Population Snapshot of Streptococcus pneumoniae Serotype 19A Isolates before and after Introduction of Seven-Valent Pneumococcal Vaccination for French Children

Abstract

Serotype 19A Streptococcus pneumoniae strains are now more frequent in French children than before the introduction of a seven-valent conjugate vaccine (PCV7). By applying multilocus sequence typing to 144 serotype 19A isolates collected before and after beginning PCV7 vaccination, we detected clonal expansion of the preexisting penicillin-intermediate sequence type 276.

Since the introduction of the heptavalent pneumococcal conjugate vaccine (PCV7), the incidence of invasive pneumococcal disease (IPD) caused by the vaccine serotypes has fallen significantly in children and adults (11, 20). However, the incidence of IPD due to serotypes not covered by PCV7 was recently reported to have risen (20). In the United States and in France, the relative incidence of IPD caused by Streptococcus pneumoniae serotype 19A strains has increased among vaccinated children (12, 13). Moreover, clones of serotype 19A showing high-level resistance to penicillin and ceftriaxone and resistant to all oral antibiotic compounds licensed for children have recently emerged as important causes of IPD and otitis in the United States (16, 17).

PCV7 was introduced in France in March 2002 and is officially recommended for immunization of infants under 2 years of age. However, the uptake of the vaccine reached more than 30% by 2003 (7). We used multilocus sequence typing (MLST) to determine the population structure and susceptibility profiles of serotype 19A S. pneumoniae isolates collected from children before and after the arrival of PCV7 in France.

A total of 144 serotype 19A S. pneumoniae consecutive clinical isolates, obtained from our pediatric network between 1992 and 2007 in towns located throughout France, were included in the study. Some of the strains studied here belong to previously described collections (3). They were isolated from children with invasive infections in 74 cases (bacteremia, meningitis, and osteoarticular infections) or with noninvasive infections in 70 cases (acute otitis media, conjunctivitis, and bronchopulmonary infections). Serogrouping was performed by latex agglutination (pool latex antisera; Staten Serum Institut, Copenhagen, Denmark), and serotyping was performed by PCR as described previously (14).

The study period was divided into the prevaccine period (1992 to 2002; 49 isolates) and the vaccine period (2003 to 2007; 95 isolates).

Penicillin and cefotaxime MICs were determined using the Etest method (AB Biodisk, Solna, Sweden). MICs were interpreted using CLSI guidelines (2). The mefA and ermB genes were detected by PCR as described previously (6).

MLST was performed as described elsewhere (4). PCR amplification and sequencing were performed by Genoscreen (Lille, France). Alleles were assigned using the MLST website (http://spneumoniae.mlst.net). All alleles not already present in the pneumococcal MLST database were verified by resequencing the gene fragment on both strands and were submitted to the MLST S. pneumoniae database for designation (see above for URL). Clonal complexes consisted of sequence types (STs) that shared six of seven alleles with at least one other ST in the complex. Clonal complexes were assigned by using the eBURST algorithm (5). MLST data were used to calculate the Hunter-Gaston diversity index (HGDI) (9). The HGDI is the probability that two strains randomly taken from a given population will have different genotypes. It can range from 0 (no diversity) to 1 (extreme diversity). The HGDI is very similar to the Simpson index of diversity but incorporates a finite sample adjustment which is generally desirable (18). It was calculated on the website at http://www.hpa-bioinformatics.org.uk/cgi-bin/DICI/DICI.pl, which allows generation of confidence intervals (CI). Two HGDIs were considered significantly different if their confidence intervals did not overlap. Proportions were compared between groups with Pearson's χ² test. P values of <0.05 were considered to denote significant differences.

The 144 isolates belonged to 25 different STs, 9 of which were new (ST3111, ST3293, ST3295, ST3259, ST3297, ST3299, ST3300, ST3301, ST3302). Fourteen STs were represented in the prevaccine period and 17 in the vaccine period (Table 1). ST276 was the main ST, representing 15/49 isolates (30.6%) in the prevaccine period and 70/95 isolates (73.6%) in the vaccine period (P < 0.05). A population snapshot of the 144 isolates from both periods, based on eBURST analysis (Fig. 1), showed, among the four clonal complexes, a major clonal complex (CC276) containing ST276 (predicted founder) and three STs recovered only in the vaccine period (ST2674 and two new STs, ST3300 and ST3302). This suggests that as the founding genotype increases in frequency in the population, it gradually diversifies into new single-locus variants. The HGDI was 0.867 (95% CI, 0.808 to 0.926) for isolates obtained during the prevaccine period and 0.468 (95% CI, 0.342 to 0.594) for those obtained during the vaccine period; the difference between these two values was significant.

