Pneumococcal Serotypes Associated with Community-Acquired Pneumonia Hospitalizations in Adults in Spain, 2016–2020: The CAPA Study

Newer higher valency pneumococcal conjugate vaccines (PCVs) have the potential to reduce the adult community-acquired pneumonia (CAP) burden. We describe the evolution and distribution of adult community-acquired pneumonia (CAP) serotypes in Spain, focusing on serotypes contained in the 20-valent PCV (PCV20). This was a prospective, observational study of chest X-ray (CXR)-confirmed CAP in immunocompetent adults hospitalized in one of four Spanish hospitals between November 2016 and November 2020. Pneumococci were isolated from cultures and detected in urine using BinaxNow® and Pfizer serotype-specific urinary antigen tests UAD1 and UAD2. We included 1948 adults hospitalized with CXR-CAP. The median age was 69.0 years (IQR: 24 years). At least one comorbidity was present in 84.8% (n = 1653) of patients. At admission, 76.1% of patients had complicated pneumonia. Pneumococcus was identified in 34.9% (n = 680) of study participants. The PCV20 vaccine-type CAP occurred in 23.9% (n = 465) of all patients, 68.4% (n = 465) of patients with pneumococcal CAP, and 82.2% (83/101) of patients who had pneumococcus identified by culture. Serotypes 8 (n = 153; 7.9% of all CAP) and 3 (n = 152; 7.8% of all CAP) were the most frequently identified. Pneumococcus is a common cause of hospitalized CAP among Spanish adults and serotypes contained in PCV20 caused the majority of pneumococcal CAP.


Introduction
The introduction of pneumococcal vaccines in childhood immunization programs has dramatically reduced pneumococcal disease due to direct and indirect effects [1][2][3][4].
The 20-valent PCV (PCV20) is the highest valency PCV currently licensed by the European Medicines Agency (EMA) for the prevention of invasive disease and pneumonia caused by Streptococcus pneumoniae in individuals 18 years of age and older [12].PCV20 Summary of national and regional pneumococcal immunization recommendations during the study period in the areas included.
Pneumococcal vaccination policies for infants have substantially reduced PCV13 vaccine-type (PCV13-VT) pneumococcal disease in both children and adults [2,8].Nevertheless, a certain burden of disease due to vaccine serotypes persists in adult populations, suggesting a plateau in the indirect protection provided by childhood vaccination [2,9].Additionally, a large proportion of CAP in adults is attributed to serotypes not covered by PCV13 vaccine types (non-PCV13-VT) [2, [9][10][11], serotypes 8, 22F, and 11A being the most frequently identified in Spain [2, 10,11].
A multicentre, observational, prospective, hospital-based study was conducted from 2011 to 2020 to determine the contribution of specific serotypes to CAP in immunocompetent adults.In this paper, we focus on the results of the last four years (2016-2020) and evaluate changes in serotype distribution, focusing primarily on PCV20-VTs and their comparison with other currently approved pneumococcal vaccines, and the association between PCV20-VTs and the presence of comorbidities and disease severity.

Participants
The study design and methodology have been previously described [10,18].Immunocompetent adults aged 18 years or older with CAP admitted between November 2011 and November 2020 to any of the following Spanish tertiary hospitals were included in the study: Hospital Galdakao-Usansolo, Bizkaia (Basque country), Hospital Clinic I Provincial, Barcelona (Catalonia), Hospital La Fe, Valencia (Valencia), and Hospital Álvaro Cunqueiro, Pontevedra (Galicia).The study protocol was approved by the ethics committee of each of the hospitals, and centrally approved by the Ethical Committee of Hospital Clinic (Barcelona), with a registration number of 2011/6799.All the participants gave written informed consent.We report data from the period 2016-2020 (Supplementary Figure S1).Inclusion criteria, microbiological studies, and definitions of complicated CAP are detailed in Supplementary Material and previous publications [10,[19][20][21].

