As part of a comprehensive systematic literature review on PCVs and viral respiratory infections, registered in the PROSPERO database (registration number: CRD42022339625), we identified four clinical studies investigating the efficacy or effectiveness of PCV against HCoVs or COVID-19 (Table 1). No other relevant studies were identified.

Table 1 Summary of randomized controlled trials and observational studies investigating pneumococcal conjugate vaccine efficacy or effectiveness against coronavirusesClinical trials assessing PCV efficacy against HCoVs

An RCT in South African infants demonstrated that an experimental PCV9 prevented 31% (95% confidence intervals[CI] 15 to 33%) of viral-associated pneumonia among hospitalized children [3]. A post-hoc analysis of this trial reported that PCV9 reduced pneumonia associated with HCoV in children < 2 years old (vaccine efficacy [VE], 33.9%; 95%CI 2.0 to 55.4%), especially in HIV-uninfected (VE, 64.0%; 95%CI 22.9 to 83.2) (Table 1), but not in HIV-infected children (VE, 13.6%; 95%CI -37.8 to 45.8) [6]. Protection against individual HCoVs in HIV-uninfected children varied, with high efficacy against HCoV-OC43 and HCoV-HKU1, but no protection for CoV-NL63 (Table 1). An exploratory analysis of an RCT in older Dutch adults using conservative 99.3% CIs reported that PCV13 did not significantly reduce HCoV-associated CAP (Table 1) [7]. For HCoV-associated suspected pneumonia, the P value was < 0.05, however 95%CIs were wide, spanning zero.

Observational studies evaluating PCV13 effectiveness against HCoVs and COVID-19

In a case–control study, Lewnard et al. evaluated PCV13 effectiveness against virus-associated LRTIs among adults in California, United States during the pre-COVID-19 pandemic period [4]. Matched, adjusted effectiveness against HCoV-associated LRTI was evaluated for several outcomes (Table 1). Significant protection was observed against LRTI or pneumonia when HCoV was detected. For outcome subcategories, all effectiveness point estimates were consistent with protection, but some 95% CIs overlapped zero.

Lewnard and colleagues also investigated PCV13 effectiveness against COVID-19 outcomes in a cohort of older adults in California during 2020 [8]. To reduce potential confounding, analyses included adjustments for multiple covariates and accounted for zoster vaccination as a negative-control exposure. PCV13 displayed significant effectiveness against COVID-19 diagnosis, hospitalization, and mortality (Table 1). In contrast, the pneumococcal polysaccharide vaccine PPV23 did not display protection against COVID-19, suggesting that reduced risk of COVID-19 in PCV13-vaccinated adults might be related to pneumococcal carriage, since, unlike PCV13, PPV23 does not affect pneumococcal carriage [8]. PCV13 effectiveness was abrogated by recent prior antibiotic usage, supporting a role of pneumococcus in the causal chain [8].

Potential explanations for clinical study findings

In both children and adults, the limited available evidence raises the possibility that PCVs reduce the risk of HCoV-associated respiratory disease. These results indicate that pneumococci might play a role in the development or severity of HCoV infections. Some of the observed efficacy might be due to prevention of pneumococcal-HCoV co-infection, particularly for the PCV9 trial. Since there are no suitable diagnostics for identifying non-bacteremic pneumococcal pneumonia in children, some of the HCoV-associated pneumonia cases may have been pneumococcal-HCoV co-infections for which only HCoV was detected, and these would be more likely to occur in the children who did not receive PCV. Pneumococcal pneumonia is also underdectected in adults due to limited sensitivity of blood culture and urinary antigen tests, so some of the reductions in coronavirus-associated disease in adults might be due to fewer pneumococcal co-infections [9]. Additionally, pneumococcal-viral interactions induce immune responses in the nasopharynx that influence the pathogenesis of respiratory infections [10, 11]. Studies in children and adults have found that pneumococcal densities in the nasopharynx are higher when a respiratory virus or influenza-like illness is present [12, 13]. Higher pneumococcal density in asymptomatic children was associated with increased risk of subsequent acute respiratory illness [13].

In adults, observational studies have reported reduced risk of COVID-19 after vaccination with other vaccines including influenza vaccines and recombinant zoster vaccine [14, 15]. Proposed mechanisms include nonspecific immune activation or confounding due to the ‘healthy vaccinee’ effect, whereby people who get vaccinated tend to be healthier than those who do not, and adjusted analyses fail to fully overcome underlying differences between vaccinated and unvaccinated participants. In a placebo-controlled RCT, bacille Calmette-Guérin (BCG) vaccination failed to protect older adults against SARS-CoV-2 respiratory infection, making the hypothesis of ‘trained immunity’ as the basis of heterologous vaccine protection less likely [16]. The elimination of benefit with prior antibiotics in the California PCV13 observational study makes this hypothesis less likely for PCV13 [8].

