Relationship between immune response to pneumococcal conjugate vaccines in infants and indirect protection after vaccine implementation

ABSTRACT

Introduction: Streptococcus pneumoniae is a leading cause of morbidity and mortality worldwide. Widespread infant vaccination with pneumococcal conjugate vaccines (PCVs) substantially reduced vaccine-serotype pneumococcal disease by direct protection of immunized children and indirect protection of the community via decreased nasopharyngeal carriage and transmission. Essential to grasping the public health implications of pediatric PCV immunization is an understanding of how PCV formulations impact carriage.

Areas covered: Using clinical evidence, this review examines how the immune response to PCVs is associated with subsequent nasopharyngeal carriage reduction in vaccinated infants and toddlers. By combining direct and indirect protection, carriage reduction results in a reduced spread of vaccine serotypes, and eventually, a decrease in vaccine serotype disease incidence in community members of all ages.

Expert opinion: The current review presents some of the aspects that influence the overall impact of PCVs on vaccine-serotype carriage, and thus, spread. The link between reduction of vaccine-serotype carriage and the eventual reduction of vaccine-serotype disease in the wider community is described by comparing data from current PCVs, specifically with respect to their ability to reduce carriage of some cross-reacting serotypes (i.e. 6A versus 6B and 19A versus 19F).

KEYWORDS:

• Streptococcus pneumoniae
• pneumococcal conjugate vaccine
• nasopharyngeal colonization
• serotype 19A
• serotype 6A

1. Introduction

Streptococcus pneumoniae is a leading cause of morbidity and mortality worldwide with almost two million deaths annually, including up to one million deaths in children <5 years old [1–3]. Vaccination is one of the most effective methods for preventing pneumococcal disease, not only via protecting vaccinated individuals (direct protection) but also by decreasing the circulation of pneumococcal serotypes included in the vaccines (vaccine serotypes [VTs]) in the population as a whole (indirect or herd protection). Though effective in healthy adults, nonconjugated pneumococcal polysaccharide vaccines do not elicit a protective immune response in children younger than 2 years [4]. To overcome this shortfall, vaccines with conjugated polysaccharides were developed. As well as protecting infants and young children against invasive pneumococcal disease (IPD), these vaccines have shown additional benefits, including the reduced nasopharyngeal carriage of vaccine serotypes. Over the past two decades, the first licensed pneumococcal conjugate vaccine (the 7-valent PCV, PCV7) was introduced into national immunization programs (NIPs) in most countries worldwide [5,6]. PCV7 includes purified capsular polysaccharide of serotypes 4, 6B, 9V, 14, 18C, 19F, and 23F. Although PCV7 protects against disease caused by serotypes in the vaccine [7], pneumococcal nonvaccine serotypes (NVTs) began filling the vacant nasopharyngeal niche [8], leading to a gradual increase in disease caused by some NVTs (a phenomenon known as serotype replacement) [9]. The next-generation 10-valent (PCV10) and 13-valent (PCV13) PCVs replaced PCV7 to combat additional serotypes. PCV10 contains PCV7 serotypes plus serotypes 1, 5, and 7F; PCV13 contains PCV10 serotypes plus serotypes 3, 6A, and 19A [10].

Epidemiologic evidence shows that IPD rates are highest in individuals <5 years of age (in particular, those aged <2 years), the primary target group for pneumococcal vaccination [11–15]. However, when all disease outcomes are considered throughout all ages, >80% of morbidity and mortality occurs in individuals aged >5 years—a population not usually targeted for PCV vaccination. Because pneumococcal transmission occurs most often via contact with infants and toddlers, widespread vaccination of this young age group has the potential to substantially reduce disease across all age cohorts [16] by reducing nasopharyngeal carriage and transmission of VTs. Thus, understanding the prevention of carriage by PCVs in healthy young children is key to predicting PCV impact on public health across all ages.

This review focuses on the available clinical evidence regarding antibody responses after PCV vaccination of infants and young children, protection against carriage of vaccine-related serotypes, and vaccination as related to reducing carriage among vaccinated children. A reduction in the carriage may decrease VT transmission and disease incidence in all ages by a combination of direct and indirect protection.

2. Search strategy and selection criteria

References and data were identified by reviewing publicly available country-specific epidemiology databases, congress materials, the author’s personal files, and PubMed searches (limits, English and 1 January 2000, to 1 January 2018; search terms included ‘IPD,’ ‘carriage,’ ‘19A,’ ‘6A,’ ‘6C,’ and ‘PCV’). Additional references cited in identified articles were reviewed.

It should be noted that serotype 3 is not included in this review, as immunogenicity is poor [17,18] and correlation with carriage therefore cannot be demonstrated.

3. Levels of protection elicited by PCVs

The protection elicited by PCVs against disease caused by VT pneumococci occurs on three levels (Figure 1). The first is the acquisition of circulating neutralizing antibodies (e.g. serotype-specific immunoglobulin G [IgG]); these antibodies protect against VT pneumococci once they penetrate the infection site. The second level refers to the reduction of nasopharyngeal VT acquisition [19]. Although this is also likely mediated through antipolysaccharide antibodies [19,20], it is not clear whether this protection is afforded by circulating antibodies [21–23] or by locally produced antibodies at the submucosal level [24,25]. Furthermore, it is not clear whether nonantibody-mediated acquired immunity is also involved [26]. However, the serotype specificity of PCV-mediated protection against carriage suggests that mechanisms other than protection by polysaccharide-specific antibodies do not play an important role in PCV-induced protection against carriage, although this may differ for naturally acquired immunity [27]. Nonetheless, reducing nasopharyngeal acquisition in vaccinated subjects affords another level of protection against disease by reducing the risk of pneumococci invading target organs regardless of circulating antibody activity. The third level of protection is afforded by reduction of person-to-person transmission of VT pneumococci, resulting in reduced exposure to these pathogens in vaccinated and unvaccinated individuals. The first two levels of protection are considered individual or direct protection, whereas the third level (reduced exposure) may be termed indirect or societal protection, commonly referred to as ‘herd protection.’ Together, these three protective levels enable PCVs to impact pneumococcal disease in all ages throughout the entire community.

Figure 1. Three levels of protection elicited by PCVs against disease caused by VT pneumococci.

(1) Acquisition of circulating neutralizing antibodies (i.e. serotype-specific immunoglobulin G). (2) Prevention of VT NP acquisition following exposure to the pathogen. (3) Prevention of person-to-person transmission of VT pneumococci. Levels 1 and 2 represent protection of the vaccinated individual, often referred to as direct protection. Level 3 is referred to as indirect or herd protection. NP = nasopharyngeal; PCV = pneumococcal conjugate vaccine; VT = vaccine serotype.

However, immune responses sufficient to afford efficacy against IPD (e.g. septicemia, meningitis, arthritis) alone may not be sufficient to afford acceptable efficacy against all pneumococcal diseases, as most pneumococcal disease is mucosal (e.g. pneumonia, otitis media [OM]). As discussed above, the impact of PCVs on pneumococcal carriage in healthy, asymptomatic children who are responsible for transmitting the bacterium is key to understanding the public health implications of PCV immunization.

The majority of pneumococcal disease is not invasive, but somewhat paradoxically, most knowledge regarding the relationship between immune response and disease relates to IPD. Further complicating this assessment, licensure of new pneumococcal vaccines is based on noninferiority of immune responses, which refers to the relationship between circulating postvaccination IgG antibodies and protection from IPD. The current noninferiority criteria do not account for the relationship between a measurable immune response and protection against carriage and mucosal disease, which is still relatively obscure.

3.1. The current surrogate for noninferiority against IPD does not predict noninferiority against nasopharyngeal carriage

Because effective pneumococcal vaccines have already been licensed, placebo-controlled efficacy studies of subsequent vaccines will meet ethical and practical difficulties; therefore, immunogenicity studies using surrogate markers of efficacy are typically used for licensure of new PCVs [28–30]. Noninferiority of immunogenicity compared with a licensed PCV is determined by comparing the proportion of responders with an IgG concentration equal to or above a surrogate value for protection measured 1 month after three infant doses. The IgG concentration needed for protection against IPD established by the World Health Organization is ≥0.35 µg/mL as measured by serotype-specific pneumococcal polysaccharide enzyme-linked immunosorbent assays (ELISAs) [31–33]. Others use ≥0.2 µg/mL as the surrogate of protection if 22F polysaccharide adsorption is included [34,35].

