The multifaceted impact of pneumococcal conjugate vaccine implementation in children in France between 2001 to 2014

ABSTRACT

In 2003, France was the first European country to recommend 7-valent pneumococcal conjugate vaccine (PCV7) for a large proportion of healthy children. With complicated recommendations, the vaccine coverage during the first 4 y of implementation was low, then progressively increased to reach 90% in 2008. The aim of this review was to describe the particular impact of PCVs in a country where the vaccine coverage was initially suboptimal.

After PCV7 implementation, the PCV7 serotypes nearly disappeared among pneumococci isolated from meningitis (−73%), other invasive pneumococcal disease (IPD; −90%) and pneumococcal carriage (−97%). Consequently, the rates of penicillin-resistant strains declined. However, because of important serotype replacement, the global effect on the incidence of meningitis (−31%) or other IPD (−14%) was modest and observed only in young children < 2 y old. After PCV13 transition, with immediate high vaccine coverage, the vaccine had an important impact on all pneumococcal disease: reduction of −20% for pneumococcal meningitis, −36% for non-meningitis IPD, −32% for community acquired pneumonia and −15% for S. pneumoniae carriage.

These findings underline the complexity of pneumococcal epidemiology and the importance of high and fast vaccination coverage to obtain the optimal effect of PCVs.

Keywords:

• Pneumococcal conjugate vaccine
• children
• impact
• carriage
• pneumococcal meningitis
• invasive pneumococcal disease
• pneumonia
• acute otitis media

Abbreviations

PCV	=	pneumococcal conjugate vaccine
Sp	=	Streptococcus pneumoniae
US	=	United States
IPD	=	invasive pneumococcal disease
InVS	=	French National Institute of Sanitary surveillance
CSF	=	cerebrospinal fluid
CNRP	=	National reference Center for Pneumococci
GPIP	=	French Group of Pediatric Infectious Diseases Pediatricians
ACTIV	=	Association Clinique Thérapeutique Infantile du Val de Marne
CI	=	Confidence interval
CAP	=	community acquired pneumonia
CRP	=	C-reactive protein
PCT	=	procalcitonin
AOM	=	acute otitis media
NP	=	nasopharyngeal
MEF	=	Middle ear fluid
PNSP	=	penicillin non susceptible pneumococci

Introduction

Streptococcus pneumoniae (Sp) is one of the several hundred bacterial species physiologically present in the rhinopharyngeal microbiome.¹ However, pneumococcus is the causative agent of a large spectrum of diseases, including invasive infections such as meningitis, bacteremia, and bacteremic pneumonia and mucosal infections such as otitis media, sinusitis and pneumonia (Fig. 1).^2,3

Figure 1. Spectrum of pneumococcal infections.

S. pneumoniae is the leading cause of infections in humans and the main pathogen that causes community-acquired pneumonia worldwide, particularly in young children. Thus, pneumococcus are responsible for great mortality and morbidity. Consequently, S. pneumoniae infection is an important cause of vaccine-preventable deaths in children.⁴

The peak incidence of pneumococcal infections occurs during the first 2 y of life. However, before age 3 to 6 months, these infections are uncommon, young infants being partially protected by maternal antibodies.⁵ Although the proportion of children with underlying diseases increases with age, whatever the age, most children with pneumococcal infection were healthy before the pneumococcal disease.

More than 90 pneumococcal serotypes were identified before pneumococcal conjugate vaccine (PCV) implementation, but only a few (5 to 11) were responsible for more than 75% of the cases of invasive pneumococcal disease (IPD) worldwide.⁶

Pneumococcus is naturally susceptible to a large variety of antibiotics including ß-lactams and macrolides. Since the beginning of the 1980s, occasional cases of resistance have been reported, and after 1990, resistance to both ß-lactams and macrolides increased. Thus, Pneumococcus infection is recognized as a major health problem and related to the level of antibiotic consumption in the country.⁷ Not all serotypes are involved in the antibiotic resistance: those more frequently carried in children and for a long period, such as 6B, 9V, 14, 19A and F, 23F, are more frequently involved.⁸

Before the PCV era, the available pneumococcal vaccine (polysaccharide 23 valents) had several limitations, including lack of efficacy in children < 2 y old, the age when the highest incidence of pneumococcal infection is observed. In the United States, after PCV7 implementation, in 2000, impressive data for effectiveness against IPD was published.⁹