Study Procedures
Demographic information, medical history, and clinical data of participants (including underlying conditions, tobacco cigarette smoking, alcohol consumption, history of previous pneumonia, history of hospitalization for at least 48 h within the 2 weeks before current admission, and pneumococcal and influenza vaccination history) were recorded as per standard clinical practice.Severity at the time of hospital admission was analysed by the CURB-65 score, Pneumonia Severity Index (PSI), and the Infectious Diseases Society of America/American Thoracic Society (IDSA/ATS) minor criteria.Patient evolution and outcome were also collected.

Microbiological Assessments
Microbiological tests were performed according to standard clinical practice [19].High-quality sputum was defined as <10 epithelial cells and >25 leukocytes per field with 100× magnification.The sputum cultures with exclusive or abundant S. pneumoniae growth and consistent Gram staining were considered positive for pneumococcus.Urine samples were collected from all the participants (n = 1948).Urine specimens (n = 1782) were tested using the BinaxNow ® S. pneumoniae Urinary Antigen Test and BinaxNow ® Legionella pneumophila Urinary Antigen Test.
Serotype-specific urinary antigen detection (UAD) tests UAD1 and UAD2 were used in all the urine samples (n = 1948) to detect serotypes contained in the 20-valent PCV and serotypes 2, 9N, 17F, and 20 [22,23].Serotyping was also performed on isolates from blood or pleural fluid cultures.The isolates and urine samples were processed as described in Torres et al. 2021 [10].
Complicated CAP was defined as the presence of one of the following on admission: bacteraemia, multilobar infiltrates, pleural effusion or empyema, respiratory failure, sepsis, and/or septic shock.Treatment failure was considered according to Spanish guidelines [19].Clinical stability with modified Halm criteria was achieved when the patient's vital signs were stable for a 24 h period, i.e., heart rate < 100 beats per minute; respiratory rate < 24 breaths per minute; axillary temperature < 37.2 • C; systolic blood pressure > 90 mmHg; oxygen saturation > 90% while breathing room air or at baseline for patients with chronic obstructive lung disease or those receiving oxygen therapy at home; good level of consciousness; tolerance to oral intake; adequate hydration and nutrition; and the absence of other active clinical or psychosocial problems requiring hospitalization [21].

Data Analyses
The baseline demographics and clinical data were summarized using the means and standard deviation, or the medians and interquartile range, as appropriate.The Pearson chi-squared test (or the exact Fisher test for 2 × 2 tables or likelihood ratio for MxN tables) was performed to compare proportions.For comparisons of independent groups, the Student's t-test (or its non-parametric equivalent Mann-Whitney U test), and the single factor ANOVA and Kruskal-Wallis H test for continuous variables were used.Assumptions of normality and homoscedasticity of the variables were studied for the use of parametric tests.Furthermore, exploratory bivariate logistic regression models were used to explore how each identified serotype influences intensive care unit (ICU) admission, severity of disease or death.Severity/complications of pneumococcal pneumonia were used as the dependent variable and the identified serotype compared to no serotype was identified as the independent variable.The R software V 4.1.1,EPIDAT 3.0, and SPSS 19.0 software were used to analyse the data.