The protective threshold of ≥0.35 µg/mL was derived from aggregating all seven serotypes in PCV7 from diverse infant populations at three study sites where infants were vaccinated with PCV7 (Northern California, Arizona) or the 9-valent PCV (PCV9), which contains PCV7 serotypes plus serotypes 1 and 5 (South Africa). However, the protective threshold of each serotype and study site varied [20,36,37]. The estimated protective concentrations against the aggregated PCV7 serotypes were 0.20, 1.0, and 0.68 µg/mL for Northern California, Arizona, and South Africa, respectively [33]. Thus, although using a uniform correlate of protection against IPD has enabled licensing of efficacious new PCVs, diversity in a population and serotype-specific correlations should be considered. A postlicensure effectiveness study in the United Kingdom emphasized such serotype diversity [38]. Regardless of the serotype or population examined, the serum antibody concentrations necessary to prevent carriage and mucosal diseases are clearly considerably higher than those needed to protect against IPD, rendering the low threshold of ≥0.35 µg/mL irrelevant for these outcomes.

Limitations of this regulatory approach were highlighted by studies demonstrating an inverse relationship between postvaccination serotype-specific antibody concentrations and the risk of nasopharyngeal pneumococcal acquisition [39–41]. A large double-blind study investigated the relationship between serotype-specific IgG concentration after a three-dose infant PCV7 or PCV13 series (vaccination at 2, 4, and 6 months of age; serum serotype-specific IgG concentrations measured at 7 months of age) and the probability of serotype-specific nasopharyngeal acquisition from age 7 to 24 months [42]. Nasopharyngeal acquisition probability decreased as IgG concentrations increased, reaching statistical significance for serotypes 6A, 6B, 14, and 23F. A nonsignificant but similar pattern was also observed for serotypes 4, 9V, 18C, and 19F. The probability of new acquisitions in children who achieved concentrations of ≥0.35 µg/mL was high, ranging from 65% to 100% for various serotypes, clearly demonstrating that the ≥0.35-µg/mL threshold does not predict protection against carriage. In fact, the predicted acquisition rate was relatively high even when reaching ≥50th percentile of antibody concentrations, and no threshold was demonstrated when reaching concentrations in the 90th percentile [41]. Similar results were observed in a study on the efficacy against acute OM in Finland [43].

Together, these data suggest three important points: (1) the protective threshold for preventing nasopharyngeal acquisition or mucosal disease is considerably higher than that for IPD; (2) variability exists between serotypes in the degree of protection achieved by specific IgG concentrations; and (3) there is an inverse relationship between post-infant series serotype-specific serum IgG concentration and efficacy against nasopharyngeal acquisition rates, with no achievable threshold concentration predicting 90% efficacy. Thus, the post-infant series antibody concentration may approximate the degree of immune stimulation that results in protection against carriage. A recent study reported that the best predictor of protection against nasopharyngeal pneumococcal acquisition is the presence of circulating serotype-specific memory B cells in the blood [44]. Thus, serotype-specific IgG concentrations measured 1 month after vaccination may represent stimulation of memory B cells. It is plausible to speculate that serotype-specific IgG concentrations measured 1 month after vaccination may predict the degree of stimulation to memory B cells.

The studies described above were conducted with a vaccine schedule of 3 + 1 (given at 2, 4, 6, and 12 months of age) and antibody concentrations measured 1 month after completion of the primary series (age 7 months). To date, no published results examining the correlation between postvaccination antibody concentrations and acquisition of carriage have been identified from studies using other vaccine schedules, although it is anticipated that these data are forthcoming [45,46]. However, studies using other schedules have shown an inverse correlation between PCV dose number and carriage acquisition [47–49]. In a study where three doses of PCV7 were given at 2-month intervals starting at age 2 months, efficacy against carriage lasted through the second year for most serotypes. Efficacy was similar in those who did or did not receive a booster dose at 12 months of age, despite a decline in circulating antibody concentrations in children not receiving a booster dose [50].

In contrast to PCVs, nonconjugated polysaccharide pneumococcal vaccines do not afford protection against carriage, despite eliciting high serum IgG concentrations. In one randomized study, a booster with nonconjugated 23-valent pneumococcal polysaccharide vaccine (PPSV23) was compared with placebo in infants previously primed during infancy with one, two, or three PCV7 doses. A brisk increase in antibody concentrations of the PCV7 serotypes was observed after PPSV23 administration, but no effect on carriage was observed [48,51], suggesting that plain polysaccharide booster vaccines do not induce protective antibodies against carriage regardless of the elicited concentrations. Results were consistent with previous studies, suggesting that immunizing vaccine-naive children with PPSV23 or administering a PPSV23 booster does not affect nasopharyngeal carriage in children (in contrast to PCV), despite being immunogenic [52,53]. This difference likely derives from different types of immune response elicited by PCV versus PPSV23 (T-cell dependent vs T-cell independent, respectively) [54].

3.2. Noninferiority, as used for licensure, predicts protection against VT IPD similar to that of a licensed comparator but may mask inferior protection against carriage

Current noninferiority testing required for PCV licensure is not sufficient to predict noninferiority against carriage; thus, the overall potential public health impact of newly licensed vaccines cannot be compared using this method. Considering the data presented above, a PCV eliciting higher serum IgG responses will logically provide better protection against carriage and VT pneumococcal spread, resulting in more effective indirect protection. Alternatively, antibody geometric mean concentrations (GMCs; measured by ELISA) or geometric mean titers (GMTs; measuring functional bactericidal antibody responses by opsonophagocytic activity [OPA]) determined shortly after vaccination are better predictors of efficacy against carriage acquisition than the proportion of vaccine recipients achieving an IgG concentration threshold of ≥0.35 µg/mL [41].

Currently, only two PCVs have been licensed based on the noninferiority principle: PCV13 and PCV10. Although both are noninferior to PCV7 for most serotypes, their effect on carriage is not similar. Studies used for PCV10 licensure that were designed to test noninferiority of PCV10 versus PCV7 [34,35] show the potential limitations of the current assessment method to predict noninferiority for protection against carriage. Specifically, the proportion of children who reached serum IgG concentrations ≥0.20 µg/mL for each serotype was examined to establish noninferiority. Although these proportions were similar for both vaccines for most serotypes (except serotypes 6B and 23F; Figure 2 (a)) [34,35], the actual serotype-specific IgG concentrations (expressed by their GMCs) showed a large difference for most serotypes, usually in favor of PCV7 (Figure 2 (b)) [34,35]. Although these data suggest that direct efficacy of PCV10 and PCV7 against IPD for common serotypes is similar, the data also suggest that PCV7 might better protect against VT carriage.

Figure 2. Differences in inferiority observed when measuring the proportion of children achieving concentrations above threshold vs geometric mean antibody concentrations: A comparison between PCV10 and PCV7.

(a) After the third primary vaccine dose, PCV10 was noninferior to PCV7 for most serotypes using the proportion of subjects achieving a titer of ≥0.20 µg/mL response; the only exceptions were serotypes 6B and 23F. (b) PCV10 was inferior to PCV7 for most serotypes when measuring the geometric mean concentrations of anti-polysaccharide immunoglobulin G; the only exception was serotype 19F. Solid bars, data from Wysocki J, et al. Pediatr Infect Dis J 2009 [34]. Hatched bars, data from Vesikari T, et al. Pediatr Infect Dis J 2009 [35]. ELISA = enzyme-linked immunosorbent assay; PCV = pneumococcal conjugate vaccine; PCV7 = 7-valent PCV; PCV10 = 10-valent PCV.

This suggestion garners support from published information on the efficacy of PCV7 (or its successor, PCV13) and PCV10 (or its precursor, PCV11; Figure 3) [42,48,55–61]. When reviewing randomized controlled studies in which efficacy against carriage was examined at 12 months of age in children fully immunized with three infant doses, a pattern toward lower efficacy against VT carriage was observed in PCV10/PCV11 recipients compared with recipients of CRM₁₉₇-conjugated vaccines (PCV7/PCV13; Figure 3 (a)). Similar patterns were observed in randomized studies in which children received either two or three doses in infancy and a booster dose toward the second year of life (2 + 1 or 3 + 1 schedule) and for whom efficacy against carriage was tested at various time points during the second year of life (Figure 3 (b)). Although these studies were not designed for direct comparison between vaccines, the findings support the predictions for lower efficacy against VT carriage derived from the immunogenicity studies.