For PCV13, another plausible mechanism exists: the observed reduction in COVID-19 risk might be due to pneumococcal carriage modifying host immune responses [8]. In a study conducted among healthcare workers and patients, pneumococcal colonization was associated with diminished antiviral immune responses to SARS-CoV-2 in both cohorts [17]. SARS-CoV-2-specific salivary IgA in healthcare workers and SARS-CoV-2-specific memory B-cells in patients were lower in pneumococcal-colonized subjects compared to those without pneumococcal colonization [17]. These results build upon human challenge models exploring the relationships between pneumococcal colonization, nasopharyngeal microbiota, and responses to live attenuated influenza vaccine [11]. Typically, viruses are thought to precipitate secondary pneumococcal infections, however surveillance data from the UK raises the possibility that pneumococcal infection might increase risk or severity of subsequent COVID-19 [18, 19]. A study in US adults demonstrated higher adjusted odds of SARS-CoV-2 infection (2.73, 95%CI 1.58 to 4.69) among pneumococcal carriers and positive associations between SARS-CoV-2 and indicators of pneumococcal density, further suggesting synergistic interactions [20]. Together, these data raise the hypothesis that PCVs might reduce secondary infections caused by coronaviruses.

Limitations, remaining scientific questions, and data gaps

The clinical studies included in our summary have their own inherent limitations and biases. The two RCTs were not designed to evaluate coronavirus outcomes, and post-hoc or exploratory analyses should be interpreted with caution. The detection of HCoVs in children with pneumonia in the PCV9 trial is insufficient to demonstrate causality, however the reductions observed in the vaccine arm, plus consistency with the finding that efficacy against vaccine-type invasive pneumococcal disease was greater in HIV-uninfected children, suggest that PCV9 protects against HCoV-pneumococcal co-infection [21]. All observational studies are subject to bias, and although the two observational studies attempted to reduce bias through adjusted analyses and either matching [4] or correction using a negative control exposure [8], some residual bias, particularly due to the ‘healthy vaccinee’ effect, may remain [22].

The potential impact of PCVs on disease due to coronaviruses, including SARS-CoV-2, remains unknown. PCV9 efficacy against HCoV-associated pneumonia seen in South African children may not apply to SARS-CoV-2, as protection against individual HCoVs varied, and HCoVs primarily affect the upper respiratory tract whilst SARS-CoV-2 tropism is broader [6]. High-quality observational studies assessing PCV effectiveness against COVID-19 in children might be difficult to conduct in countries with high PCV uptake through pediatric national immunization programs. Nonetheless, future pediatric studies could explore the relationship between SARS-CoV-2, pneumococcal colonization and density, and respiratory infections.

In adults, evidence of PCV13 effectiveness against COVID-19 stems from a single cohort study conducted prior to COVID-19 vaccine availability [8]. Current data leave many questions unanswered in an era with effective SARS-CoV-2 vaccines. For example, it is unknown whether PCVs could supplement COVID-19 vaccines by reducing vaccine failures occurring because of waning host responses or SARS-CoV-2 variants. Observational studies investigating whether COVID-19 vaccines can prevent pneumococcal disease outcomes could further explore potential relationships between these two pathogens. We do not know of data to indicate whether vaccines against SARS-CoV-2 might also reduce pneumococcal outcomes; however, such an effect is plausible since it is well-known that viruses such as influenza can cause secondary pneumococcal pneumonias. Additionally, no data exist on whether interactions with coronaviruses vary by pneumococcal serotype. There is evidence from studies of other respiratory viruses (respiratory syncytial virus in children and influenza in adults) that viral interactions with pneumococci can differ by serotype [23, 24]. If PCVs have a specific effect on coronavirus disease, we hypothesize that this would be mediated through the reduction of vaccine-type pneumococci, either by reducing co-infection between vaccine-type pneumococci and coronaviruses, or by reducing carriage of vaccine serotypes that may play a role in the causal chain of coronavirus disease. A current gap in understanding is how PCVs might potentiate these effects in settings like the United States where the identification of vaccine-type pneumococci has been relatively low (either because of low carriage prevalence, low density carriage, or inadequate testing methodology), in contrast to the South African PCV9 trial conducted prior to routine use of PCV. Recent evidence indicates that pneumococcal carriage in adults is underestimated, and use of molecular methods, multiple sample types, and longitudinal sampling substantially increases detection of pneumococci, including PCV13 serotypes [25, 26]. Lastly, while pneumococcal carriage leads to changes in host antiviral immune responses, these changes have not been linked to clinical outcomes.

Because PCVs in adults are not widely used, the opportunity exists for high-quality observational studies that incorporate design elements to mitigate risk of bias to address some remaining questions [21]. In vitro and in vivo models could be used to explore the role of serotype in pneumococcal-coronavirus interactions. Extending human challenge models offers another efficient pathway for further research.