Patient Characteristics by Serotype
The PCV20 serotypes accounted for a higher proportion of all CAP among patients aged 18-64 years than in patients aged ≥ 65 years (27.6% vs. 21.2%,p = 0.001) (Table 3).Among individual serotypes, serotypes 3 and 17F were more frequent in patients aged ≥ 65 years than in younger patients (p = 0.048 and p = 0.02, respectively).Serotypes 8 and 12F were more common among patients aged 18-64 years than in older patients (p < 0.001 for both serotypes).Figure 2 shows the distribution of specific serotypes in patients with pneumococcal CAP by age group.In both age groups, serotypes 8 and 3 were the most frequent, although eight was more common in patients aged < 65 years compared to those 65 years or older (34.6% vs. 13.5%,p < 0.001).Serotype distribution by severity on admission is shown in Supplementary Table S1.Patients with complicated pneumonia had a higher percentage of serotypes included in PCV20 than those with non-complicated pneumonia (25.1% vs. 20.1%,p = 0.027).Serotype 3 was more frequently identified in participants with complicated pneumonia compared to those with non-complicated pneumonia on hospital admission (8.7% vs. 5.0%, p = 0.009).
Similar rates of CAP caused by vaccine-type (VT) serotypes were observed in adults with no comorbidities and with at least one comorbidity.Nevertheless, for adults with comorbidities, the percentage of VT cases differed by number and type of underlying diseases (Table 4 and Supplementary Table S2).
When VT serotypes were analysed by the presence of underlying conditions and age Serotype distribution by severity on admission is shown in Supplementary Table S1.Patients with complicated pneumonia had a higher percentage of serotypes included in PCV20 than those with non-complicated pneumonia (25.1% vs. 20.1%,p = 0.027).Serotype 3 was more frequently identified in participants with complicated pneumonia compared to those with non-complicated pneumonia on hospital admission (8.7% vs. 5.0%, p = 0.009).
Similar rates of CAP caused by vaccine-type (VT) serotypes were observed in adults with no comorbidities and with at least one comorbidity.Nevertheless, for adults with comorbidities, the percentage of VT cases differed by number and type of underlying diseases (Table 4 and Supplementary Table S2).When VT serotypes were analysed by the presence of underlying conditions and age group (Supplementary Table S3), we found that among patients with chronic obstructive pulmonary disease (COPD), PCV20-VT serotypes were significantly more frequent in younger patients (aged 18-64 years) than in those aged 65 years or older (36.4% vs. 20.4%,p = 0.009).Similar findings were observed in patients with diabetes and chronic renal failure.These differences were due mainly to the higher percentage of cases caused by serotypes 3 and 8 in younger patients with comorbidities.
A comparison of the evolution of pneumococcal serotypes in two locations with different pneumococcal vaccination policies (Catalonia and Galicia) is shown in Table 5 [6,24,25].Hospital Clinic is located in Catalonia, a region that follows national recommendations and advice for the use of the PPV23 vaccine in adults aged at least 65 years old or in adults with underlying conditions and where PCV13 was not introduced in the immunization program for children until 2016.Hospital Álvaro Cunqueiro is located in Galicia, a region that introduced vaccination with the PCV13 vaccine in adults aged 65 years in July 2017 and where PCV13 was included in the childhood immunization program in 2011 [6,24,25].The proportion of cases due to PCV20-VT was similar in both regions (24.5% in Catalonia vs. 24.9% in Galicia), although there were significant differences in the total number of all CAPs in which serotypes 14 and 8 were identified (2.5% vs. 0.2%, respectively, p = 0.002; and 4.3% and 10.8%, respectively, p = 0.001).

Clinical Outcomes by Serotype
Supplementary Tables S4 and S5 show serotype distribution by complications and ICU admission during hospitalization, respectively.PCV20-VTs were slightly more common in CAP patients developing complications during hospitalization (27.7% vs. 21.7%,p = 0.003) and substantially more common among patients admitted to ICU (34.4% vs. 22.6%, p < 0.001, respectively).Serotype 3 was the most common individual serotype identified in patients with complications (10.0%) or admitted to ICU (16.5%).

Discussion
We identified S. pneumoniae in 35% of immunocompetent hospitalized patients with CAP in Spain between 2016 and 2020, while 24% of all the CAP cases were due to pneu-

