Current epidemiology of invasive pneumococcal disease in Europe

In the UK, PCV7 became part of the NIP in September 2006. A recently published review reported that cumulative weekly reported cases of PCV7 vaccine-serotype IPD in children <5 years of age fell from 400 in 2006 to 25 in 2010. Cases caused by non-vaccine serotypes increased from 150 to 375 over the same period, for an overall reduction in IPD of 150 cases in 2010 compared to 2006 [13]. Reported non-vaccine serotype IPD cases among children < 2 years of age increased between 2007 and 2010 [10, 13, 16]. By 2011 they had returned to the numbers reported in 2008, a decrease that followed the introduction of the PCV13 vaccine [10, 13, 16].

PCV7 was introduced in Germany in July 2006. In children <2 years of age, approximately 100 cases of vaccine-type IPD were reported annually between 1997 and 2007. Within 2 years after widespread PCV7 vaccination, approximately 10 cases/year of vaccine-serotype IPD were reported. Most of these IPD infections occurred in unvaccinated children or in those too young to be vaccinated (German National Reference Center for Streptococci [GNRCS]; Personal communication, M. van der Linden).

German capture-recapture data show the overall incidence of IPD in those <2 years of age fell from about 20 cases/100 000 child years between 1997 and 2003 to 10/100 000 between 2007 and 2009 [17] The mean annual incidence of PCV7 serotype IPD was 13.5/100 000 child-years between 1997 and 2003, 3.5/100 000 in 2007–08 and 2/100 000 in 2008–09. Changes in incidence for non-PCV7 serotypes were not observed in the first year after implementation of PCV7 vaccination (approximately 6.5 cases/100 000), but increases were observed for some non-PCV7 serotypes, namely 1, 7 F and 19A in the second year [17, 18].

Serotype coverage of higher valency PCV vaccines

Higher valency PCV vaccines have recently been introduced in Europe, first in Germany and subsequently in Belgium, Denmark, France, Greece, Hungary, Ireland, Italy, the Netherlands, Norway, Slovakia, Sweden, the UK and Finland. They are available in Spain and Portugal, but not yet included in the NIP [14, 19]. In Germany, PCV10 was introduced in April 2009 and PCV13 was introduced in December 2009. Approximately half of German children were being given PCV10 in December 2009, when PCV13 was introduced, and by 2011 about 85% were receiving PCV13 (Personal communication M. van der Linden). German IPD surveillance data are available for the non-PCV7 serotypes 1, 3, 5, 6A, 7 F and 19A from 1997 through June 2011. In 2009–2010, before introduction of PCV10, there were 107 reported cases of IPD in children <2 years of age. Of these, PCV7 serotypes were isolated from 13 cases (12%), PCV10 serotypes from 40 (37%), and PCV13 serotypes from 73 (68%). The picture was different for severe AOM in Germany at that time. Higher frequencies of serotype 3, and 19A isolates resulted in slightly increased coverage, compared to IPD, of the 13-valent vaccine relative to the 7- or 10-valent vaccines. Thus not all serotypes were represented in similar proportions in different types of pneumococcal disease. Reports of children having AOM with efflux decreased from 459 in 2008–9 to 213 in 2010–11 as the estimated vaccination rate increased from 71.9% to 84.5%. Pathogens were identified in 65 of 213 cases and S. pneumoniae was found in 12/65 (18.5%) [20]. Vaccine coverage of serotypes isolated from nasopharyngeal carriage was not higher than 30% for any of the three available PCV vaccines. In the UK during 2011, the cumulative weekly cases of IPD caused Xby the six new serotypes in PCV13 in children < 2 years of age fell to a lower level than before introduction of PCV7. Reports of IPD caused by any serotype not in PCV7 fell to levels similar to those in 2006 when PCV7 was introduced (Figure 1) [16]. Estimated serotype coverage of PCV7, PCV10 and PCV13 in several European countries is shown in Table 3. Recent data on the serotype distribution of IPD in European children < 2, < 5 or ≤ 15 years of age are summarized in Table 4.