In 2003, France was the first European country to recommend PCV7 for a large proportion of healthy children: children < 2 y old at risk of IPD in relation to medical or living conditions, that is, children cared for more than 4 hours/week with at least 2 other children or belonging to a family with > 2 children and being breast-fed for less than 2 months, which represented about 80% of each birth cohort.¹⁰ The consequence of these very complicated recommendations, even if PCV7 was reimbursed by the French insurance system, was low vaccine coverage during the first 4 y of implementation.¹¹ PCV coverage is monitored through the Permanent Sample of Beneficiaries, extracted from the National Health Insurance Scheme (NHIS) database, which covers the whole French population. The PCV7 vaccination coverage increased slowly: in 2004, only 56% of children < 1 y old had received at least 1 dose of PCV7.¹² In June 2006, the vaccine was recommended as a universal vaccination for all children < 2 y old.¹³ After the universal recommendation, the PCV coverage for primary immunization increased quickly and surpassed 80% in 2008. The 3 + 1 vaccination schedule (2, 3, 4 months, and a booster at 12–15 months) was changed in 2008 to a 2 + 1 schedule (2, 4 and 12 months). In 2010, as soon as the PCV13 vaccine was marketed in France, the recommendation was to shift from PCV7 to PCV13 (without a catch-up program for children 2 to 5 y old) and PCV7 was withdrawn from the market in July. Since 2010, the PCV13 coverage in France is high (> 90%) for children < 2 y old.^14,15

To determine the impact of the PCVs in France, several surveillance systems of pneumococcal infection and carriage were established before and after PCV implementation.^11,13,15-29 All these studies were approved by the French National Data Protection Commission (CNIL) and/or the French Ethics Committee. Most of the studies were requested by the European Agency for the Evaluation of Medicinal Products as a post-licensing commitment. Here, we report the multifaceted impact of PCV implementation in children in France according to the results of these surveillance systems.

Impact of PCV Implementation on Pneumococcal Meningitis

The earliest French surveillance network, Epibac, was established in 1987 by the French National Institute of Sanitary surveillance (InVS). Epibac collects data on 6 severe invasive bacterial diseases including pneumococcal infections for adults and children.^13,15 IPD cases are defined as the isolation of pneumococcus or the detection of pneumococcal DNA from cerebrospinal fluid (CSF) or blood. Among IPD, meningitis accounted for 20% of cases.¹⁵

Data collected include age, sex, and site of isolation. Epibac covers more than 74% of the French metropolitan population, and the participating hospitals are distributed evenly across the national territory.¹⁵ Pneumococcal strains isolated from CSF in children (0–15 y of age) were collected from hospital laboratories and sent to the national reference center for pneumococci (CNRP) for serotyping performed by using latex particles sensitized with antisera marketed by the Statens Serum Institute and antibiotic susceptibility.

In 2001, another national network was established by the French Group of Pediatric Infectious diseases Pediatricians (GPIP) and the research group, ACTIV.^11,25,26 The main objective of this network was to complete the clinical data of the Epibac and CNRP networks. Data collected were date of birth, gender, CSF and blood culture results, vaccine status, underlying conditions, previous antibiotic treatment, clinical signs, treatment of the current episode and mortality. Indeed, 227 pediatric wards working with 168 microbiology departments throughout France were asked to report all cases of pneumococcal meningitis: the clinical investigator (pediatrician) in each participating ward was contacted for information on new cases or to confirm the lack of such cases.^11,25,26 The exhaustiveness of this surveillance system was evaluated by capture–recapture analysis of the 3 French national pneumococcal meningitis surveillance systems, and was estimated to be 61% (95% CI: 60%–66%).³⁰

Figure 2 presents the trends for number of cases and incidence of pneumococcal meningitis recorded by GPIP/ACTIV and Epibac. After PCV7 implementation, from 2001–2002 to 2008–2009, the incidence of meningitis significantly decreased by 31% in children < 2 y old and significantly increased in children 5–15 y old children.¹⁵ After PCV13 implementation, the incidence of meningitis decreased by 20%, 55%, and 40% in children < 2, 2–4 and 5–15 y old, respectively.¹⁵ Similar results were observed for number of cases by age recorded by the GPIP/ACTIV network.²⁶ In 2013, according to the GPIP/ACTIV network, the decrease in cases continued for children < 2 y old, significantly decreasing by 39% from 2009 to 2013, but remained stable for older children.²⁵

Figure 2. Trends in number of cases and incidence of pneumococcal meningitis in France from 2001 to 2012.