Discussion
We identified S. pneumoniae in 35% of immunocompetent hospitalized patients with CAP in Spain between 2016 and 2020, while 24% of all the CAP cases were due to pneumococcal serotypes included in PCV20, 12% to serotypes included in PCV13, 14% to serotypes included in PCV15, and 25% to PPSV23 serotypes.The PCV20 serotypes were more common than other serotypes among individuals younger than 65 years old and lead to invasive disease, complications at hospital admission, and ICU admission.The most frequent serotypes found in our study were 3 and 8, followed by 22F.
In Spain, the PCV13 introduction in the childhood National Immunization Program has been linked to quick and sustained decreases in total and VT IPDs in young children and older unvaccinated cohorts [2].Recent data, however, indicate that the indirect effects have plateaued in many regions, including those with great coverage, specifically for serotypes 3, 7F, 19A, and 19F [26].Thus, the prevalence of PCV13 serotypes in Spanish adults showed that childhood vaccination provided insufficient indirect protection [10].Moreover, non-PCV13 serotypes, especially serotype 8, have a rising trend in the country [2].Our study describes the evolution and distribution of pneumococcal serotypes causing CAP in adults, focusing on PCV20, the highest valency PCV currently licensed, which includes serotype 8 and could help reduce the CAP burden in adult populations.
We observed a rise in the number of CAP cases detected due to S. pneumoniae in Spain during the last four years, compared to our previously published data (35% vs. 29%) [10].The introduction of the UAD-2 diagnostic test, which detects serotypes 2, 8, 9N, 10A, 11A, 12F, 15B, 17F, 20, 22F, and 33F, might be the main reason for this apparent increase since it is more sensitive for identifying pneumococcal serotypes than either blood culture or BinaxNow ® while maintaining specificity [23].The improved diagnostic sensitivity for the non-bacteraemic CAP suggests that our results more accurately reflect the burden of pneumococcal CAP.The inclusion of additional serotypes in future iterations of the UAD should further improve the pneumococcal CAP burden estimates.
Several studies from Spain and other European countries, such as the United Kingdom, reported similar proportions of CAP due to S. pneumoniae [11,27].In contrast, a lower proportion of the CAP cases associated with S. pneumoniae was reported from two studies in the USA (9.9% and 12.3%), and Greece (13.5%), even though they also used UAD1 and UAD2 [9,28,29].These differences could be partially explained by the pneumococcal vaccination policy for children and adults.As in most other European countries, Spain recommends a 2 + 1 PCV13 schedule for children.During the study period, PPSV23 was preferentially recommended in Spain for adult immunization.In Greece, the paediatric immunization program included PCV13 in a 3 + 1 schedule until June 2019 and 2 + 1 thereafter, and uses both PCV13 and PPSV23 for adult immunization, while the USA has consistently recommended a 3 + 1 schedule for children and, since 2014, PCV13 for adults aged 65 years and older [30].Hence, it is possible that patients in Greece and the USA had improved herd immunity as a result of the 3 + 1 infant schedule or better direct protection because of the PCV13 vaccination of older adults.
Serotype 3 remains one of the most common serotypes in Spain.Numerous studies have found that PCV13 provides protection against IPD and CAP among directly vaccinated persons [31].However, its ability to reduce carriage may be lower than for other vaccine serotypes, thus offering less or no indirect protection from childhood immunization [32].Moreover, a new serotype 3 clade with high levels of antibiotic resistance and altered subcapsular proteins may increase community transmission and raise the risk of disease, particularly among unvaccinated individuals [33,34].In Spain, serotype 3 disease risk among older adults was probably compounded by low pneumococcal vaccination coverage (including with PCV13) for this target group during the study period [2].
As in other European sites [35], the rising trend previously observed in Spain for serotype 8 was maintained [10].In Spain, serotype 8 is mostly associated with two main clonal complexes (CC53 and CC404), both related to penicillin-susceptible isolates [36].
The fact that this serotype was less frequent among older adults may be associated with this antimicrobial susceptibility pattern, as antimicrobial-resistant pneumococci appear more frequently in older patients with comorbidities [11].Previous studies have linked the presence of serotype 8 to highly invasive disease and moderate CFR [15].If Spain implements PCV20 among older adults and young children, the degree to which serotype 8 is reduced will depend on serotype-specific efficacy against disease and carriage, as well as vaccine uptake, and will require ongoing monitoring of IPD and CAP, including non-bacteraemic CAP.
In our study, we identified similar numbers of CAP cases among persons aged 18-64 years with comorbidities as in adults aged 65 years and older, and PCV20-VTs were more common in the former.This raises the possibility that new, higher valency PCVs will provide a similar public health benefit in both populations, a conclusion reached by the United States Advisory Committee on Immunization Practices (ACIP) when they recommended routine PCV immunization of both groups [37] In our study, PCV20-VT serotypes were associated with more severe clinical manifestations, including invasive disease, complications during hospitalization, and ICU admission.Previous reports have also shown a correlation between serotypes included in PCV20 and increased disease severity [13,14,38,39].This is to be expected, as the serotypes in PCV13 and subsequently, PCV20 were chosen in part for their association with IPD and severe disease [13,[38][39][40].
This investigation has some limitations, as described in the previously published results of the CAPA study [10].Firstly, the proportion of the CAP cases due to vaccine serotypes may have been overestimated, because no serotype-specific antigen exists for most serotypes not contained in PCV20.However, we found that the percentage of patients with pneumococcus infection identified by culture who had a PCV20-VT was 82.2% (83/101), indicating that the overestimation was not substantial.Moreover, the sensitivity of the UAD assays for non-bacteraemic pneumonia remains undefined due to the lack of a gold standard, and this value may be lower than that reported for bacteraemia pneumonia [41].Furthermore, we did not calculate CAP incidence, and we could not identify the microorganism responsible for CAP in 49.3% of the cases.We only investigated the burden of CAP in hospitalized, immunocompetent patients, and the results may not be representative of the outpatient setting or immunocompromised patients.The characteristics of our patient population also may explain the relatively low mortality rate found in the study.Lastly, our study included data collected during the peak of the COVID-19 pandemic, during which non-pharmaceutical interventions, such as mask-wearing and restrictions on social mingling, may have altered serotype epidemiology in a way that is not reflective of non-COVID-19 periods; this possibility will need to be assessed with ongoing surveillance.