Figure 3. Carriage studies examining the efficacy of CRM₁₉₇-conjugated vaccines (i.e. PCV7/PCV13) and protein-D conjugated vaccines (i.e. PCV10/PCV11).

(a) All published studies in which PCV was administered as three infant doses, and carriage was measured around 12 months of age, before the toddler dose. (b) All published studies in which PCVs were administered at either two or three doses during infancy, and a booster was administered at around 12 months of age. The age when carriage was measured appears on the y-axis. Several age points were often used for the same study. The circled endpoints are for studies in which children received two doses in infancy. For all others, children received three infant doses. Each circle represents the endpoint efficacy of the vaccine measured against placebo or control vaccine; error bars represent CIs. Adapted from Fleming-Dutra KE, et al. Pediatr Infect Dis J 2014 [55]. Studies on y-axes correspond to: (a) Prymula R, et al. Vaccine 2009 [57]; (b) GlaxoSmithKline COMPAS Study 109563 (10PN-PD-DIT-028) [61]; (c) Vesikari T. European Congress of Clinical Microbiology and Infectious Diseases 2013, Berlin, Germany [56]; (d) Cheung YB, et al. Pediatr Infect Dis J 2009 [58]; (e) Russell FM, et al. Clin Vaccine Immunol 2010 [48]; (f) Dagan R, et al. Clin Infect Dis 2013 [42]; (g) O’Brien KL, et al. J Infect Dis 2007 [59]; (h) Prymula R, et al. Vaccine 2011 [62]; and (i) van Gils EJ, et al. JAMA 2009 [60]. *Data are for the additional six serotypes in PCV13 plus 6C combined. PCV = pneumococcal conjugate vaccine; PCV7 = 7-valent PCV; PCV9 = 9-valent PCV; PCV10 = 10-valent PCV; PCV11 = 11-valent PCV; PCV13 = 13-valent PCV; VT = vaccine serotype.

4. Serotype-specific efficacy against pneumococcal carriage of a cross-reacting serotype: lessons learned from serotypes 6B versus 6A and 19F versus 19A

4.1. Prelicensure studies

PCV7 and PCV10 were developed with the expectation that the protective anticapsular antibodies induced by serotypes in the vaccine could also protect against related, potentially cross-reacting serotypes. Specifically, antibodies generated against vaccine serotypes 6B and 19F present in PCV7 and PCV10 were anticipated to be sufficient to protect against colonization and disease caused by the related serotypes 6A and 19A. Such cross-protection was shown for serotype 6A in infants fully vaccinated with PCV7 or PCV10 against IPD [63–65]; in addition, some limited data suggested that PCV10 may protect against IPD caused by serotype 19A [65–67]. However, because cross-protection needs higher antibody levels of the parent polysaccharide antigen (i.e., protection against serotype 19A carriage by anti-serotype 19F antibodies) and because an overall higher level of antibodies is needed to protect against carriage than against IPD, the cross-protection afforded against IPD that is needed may not suffice to protect against carriage.

Head-to-head immunogenicity studies comparing immune responses directed against serotypes 6A and 19A elicited by PCV7, PCV10, and PCV13 clearly demonstrated that antibody responses elicited by PCV13 were higher than PCV7 and PCV10 as measured by OPA GMTs (Table 1) and IgG GMCs [34,35,68–76]. Responses to PCV7 were somewhat better than PCV10 for cross-reactive immunity against serotype 6A; conversely, responses to PCV10 were somewhat better than PCV7 for cross-reactive immunity against serotype 19A (Table 1) [34,35,68,77]. However, responses elicited by PCV13 were an order of magnitude higher against both serotypes 6A and 19A compared with PCV7 and PCV10 [72]. In addition, a recent study comparing the 2 + 1 schedule of PCV10 and PCV13 showed that postbooster opsonic activity was significantly higher for PCV13 for five of the 10 common serotypes (i.e. 4, 6B, 7F, 9V, and 23F; PCV10/PCV13 ratios 0.22–0.72) and for the additional PCV13 serotypes (0.08, 0.03, and 0.04 for serotypes 3, 6A and 19A, respectively) versus PCV10 [76]. As the effect on mucosal infections and carriage may depend on higher antibody concentrations than for IPD, these differences suggest a higher potential immune response for PCV13 compared with PCV10 for the six common and three additional vaccine serotypes [76]. Moreover, a recent study examining IgG and B-cell responses among humans challenged with S. pneumoniae suggested that serotype-specific circulating memory B-cell responses (but not IgG) are associated with natural protection against carriage [44]. Importantly, serotype 6A and 19A memory B cells were frequently higher in PCV13 recipients than in PCV10 recipients both before and after booster dose administration [78].

Table 1. Immunogenicity of serotypes 19A and 6A using OPA assays from clinical studies comparing PCV7, PCV10, and PCV13.

	Serotype 19A OPA GMT (95% CI)		Serotype 6A OPA GMT (95% CI)
Vaccine Administered	Primary Series	Booster Dose	Primary Series	Booster Dose
PCV7 vs PCV10 Studies
PCV10 [68]	10.9 (8.4, 14.3)	64.9 (44.6, 94.5)	188.9 (139.8, 255.3)	196.1 (134.6, 285.9)
PCV7 [68]	5.3 (4.2, 6.7)	15.5 (8.1, 29.9)	358.0 (213.4, 600.5)	1415.7 (891.2, 2248.9)
PCV10 [35]	8.6 (7.1, 10.5)	29.2 (22.3, 38.3)	40.9 (31.4, 53.3)	254.8 (202.8, 320.1)
PCV7 [35]	4.5 (3.9, 5.3)	11.1 (7.3, 17.0)	83.3 (51.6, 134.3)	872.7 (597.1, 1275.7)
PCV10* [34]	7.1 (5.6, 9.0)	18.9 (13.0, 27.4)	100.9 (65.9, 154.6)	293.7 (217.9, 395.8)
PCV7* [34]	4.1 (3.9, 4.2)	8.7 (6.6, 11.4)	231.5 (155.1, 345.4)	1166.4 (877.2, 1550.9)
PCV10^† [74]	10.6 (7.9, 14.2)	89.3 (58.9, 135.4)	-	-
PCV7^† [74]	4.2 (3.8, 4.7)	8.7 (5.4, 14.2)	-	-
PCV10^‡ [74]	10.1 (7.8, 13.1)	70.7 (45.3, 110.4)	-	-
PCV7^‡ [74]	4.0 (4.0, 4.0)	20.3 (8.8, 46.8)	-	-
PCV10 [75]	9.0 (6.9, 11.7)	-	-	-
PCV7 [75]	4.9 (4.4, 5.4)	-	-	-
PCV7 vs PCV13 Studies
PCV13 [69]	152 (105, 220)	1024 (774, 1355)	980 (783, 1226)	2242 (1707, 2945)
PCV7 [69]	7 (5, 8)	268 (164, 436)	100 (66, 152)	539 (375, 774)
PCV13 [70]	442 (361, 543)	1349 (1089, 1672)	1228 (883, 1708)	2502 (1916, 3266)
PCV7 [70]	7 (5, 9)	59 (37, 94)	122 (74, 202)	501 (341, 736)
PCV13 [71]	557 (493.1, 628.9)	1287 (1109.0, 1494.1)	4882 (4206.4, 5665.3)	5799 (4938.7, 6808.4)
PCV7 [71]	9 (7.3, 12.0)	24 (17.3, 34.5)	456 (301.0, 691.3)	1047 (711.2, 1542.0)
PCV10 vs PCV13 Studies
PCV13 [72]	-	-	2832.0 (2212.8, 3624.4)	5200.7 (4134.9, 6541.3)
PCV10 [72]	-	-	36.5 (22.6, 59.0)	146.4 (87.6, 244.6)
PCV13 [73]	-	4516 (3031, 6729)	-	18,094 (13,293, 24,629)
PCV10 [73]	-	285 (139, 585)	-	2888 (1812, 4602)

GMT = geometric mean titer; OPA = opsonophagocytic activity; PCV = pneumococcal conjugate vaccine; PCV7 = 7-valent PCV; PCV10 = 10-valent PCV; PCV13 = 13-valent PCV.

*Reported with coadministered Haemophilus influenzae type b meningococcal serogroup C (Hib-MenC).

^†Data from Filipino infants.

^‡Data from Polish infants.

