work were those previously established, although the ranking of these serotypes varied [42]. Serotype 18C was the most frequently isolated serotype, although it was not among the top three serotypes in the previous studies [42]. Only 5.5% of the strains studied belonged to serogroup 23, while previously, 10.75% of the invasive strains from paediatric patients belonged to this serogroup [42]. Three strains belonged to serotype 1, ranking this serotype eighth in frequency. All strains were fully susceptible to penicillin. This serotype could be of concern because it has been shown to spread quickly inside a community or a country [7, 43]. Moreover, serotype 1 seems to be particularly virulent: two recent studies emphasized the role of this serotype in paediatric empyema, one of the complications of bacterial pneumonia [44, 45, 46]. Serotypes 6B, 9V and 23F were isolated predominantly in children younger than 2 years of age, and serotype 18C was isolated predominantly in children older than 2 years of age. This difference between these two age groups was reported previously [4, 42]. The ranking of serotype frequency among PNSSP was as established previously: 14, 9, 6, 19, 23 [6, 47]. In the two most frequent serogroups, the percentage of PNSSP was higher than 80%. Of the strains belonging to serotype 9V, only one was highly resistant to amoxicillin. In terms of serogroup, the rates of coverage by the conjugated heptavalent vaccine were similar to those reported previously in France [4]. Moreover, cross-protection against the entire serogroup due to immunization against one serotype was still debated [9].

According to the results of PFGE, the analysis of genetic relatedness between strains showed a varying degree of identity depending on serotype. Strains belonging to the major previously established PNSSP serotypes were closely related (6B, 9, 14, 23F) and probably belong to internationally distributed multiresistant clones as described previously [5, 6, 48, 49, 50]. On the other hand, penicillin-susceptible isolates belong to sporadic clones. This difference has been described previously in France [6, 50].

Our study is limited to one geographic area, whereas serotype prevalence and patterns of resistance may vary from region to region. The number of strains studied did not allow us to compare the characteristics of strains obtained from CSF versus blood cultures, characteristics which could be significantly different [46]. Nevertheless, this work is the first study that (i) provides recent data about serotype distribution in the main region of France and (ii) supplies information on the intense circulation of some resistant clones in this area. Moreover, accurate data are provided about penicillin and vancomycin tolerance as well as resistance to fluoroquinolones and macrolides in invasive isolates obtained from French children. Further studies about the impact of new vaccine strategies are needed.

In accordance with previously published data, the rate of PNSSP in the Ille-de-France area was found to be high, but the French recommendations for the antibiotic therapy of pneumococcal infections seem to be appropriate. The use of the seven-valent conjugated vaccine seems to be a critical measure for prevention of invasive pneumococcal infections in children in this area of France. Similar studies are needed to evaluate the impact of vaccination on the epidemiology of Streptococcus pneumoniae infections.

Copyright information

© Springer-Verlag 2004