TABLE 1.

Antimicrobial susceptibilities and prevalences of erm and mef in serotype 19A isolates, according to ST and prevaccine or vaccine period

Clonal complex and ST (na)	No. of isolates fromb:		MIC50 [μg/ml] (range) of:		mefA/ermB determinant (no. of isolates harboring determinant/total isolates)	No. (%) of isolates obtainedc:		P value
	Invasive infections	Noninvasive infections	Penicillin	Cefotaxime		Before 2003	In 2003 and later
CC63 (2)	0	2	0.5 (0.5-0.75)	0.25 (0.25-0.5)	ermB (2/2)	0	2 (2.1)
ST63 (1)	0	1	0.5	0.25	ermB (1/1)	0	1 (1)
ST3299 (1)	0	1	0.75	0.5	ermB (1/1)	0	1 (1)
CC276 (90)	35	55	0.5 (0.25-1)	0.5 (0.125-1)	ermB (90/90)	15 (30.6)	75 (78.9)	<0.05
ST276 (85)	34	51	0.5 (0.25-1)	0.5 (0.125-1)	ermB (85/85)	15 (30.6)	70 (73.6)	<0.05
ST2674 (2)	0	2	0.5 (0.5-0.75)	0.5	ermB (2/2)	0	2 (2.1)
ST3300 (2)	0	2	0.75 (0.75-1)	0.5 (0.5-0.75)	ermB (2/2)	0	2 (2.1)
ST3302 (1)	1	0	1	0.5	ermB (1/1)	0	1 (1)
CC416 (9)	6	3	0.032 (0.016-0.032)	0.032		6 (12.2)	3 (3)
ST416 (7)	4	3	0.032 (0.016-0.032)	0.032		4 (8.1)	3 (3)
ST2343 (1)	1	0	0.032	0.032		1 (2)	0
ST3295 (1)	1	0	0.032	0.032		1 (2)	0
CC202 (2)	2	0	0.016	<0.016	ermB (1/2)	1 (2)	1 (1)
ST202 (1)	1	0	0.016	<0.016		0	1 (1)
ST3259 (1)	1	0	0.016	<0.016	ermB (1/1)	1 (2)	0
ST62 (1)	0	1	0.5	0.5	ermB (1/1)	0	1 (1)
ST81 (8)	7	1	1.5 (1-3)	1 (0.5-2)	ermB (7/8)	7 (14.2)	1 (1)
ST193 (3)	3	0	0.016 (0.016-0.064)	0.032 (0.032-0.25)	ermB (3/3)	3 (6.1)	0
ST199 (4)	4	0	0.064 (0.032-0.064)	0.032		4 (8.1)	0
ST277 (1)	0	1	0.5	0.5		0	1 (1)
ST320 (3)	0	3	1 (1-1.5)	1 (0.75-1)	ermB/mefA (3/3)	0	3 (3.1)
ST994 (4)	3	1	0.016 (0.016-0.032)	0.032 (0.016-0.032)	ermB (1/4)	2 (4.1)	2 (2.1)
ST1201 (8)	7	1	0.016 (0.016-0.032)	0.032	ermB (1/8)	7 (14.2)	1 (1)
ST2013 (4)	2	2	0.25 (0.064-0.5)	0.25 (0.064-0.25)		1 (2)	3 (3.1)
ST2102 (1)	1	0	0.016	0.032		1 (2)	0
ST3111 (1)	1	0	0.25	0.125	ermB/mefA (1/1)	0	1 (1)
ST3293 (1)	1	0	0.032	0.032		1 (2)	0
ST3297 (1)	1	0	0.75	0.75	ermB (1/1)	0	1 (1)
ST3301 (1)	1	0	0.032	0.032		1 (2)	0

^an, no. of isolates.

^bA total of 74 isolates were from cases of invasive infection, and 70 were from cases of noninvasive infection.

^cA total of 49 isolates were obtained before 2003, and 95 were obtained during or after 2003.

FIG. 1.