Nevertheless, it is important to note that studies suggest that PCV7 vaccination does provide some protection against serotype 6A carriage [42,79] and OM [42,80]. However, given markedly higher antibody responses to serotypes 6A and 19A by PCV13 compared with PCV7 and PCV10, it is predicted that PCV13 will provide the best protection against serotype 6A and 19A carriage [42,69,70,81,82].

Still, the question remains: Are the lower immune responses elicited by cross-reacting serotypes in PCV10 sufficient to reduce carriage of serotypes 6A and 19A? A convincing answer to this question derives from randomized studies comparing efficacy against serotype 6A and 19A carriage by PCV13 versus PCV10 (in Papua New Guinea), PCV10 versus PCV7 (in the Netherlands), and PCV13 versus PCV7 (in Israel) (data for serotype 19A shown in Figure 4) [42,83,84]. In the Papua New Guinea study [84], PCV10 or PCV13 was given as a three-dose series at 1, 2, and 3 months of age; carriage was examined prior to and at 1 and 6 months following the third vaccine dose. Compared with PCV13 recipients, PCV10 recipients had a higher prevalence of 19A carriage at both post-vaccination time points. However, the number of isolates was too small to determine statistical significance. In the Netherlands and Israel studies, PCVs were given as a three-dose infant series, with a booster toward the second year of life (3 + 1); carriage was examined at multiple age points. Whereas serotype 19A carriage acquisition and prevalence were significantly lower in PCV13 than in PCV7 recipients (Figure 4 (a)) [42], serotype 19A carriage was not significantly different at any point in PCV10 compared with PCV7 recipients (Figure 4 (b)) [83]. Similar findings were found for serotype 6A [42,83]. Studies directly comparing the efficacy of PCV13 versus PCV10 against 6A and 19A carriage with significant carriage outcome to permit statistical significance are lacking. However, since the studies in Israel and the Netherlands compared PCV13 and PCV10, respectively, each to PCV7, this allowed use of PCV7 as the bridging element for comparison. In a double-blind, placebo-controlled study in Finland, PCV10 was more effective than placebo against serotype 19A in one single age point with a 3 + 1 schedule but at no age point with a 2 + 1 schedule. A summary of other studies supports the suggestion that PCV10 may have negligible or even detrimental effects on carriage of serotype 19A [85]. For serotype 6A, no reduction in carriage at any age point was demonstrated with any of these schedules [86]. Importantly, PCV13 reduced serotype 6C carriage due to higher titers of cross-protective antibodies against serotype 6A [42]; this could not be achieved by PCV7 or PCV10, which only contain antigen against serotype 6B and not 6A. These studies strongly suggest that the antibody responses to 6B and 19F in PCV10 and PCV7 may not be sufficient to reduce serotype 6A and 19A nasopharyngeal carriage to a similar level as the actual 6A and 19A antigens in PCV13.

Figure 4. Cross-protection induced by cross-reactive serotype antigens does not provide a similar level of protection to that elicited by the specific serotype antigen against serotype 19A carriage: results of two double-blind studies.

(a) Results of a double-blind study comparing PCV13 and PCV7 for 19A carriage in Israel. Adapted from Dagan R, et al. Clin Infect Dis 2013 [42]. (b) Results of a study comparing the efficacy of PCV10 and PCV7 on 19A carriage in the Netherlands. Adapted from van den Bergh MR, et al. Clin Infect Dis 2013 [83]. PCV = pneumococcal conjugate vaccine; PCV7 = 7-valent PCV; PCV10 = 10-valent PCV; PCV13 = 13-valent PCV.

4.2. Postlicensure studies

The results of the studies reviewed above examining immunogenicity and efficacy against carriage are consistent with post-PCV introduction carriage studies (Table 2) [42,83,86–111]. For example, in Kenya, PCV10 was introduced into the NIP in January 2011 [87]. PCV10 was not only unable to reduce serotype 6A and 19A carriage, but carriage of these two serotypes actually increased after PCV10 introduction, as expected from an NVT serotype [87]. Following the introduction of PCV10 in Fiji in October 2012, annual cross-sectional studies to 2015 showed that carriage of 19A increased in infants aged 5 to 8 weeks and 12 to 23 months but did not change significantly in children aged 2 to 6 years [108]. In contrast, carriage of serotype 6A remained unchanged in the infant and toddler age groups and decreased significantly in 2- to 6-year-old children over the same time period. A surveillance study in the Netherlands examining carriage 7 years after the implementation of PCV7/PCV10 did not support PCV10-induced cross-protection against serotype 19A carriage in children or their parents. At 2 years after PCV10 introduction, serotype 19A was mainly reduced in 24-month-old children vaccinated with PCV7, rather than in younger children who received PCV10 [112]. Three years after PCV10 introduction in Brazil, serotypes 6C, 6A, and 19A were the first, second, and sixth most prevalent carriage serotypes in healthy children aged 12 to 23 months [103]. Similarly, a study in New Zealand evaluating pneumococcal carriage 3 years after PCV10 introduction reported increased carriage of serotypes 19A and 6C compared with the PCV7 period. These two serotypes ranked first and second in carriage prevalence in healthy children <3 years of age [101].

Table 2. Postlicensure PCV10 and PCV13 studies assessing the effect of vaccination on serotype 19A, 6A, and 6C carriage.