Decreases in IPD cases caused by the six new vaccine serotypes contained in PCV13 were also reported in Germany by capture recapture surveillance. The number of reported IPD cases caused by serotypes 1, 3, and 7 F decreased. However, an increase in cases caused by serotype 19A that began in 2006–07 continued until 2010–11, when the rate of increase slowed. The total number of IPD cases caused by these non-PCV7 serotypes decreased from 60 in 2009–10 to just over 40 through June 2011, approximately 18 months after exclusive use of PCV13 (Figure 2) [18]. From July-December 2011 only 6 non-PCV7 serotype cases were reported (1:n = 1, 3:n = 2, 5:n = 0, 6A:n = 1, 7 F:n = 1, 19A:n = 1) [32]. It is difficult to separate the impacts of different licensed higher valency vaccines in Germany as the commercially available vaccines changed (first PCV10 only, then PCV10 and PCV13 concomitantly and finally a predominance of PCV13. Preliminary evidence, however, suggests that increased serotype coverage of higher valency vaccines is having an effect on pneumococcal infections caused by non-PCV7 serotypes, including those caused by 19A, 7 F and 6C [18, 20, 32, 33].

The burden of CAP in children

Childhood community acquired pneumonia (CAP) has significant morbidity and mortality. The burden of CAP in developed countries is 10–15 cases/1000 children per year with an annual hospital admission rate of 1–4/1000 [34]. The incidence is higher in children less than 5 years of age; 6-40/1000 compared to 7–16/1000 in children from 5–15 years of age [34]. Most deaths occur in children younger than 4 years of age [23]. Influenza, respiratory syncitial virus (RSV) and parainfluenza 1, 2 and 3 are frequent viruses causing CAP; S. pneumoniae is the most common bacterial cause [35].

The increasing use of pneumococcal vaccines underscores the need for accurate diagnosis in order to define the burden of pneumococcal disease and monitor vaccine effectiveness. However, the clinical differentiation of pneumonia from other lower respiratory tract infections, such as bronchitis and bronchiolitis, can be challenging because its clinical signs and symptoms, including fever, cough and tachypnea are not specific. Often children present at general practices where chest X-ray and other diagnostic methods with increased specificity are not available. The more defined the clinical endpoint, e.g., lobar pneumonia, empyema; the smaller the aetiologic diversity. Finally, there are much more published data from children hospitalized with severe or complicated CAP confirmed by X-ray and from identification of the aetiologic agent by blood culture than there are from children with less severe pneumonia.

Aetiology of childhood CAP

An extended discussion of methodologies used to identify causative agents of CAP is beyond the scope of this article, but it is interesting to look briefly at some studies that have compared blood culture, PCR and serology for estimating the proportion of CAP cases that are attributable to S. pneumoniae. Blood culture has some well-known limitations. It is not routinely carried out in either general practice or hospitals and has a low sensitivity of 10-30% in childhood CAP, which decreases with prior antibiotic use.

In a prospective study, PCR confirmed pneumococcal infection in 80 (10.6%) of 753 children hospitalized with a diagnosis of community acquired bacteremic pneumonia [34]. PCR allowed serotyping in blood samples from 74 of the 80 children (92.5%). More than two-thirds of those infections were caused by non-PCV7 serotypes, with serotypes 1, 3 and 19A accounting for more than half. Serotype 1 was the most frequent serotype (26/80, 32.5%) and was associated with complications (50.0% in patients with complicated pneumonia versus 18.2% in patients with uncomplicated pneumonia) and older age. Serotype 19A was second in frequency (15.0%) and was associated with younger age. Blood culture and PCR were performed simultaneously in 292 of the 753 patients. S. pneumoniae was identified in 47 of the 292 by one or both of the methods; in 45/292 (15.4%) by PCR but only in 11/292 (3.8%) by culture [31].

In another study, serology indicated a pneumococcal aetiology (titration of anti-pneumococcal pneumolysin and anti-pneumococcal C-polysaccharide IgG) in 18 of 101 children with CAP (18%), but blood culture was positive in only one of them [36]. In yet another study of 99 children with CAP, a pneumococcal aetiology was established in 46% by serology, in 22% by PCR and 1% by culture [37].

