Invasive Pneumococcal Disease in Adults in Portugal: The Importance of Serotypes 8 and 3 (2015–2018)

Increasing the uptake of the 13-valent pneumococcal conjugate vaccine (PCV13) in children is expected to alter the serotypes causing invasive pneumococcal disease (IPD) in adults due to herd protection. We characterized 2172 cases of adult IPD in 2015–2018 in Portugal after the introduction of PCV13 in the national immunization plan of 2015. Among the 58 detected serotypes, serotypes 8 (n = 413; 19%), 3 (n = 334; 15%), 22F (n = 148; 7%), 14 (n = 138; 6%), and 19A (n = 116; 5%) were the most frequent. Among PCV13 serotypes, 7F and 19A IPD decreased, but serotype 3 IPD remained stable. The non-PCV13 serotypes were a heterogeneous group, with serotypes 23A and 23B enriched among CSF cases; serotype 8 associated with younger patients; and serotypes 22F, 6C, and 31 associated with older patients. The continued increase of serotype 8 IPD was one of the drivers for the increased coverage of the 23-valent pneumococcal polysaccharide vaccine (PPV23; 80% in 2015–2018). Antimicrobial resistance was associated with older age and serotypes 6C, 11A, 14, 15A, 19A, and 19F. Three years after the introduction of PCV13 in the NIP with an uptake of >95%, the proportion of PCV13 serotypes causing IPD in adults stabilized in Portugal. The direct vaccination of adults may be important in preventing IPD in this age group.


Introduction
The introduction of pneumococcal conjugate vaccines (PCVs) in children led to changes in the incidence and serotypes causing invasive pneumococcal disease (IPD), not only in vaccinated children but also in adults. In fact, initial data from the United States showed that more cases were averted by this herd protection than by the effects on the groups targeted by vaccination. The increasing use of the 23-valent polysaccharide vaccine (PPV23) and the approval of the 13-valent conjugate vaccine (PCV13) for use in adults also has the potential to further impact IPD in this age group. In Portugal, as elsewhere, the use of increasing valency PCVs in children led to changes in the serotypes causing IPD in adults [1][2][3]. However, although reaching a substantial vaccine coverage (around 63%), the use of the PCVs through the private market and without reimbursement may have contributed to the persistence of vaccine serotypes as causes of IPD by creating an uneven distribution of the vaccine. The use of the 7-valent PCV (PCV7) in children led to a decrease in the proportion of PCV7 serotypes in adult IPD in Portugal in 2006-2008, with the most frequent serotypes being 3, 1, 7F, 19A, and 14 [3]. After the introduction of PCV13 for children vaccinations in 2010, a decline in the proportion of the additional serotypes included in PCV13 was observed in adult IPD, mostly driven by decreases in the proportion

Materials and Methods
Since 1999, the Portuguese Group for the Study of Streptococcal Infections has been monitoring invasive pneumococcal disease in Portugal through a laboratory-based surveillance system of 31 microbiology laboratories involving the collection and shipment of all IPD isolates to a central laboratory for characterization [12]. Though all laboratories were periodically contacted to submit isolates to the central laboratory, no audit was performed to ensure compliance, which may be variable in this type of study. A case of IPD was defined by the isolation of pneumococci from a normally sterile fluid, such as blood, cerebrospinal fluid (CSF), or pleural fluid, or the detection of pneumococcal DNA in a normally sterile body fluid, excluding blood. Isolates included in this study were from adult patients (≥18 years) with IPD between January 2015 and December 2018, with only one isolate from each episode being considered. Isolates were identified as pneumococci by colony morphology on blood agar plates, optochin susceptibility, and bile solubility.