Study Location	Study Period	Age Group, n	Serotype 19A Carriage	Serotype 6A Carriage	Serotype 6C Carriage
PCV10 Studies
Kenya [87]	2009–2012 (PCV10 included in NIP in 2011)	All age groups, 2031	Carriage increased from 2009–2012*	Carriage increased from 2009–2012*	Only two isolates detected during study period
Brazil [88]	Dec 2010–Feb 2011 (PCV10 included in NIP in 2010)	7–11 mo and 15–18 mo, 1287	Serotype 19A was second highest among non-PCV10 serotypes (6.3%)	Serotype 6A was highest among non-PCV10 serotypes (9.8%)	Serotype 6C was third highest among non-PCV10 serotypes (5.9%)
Brazil [103]	Mar–Aug 2010 and Jun 2013 (PCV10 included in NIP in 2010)	12–23 mo, 901	Pre-PCV10: 1.8% Post-PCV10: 5.1%	Pre-PCV10: 4.2% Post-PCV10: 4.0%	Pre-PCV10: 1.8% Post-PCV10: 11.2%
Australia [89]	Sep 2008–Dec 2012 (PCV10 included in NIP in 2009 in Northern Territories)	0–6 y, 421 (PCV7) and 443 (PCV10)	PCV7 group: 10% PCV10 group: 8%	Serotype 6A was sixth most common serotype identified in PCV7 group; not in top 10 serotypes in PCV10 group	Serotype 6C was fifth most common serotype in both groups
New Zealand [101]	May–Nov 2011 May–Nov 2014 (PCV10 included in NIP in late 2011)	<36 mo, 935	May–Nov 2011: 15% May–Nov 2014: 22%	May–Nov 2011: <5% May–Nov 2014: <5%	May–Nov 2011: 5% May–Nov 2014: 12%
Netherlands [83]	Apr 2008–Dec 2010 (PCV7 included in NIP in Mar 2006; PCV10 introduced in Mar 2011)	5–24 mo, 260 (PCV7) and 520 (PCV10)	PCV7: 30.8% PCV10: 28.1%	Similar between PCV7 and PCV10 groups	NR
Netherlands [102]	Autumn/winter 2010/2011 and autumn/winter 2012/2013 (PCV7 included in NIP in Mar 2006; PCV10 introduced in Mar 2011)	11–24 mo, 1169	In 2012/2013, 19A carriage reduced significantly in children aged 24 mo vaccinated with PCV7, but not significant in children aged 11 mo vaccinated with PCV10	NR (assay did not distinguish between 6A and 6B)	NR (assay did not distinguish between 6C and 6D)
Finland [86]	Feb 2009–Dec 2011 (PCV10 included in NIP in Sep 2010)	6 wk–18 mo, 6178	PCV10 (3 + 1 schedule), across all visits: 1.8% Control across all visits: 3.4%	PCV10 (3 + 1 schedule), across all visits: 6.7% Control across all visits: 6.2%	PCV10 (3 + 1 schedule): <5% Control: 1.4%
Fiji [108]	Jul 2012–Dec 2012 Jul 2013–Dec 2013 Jul 2014–Dec 2014 Jul 2015–Dec 2015 (PCV10 included in NIP in Oct 2012)	5–8 wk, 2006 12–23 mo, 2004 2–6 y, 2052 (Approx. 500 subjects were sampled each year in each age group)	5–8 wk: 2012: 0.42%; 2015: 2.08% 12–23 mo: 2012: 1.01%; 2015: 2.83% 2–6 y: 2012: 1.37%; 2015: 2.96%	5–8 wk: 2012: 1.04%; 2015: 1.66% 12–23 mo: 2012: 3.85%; 2015: 3.04% 2–6 y: 2012: 3.33%; 2015: 1.38%	NR
Iceland [109]	2009–2015 (PCV10 included in NIP in 2011)	<4 y, 1224	2009–2012: 5.0% 2013–2015: 3.6%	2009–2012: 8.3% 2013–2015: 5.7%	2009–2012: 0.9% 2013–2015: 6.5%
Brazil [110]	2010–2013 (PCV10 included in NIP in 2010)	12–23 mo, 901	2010: 1.8% 2013: 5.1%	2010: 4.2% 2013: 4.0%	2010: 1.8% 2013: 11.2%
Kenya [111]	2009–2017 (PCV10 included in NIP in 2011)	<5 y, 1409	2009–2010: 1.75%* 2011–2017: 6.25%*	2009–2010: 6.75%* 2011–2017: 6.25%*	NR
PCV13 Studies
United States [90]	2001–2011 (PCV13 included in NIP in 2010)	<7 y, 5380	Decline from 2009–2011*	Decline from 2009–2011*	Decline from 2009–2011*
United States [91]	Jul 2010–Jun 2012 (PCV13 included in NIP in 2010)	<5 y, 4330	Year 1: 3 per 100 children Year 2: <1 per 100 children	2010: <5 per 1000 children; after beginning of 2012: no further 6A carriage identified	2 per 100 children; no apparent increase or decrease over time
United States [92]	Jul 2010–Jun 2013 (PCV13 included in NIP in 2010)	6–59 mo, 2048	Jul–Dec 2010: 26% Jan–Jun 2013: 3%	Jul–Dec 2010: 0.8% Jan–Jun 2013: not detected	Jul–Dec 2010: 8% Jan–Jun 2013: 1%
Italy [93]	Sep–Dec 2011 (PCV13 included in NIP in 2010)	3–59 mo, 1312^†	Unvaccinated: 9% PCV7: 12% PCV13: 0%	Unvaccinated: 5% PCV7: 1% PCV13: 0%	Unvaccinated: 14% PCV7: 8% PCV13: 5%
France [94]	Oct 2010–May 2011 (PCV13 included in NIP in 2010)	6–24 mo, 943	PCV7: 15.4% PCV13: 7.5% (P< 0.001)	PCV7: 0 PCV13: 0.5%	PCV7: 8.4% PCV13: 3.7% (P= 0.003)
France [95]	1999–2012^‡ (PCV13 included in NIP in 2010)	3–40 mo, 343 (2008) and 365 (2012)	2008: 15.5% 2012: 6.5%	NR	2008: NR 2012: 5.3%
Portugal [96]	2006–2011 (PCV13 licensed in 2010)	0–6 y, 448	2006: 7.7% 2007: 11.8% 2009: 14.6% 2010: 13.1% 2011: 5.3%	2006: 12.0% 2007: 7.3% 2009: 1.9% 2010: 1.0% 2011: 2.8%	2006: 4.1% 2007: 5.4% 2009: 8.1% 2010: 17.6% 2011: 7.4%
Israel [107]	Jul 2009–Dec 2014 (PCV7 included in NIP in Jul 2009; PCV13 gradually replaced PCV7, without a further catchup, in Nov 2010)	<5 y, 10,702	Nov 2009–Jun 2010: 6.3% Jul 2014–Dec 2014: 2.0%	Nov 2009–Jun 2010: 5.1% Jul 2014–Dec 2014: 0.1%	NR
Israel [42]	Feb 2008–Sep 2011 (PCV7 included in NIP in Jul 2009; PCV13 gradually replaced PCV7, without a further catchup, in Nov 2010)	2–24 mo, 934 (PCV7) and 932 (PCV13)	PCV7: 24.6% PCV13: 14.3%	PCV7: 16.5% PCV13: 9.9%	PCV7: 6.2% PCV13: 2.9%
Gambia [98]	Mar–Jun 2011 (in children who received three PCV7 doses) and Mar–Jun 2012 (in children who received three PCV13 doses); PCV7 introduced into EIP second half of 2009; PCV13 replaced PCV7 in Jun 2011	6–11 mo, 339 (PCV7) and 350 (PCV13)	PCV7: 8.3% PCV13: 6.3% (3 y after PCV7 introduction)	PCV7: 15.3% PCV13: 5.7%	NR
Sweden [104]	2004 and 2011–2013 (PCV7 introduced in NIP in Oct 2007; replaced with PCV13 in Jan 2010)	1–5 y, 719 at daycare centers (2004); 0–5 y, 2161 (2011–2013)	Slight (<1%) increase* (<5% of isolates both before and after PCV introduction)	>10% decrease* (~12% of isolates before PCV introduction and <5% after PCV introduction)	NR for pre-PCV period and ~5% of isolates after PCV introduction*
United Kingdom [99]	Jul 2012 and Mar 2013 (PCV7 introduced into NIP in 2006; replaced by PCV13 April 2010)	1–5 y and their household members, 217 households (683 participants)	Carried by only one adult (26 y; unvaccinated)	NR	NR
United Kingdom [106]	Oct–Mar 2006/2007 to Oct–Mar 2012/2013 (PCV7 introduced into NIP in 2006; replaced by PCV13 April 2010)	≤4 y, 2267 swabs	2006/2007: ~5% 2012/2013: not detected*	2006/2007: ~10%* 2012/2013: 2.6%	2006/2007: ~5% 2012/2013: ~5%*
Australia [100]	Feb 2010–Aug 2013 (PCV13 included in NIP in 2011 in Northern Territories)	0–6 y, 511 (PCV10) and 140 (PCV13)	PCV10: 8% PCV13: not detected	PCV10: not detected PCV13: not detected	PCV10: 5% PCV13: not detected
Australia [105]	Aug 2008–Nov 2014 (PCV7 included in NIP for Aboriginal children in 2001; replaced by PCV13 in Jul 2011)	All Aboriginal age groups; 2824 swabs, 60 nose-blown samples, 1 unknown	Pre-PCV13: 6.6% Post-PCV13: 3.4%	Pre-PCV13: 4.3% Post-PCV13: 2.5%	Pre-PCV13: 8.5% Post-PCV13: 4.1%
Lao PDR [115]	Nov 2013 – Feb 2014 Nov 2015 – Feb 2016 (PCV13 was included in the Lao PDR NIP in late 2013)	5–8 wk, 498 (pre-PCV13) and 502 (post-PCV13) 12–23 mo, 503 (pre-PCV13) and 507 (post-PCV13)	5–8 wk: No significant change 12–23 mo: No significant change	5–8 wk: Pre-PCV13: 2.2% Post-PCV13: 0.6% 12–23 mo: Pre-PCV13: 6.4% Post-PCV13: 3.0%	5–8 wk: No significant change 12–23 mo: Pre-PCV13: 3.2% Post-PCV13: 1.2%

EIP = expanded immunization program; NIP = national immunization program; NR = not reported for individual serotype; PCV = pneumococcal conjugate vaccine; PCV7 = 7-valent PCV; PCV10 = 10-valent PCV; PCV13 = 13-valent PCV.

*Trend based on graph; specific numbers not provided in the reference.

^†From these children, 1295 nasopharyngeal specimens were collected.

^‡Only results for 2008 and 2012 are presented here.

In contrast to PCV10, the results of several PCV13 postlicensure studies demonstrated a rapid impact on the carriage of serotypes 19A, 6A, and 6C (Table 2). Following PCV13 licensure in the United States in February 2010 and introduction into national recommendations in March 2010 for children aged 6 weeks through 71 months [113], a study showed a substantial decrease in serotype 19A carriage from 26% in 2010 to 3% in 2013 among children aged 6 to 59 months presenting to an emergency department in Atlanta, Georgia [92]. Similarly, in Italy, where PCV13 was introduced in mid-2010, serotype 19A was identified at the end of 2011 in only 9% of nasopharyngeal swab isolates from unvaccinated children, 12% from PCV7 recipients, and none from PCV13 recipients; for serotype 6A, these percentages were 5%, 1%, and 0%, respectively [93].