Population snapshot of 144 isolates from France (1992 to 2007), based on eBURST analysis. Each ST encountered is indicated by a circle, with diameters proportional to numbers of isolates (extracted from Table 1). The blue circle for ST276 corresponds to the founder of the clonal complex 276. The lines join single locus variants (SLVs). Two data sets were compared: a data set with the STs of the prevaccine period (1992 to 2002) and a data set with the STs of the vaccine period (2003 to 2007). STs are colored differentially depending on their membership in the two data sets: black (STs found only during the prevaccine period), green (STs found only during the vaccine period), or cyan (STs found during both periods). For ST276, the percentages of isolates before and during the vaccine period are indicated. Some STs of this study are related to STs of major clones described in the Pneumococcal Molecular Epidemiology Network (http://www.sph.emory.edu/PMEN) and indicative of potentially capsular switching. For each of these 10 STs, the name of the clonal type was indicated (e.g., Greece²¹-30). When a clonal type was related but not identical to an ST of this study, the degree of relatedness between them was specified (SLV or double locus variant [DLV]) and the ST no. of the clonal type was followed by an asterisk if it was not present in the study data set. ST320, ST202, and ST3259 are linked to the clonal type Taiwan^19F-14. However, ST202 and ST3259 could be considered relatives of ST320 in this context because it is known that ST320 already exists as a 19A serotype and it is more parsimonious to assume fewer cases of serotype switching.

In the prevaccine period, 27 isolates (55.1%) were susceptible to penicillin, 16 (32.6%) were intermediate, and 6 (12.2%) were resistant. In the vaccine period, 7 isolates (7.3%) were susceptible to penicillin, 87 (91.5%) were intermediate, and 1 was resistant (1%). The proportion of isolates susceptible to penicillin was significantly lower after implementation of PCV7 vaccination (prevaccine, 55.1%; vaccine period, 7.3%; P < 0.01). Twenty-five isolates (51%) were resistant to erythromycin in the prevaccine period, compared to 86 isolates (90.5%) in the vaccine period. All of the CC276 isolates showed intermediate resistance to penicillin (MIC, 0.25 to 1 μg/ml) and carried ermB. Only four isolates carried both ermB and mefA; these belonged to ST320 and ST3111 (Table 1).

In the prevaccine period, 19A was the second-leading serotype in children with pneumococcal meningitis in France (1) and the fourth in bacteremic children less than 2 years old (19). After the implementation of PVC7 vaccination, in 2005, S. pneumoniae 19A was the leading serotype in children with meningitis and also the leading serotype in bacteremic children less than 2 years old in France (1, 19).

We found that preexisting CC276 strains moderately resistant to penicillin contributed disproportionately to the increase in isolation of serotype 19A S. pneumoniae from French patients with invasive and noninvasive pneumococcal disease, which rose from 30.6% to 73.6% of isolates (P < 0.05), suggesting clonal expansion after a relatively short period of PCV7 use. In the same way, the HGDI of the pneumococcal isolates was greatly and significantly reduced after PCV7 vaccination. However, longitudinal fluctuations in 19A clones before vaccination may have occurred, and CC 276 could have been more common toward the end of the prevaccine period. Clonal expansion of ST199 contributed to the emergence of serotype 19A in the United States after the introduction of PCV7 (13, 15, 16). In the United States, ST320 and ST2722 were associated with high-level resistance to penicillin, amoxicillin, and ceftriaxone and could represent a challenge for conventional therapy (16, 17). In contrast, ST320 strains circulating in France are moderately resistant to penicillin and resistant to oral cephalosporin (data not shown) and harbor both mefA and ermB, like their Korean counterparts (10). The difference in the level of penicillin resistance among ST320 strains in the United States and France could be related to the fact that amoxicillin and cefpodoxime, respectively, were the most commonly prescribed antibiotics in the two countries (8, 21). Indeed, in the United States, a high level of prescription of the former drug with PCV-7 could coselect highly penicillin-resistant strains. In contrast, in France, wide use of cefpodoxime may be associated with the emergence of strains resistant to cefpodoxime but only moderately resistant to penicillin. No highly penicillin-resistant clones have so far spread in France. Highly penicillin-resistant isolates were limited to ST81, related to Spain^23F-1, no representatives of which have been recovered since 2004. This finding might be associated with the general fall in antibiotic consumption observed between the two periods of our study (8). However, the success of ST276 and the nonsuccess of ST81 can also be explained in terms of stochastic effects and do not necessarily imply higher virulence, higher transmissibility, or better adaptation to antibiotic selection pressure.