Minimal inhibitory concentrations (MICs) for penicillin and cefotaxime were determined using Etest strips (Biomérieux, Marcy l'Étoile, France). Unless otherwise stated, we used the CLSI-recommended breakpoints for oral penicillin as epidemiological breakpoints that allowed for comparison with previous studies [15]. Isolates were further characterized by determining their susceptibility to erythromycin, clindamycin, vancomycin, Microorganisms 2021, 9, 1016 3 of 14 linezolid, tetracycline, levofloxacin, trimethoprim-sulfamethoxazole, and chloramphenicol by the Kirby-Bauer disk diffusion technique according to the CLSI recommendations and interpretative criteria [15].
Macrolide resistance phenotypes were identified using a double disc test with erythromycin and clindamycin, as previously described [16]. Simultaneous resistance to erythromycin and clindamycin defines the MLS B phenotype (resistance to macrolides, lincosamides, and streptogramin B), while non-susceptibility to only erythromycin indicates the M phenotype.
Simpson's index of diversity (SID) and respective 95% confidence intervals (CI95) were used to measure population diversity [17]. Differences were evaluated by Fisher's exact test, and the Cochran-Armitage test (CA) was used for trends with the false discovery rate (FDR) correction for multiple testing [18]. A p-value of < 0.05 was considered significant for all tests.

Isolate Collection
Between 2015 and 2018, a total of 2172 isolates responsible for adult IPD were collected in Portugal, distributed as follows: 529 in 2015, 501 in 2016, 576 in 2017, and 566 in 2018. One additional case was identified and directly serotyped by molecular methods from patient CSF in 2015. In most cases, the pneumococcus was identified in blood (93.1%; n = 2013), as well as in the CSF (5.2%; n = 113), pleural fluid (1.3%; n = 29), peritoneal fluid (0.5%; n = 11), synovial fluid (0.2%; n = 5), and pericardial fluid (0.1%; n = 2). The distribution of the cases among the different age groups was as follows: 397 isolates were recovered from patients aged 18-49 years, 497 were recovered from patients 50-64 years, and 1279 were recovered from older patients (≥65 years).
Overall serotype diversity was high (0.920, CI95 = 0.913-0.926), with no differences in SID between the four years included in this study. Serotype diversity was also high among the three age groups, although a difference was noted between isolates recovered in patients        There were differences in serotype distribution between the age groups. The proportion of isolates expressing the 24 major serotypes (n > 20 isolates) in each age group is represented in Table 1. Among these serotypes, some decreased in importance with increasing age group, such as serotype 8 and 4, while the opposite trend was noted for serotypes 3, 22F, 6C, 14, and 31 (all significant after FDR correction).  Overall serotype diversity was high (0.920, CI95 = 0.913-0.926), with no differences in SID between the four years included in this study. Serotype diversity was also high among the three age groups, although a difference was noted between isolates recovered in patients aged 18-49 years (SID = 0.884; CI95 = 0.859-0.908) and those aged ≥65 years (SID = 0.927; CI95 = 0.920-0.933) (p = 0.015). Isolates recovered from patients aged 50-64 years had an intermediate SID value between these two: 0.905 (CI95 = 0.888-0.921).
There were differences in serotype distribution between the age groups. The proportion of isolates expressing the 24 major serotypes (n > 20 isolates) in each age group is represented in Table 1. Among these serotypes, some decreased in importance with increasing age group, such as serotype 8 and 4, while the opposite trend was noted for serotypes 3, 22F, 6C, 14, and 31 (all significant after FDR correction).   During the study period, there were also changes in the proportion of IPD cases caused by vaccine serotypes, as represented in Figure 4. PCV13 serotype IPD decreased from 40.0% in 2015 to 33.6% in 2018, mostly driven by a decrease in the proportion of PCV7 serotype IPD, which decreased from 15.5% in 2015 to 11.3% in 2018, both of which were significant after FDR correction (p = 0.01 and p = 0.02, respectively). The proportion of IPD caused by isolates expressing addPCV13 serotypes also decreased, though less significantly, from 24.5% in 2015 to 22.3% in 2018. The proportion of PPV23 serotype IPD slightly increased from 78.3% in 2015 to 81.3% in 2018, with the cases caused by the addPCV15, addPCV20, and addPPV23 serotypes considered together increasing from 39.1% in 2015 to 48.4% in 2018 (p < 0.001, which was significant after FDR correction). The proportion of NVT IPD decreased slightly from 20.9% in 2015 to 18.0% in 2018. The variations in PCV15 serotype IPD closely tracked those of PCV13 serotype IPD, whereas that of PCV20 serotype IPD closely followed that of PPV23 serotype IPD (Figure 4). When also considering the previous study period (2012)(2013)(2014) in the analysis, the same trends were significant, but, in this case, the decrease in PCV13 serotype IPD was mostly driven by a decrease in the proportion of IPD cases caused by the addPCV13 serotypes from 37.6% in 2012 to 22.3% in 2018 (p < 0.0001, which was significant after FDR correction).