The distribution of serotypes markedly changed after PCV implementation. Indeed, the number of cases involving PCV7 serotypes significantly decreased by 90% between 2001 and 2012²⁶ and that for the 6 additional PCV13 serotypes (particularly 19A and 7F) increased between 2002 and 2009 and decreased sharply by 67% from 2009 to 2012.²⁶ The frequency of meningitis with non-vaccine types remained stable during the same period. The non-PCV13 serotypes reported in 2012 by GPIP/ACTIV accounted for 68% of cases and were mainly represented by serotypes 12F (15%), 24F (15%), 22F (7%) and 15B/C (7%).²⁶

Trends in other Non-meningitis IPD

The Epibac network database allowed for analyzing the trends in non-meningitis IPD.^13,15 After PCV7 implementation, from 2001–2002 to 2008–2009, the incidence of non-meningitis IPD in children < 2 y old decreased slightly but significantly (−14%) and increased in older children.¹⁵ After PCV13 implementation, the incidence of non-meningitis IPD decreased in the entire pediatric population. The number of cases decreased between 2008–2009 and 2012, by 36%, 37% and 51%, in children < 2, 2–4 and 5–15 y old, respectively (Fig. 3).¹⁵ Globally, the incidence of non-meningitis IPD decreased in all age groups from 2001–2002 to 2012 but was only significant for children < 2 y old.¹⁵

Figure 3. Trends in number of cases and incidence of non-meningitis invasive pneumococcal disease (Epibac network) from 2001 to 2012.

Impact of PCV Implementation on IPD

The distribution of serotypes changed greatly after PCV implementation: from 2001–2002 to 2008–2009, the incidence of all IPD cases due to PCV7 serotypes decreased by 90% in children < 5 y old and by 61% in children aged 5–15; during the same period, the incidence of IPD due to non-PCV7 serotypes increased in the entire pediatric population (<15 years).¹⁵ Moreover, just before PCV13 implementation, the additional serotypes accounted for 63%, 73%, and 66% of IPD cases in children < 2, 2–4, and 5–15 y old, respectively, with 19A and 7F serotypes the most frequent. In the last year of PCV7 use, PCV13 covered 74% of IPD cases in children < 5 y old and 76% in children aged 5–15. Consequently, with high PCV coverage, PCV13 implementation has had a strong impact on the incidence of IPD. The incidence of infection with the most prevalent serotypes decreased significantly: 83% for 19A, 77% for 7F, 96% for 1 and by 85% for 3. This reduction in infection due to non-PCV7 serotypes was also observed in older children. During the same period, the incidence of IPD with non-PCV13 serotypes almost doubled in children < 2 y old; infection with 24F and 12F serotypes was the most frequent. The incidence of IPD with non-PCV13 serotypes increased in children 2–4 y old also but not in older children.¹⁵

The IPD survey of GPIP/ACTIV adds information regarding clinical data (history, immunization status and clinical presentation). This complementary network was established in 2011.²⁵ The databases of the CNRP for microbiological data and the GPIP/ACTIV for clinical data were cross-checked for IPD and the results were combined for analysis of clinical and biological features of IPD from 106 pediatric wards. In this large cohort of 594 children with IPD recorded after PCV13 implementation, the associations among serotype, pathology, and underlying condition were analyzed.²⁵ Figure 4 shows that the proportion of type of IPD and underlying conditions varied significantly by serotype. Moreover, patients with pneumococcal meningitis more frequently had non-vaccine types than PCV13 vaccine types (+6C) (39% vs 18.6%, p < 0.001), were younger (mean age, 38.6 months [median 15.9] vs 57 months [median 50.6], p < 0.001) and had more underlying conditions (20.8% vs 12%, p = 0.005). Thus, following the changes in serotype distribution induced, PCVs had not only reduced the incidence of IPD but also, they had changed their profiles: among the remaining IPD cases, the spectrum of IPD and the rate of patients with underlying conditions have also been modified.²⁵

Figure 4. Distribution of pneumococcal diseases and underlying condition by serotype from 2011 to 2014 (GPIP/ACTIV network).

Trends in Pneumonia and Empyema

Since June 2009, the GPIP/ACTIV has collected data from 8 French tertiary pediatric emergency departments.¹⁶ All patients between 1 month and 15 y old with chest X-ray–confirmed community acquired pneumonia (CAP) have a case report form completed by a designated clinical investigator; the completed forms are regularly sent to the investigating center (GPIP/ACTIV). To assess the impact of PCV13 in France, the surveillance started during the period of exclusive use of PCV7, followed by the transition from PCV7 to PCV13, to exclusively PCV13. CAP was defined by the association of fever and a chest X-ray showing consolidation and/or pleural effusion, diagnosed by a pediatrician and confirmed by a pediatric radiologist. The collected data include age, sex, risk factors for invasive pneumococcal disease, pneumococcal vaccine status, presence of pleural effusion, biological results such as C-reactive protein (CRP) and procalcitonin (PCT) levels, blood cultures and pleural sample analysis, admission or discharge, treatment and short-term outcome.