Impact of PCV7 on CAP

The impact of vaccination on CAP incidence can help define the contribution of pneumococcal vaccine serotypes to the overall disease burden. In the UK, hospital admissions for CAP among children younger than 2 years of age rose by 31%, from 1022 to 1340/million child-years between 1997 and 2006. CAP admissions then decreased by 19% to 1079/million in the two years following introduction PCV7 against a background of increasing hospital admissions of children for RTI [38].

Admissions for CAP also declined during the same period in older children who had not been vaccinated. Some of that decrease could have resulted from natural variation in the disease incidence [39]. However, herd protection likely accounted for some of the decrease, as the PCV7 vaccination coverage among those ≤2 years of age in the UK was at least 84%. Other reports have also suggested an additional impact of PCV7-vaccination on CAP incidence through herd protection [40].

In Italy, a retrospective study compared the incidence of hospital admissions in the first 2 years of life for childhood CAP in a group born in 2000–2002 before the introduction of PCV7 with a group born in 2003–2005, after its introduction [41]. For all-cause pneumonia, the incidence was 64.2/10 000 child-years among those born in 2000–2002 compared with 54.4 for those born in 2003–2005, a vaccine effectiveness of 15.2%. The corresponding results for pneumococcal pneumonia were 1.91 cases/10 000 child years versus 0.56 cases, a vaccine effectiveness of 70.5%. Only 2.9% of the all-cause pneumonia admissions in 2000–2002 were coded as pneumococcal pneumonia, but a 15.2% reduction in pneumonia by PCV7 highlights the difficulties in finding an aetiologic diagnosis of childhood CAP.

Empyema: complicated CAP

Empyema is an invasive complication of pneumonia. The incidence of empyema in children with CAP began to increase in both North America and in Europe in the mid to late 1990s. It has continued to increase even in those countries where PCV7 has been introduced [42–46]. Studies in Europe and the USA have consistently shown a small number of serotypes - included in PCV7 to be associated with empyema. The most frequently associated are serotypes 1, 3, 5, 7 F and 19A; which account for 50 to 85% of empyema cases, and serotype 14, accounting for up to 13% and occurring in children not vaccinated with PCV7 [47–50].

Estimated serotype coverage of PCVs in childhood CAP

From what is known of the pneumococcal serotypes currently associated with childhood CAP [35, 51], both licensed higher valency vaccines have substantially broader serotype coverage than PCV7, i.e., more than 80% compared with 20-40%. The CAP coverage offered by PCV7 decreases from about 60% in children younger than 2 years of age to 40% in those from 2 to 5 years of age and to approximately 10% in those older than 5 years [51]. This decrease in estimated coverage did not occur for the higher valency vaccines among older children. Coverage was also estimated to be higher in complicated than in uncomplicated CAP [35]. PCV’s of higher valency thus have the potential to decrease the remaining burden of CAP and especially complicated CAP.

To summarize, CAP aetiology remains a difficult clinical diagnosis with low specificity, and wider use of diagnostic tools having increased specificity is needed. Consequently, the prevalence of CAP attributable to S. pneumoniae is probably underestimated. Recent reports do indicate a continuous impact of PCV7 on hospitalisations for vaccine serotype CAP. However, variations in serotype circulation and occurrence of CAP caused by non-PCV7 serotypes highlight the potential of higher valency PCVs to decrease the remaining burden of CAP.

