Invasive Pneumococcal Disease After 2 Decades of Pneumococcal Conjugate Vaccine Use Free

A population-based statewide surveillance program for invasive Streptococcus pneumoniae infections in children has been in place in Massachusetts, United States, since October 2001.^3,18  For this current study, cases identified during January 1, 2002, to December 31, 2021, covering a 2-decade period with PCV use were included. Isolates of S. pneumoniae from normally sterile body sites including blood, pleural fluid, and cerebrospinal fluid collected from Massachusetts residents aged <18 years are submitted to the Massachusetts Department of Public Health. Next, epidemiologists collect demographic and clinical information, including previous receipt of PCV, via a subsequent phone interview with the parents/guardians and/or primary care providers using a standardized case report form. Bacteremia is defined as cases of pneumococcal blood stream infections without focal infection, pneumonia is defined as cases with bacteremia and clinically or radiographically identified pneumonia or empyema with a culture positive pleural fluid, and meningitis is defined as cases clinically diagnosed with central nervous system infection with culture-positive cerebrospinal fluid. Pneumococcal isolates are then transferred to the Maxwell Finland Laboratory for Infectious Diseases at Boston University Medical Center for further analysis.^2,18  We defined protective vaccination status as having received 2 or more doses for children <12 months of age, or 3 doses in the first year of life or 1 dose after the age of 12 months in children aged >12 months, ≥14 days before the diagnosis of IPD. Breakthrough infection was defined as IPD in a child with sufficient doses of PCV to have achieved protective vaccination status.

Isolates were confirmed as S. pneumoniae by using standard microbiologic methods according to guidelines from the Clinical and Laboratory Standards Institute, including optochin sensitivity (5 mm inhibition) and bile solubility.¹⁹  We included different β-lactam antibiotics such as penicillin, amoxicillin, ceftriaxone, and erythromycin as representative macrolides, and levofloxacin to represent fluoroquinolones in our antibiotic panel. Penicillin intermediate resistance was defined by minimum inhibitory concentration (MIC) values between 0.12 and 1.0 mg/mL, and high-level resistance was defined as MIC values ≥2.0 mg/mL. Multidrug resistance was defined as resistance to ≥3 antimicrobial classes. We used Quellung reaction and pneumococcal antisera (Statens Serum Institute, Copenhagen, Denmark) for serotyping.²⁰  Isolates found to be unencapsulated were confirmed to be S. pneumoniae by lytA gene amplification using real-time polymerase chain reaction and defined as nontypeable.

We used results from respiratory pathogen panel multiplex polymerase chain reaction tests (Biofire Diagnostics, Salt Lake City, Utah, United States) collected from children with respiratory symptoms at Boston Medical Center to track the presence of seasonal respiratory viruses (respiratory syncytial virus [RSV], human metapneumovirus [HMPV], parainfluenza viruses, influenza virus, and rhinovirus) in our community between January 2020 and December 2021.²⁰ 

We calculated annual IPD incidence rates by dividing the number of IPD cases by the population of Massachusetts residents <18 years of age defined as the US Census Bureau estimates for each calendar year,²¹  and 95% confidence intervals (CIs) around estimates were calculated using Wilson method with an α of .05. Averted IPD cases in post-PCV13 era was calculated as the difference between the cases expected without PCV13 based on 2009 observed cases and the cases observed in PCV13 era between 2011 and 2021. Statistical analyses were conducted in SAS version 9.3.1 software (SAS Institute, Gary, Indiana). We used χ² and t tests or analysis of variance to compare proportions and continuous variables as appropriate, and the Cochran–Mantel–Haenszel test of trend to evaluate changes over time. We used 2-sided P values <.05 for statistical significance. Approval for this study was granted by the institutional review boards of both Boston University Medical Center and Massachusetts Department of Public Health.

We identified 1347 IPD cases among Massachusetts children between January 2002 and December 2021 (Table 1). After implementation of PCV7, incidence of IPD in children aged <18 years declined 72% between 2002 and 2021 (incidence rate ratio 0.28, 95% CI 0.18–0.45). The incidence of IPD in children aged <2 years declined from 33.5 per 100 000 (95% CI 24.5–42.4 per 100 000) in 2002 to 8.6 per 100 000 (95% CI 3.9–13.3 per 100 000) in 2021 achieving a nearly 75% reduction after 2 decades of PCV immunization in successive birth cohorts (Fig 1A) (Table 2). The reduction in IPD rates continued after replacement of PCV7 with PCV13 with a decline from 6.7 per 100 000 (95% CI 5.3–8.0 per 100 000) in 2010 to 1.7 per 100 000 (95% CI 1.0–2.3 per 100 000) in 2021 in children aged <18 years reflecting a 74% decline and ∼541 cases averted since 2010.