In the United Kingdom, PCV7 was replaced by PCV13 in April 2010. In 217 households from July 2012 to March 2013, only one unvaccinated adult was found to carry serotype 19A; the respective data for serotype 6A were not reported [99]. Another study comparing carriage 4 years before and 3 years after PCV13 introduction reported a significant decrease in the carriage of serotypes 19A and 6A in children ≤4 years of age by 2013. Serotype 6C carriage fluctuated but returned to 2006 levels by 2013 [106]. In Israel, the proportion of children carrying 6A decreased by 90% from the early PCV7 period to the post-PCV13 period. For the five additional PCV13 serotypes excluding 6A (i.e. 1, 3, 5, 7F, and 19A), carriage declined by 66% [107]. In The Gambia, serotype 6A pneumococcal carriage in infants after PCV13 introduction decreased from 15.3% to 5.7%, although 19A carriage decreases were more modest (8.3% to 6.3%) [98]. In the Aboriginal population in Australia, PCV7 was recommended from 2001 to 2009 (three doses, with a PPSV23 booster). A 3 + 1 PCV10 schedule was used in the Northern Territory only from October 2009 to October 2011, while all other states and territories used PCV7 until they switched to a 3 + 0 PCV13 series beginning in 2011 [114]. In a study comparing carriage in young children across the three periods (all children in the study received the respective vaccines in each period), serotypes 19A, 6A, and 6C were not reduced in the PCV10 period when compared with the PCV7 period, but they were significantly reduced after PCV13 introduction [89,100].

Shortly after the introduction of PCV13 to Lao PDR in late 2013, carriage of serotype 6A decreased both in unvaccinated infants (i.e. aged 5 to 8 weeks) and in toddlers, whereas 6C decreased only in toddlers and 19A carriage did not change significantly in either age group [115].

5. Reduced efficacy against VT carriage predicts lower impact on VT disease across ages

Given the studies reviewed above, a reduced immune response to a PCV may logically result in reduced efficacy against VT carriage, which in turn may lead to a lower ability to afford indirect protection. To date, no head-to-head comparative studies have been conducted between PCVs to validate this hypothesis. However, insights can be garnered by comparing post-PCV implementation data on IPD in all ages. If the above hypothesis is correct, in the presence of similar vaccination rates in infants and toddlers, the more immunogenic vaccine should have a greater impact on VT IPD through indirect protection in all ages.

Widespread PCV7 introduction was associated with a substantial reduction in diseases caused by VT serotypes, but a more complex picture was observed for the cross-protecting serotypes 6A and 19A. For example, PCV7 rapidly reduced serotype 6A disease in the vaccinated population, but disease reduction in the nonvaccinated population was incomplete, demonstrating a reduced impact on serotype 6A carriage [116,117]. In most countries using PCV7, a clear increase in serotype 19A carriage and disease was observed; serotype 19A was actually the most successful replacing serotype in all age groups [63,116–118]. Furthermore, shortly after PCV7 introduction, it was discovered that what was thought to be a pure serotype 6A actually consisted of two cross-reacting serotypes, 6A and 6C [119,120]. Although antibodies to serotype 6B were protective against serotype 6A IPD, and to a lesser extent carriage [116,117], antibodies to serotype 6B were not of sufficient concentration to protect against disease or carriage of serotype 6C [120]. This resulted in significant post-PCV7 increases in 6C carriage and disease in all ages [121–124].

For the cross-reacting serotype 19A, some cross-protection was described against IPD caused by this serotype in children fully vaccinated with PCV10 [118,125,126]; however, this was not shown for mucosal disease or carriage. In fact, while early post-PCV10 introduction studies in the Netherlands [127] and Finland [65] suggested a reduction of serotype 19A diseases in young children, longer follow-up showed that for both children and adults, serotype 19A IPD rates were higher after PCV10 introduction [126,128]. In contrast to PCV7 and PCV10, after implementation of PCV13, a rapid and marked reduction in disease from and carriage of serotypes 6A and 19A was observed, as well as a reduction in cross-reactive serotype 6C [38,90,97,129–139].

When attempting to compare PCV13 and PCV10 for the differences in the impact on serotypes 19A and 6A in all ages, one must include only countries for which data on IPD are thoroughly collected and vaccine uptake is high. Examining such studies clearly demonstrated the dichotomy between the two vaccines regarding their impact on serotypes 19A, 6A, and 6C IPD in all ages (Table 3) [104,118,126,127,134,136,140–151]. In Australia and Finland, for example, IPD caused by serotypes 19A and 6C increased after PCV10 introduction; however, PCV13 was associated with a rapid and marked decrease in IPD caused by these serotypes in multiple countries among all ages [104,134,136,142,143]. An exception is Ireland, where serotype 19A IPD remained unchanged in children <5 years of age and increased slightly in those ≥5 years of age after PCV13 implementation, the reason for which is unknown [144]. For serotype 6A, IPD was generally reduced with both vaccines in young children, but the impact on disease in the elderly was less apparent for PCV10 than for PCV13 [104,118,126,134,136,140–144].

Table 3. Impact of PCV10 and PCV13 against IPD caused by serotypes 19A, 6A, and 6C across all ages in countries with robust surveillance.