Evolution of AOM

The evolution of pneumococcal AOM is thought to begin most often with pneumococcal colonization and if local spread occurs with mucosal infection, AOM can result. The first cases of AOM are often caused by pneumococcus and are responsive to antibiotic treatment. Recurrent cases are increasingly likely to involve NTHi. Some of these cases respond to treatment, but others become chronic, with increasing likelihood of NTHi involvement [53–55]. Chronic AOM that does not eventually resolve with treatment can evolve into serious otitis media or chronic suppurative otitis media, which may result in hearing impairment or require tube insertion or surgery. The aetiology of pneumococcal AOM in over 5000 children in Israel who underwent tympanocentesis or presented with otorrhea revealed a (38%) rate of mixed pneumococcal plus NTHi infection, 59% single pneumococcal infections and 3% mixed infections with pneumococcus and other bacteria [56]. Mixed infections were associated with increasing age, previous AOM episodes, bilaterality and being a Bedouin child (possibly because of crowding in extended family households and increased carriage) [56]. This aetiology of mixed AOM is not unique to Israel, as mixed pneumococcal and NTHi infections have been observed in 19%, 29% and 40% of complicated AOM cases in studies conducted in French, Argentine and Australian aboriginal children, respectively (Cohen, personal communication, [57, 58]. Determination of bacterial growth from tube otorrhea samples over an 8-day follow up of placebo or antibiotic treatment confirmed that single infection with either H. influenzae or S. pneumoniae resolved more quickly and were more responsive to antibiotics than mixed infections [59]. The patient characteristics, difficulty of treatment and survival advantages in mixed AOM are suggestive of biofilm formation, which has also been shown to play a role in other otorhinolaryngological infections [60]. In vitro and animal models have shown that NTHi can passively confer beta-lactam resistance to S. pneumoniae by forming biofilms or by production of beta-lactamase [61].

It is likely that NTHi is often associated with a biofilm formation in the middle ear, especially in conjunction with an initial S. pneumoniae infection [55, 62–64]. The bacteriologic steps in AOM progression might begin with a simple S. pneumoniae infection at first. However, repeated, recurrent and chronic AOM could become increasingly associated with NTHi and decreasing proportions of S. pneumoniae and with biofilm formation.

The role of PCV

PCV vaccination reduces carriage of the pneumococcal serotypes that are present in the vaccine and thus prevents local spread that would otherwise cause early AOM infections [65]. If the incidence of early, simple AOM is reduced, then one would expect that there should also be a reduction in mixed and complicated AOM. In fact, looking again at an early PCV7 efficacy trial, there was relatively low overall efficacy against any AOM, but up to nearly a 23% reduction in recurrent AOM [66]. The importance of timely vaccination to prevent the first cases of AOM is highlighted by reports that while vaccination of children with a history of recurrent AOM pneumococcal vaccination reduced carriage of vaccine serotypes, it did not protect against subsequent recurrences of AOM [67–69]. A randomized trial of an investigational PCV11 vaccine has shown (by microbiological analysis of middle ear fluid) that PCV vaccination reduced the incidence of mixed pneumococcal and NTHi infection by more than 30% [70]. When comparing the pathogens recovered from middle ear fluid of PCV vaccinated versus control children with an epidemiology suggestive of complicated AOM, there were reductions of 58% in pneumococcal vaccine serotypes, 51% in non-vaccine serotypes and 35% in H. influenzae, which would be expected if pneumococcal vaccination reduced initial AOM infections that eventually became recurrent or chronic and mixed [70]. The PCV method included an H. influenzae-derived protein carrier, therefore at least some role of the NTHi antigen could not be ruled out. Two other recent observational studies showed that PCV reduced emergency room visits for AOM with mixed infection and complicated by otorrhea. They also showed that there were significant reductions in pneumococcal and NTHi isolates from middle ear fluid compared to fluid obtained prior to PCV vaccination [71]. Studies that have reported on the clinical effectiveness of PCV7 against recurrent, complicated or antibiotic-resistant AOM have recently been reviewed [72].

Preliminary results from PCV efficacy studies and post PCV epidemiology data suggest that considerable reduction of both S. pneumoniae and NTHi infections may be achieved by prevention of early pneumococcal carriage and AOM, decreasing the number of children with severe disease [73]. One study has already shown that PCV vaccination resulted in 40% reductions in ambulatory visits and antibiotic use for AOM [74]. Whether the presence of the H. influenzae-derived protein D as a carrier in PCV10 may further decrease the role of NTHi in AOM still remains to be seen. Given the evidence for serotype replacement, extension of the PCV serotype spectrum beyond PCV7 may provide additional benefit in prevention the evolution of AOM. Broadened prevention of AOM would have an important effect on cost effectiveness of vaccination [75–77].