To achieve the goals of this study, three different approaches were conducted at Hospital Universitari de Bellvitge (HUB), a teaching hospital located in the southern Barcelona area. The Microbiology Service receives samples for diagnosis from primary care centers and from patients who come to the hospital ( Supplementary Figure S1 ).

The first approach was to analyze changes in invasive pneumococcal diseases (IPD) during the COVID-19 pandemic. All IPD episodes in adults (≥18 years old) from January 2019 to December 2023 were retrospectively collected and available pneumococcal isolates were subjected to antibiotic susceptibility testing and WGS. IPD was defined as the isolation from a sterile body site in a patient with signs and symptoms of infection. To contextualize the monthly trends in IPD including the pandemic period, data from 2010 to 2023 was analyzed.

The second approach was to investigate the relationship between pneumococcal disease and COVID-19. All patients with SARS-CoV-2 who had a S. pneumoniae isolate from a clinical sample between March 2020 and December 2023 were included (n = 66). Available pneumococcal isolates were sequenced (WGS, n= 57) and clinical charts were reviewed.

Third approach was to assess S. pneumoniae colonization in patients with SARS-CoV-2. A random selection of SARS-CoV-2 positive nasopharyngeal samples (routinely diagnostic procedures) of patients were collected between May and October 2023 (n = 461). A third of the samples were collected from patients attending primary care centers and the rest from hospital patients (only one sample per patient was included). COVID-19 disease was defined as the detection of SARS-CoV-2 in a patient presenting symptoms of infection such as cough, fever, shortness of breath, sudden onset of anosmia, ageusia or dysgeusia. Patients that did not require further hospitalization were considered mild cases while those requiring hospital care were classified as severe. All samples were prospectively stored at -80°C for pneumococcal colonization detection by RT-PCR and viral genome sequencing. Clinical charts were reviewed for the positive pneumococcal colonization patients and representative negative control group.

The identification of S. pneumoniae isolates was routinely performed using standard microbiological procedures (MALDI-Biotyper, bile solubility and/or optochin susceptibility). Antibiotic susceptibility was determined by disk diffusion and E-test in accordance with EUCAST recommendations and criteria.

Pneumococcal colonization was studied using a dual-target RT-PCR assay (lytA and piaB genes) in nasopharyngeal samples. Colonization was confirmed when both targets were detected and the cycle threshold (Ct) values were similar (differences <3 Ct) (Simões et al., 2016; Miellet et al., 2023).

Nasopharyngeal samples were considered positive for SARS-CoV-2 if detected by RT-PCR with a Ct value <37 (Alinity m Resp-4-plex assay) (S. T. Almeida et al., 2020).

For library preparation and sequencing, S. pneumoniae was cultured overnight on 5% sheep blood agar at 37°C with 5% CO2. DNA was extracted using the QIAamp DNA Mini Kit (Qiagen, Germany) and quantified with the Qubit dsDNA HS Assay Kit (Thermo Fisher, USA). Illumina paired-end libraries (2x300 bp) were prepared using the DNA Prep kit and sequenced on the Illumina MiSeq Platform (Illumina, USA).

For bioinformatic analysis, quality assessment and genome assembly were performed using the Bactopia pipeline. Reads quality control was conducted with Bactopia’s preprocessing module, and genome assembly was performed using the assembly module. The Bactopia MLST module was used for in silico MLST determination using PubMLST database (web PubMLST). In silico serotyping was performed with SeroBA (github.com/sanger-pathogens/seroba) ( Supplementary Table S1 ). The reads were deposited in the European Nucleotide Archive (ENA), and the metadata is outlined in Supplementary Table S2 . Antibiotic resistance due to mutations in resistance-related genes, such as pbp1a, pbp2b and pbp2x (β-lactams), parC, parE and gyrA (quinolones), or folA and folP (cotrimoxazole), was analyzed and compared with the reference genome (S. pneumoniae R6, NC_003098.1) using Geneious. Acquired resistance genes were identified using AMRfinder from Bactopia workflow.

For in-depth analysis of serotype 12F recombination events, 14 genomes from pneumococci isolated in our hospital between 2008 and 2023 were selected. Recombinant blocks were detected by Gubbins using NZ_LS483450 as reference and default parameters.