Country	Year Vaccine Introduced	Age Group	Cases or Incidence*
			Serotype 19A IPD (Year)	Serotype 6A IPD (Year)	Serotype 6C IPD (Year)
PCV10
Finland [126]	2010	<5 y	Cases (2009) 8 (2016) 15	Cases (2009) 4 (2016) 0	Cases (2009) 0 (2016) 1
		≥5 y	Cases (2009) 21 (2016) 151	Cases (2009) 22 (2016) 117	Cases (2009) 6 (2016) 44
Netherlands [127,146]	2011	<5 y	Incidence (2010–2011) <2 (2015–2016) <1.5 Cases (2009–2011) 7 (2013–2014) 0	Incidence (2010–2011) 0 (2015–2016) 0 Cases (2008–2011) 2 (2011–2014) 0	Incidence (2010–2011) 0 (2015–2016) 0 Cases (2008–2011) 0 (2011–2014) 1
		≥65 y	Incidence (2010–2011) <5 (2015–2016) 5 Cases (2009–2011) 63 (2013–2014) 33	Incidence (2010–2011) <1 (2015–2016) 0	Incidence (2010–2011) 0.5 (2015–2016) <1.5
Chile [140,141]	2011	<2 y	Cases (2007) 12 (5%) (2015) 15 (19%)	Cases (2007) 15 (6%) (2015) 3 (4%)	NR
		All ages	Cases (2007) 28 (3%) (2015) 78 (12%)	Cases (2007) 39 (4%) (2015) 13 (2%)	NR
Australia [142,143]	2009	<5 y	Cases (2009) 100 (2016) 20	Cases (2009) 5 (2016) 1	Cases (2009) 5 (2016) 1
		≥5 y	Cases (2009) 221 (2016) 105	Cases (2009) 22 (2016) 2	Cases (2009) 53 (2016) 50
New Zealand [118]	2011	<5 y	Cases (2011) 13 (2016) 15	Cases (2011) 1^† (2016) 0	Cases (2011) NR (2016) 1
		≥5 y	Cases (2011) 50 (2016) 63	Cases (2011) 23^† (2016) 1	Cases (2011) NR (2016) 17
Brazil [149]	2010	<5 y	Cases (2005–2009) 44 (2011–2015) 126 (2016–2017) 92	NR	NR
		5–49 y	Cases (2005–2009) 43 (2011–2015) 104 (2016–2017) 87	NR	NR
		≥50 y	Cases (2005–2009) 17 (2011–2015) 87 (2016–2017) 73	NR	NR
Austria [148]	2012	<5 y	IRR (2009–2011) 0.42 (2013–2014) 0.73 (2015–2016) 1.87	IRR (2009–2011) 0.21 (2013–2014) 0.98 (2015–2016) NR	NR
		≥50 y	IRR (2009–2011) 0.34 (2013–2014) 1.17 (2015–2016) 1.63	IRR (2009–2011) 0.26 (2013–2014) 0.63 (2015–2016) 0.55	NR
Colombia [151]	2011^‡	<12 mo	Isolates (2009) 2 (2015) 10	Isolates (2009) 3^† (2015) 4	Isolates (2009) NR (2015) 2
		12–23 mo	Isolates (2009) 0 (2015) 5	Isolates (2009) 0^† (2015) 1	Isolates (2009) NR (2015) 0
		24–59 mo	Isolates (2009) 3 (2015) 16	Isolates (2009) 1^† (2015) 2	Isolates (2009) NR (2015) 1
PCV13
United States [134]	2010	<5 y	Cases (2009–2010) 205 (2012–2013) 13	Cases (2009–2010) 1 (2012–2013) 0	Cases (2009–2010) 8 (2012–2013) 3
		5–17 y	Cases (2009–2010) 19 (2012–2013) 7	Cases (2009–2010) 2 (2012–2013) 1	Cases (2009–2010) 7 (2012–2013) 3
		≥18 y	Cases (2009–2010) 473 (2012–2013) 161	Cases (2009–2010) 44 (2012–2013) 6	Cases (2009–2010) 185 (2012–2013) 152
Israel [145]^§	2010	<5 y	Cases (2009–2010) 33 (2014–2016) 1	Cases (2009–2010) 15 (2014–2016) 2	NR
		≥18 y	Incidence (2009–2010) 0.65 (2014–2015) 0.50 Cases (2009–2010) 31 (2014–2015) 20	Incidence (2009–2010) 0.35 (2014–2015) 0.05 Cases (2009–2010) 16 (2014–2015) 2	Incidence (2009–2010) 0.10 (2014–2015) 0.20
United Kingdom^‖	2010	<5 y	Cases (2009) 93 (2014) 14	Cases (2009) 2 (2014) 0	NR
		≥5 y	Cases (2009) 511 (2014) 268	Cases (2009) 96 (2014) 16	NR
New Zealand [118]	2014	<5 y	Incidence (2014) 5.5 (2016) NA Cases (2014) 17 (2016) 15	Incidence NA Cases (2014) NR for single serotype (2016) 0	Incidence NA Cases (2014) NR for single serotype (2016) 1
		≥5 y	Incidence (2014) 5–64 y, 1.1 ≥65 y, 4.6 (2016) NA Cases (2014) 70 (2016) 63	Incidence NA Cases (2014) NR for single serotype (2016) 1	Incidence NA Cases (2014) NR for single serotype (2016) 17
Sweden [104]	2010	All ages	Incidence (2009–2010 vs 2011–2014) RR = 0.67; 95% CI: 0.44, 1.04	Incidence (2009–2010 vs 2011–2014) RR = 0.66; 95% CI: 0.34, 1.29	Incidence (2009–2010 vs 2011–2014) RR = 0.85; 95% CI: 0.44, 1.63
Australia [142,143]	2011	<5 y	Cases (2011) 106 (2016) 20	Cases (2011) 1 (2016) 1	Cases (2011) 8 (2016) 1
		≥5 y	Cases (2011) 273 (2016) 105	Cases (2011) 8 (2016) 2	Cases (2011) 91 (2016) 50
Norway [136]	2011	<5 y	Isolates (2009) 5 (2010) 7 (2011) 10 (2012) 0	Isolates (2009) 0 (2010) 1 (2011) 0 (2012) 0	Isolates (2009) 0 (2010) 2 (2011) 3 (2012) 1
		≥5 y	Isolates (2009) 50 (2010) 72 (2011) 81 (2012) 53	Isolates (2009) 29 (2010) 13 (2011) 10 (2012) 6	Isolates (2009) 7 (2010) 30 (2011) 38 (2012) 24
Ireland [144]	2010	<5 y	Isolates (2009) 3 (2016) 3	Isolates (2009) NR (2016) 0	Isolates (2009) NR (2016) 0
		≥5 y	Isolates (2009) 18 (2016) 26	Isolates (2009) NR (2016) 0	Isolates (2009) NR (2016) 6
South Africa [147]	2011	<2 y	Incidence (2005–2008) 4.5 (2011) 3.1 (2012) 1.3 Isolates (2005–2008) 94 (2011) 67 (2012) 29	Incidence (2005–2008) 6.3 (2011) 2.4 (2012) 0.9 Isolates (2005–2008) 131 (2011) 52 (2012) 20	NR
		25–44 y	Incidence (2005–2008) 1.2 (2011) 1.2 (2012) 0.8 Isolates (2005–2008) 166 (2011) 184 (2012) 128	Incidence (2005–2008) 0.8 (2011) 0.7 (2012) 0.4 Isolates (2005–2008) 110 (2011) 117 (2012) 66	NR
Taiwan [150]	2011	≤5 y	Incidence (2008–2009) 2.19 (2010–2012) 9.01 (2013–2017) 2.77	Incidence (2008–2009) 0.68 (2010–2012) 1.02 (2013–2017) 0.21	NR

IPD = invasive pneumococcal disease; IRR = incidence rate ratio; PCV = pneumococcal conjugate vaccine; PCV10 = 10-valent PCV; PCV13 = 13-valent PCV; NA = not available; NR = not reported.

*Incidence reported per 100,000 population.

^†6A and 6C not distinguished for analysis.

^‡As reported in Leal et al. [152].

^§R. Dagan, The Pediatric Infectious Disease Unit, Soroka University Medical Center, Beer-Sheva, Israel, personal communication, 4/9/2018.

^‖E. Miller, Immunisation Department Health Protection Agency, Center for Infections, London, UK, personal communication, 4/18/2016.

The differential impact of PCV10 and PCV13 on serotype 19A IPD in six countries is illustrated in Figure 5 [118,125,126,134]. The figure clearly shows a sharp decline in IPD caused by serotype 19A both in children <5 years of age (mainly by direct protection through vaccination) and in those >5 years of age (primarily elderly individuals ≥65 years of age by indirect protection) in the three countries using PCV13 (Israel, United Kingdom, and United States). In contrast, in the three countries using PCV10, IPD caused by serotype 19A increased in both children and the elderly, with a sharper and more dynamic increase in the elderly resulting from the increase in carriage of serotype 19A in vaccinated children. A similar, although less dramatic, pattern was observed with serotype 6A (Supplementary Figure 1) [125,126,134], which was reduced in young children but less clearly in those >5 years of age (primarily adults).

Figure 5. Differential impact on serotype 19A IPD post-PCV10 vs post-PCV13 implementation in six countries: three countries post-PCV13 introduction (Israel, United States, United Kingdom) and three countries post-PCV10 introduction (Chile, New Zealand, Finland).

(a) Number of cases of serotype 19A IPD in children <5 years of age between 2004 and 2016. (b) Number of cases of serotype 19A IPD in individuals ≥5 years of age between 2004 and 2016. Vertical dashed lines denote the implementation year of PCV7, PCV10, or PCV13. Vaccination with PCV10 was conducted in Chile and Finland per 3 + 1 and 2 + 1 schedules, respectively. Data are adapted from (a) Israel National Surveillance. Personal communication to Ron Dagan; (b) Moore MR, et al. Lancet Infect Dis 2015 [134]; (c) S. Ladhani, Pediatric Infectious Disease, Public Health England, London, UK, personal communication, 12/30/2017; (d) Pan American Health Organization. SIREVA II (Sistema de Redes de Vigilancia de los Agentes Responsables do Neumonias y Meningitis Bacterianas) [125]; (e) Public Health Surveillance. Information for New Zealand Public Health Action. Invasive Pneumococcal Disease Reports [118]; and (f) National Institute for Health and Welfare. Incidence of Invasive Pneumococcal Disease in Finland [126]. *Number of cases (scale) per country differs. ^†Threshold for Chile is <2 years and ≥2 years of age. ^§Surveillance years may span January–December or July–June. IPD = invasive pneumococcal disease; PCV = pneumococcal conjugate vaccine; PCV7 = 7-valent PCV; PCV10 = 10-valent PCV; PCV13 = 13-valent PCV.

As noted above, there are substantial differences between PCV10 and PCV13 in their impact on the carriage of serotypes 19A, 6A, and 6C in vaccinated children, and thus on indirect protection. These observations can be used as the proof-of-concept for the direct link between PCV-induced antibody concentration, carriage, and overall community protection against pneumococcal diseases. The overall disease caused by these three serotypes is sizeable, especially in countries using PCV10, but much of the IPD in all ages is still caused mainly by NVTs that increase after PCV introduction [126,131]. In a phenomenon known as serotype replacement, VT serotype carriage decreases and NVT carriage increases after PCV implementation [8,9]. Indeed, IPD caused by NVT increased post-PCV implementation, especially in the elderly [126,134]. Currently, more than 6 years after PCV13 and PCV10 implementation, the disease and carriage replacement by NVT has not yet reached steady-state [102,141,145]. Therefore, it is possible that overall IPD rates in the frailest elderly and those at highest risk for IPD may not reflect the benefits derived from the infant vaccination programs until a steady-state is achieved.