In the 3 first years of surveillance, 5,645 patients met the criteria for CAP, which included 365 cases with pleural effusion. Figure 5 presents the trends in number of cases of overall pneumonia, pneumonia in children < 2 y old, those with CRP level > 120 mg/L, and those with pleural effusion or positive blood culture. The reduction was significant for all variables but more marked with CRP level > 120 mg and pleural effusion with documented pneumococcal pneumonia. PCV7 serotypes were identified in only 2 patients during the 3 periods. Additional PCV13 serotypes were predominant in 51 isolates (82%), but their frequency decreased by 74%, from 27 to 7 cases, between the pre- and post-PCV13 periods. Serotype 1 was predominant, accounting for 27 cases, and decreasing from 12 to 5 cases between the pre- and post-PCV13 periods. The serotype 19A was the second most frequent (n = 13 cases) and decreased from 8 to 1 cases between the periods. Serotype 1 was isolated mainly in children > 5 y old and 19A in children < 2 y old. The number of non-PCV13 serotype cases did not increase throughout the study period.¹⁶

Figure 5. Trends in proportion of community-acquired pneumonia (CAP) (bacteremic and non bacteremic pneumonia) from 2009 to 2012.

Impact of PCVs on Pneumococcal Carriage in Children with Acute Otitis Media (AOM)

In France, AOM incidence data are not available and tympanocentesis for bacterial culture are not used in clinical practice. Because the causative bacteria of AOM often grow alongside other organisms in nasopharyngeal (NP) cultures, NP samples (NPS) have poor positive predictive value for the causative agent and are not recommended in clinical practice.³¹ However, because NPS are easy to obtain, painless, non-traumatic, and they have an epidemiological value for monitoring serotypes and antibiotic resistance changes, we have scheduled to follow the modifications induced by PCVs implementation in children with AOM. In the early 2000s, a national NP study has started to complete the epidemiological surveillance of pneumococcus.^18-23,28

Since 2001, 121 pediatricians who are part of the research and teaching ACTIV network throughout France have participated in a prospective NP carriage study.²³ From October to June of each year, 700 children of both sexes with suppurative AOM with fever and/or otalgia (to increase the probability of pneumococcal AOM), 6 to 24 months old, were enrolled by pediatricians in pediatric outpatient clinics. Diagnostic criteria for AOM included the Paradise algorithm for acute suppurative otitis media (effusion + marked redness or marked bulging or moderate redness and bulging).³² Standardized history and physical examination findings were recorded, and information gathered at study entry included sex, age, daycare modalities, recent antibiotic treatment (within 3 months before enrolment), history of AOM, and clinical symptoms and signs of AOM.²³ NP specimens were obtained with use of cotton-tipped wire swabs, which were immediately inoculated in transport medium (Copan Venturi Transystem, Brescia, Italy), stored at room temperature and sent within 48 hours to the CNRP at G. Pompidou European Hospital and to Robert Debré Hospital, in Paris.²³

Four periods were distinguished: period 1 (October 2001 to June 2006), representing the absence of generalized PCV7 vaccination recommendation and poor vaccine coverage (<60%); period 2 (October 2006 to June 2010), generalized recommendation and high PCV7 vaccine coverage (90%); period 3 (October 2010 to June 2011) PCV7–PCV13 transition; and period 4 (October 2011 to June 2014), high PCV13 coverage (>99%).²³ For 7,991 children enrolled over 13 years, several results were reported in this recent study (Fig. 6): a slight but significant decrease in overall pneumococcal carriage (−15%, 71.2% to 56.2% from 2001 to 2014), a quasi-eradication of carriage of PCV7 vaccine types plus serotype 6A, and an increase in carriage of 6 additional serotypes plus 6C during the PCV7 period, followed by sharp decrease in carriage of these latter serotypes after PCV13 implementation. Among the PCV7 vaccine types, only 19F was not completely eradicated. Interestingly and as expected, the carriage of some non-vaccine types increased and the most frequently carried non-PCV13 serotypes in the last year of the study were serotypes 15B/C (9.4%), 11A (7.1%), 15A (3.5%), and 35B (3%).²³

Figure 6. Pneumococcal carriage and serotype distribution in children with acute otitis media (AOM) from 2001 to 2014.