During the coronavirus disease 2019 pandemic years, 2020 to 2021, the rate of IPD among children aged <18 years reached 1.6 per 100 000, the lowest incidence observed, over the 20-year study period. Additionally, the usual seasonal pattern was disrupted and an unexpected paucity of IPD cases were observed during the 14 months between May 2020 and June 2021 (Fig 1B). This represents a period when severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) surged in Massachusetts and seasonal respiratory viruses (RSV, influenza virus, HMPV, parainfluenza virus) were virtually absent in the community as documented by their disappearance from complete respiratory panels processed from children with respiratory symptoms at Boston Medical Center (Fig 1C).

Demographic and clinical characteristics of IPD cases are presented in Table 1. One-fourth of all cases (313, 25.9%) reported at least 1 underlying clinical condition considered at risk/high risk for pneumococcal infections. Children with an underlying comorbidity were, on average, 10 months older at the time of IPD diagnosis compared with the cases seen among otherwise healthy children (mean age 55.2, 95% CI 49.1–51.3 months versus 45.8, 95% CI 42.6–46.8 months). Older children diagnosed with IPD were more likely to have an underlying risk factor for pneumococcal infections (P = .004) and nearly one-third of the cases in children aged >5 years had at least 1 underlying condition (98 cases, 30.3%). The most common conditions included neuromuscular disorders/seizures (14.7%), asthma (13.4%), prematurity (11.8%), hematologic malignancy (10.7%), organ transplant (9.0%), sickle cell disease (8.0%), and congenital heart disease (7.7%).

After the introduction of PCV13 in 2010, the proportion of pneumonia cases among all IPD cases (26.1% vs 33.7%, P < .05) declined compared with pre-PCV13 era (Table 1). Over the 2 decades of the study, we identified 35 deaths with IPD (2.8%), and 22 (62.9%) children who died met the definition for protective vaccine status. Case fatality rate (CFR) increased from 2% in the pre-PCV13 period to 4.6% (P < .008) in the post-PCV13 era. Eleven (35.5%) of these cases had at least 1 underlying comorbidity defined as a risk factor for IPD (2 asthma, 2 prematurity, 1 sickle cell disease, 1 congenital heart disease, 5 hematologic disease on steroid/malignancy/transplant). Among children with underlying conditions, CFR was 3.8% and was similar to otherwise healthy children (2.4%, P > .05).

Serotype information was available for 1062 (78.9%) cases. Serotypes 19A (268, 25.2%), 7F (118, 11.1%), 22F (72, 6.8%), 15BC (70, 6.6%), 3 (61, 5.7%), and 35B (60, 5.6%) caused 61.1% of all cases (Fig 2). The 6 additional serotypes that are included in PCV13 were responsible for 57.6% (411) of all IPD cases with serotype information before implementation of PCV13 in Massachusetts (Fig 3A). Serotypes 15B/C (39, 10.8%), 3 (31, 8.6%), 33F (29, 8.0%), 23B (21, 5.8%), and 35B (17, 4.7%) were responsible for 37.8% of all cases diagnosed in post-PCV13 era (Fig 3B). The 2 additional serotypes included in PCV15 (ie, 22F and 33F) caused 16.9% (61) and the 5 additional serotypes included in PCV20 (ie, 8, 10A, 11A, 12F, 15BC) caused an additional 18.1% (69) of remaining IPD cases diagnosed between 2011 and 2021.

After PCV13 implementation, penicillin nonsusceptibility continued to decline (9.8% vs 5.3% in pre- versus post-PCV13 era, respectively, P = .003), however was more common in serotypes not included in PCV13 (14.8% vs 1.4% in non-PCV13 serotypes versus PCV13 serotypes, respectively, P < .001). Serotypes 9N, 15A, and 35B had higher MIC values (1.86 ug/mL, 95% CI 1.26–2.45) compared with the serotypes included in PCV13 (0.89 ug/mL, 95% CI 0.71–1.07). A total of 124 of 1006 (12.3%) isolates were resistant to a β-lactam and at least 1 other class of antibiotics (13.1% in 2002–2010 vs 10.7% in 2011–2021, P = .28), and 94 of 1006 (6.3%) were multidrug resistant (2 additional antibiotic classes).