The evolution of individual serotypes responsible for IPD in adults from 2012 to 2018 is represented in Table 2 for serotypes expressed by >3 isolates in at least one of the considered years. In the current study period (2015-2018), the proportion of IPD caused by serotypes 8 and 12F increased from 14.9% and 0.9% in 2015 to 22.1% and 2.7% in 2018, respectively, both unsupported after FDR correction. The opposite trend was detected for serotypes 7F (from 3% to 1.2%), 19A (from 6.8% to 2.8%), and 6C (from 3.8% to 1.8%), although none of these were supported after FDR correction. However, when considering data from 2012 to 2018, several significant changes were detected, including an increase in proportion of serotype 8 IPD (from 8.4% to 22.1%; CA p < 0.001) and serotype 33F IPD (from 0.2% to 1.6%; CA p = 0.001), as well as decreases in the proportion of serotype 7F (from 8.2% to 1.2%; CA p < 0.001), 1 (from 3.0% to 0%; CA p < 0.001), and 19A IPD (from 9.7% to 2.8%; CA p < 0.001). During the study period, there were also changes in the proportion of IPD cases caused by vaccine serotypes, as represented in Figure 4. PCV13 serotype IPD decreased from 40.0% in 2015 to 33.6% in 2018, mostly driven by a decrease in the proportion of PCV7 serotype IPD, which decreased from 15.5% in 2015 to 11.3% in 2018, both of which were significant after FDR correction (p = 0.01 and p = 0.02, respectively). The proportion of IPD caused by isolates expressing addPCV13 serotypes also decreased, though less significantly, from 24.5% in 2015 to 22.3% in 2018. The proportion of PPV23 serotype IPD slightly increased from 78.3% in 2015 to 81.3% in 2018, with the cases caused by the addPCV15, addPCV20, and addPPV23 serotypes considered together increasing from 39.1% in 2015 to 48.4% in 2018 (p < 0.001, which was significant after FDR correction). The proportion of NVT IPD decreased slightly from 20.9% in 2015 to 18.0% in 2018. The variations in PCV15 serotype IPD closely tracked those of PCV13 serotype IPD, whereas that of PCV20 serotype IPD closely followed that of PPV23 serotype IPD (Figure 4). When also considering the previous study period (2012-2014) in the analysis, the same trends were significant, but, in this case, the decrease in PCV13 serotype IPD was mostly driven by a decrease in the proportion of IPD cases caused by the addPCV13 serotypes from 37.6% in 2012 to 22.3% in 2018 (p < 0.0001, which was significant after FDR correction).  Table 2 for serotypes expressed by >3 isolates in at least one of the considered years. In the current study period (2015-2018), the proportion of IPD caused by serotypes 8 and 12F increased from 14.9% and 0.9% in 2015 to 22.1% and 2.7% in 2018, respectively, both unsupported after FDR correction. The opposite trend was detected for serotypes 7F (from 3% to 1.2%), 19A (from 6.8% to 2.8%), and 6C (from 3.8% to 1.8%), although none of these were supported after FDR correction. However, when considering data from 2012 to 2018, several significant changes were detected, including an increase in proportion of serotype 8 IPD (from 8.4% to 22.1%; CA p < 0.001) and serotype 33F IPD    (for the purpose of this paper, we assumed that protection was afforded against IPD by the 15B/C group of isolates); addPPV23: the additional 3 serotypes exclusively present in the 23-valent pneumococcal polysaccharide vaccine-9N, 17F and 20; NVT: serotypes not included in any of the currently available pneumococcal vaccines. 2 CA: Cochran-Armitage test for trend. Values in bold were significant after FDR correction.