Conclusions
Pneumococcus is a common cause of CAP in older adults in Spain and PCV20-VTs are a common cause of pneumococcal CAP.Newer, expanded valency PCVs, particularly those that contain the highly prevalent serotype 8, have the potential to substantially reduce disease burden.The specific target groups for adult vaccination could be defined by formal cost-effectiveness analyses.

Microorganisms 2023 , 18 Figure 3 .
Figure 3. Serotypes associated with severity on admission or during hospitalization in bivariate analysis Only serotypes with significant values are shown in the figure.For comparisons, no serotype identified is the reference parameter.ICU: intensive care unit; OR: odds ratio; PSI: pneumonia severity index.

Figure 3 .
Figure 3. Serotypes associated with severity on admission or during hospitalization in bivariate analysis Only serotypes with significant values are shown in the figure.For comparisons, no serotype identified is the reference parameter.ICU: intensive care unit; OR: odds ratio; PSI: pneumonia severity index.

Table 1 .
Characteristics of the study population in all community-acquired pneumonia cases, 2016-2020.

Table 2 .
Distribution of vaccine serotypes in all community-acquired pneumonia and pneumococcal community-acquired pneumonia by study period, 2016-2020.

Table 3 .
Vaccine serotype distribution in all community-acquired pneumonia and pneumococcal community-acquired pneumonia by age group and study period, 2016-2020.

Table 4 .
Vaccine serotype distribution in all community-acquired pneumonia and pneumococcal community-acquired pneumonia according to the presence of at least one underlying condition, 2016-2020 a .
a Patients might have more than one underlying condition.1Confirmedpneumococcal CAP (according to UAD1/UAD2 or BinaxNow ® tests) for which blood and/or pleural fluid culture results were negative; 2 S. pneumoniae isolated in blood and/or pleural fluid culture.Four isolates of invasive pneumococcal pneumonias were not serotyped due to lack of sample or lysis.CAP: community-acquired pneumonia; COPD: chronic obstructive pulmonary disease; PCV13: 13-valent pneumococcal conjugate vaccine; PCV15: 15-valent pneumococcal conjugate vaccine; PCV20: 20-valent pneumococcal conjugate vaccine; PPSV23: 23-valent pneumococcal polysaccharide vaccine.