6. Conclusion

This review provides links between the degree of immune responses to PCVs, acquisition of carriage, and resulting reductions in vaccine serotype carriage and circulation in the community. The differential impact of PCV10 versus PCV13 on serotype 19A, 6A, and 6C carriage can serve as the proof-of-concept for the relationship between immune response to PCVs and impact on public health. Considering the overall dynamics of pneumococcal serotypes with other microbiota, antibiotics in the community, and environmental conditions, judging the net effect of PCVs on serotype dynamics is difficult. Vigilant and long-term follow-up is needed before firm conclusions can be made regarding the overall impact of PCVs in the community, especially for the frailest individuals.

7. Expert opinion

S. pneumoniae causes substantial morbidity and mortality globally, leading to nearly two million deaths annually and up to one million deaths among those aged <5 years [1–3]. Nasopharyngeal carriage is thought to be a prerequisite for clinically symptomatic pneumococcal infections but is also responsible for the circulation of the pathogens in the community [153]. Young children are the main carrier of vaccine-serotype pneumococci (the serotypes that are most commonly responsible for pneumococcal disease), and thus, are the main source of dissemination of these serotypes in the community [154]. Although the highest incidence of pneumococcal disease is in children <5 years (mainly <2 years of age) [16,153], this age group constitutes only a small cohort; however, by spreading S. pneumoniae in the community, this age group is responsible for high rates of pneumococcal disease in other age cohorts [155] and in the immunocompromised [156].

The introduction of PCV7 (the first licensed PCV) was received with great enthusiasm since it not only protected against IPD caused by the 7 serotypes in the vaccine but also against mucosal disease [79]. Most importantly, PCV7 reduced carriage and spread of the VT7 serotypes, leading to increased protection of all ages [16]. However, at the same time, it rapidly became clear that two main limitations were associated with PCV7 that reduced long-term expectations for the vaccine. The first was the limited number of serotypes in the vaccine (only 7 of the existing >90). The second was the serotype replacement phenomenon, where non-PCV7 serotypes rapidly occupied the nasopharyngeal niche after the seven vaccine serotypes were reduced [8]; this was also associated with an increase in diseases caused by the non-VT serotypes, especially among the immunocompromised [9,156]. These phenomena resulted in extensive investment in research in an attempt to expand the valency of PCVs. However, to date, close to 20 years past implementation of PCV7, the two currently licensed vaccines are still limited to 10 and 13 serotypes (PCV10 and PCV13, respectively).

In this regard, the potential effect of these vaccines on serotypes closely resembling but not identical to those in the vaccines (the so-called ‘cross-reacting serotypes’) generated great interest. By far, the two most common vaccine-related serotypes both in carriage and disease were serotypes 6A and 19A [157,158]. It was clear that cross-protection against serotypes 6A and 19A (which are not constituents of PCV7) can be stimulated by serotypes 6B and 19F (which are included in PCV7), respectively. However, it was not clear whether this cross-protection was sufficient to result in the reduction of carriage and disease caused by serotypes 6A and 19A.

The data discussed in the current review clearly demonstrate that the lack of serotype 19A as an integral antigen in PCV formulations resulted in an insufficient impact on the carriage of 19A, leading to serotype 19A replacement in community-wide carriage and disease. The lack of serotype 6A in PCV formulations led to ambiguous results after PCV7 or PCV10 use because 6A carriage was often but not universally reduced [42,79,83,87]. A recent discovery revealed the emergence of serotype 6C [119], a new serotype that could be cross-protected by serotype 6A but not serotype 6B. Such cross-protection by serotype 6A is possible only if serotype 6A is included in the vaccine [42,120]. Thus, an inherent difference became clear between PCV10, which does not contain serotype 6A, and PCV13, which does contain serotype 6A. Specifically, PCV13 implementation reduced carriage and disease caused by serotypes 19A and 6C in the community in all ages, while PCV10 implementation resulted in an increase in the prevalence of these serotypes. For serotype 6A, the differences are more subtle.

To date, the factors responsible for reduced carriage acquisition of vaccine serotypes afforded by PCVs are not fully understood, but it is clear that the intensity of the immune response (as measured by serotype-specific IgG and memory B cell responses) plays a major role [42,44,69,70,81,82]. Thus, a vaccine that contains more serotype antigens represents one side of the equation for success in reducing diseases in the entire community; on the other side is vaccine immunogenicity as related to serotype-specific memory B cells and antibody stimulation.

Efforts to expand the serotype range in the vaccine are ongoing, as currently extended serotype PCVs are being developing (from PCV15 [159] to PCV20 [160]). Inclusion of additional serotypes will aid further reduction of disease in the community, if the additional serotypes behave similarly to those in the current PCVs. However, the vaccines under development include additional serotypes that are all conjugated to a single carrier (mainly CRM₁₉₇) [159]. Carrier protein overload with CRM₁₉₇, which is antigenically related to diphtheria toxoid, presents the risk of interference, especially through the mechanism of carrier-induced epitope suppression [161]. Thus, in theory, the inclusion of more serotypes in a conjugate vaccine may result in lower antibody responses for some serotypes, resulting in reduced protection against carriage. Therefore, studies of vaccines under development must specifically investigate serotype-specific immune responses and their effect on the acquisition of vaccine serotypes. Such analyses will ensure continued community-wide reduction in disease.

Continued surveillance after the implementation of current and future vaccines is crucial to ensure the best protection of the community from pneumococcal diseases.

8. Five-year view

Infant pneumococcal vaccination programs directly protect vaccinated subjects against IPD caused by VTs and confer indirect protection to others through the decreased likelihood of transmission of VT pneumococci. Increased prevalence of IPD caused by NVTs after vaccine introduction places vulnerable populations, such as the elderly, at risk for IPD caused by NVTs due to serotype replacement in younger individuals who are carriers. As higher-valency PCVs are developed and introduced, carriage of all serotypes must be monitored. Similar to the observations regarding PCV13, the greatest protection against carriage for a given serotype is likely to be afforded by a vaccine containing a higher number of the most important serotypes globally. Such a vaccine must also contain the common serotypes included in any older vaccine formulation to maintain protection against the broadest number of serotypes.

Article highlights

• Pneumococcal conjugate vaccines (PCVs) protect against invasive pneumococcal disease (IPD), both directly through the acquisition of antibodies and decreased nasopharyngeal carriage and indirectly through decreased person-to-person transmission.

• The efficacy of PCVs in reducing acquisition of vaccine-serotype carriage depends on the innate response to PCVs, namely the serotype-specific immunoglobulin G (IgG) response and the serotype-specific memory B-cell response.

• The immune responses elicited by cross-reacting serotypes may be insufficient to decrease carriage of vaccine-related serotypes (i.e. serotypes 6A and 6C by serotype 6B in the vaccine, and serotype 19A by serotype 19F in the vaccine).

• Considerable differences between PCV10 and PCV13 in the carriage of serotypes 19A and 6C, and to a lesser extent serotype 6A, have been seen in vaccinated children, resulting in differences in indirect protection.

• Incidence of IPD caused by nonvaccine serotypes has increased after PCV implementation, particularly in the elderly, necessitating extended-spectrum PCVs—especially those that maintain robust immune response to each serotype in these vaccines.

• Assessing the impact of infant PCV programs on nasopharyngeal carriage requires further long-term follow-up.

Declaration of interest

R Dagan has received research grants and consulting and speaker fees from Pfizer; grant and consulting fees from MSD; and consulting fees from MeMed. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose

Acknowledgments

This work was supported by Pfizer Inc. The sponsor had no involvement in the study design; collection, analysis, and interpretation of the data; or the writing of the report. R Dagan confirms he had full access to all data and has the final responsibility for the decision to submit the manuscript for publication. Editorial/medical writing support was provided by Nicole Gudleski O’Regan, PhD, and Jill E. Kolesar, PhD, at Complete Healthcare Communications, LLC (North Wales, PA), a CHC Group company, and was funded by Pfizer Inc.

Supplemental material

Supplemental data for this article can be accessed here.

Additional information

Funding

This work was supported by Pfizer Inc. The sponsor had no involvement in the study design; collection, analysis, and interpretation of the data; or the writing of the report.