Discussion
It is well established that vaccinating children leads to a decrease of vaccine-type IPD in adults through herd immunity [19,20]. This is thought to be related to the proportion of vaccinated children [19], so the increase in uptake anticipated from the introduction of PCV13 in the NIP in Portugal was expected to result in further effects in adult IPD. Moreover, the emergence of NVT disease could further influence the serotypes in IPD and even partially erode the benefits of the herd effect of vaccination on the overall burden of disease [20,21].
The decrease of PCV7 serotypes, felt since a few years after PCV7 introduction [1,2], continued to occur but at a very slow pace and with all PCV7 serotypes still being present as causes of IPD in 2018. In contrast, the decrease of the addPCV13 serotypes was underpinned by serotypes 1 and 5, which were not found among IPD cases in the last years of the study, and decreases in serotypes 7F and 19A. In fact, the decrease in addPCV13 serotypes was not more pronounced due to an increase in serotype 3 IPD. Significant increases in the incidence of serotype 3 IPD were also seen in France and England in recent years [21,22]. In fact, the leading serotypes causing adult IPD in Portugal in 2015-2018 were similar to those found in other European countries [4,[21][22][23], although their rank order could be different. Among the PCV serotypes, serotype 3 was always found to be a leading cause of adult IPD in Europe in recent years, together with serotypes 19A and 7F [4,[21][22][23], although the latter two have frequently been found to be decreasing of late, as seen in Portugal. Serotype 14 persists in Portugal and Spain [4,23] but is less frequent elsewhere [21,22]. Among the serotypes not included in any PCV, serotype 8 is consistently found among the three leading causes of adult IPD, if not the most prevalent, while serotypes 9N, 12F and 22F are also leading causes of IPD, albeit with a more variable rank order [4,[21][22][23]. For instance, serotype 12F is one of the three most prevalent serotypes in Spain and England [4,22,23], but it is much less frequent in Portugal and France [21]. Among the other non-PCV serotypes responsible for >2% of adult IPD in Portugal in 2015-2018-11A, 20, 15A, and 6C-most were also recently found to be important causes of adult IPD in Spain [4,23], and serotype 15A was also important in France and England [21,22], but serotype 20 was not a significant cause of disease in any of these countries. Though we have considered serotype 6C an NVT, cross-protection from the 6A and 6B components in PCV13 has been suggested, leading to a future potential reduction of this serotype despite its current persistence as a cause of adult IPD in several countries. In Ontario, Canada, IPD was found to have a similar serotype distribution to that of these European countries, with a smaller share of serotype 8 IPD [24]. In Argentina, five serotypes are responsible for over 5% of all adult IPD: 3, 8, 12F, 7F, and 1 [25]. Most of these are also among the most frequent in Europe, but, in contrast to Argentina, serotype 1 has greatly decreased or is even absent from adult IPD in European countries. Taken together, these results suggest that a similar set of serotypes is emerging in most countries using PCV13 in children as the leading causes of adult IPD, albeit with some notable regional differences in the persistence of PCV13 serotypes and the prevalence of some emerging serotypes. In contrast to the relative homogeneity among these countries, a recent regional study from Japan did not find any case of serotype 8 or 9N IPD [26]. Another prominent difference is the USA, where no increases in non-PCV13 serotype IPD were evident in either children <5 years or adults ≥ 65 years and serotypes 8, 9N, 12F, and 15A were found to be responsible for a minority of IPD [27]. The reasons behind these differences continue to be a matter of debate [27].