Impact of PCVs on AOM Treatment Failure

For 22 years, from 1992 to 2014, our group and others in France have performed several studies on AOM non-response to antibiotic treatment in children.^{28,29,33–36} Antibiotic treatment failure was defined as the persistence of AOM symptoms after at least 48 hours of antibiotics or their recurrence within 4 d after the end of treatment.²⁹ Middle ear fluid (MEF) was obtained by tympanocentesis and/or by sampling spontaneous discharge according to recommended clinical practice guidelines.^29,33-36

These studies performed before and after PCV implementation allow for analyzing the epidemiology of pathogens involved in AOM treatment failure. Figure 7 presents the pathogens recovered by period considered in these different studies.^29,33-35 Before PCV7 implementation, from 1996 to 1998, the main pathogen isolated from MEF was S. pneumoniae (60.8%), then H. influenzae (39.2%). After PCV7 implementation, S. pneumococcus and H. influenzae were equally represented. Probably because of the role of biofilm, AOM treatment failure due to H. influenzae was generally not related to antibiotic resistance.^28,37 By contrast, for pneumococcus infection, most strains isolated were penicillin non-susceptible.^28,29 After PCV7 implementation, serotype 19A became the most frequently isolated serotype with AOM treatment failure.²⁹ This finding is not surprising because of the role of serotype 19A in AOM,^20,21 its frequent multi-resistance to antibiotics³⁸ and its emergence after PCV7 implementation. Another recent French study evaluated the trends in antibiotic resistance and serotype distribution of pneumococci isolated from MEF in children with AOM from 2001 to 2011.³⁶ The authors reported a marked reduction in infection with vaccine serotypes after PCV7 implementation, from 63.0% in 2001 to 13.2% in 2011, with increased incidence of infection with the additional 6 serotypes included in PCV13 during the same period, with a particularly high proportion of 19A isolates.³⁶ After PCV13 implementation, H. influenzae was the most frequent otopathogen isolated and accounted for 75.9% of positive MEF samples.³⁵

Figure 7. Pathogens implicated in AOM treatment failure before and after PCV implementation in children from 1992 to 2014.

Impact of PCVs on Antibiotic Resistance

Because most pneumococcal-resistant strains belonged mainly to PCV7 and then PCV13 serotypes, the implementation of these vaccines reduced antibiotic resistance in all cases of S. pneumoniae infection.^{22,26,36,39,40} Indeed, we previously demonstrated the strong impact of the PCV7 implementation, combined with reduced antibiotic use, on the carriage of penicillin non-susceptible pneumococci in children with AOM.²²

After PCV13 implementation, because selective pressure continues to occur with excessive antibiotic consumption in France, some non-vaccine serotypes are frequently resistant.²³ We recently published the serotype distribution by resistance profile among pneumococcal carriers in children with AOM before and after PCV13 implementation.²³ Penicillin non-susceptible pneumococci (PNSP) was mostly represented by PCV7 serotypes (14, 23F, 19F, 6B, 9V), 6A and 19A before PCV13 implementation. By contrast, in the PCV13 era, PNSP isolates were predominantly represented by serotypes 19A, 19F and non-PCV serotypes, particularly 11A, 15A, 15B/C, 29 and 35B.²³ Interestingly, among PNSP isolates, the frequency of serotype 11A was increased, whereas that of non-susceptible strains 15A and 35B were already high and stable.²³

Conclusions

As reported in several countries, in France, we also observed a strong impact of PCVs on vaccine serotypes isolated in invasive or mucosal pneumococcal disease.^{15,23,26,40,41} However, France saw important serotype replacement after PCV7 implementation. The prevalence of non-vaccine serotypes before PCV implementation was probably a major factor in determining the replacement.¹¹ Moreover, the relatively low vaccination uptake, together with the lack of a catch-up program, could have accelerated the serotype replacement.⁴² By contrast, the implementation of PCV13 with a high and fast vaccination coverage did not seem to cause the same phenomenon. Indeed, the serotype replacement we report here is so far limited. These findings underline the complexity of pneumococcal epidemiology and the importance of high and fast vaccination coverage. The need for long-term surveillance of both IPD and carriage to determine the formulation of extended future PCVs is crucial and the French surveillance systems appear to be effective.

Disclosure of Potential Conflicts of Interest

Grants for ACTIV from Pfizer, Novartis, Sanofi and GSK during the conduct of the study.

Pr Cohen Robert reports personal fees from Pfizer, GSK, Sanofi and Novartis outside the submitted work. Dr Levy Corinne reports personal fees from Pfizer and Novartis outside the submitted work.