The reduction in IPD observed in Massachusetts children after replacement of PCV7 with PCV13 has been impressive, with a decline from 6.7 to 1.7 (72%) per 100 000 in children aged <18 years, resulting in an estimated 541 cases averted since 2010. Reductions in IPD were achieved in all age groups. The rate of decline was greatest in the first 5 years after introduction of PCV13 and more modest since 2017. We observed a modest increase in age and an increase in proportion of disease manifesting without a focus. One potential explanation is that serotype 3 is now a greater proportion of IPD disease burden and was reported to cause IPD more often in children aged >2 years in PCV13-immunized children.¹¹  A second speculation is that the emerging non-VSTs are more likely to cause sepsis than focal disease; however, the biologic explanation remains to be defined. CFRs increased from 2% in the pre-PCV13 period to 4.6% (P < .008) in the post PCV13 era. A greater percentage of children who died had comorbid conditions; however, CFR among individuals with comorbidity was not statistically significantly higher. Similar to reports from other geographical locations, the incidence of pneumococcal disease reached a nadir during the SARS-CoV-2 pandemic.^22,23  As previously reported, our surveillance studies suggest the absence of seasonal respiratory viruses in part explains the paucity of IPD cases in 2020 to 2021.^22,24 

In the 2 years before introduction of PCV13, serotypes 3, 7F, and 19A were isolated in nearly 60% of IPD cases. After PCV13 introduction, both 7F and 19A cases have dramatically declined; however, serotype 3 continues to cause a large proportion of breakthrough infections, suggesting the need for further research addressing immune protection for serotype 3. A prospective cohort study in Israel reported a 67% decline in IPD cases among children post-PCV13 era compared with the pre-PCV13 period, with VSTs 3, 14, and 19A continuing to cause breakthrough infections.²⁵  Serotype 3 and a related clonal complex (CC), CC180, was reported to increase after the introduction of the PCV13 vaccine in Spain.²⁶  Several studies suggest modest direct protection was observed after immunization with PCV13 against IPD because of serotype 3, but no reduction in colonization and persistent circulation in the community.^27,28 

In addition to breakthrough infections with VSTs, our study identified the increased circulation of non-PCV13 serotypes over the past 2 decades. Of importance, we observed disease because of serotypes 15A, 15BC, 23B, and 35B, which are not included in PCV13 (Fig 3b), suggesting potential serotype replacement. This increase in non-PCV13 serotypes suggests the importance of considering these serotypes in future vaccine formulations and continued surveillance of emerging serotypes. Next-generation PCV15 includes serotypes 22F and 33F, and PCV20 includes serotypes 8, 10A, 11A, 12F, and 15B, which in theory will extend coverage by 20% and 35%, respectively, on the basis of current serotype distribution in our study. Other studies have identified an increase in non-PCV13 serotypes circulating, as well.^29,30  A study in the United Kingdom identified that serotypes 8, 9N, 10A, 12F, 15B/C, 23B, and 24F, with 8 and 12F were causing the greatest increase.³¹  Some of these serotypes (eg, serotype 24F) are reported to have multiple CCs where only serotype strains of a specific CC are associated with invasive disease. Similarly, the same virulent CC may be found with multiple non-VSTs, enabling such combinations to be invasive³²  and reducing the overall decline in IPD in the community. In addition, specific CCs known to be linked to multidrug-resistant pneumococcal lineages, such as CC230 (expressed by serotype 14 and 19A), are now being reported in non-PCV13 serotypes (eg, 24F) that have been reported to cause increased IPD in Europe and have been identified in outbreaks affecting young children.³³  Although we did not observe an increase in serotypes 8, 12F, or 24F in Massachusetts, these more virulent CCs are circulating in other geographic regions and require ongoing surveillance for potential emergence in the United States.

Before introduction of PCV7, penicillin resistance clustered in a limited number of pneumococcal serotypes, mostly VST. As reported by others, β-lactam resistance is now emerging in non-VSTs.^34,35  High consumption of penicillin and/or macrolides has been associated with the increase in multidrug-resistant CCs in the emerging non-PCV13 serotypes.^36,37  We identified pneumococci serotypes 15A, 35B, and 9N with mean MICs of 1.8 ug/mL (minimum 0.02, maximum 16.0). The greater prevalence of penicillin resistance in non-PCV13 serotypes suggests the importance in targeting these specific strains in next-generation vaccines. In our study, we saw intermediate to high-level nonsusceptibility among serotypes 9V, 14, 19F, 19A, 6B, 35B, 15A, 6A, 7F, and 9N (Fig 4). Antibiotic-selective pressure will vary with the type and amount of antibiotic consumed in each population and continuous monitoring of circulating serotypes, and susceptibility changes are necessary to identify the trends in each setting.