Similarly to what was found in England [22], serotype 3 IPD in Portugal was associated with older age. This was also the case for serotypes 14, 22F, and 31 IPD. The increased importance of IPD by these serotypes with age could be associated with infections in adults with underlying conditions, whose proportion also increases with age [28], but our study was not designed to address this. The increased case fatality rate of serotype 3 IPD reported in England [22] and its increased prevalence in older-aged adults could also be partly responsible for the increasing mortality of pneumonia with age in Portugal [28].
Despite two decades of PCV7 use in children, PCV7 serotypes are still present as causes of IPD in adults, and their demise has not accelerated after four years of the introduction of PCV13 in the NIP with an uptake > 95%. Serotypes 14 and 19F (both PCV7 serotypes) are among the leading antimicrobial-resistant serotypes, suggesting that antimicrobial use could partly counter the selective force imposed by vaccination and justify their persistence. The proportion of IPD caused by PCV13 serotypes did decrease after introduction in the NIP but seems to have stabilized in the two most recent years, as in other European countries [20]. The future PCV15 vaccine covers and additional 8% of adult IPD relative to PCV13, and PCV20 covers an additional 28% relative to PCV15 (only 8% less than PPV23). The changes in the potential coverage of PCV20 and PPV23 varied in parallel, attesting to the relevance of the addPCV20 serotypes, which increased in importance in the ≥65 years group during the study period. The large proportion of vaccine-preventable disease in Portugal highlights the potential benefits of increasing the uptake of pneumococcal vaccines in adults, with the forthcoming PCV15 and, particularly, PCV20 vaccines broadening the potential coverage by PCVs and being welcome additions to the prevention of pneumococcal disease.
The average number of isolates from adult IPD sent to us per year in this study period (n = 543) substantially increased when compared to 2008-2014 (n = 405) [1]. Though this could have been due to increases in adult IPD, as reported elsewhere [21], it could also have been influenced by increased reporting. Consistent with the latter idea, there were no differences in the proportion of isolates of each age group between the two periods. Since our study was not population-based, we could not further clarify this point.
Our study had several limitations. Our surveillance was not population-based, and it was not designed to estimate the incidence of IPD because it was an exclusively laboratory surveillance and lacked compliance audits. A survey of hospitalized pneumonia cases among adults in Portugal in 2015 found that 1328 had pneumococcal etiology [28]. Considering that most cases of IPD are from invasive pneumonia (e.g., 70% of cases in the recent report from England [22]) and that invasive pneumonia is ≈10% of all pneumonia cases, the 530 isolates we received in 2015 represent a very high fraction of all adult IPD cases in Portugal. The stability of our surveillance network and its active nature are consistent with the identification of most IPD cases, as suggested by the available data for 2015, so we expect our sample to be representative of Portugal. The study was also not designed to collect information that is important to evaluate the severity of the infections caused by the different serotypes (e.g., hospitalization, ICU admission, and 30-day mortality) or relevant patient information (e.g., vaccination and comorbidities), which would have been important to better understand the changes in IPD accompanying the described serotype dynamics and the potential benefits of the currently used vaccines.

Conclusions
Despite the introduction of PCV13 in the NIP for children four years ago, the reduction in the PVC13 serotype adult IPD was modest and seems to have stabilized. At best, the herd effect may cause an ongoing slow decrease of addPCV13 serotype IPD like what we saw for the PCV7 serotypes, with almost two decades of use of PCVs targeting these serotypes with an uptake > 60%. This would mean that a substantial fraction of potentially vaccine-preventable disease would continue to occur in the coming years. The advent of new PCVs with a higher valency (PCV15 and PCV20), together with PPV23, affords new opportunities to prevent IPD in adults. An increase in vaccine uptake in adults could potentially lead to important reductions in the 12% mortality estimated for pneumococcal pneumonia requiring hospitalization in Portugal in 2015 [28].