Through this study, we observed that one-fourth of IPD cases involved children with comorbid conditions both pre- and post-PCV13 implementation (Table 1). This suggests that those with underlying conditions remain at increased risk of pneumococcal disease, including non-VSTs.³⁸  Further research should go into interventions to protecting this population. In a systematic review of vaccine failure, it was observed that about half of the cases of IPD among children within the study had a comorbid condition.³⁹  This highlights the importance of continued surveillance of vaccine failure to understand who may escape protection from vaccine, as well as which serotypes are most likely to cause vaccine failure in these populations. The most common comorbid conditions observed were asthma, prematurity, and neuromuscular disorders. It is important to recognize that these comorbidities, which are prevalent in the pediatric population, contribute more to the burden of pneumococcal disease than the relatively uncommon high-risk conditions such those associated with immune deficiency. In a recent systemic review, 26% of the primary IPD cases seen in children aged >2 years were found to have an underlying primary immunodeficiency, with higher risk among patients with meningitis or complicated pneumonia.⁴⁰  Chronic inflammation has been hypothesized as the cause for increased susceptibility to pneumococcal disease in children with comorbidities such as asthma, but the precise mechanisms have not been defined. In addition to PCV13/PCV15, pneumococcal polysaccharide vaccine 23 is currently recommended for children with at-risk/high-risk conditions; however, coverage rates vary from 20% to 72% between different countries, and experience suggests only a small proportion of children with at-risk or high-risk profiles have received pneumococcal polysaccharide vaccine 23.^41–44 

Between May 2020 and June 2021, we observed a significant decrease in pediatric IPD coinciding with the presence of SARS-CoV-2 in Massachusetts, whereas seasonal respiratory viruses such as RSV, influenza, HMPV, and parainfluenza virus were absent. This trend was also observed in a prospective cohort study in Israel, where, from October 2020 to February 2021, IPD rates were lower compared with what would normally have been expected, as well as a decrease in other respiratory infections.²²  A similar trend was reported from Hong Kong.⁴⁵  It is likely that school closures contribute to the decrease in viral circulation, but the observations of decreased respiratory virus were reported globally, including in the United Kingdom where school closure was limited to the early phase of the pandemic. Masking may also contribute, but again, there was widespread reliance on cloth masks, which subsequently were reported to be of limited value. Viral competition/interference has also been reported for influenza viruses and recently identified that influenza and RSV competes with SARS-CoV-2.⁴⁶  How much each of these contribute is not known. Studies revealing no substantial change in pneumococcal colonization during the pandemic add to the evidence that the paucity of seasonal virus circulation was a major contributor to the decline in IPD.^22,24 

The limitations of this study relate to the reliance on surveillance data, which have the potential to underreport cases. However, IPD is reportable in Massachusetts for children aged <18. It is based on microbiology laboratories reporting and sending isolates to Massachusetts Department of Public Health; therefore, we believe high rate of compliance are likely. The same surveillance system for laboratory-confirmed IPD cases followed by a phone interview with the reporting providers using a standard case report form has been in place for >2 decades in our state. We do not have access to medical records, including vaccination history, and we rely on the reported information regarding comorbidities and are unable to capture the severity of comorbidities or other potentially modifying factors, which can lead to misclassification. Additionally, our analysis does not account for immigration in and out of the state, and our results may not be generalizable to other populations with varying vaccination coverage. Lastly, we included the 3-dose primary series without a booster and 1-dose regimen after 12 months of age as protective; however, the durability of such regimens is likely different than 2 + 1 and 3 + 1 regimens.

In conclusion, our study provides further evidence for the substantial benefits and effectiveness of the pneumococcal vaccines against the disease caused by the serotypes included in the vaccine. However, our findings also highlight the importance of ongoing surveillance in being able to develop better vaccines targeting circulating strains under vaccine-selective and antibiotic-selective pressures. We identified emerging non-PCV13 serotypes in the Massachusetts, community as well as antibiotic resistant serotypes.

CC

clonal complex

CFR

case fatality rate

CI

confidence interval

HMPV

human metapneumovirus

IPD

invasive pneumococcal disease

MIC

minimum inhibitory concentration

PCV

pneumococcal conjugated vaccine

PCV7

sequential 7-valent pneumococcal conjugate vaccine

PCV13

sequential 13-valent pneumococcal conjugate vaccine

RSV

respiratory syncytial virus

S. pneumoniae

Streptococcus pneumoniae

SARS-CoV-2

severe acute respiratory syndrome coronavirus 2

VST

vaccine serotype