Genomic sequencing of SARS-CoV-2 was performed following the ARTIC amplicon sequencing protocol (artic.network/ncov-2019).Total nucleic acid extraction was performed with MagMAX Viral/Pathogen II Nucleic Acid Isolation Kit on a KingFisher Flex purification system (Applied Biosystems, USA). The Ct was determined using the TaqPath COVID-19 RT-PCR assay on a QuantStudio 5 (Applied Biosystems, USA), and only samples with a Ct value <30 were sequenced. RNA retrotranscription was carried out with LunaScript (New England BioLabs, USA), followed by amplification of the 30 kb viral genome using the xGen SARS-CoV-2 ARTIC Amplicon Panel (v4.1), which generates 400 bp amplicons (IDT, USA). DNA quantification was performed using the Qubit dsDNA HS Assay Kit (Thermo Fisher, USA). For library preparation and sequencing, Illumina paired-end libraries (2x200 bp) were prepared using the Illumina DNA Prep kit and sequenced on the Illumina MiSeq Platform (Illumina, USA).

Bioinformatic analysis was conducted using the DRAGEN COVID Lineage App from BaseSpace (basespace.illumina.com). Pangolin was used to assign lineages to the COVID-19 sequences. Consensus genomes and demographic data were deposited at the GISAD web page (https://www.gisaid.org) for public access.

Chi-square test and Fisher’s exact test were applied to assess the association between categorical variables, set at α < 0.05 (two-tailed) for significance. Chi-square test was applied by default, while Fisher’s exact test was applied when the expected frequency in any cell was less than 5. Tests were performed using the chisq.test and fisher.test functions in R (version 4.3.2), respectively. To analyze changes in the serotype and clonal distributions of invasive isolates, three periods were considered: pre-pandemic (January 2019-March 2020), pandemic (April 2020-June 2022), and post-pandemic (July 2022-December 2023).

From January 2019 to December 2023, a total of 355 IPD episodes were recorded. Of these, 150 episodes occurred in adult patients aged 18 to 64 years and 205 in those over 64 years old. The monthly number of IPD episodes detected from 2010 to 2023 is shown in Figure 1A . The seasonal fluctuation of IPD was interrupted during the pandemic, with a sharp decrease in the number of IPD episodes (from 105 episodes in 2019 to 44 in 2020 and 31 in 2021). The lockdown due to the SARS-CoV-2 pandemic began in Spain in March 2020. Following this, the number of IPD episodes decreased, with no winter season peak during 2020-2021 (n = 11). Subsequently, winter seasons were characterized by a limited recovery in IPD numbers during 2021-2022 (n = 30) and 2022-2023 (n = 65).

The decline and recovery of IPD cases varied by age group ( Supplementary Figure S2 ). In the 18–64 age group, there was a sharp decline from 2019 to 2021, representing a decrease of 68.3%. Recovery began in 2022 but rebounded significantly in 2023 with a total number exceeding the pre-pandemic levels (+26.8%). For patients over 64 years old, the decline was even more evident and pronounced. Cases dropped from 2019 to 2021 (70.6% reduction). After limited recovery in 2022, the figure at the end of 2023 was similar to 2019 (–6.2%).

To identify resilient serotypes during the lockdown ( Figure 1B ), we analyzed the serotype distribution of IPD episodes. We compared three periods: pre-pandemic (January 2019-March 2020), pandemic (April 2020-June 2022), and post-pandemic (July 2022-December 2023). Overall, the serotype distribution remained stable across all three periods, with predominance of serotypes 3 and 8. Notably, although the numbers were low, serotype 12F disappeared during the lockdown, while serotype 9N emerged. The proportion of IPD cases covered by PCV15, PCV20 and PCV21 in the three periods was: 32.8%, 60.8% and 64.0% in the pre-COVID period; 28.0%, 54.7% and 61.3% during the COVID-19; and 35.9%, 62.8% and 65.5% in the post-COVID period, respectively.

Regarding the Global Pneumococcal Sequencing Cluster (GPSC) and Sequence Type (ST) distribution of the main serotypes, different patterns were observed. For example, GPSC3-ST53 was predominant in serotype 8 isolates across all periods ( Figure 2 ). Among serotype 3 isolates, GPSC12-ST180 was nearly as prevalent as GPSC83-ST260 before the pandemic, and it became clearly dominant in the post-pandemic period. Regarding serotype 12F, a notable change was observed. Before the pandemic, this serotype consisted of GPSC26-ST989 (n = 3) and GPSC55-ST8060 (n = 2). This serotype was not detected during the pandemic and re-emerged in the post-pandemic period in association with a new sub-lineage of GPSC26-ST3377 exhibiting a multidrug resistant phenotype (MDR, penicillin and macrolide resistance, see below). Among serotype 9N isolates, genetic diversification was observed. During the first two periods, including part of the pandemic, this serotype was exclusively composed of GPSC16-ST66. In the post-pandemic, the number of cases increased, and this serotype was represented by three GPSCs: GPSC16-ST66 (n = 4), GPSC699-ST19827 (n = 2), and GPSC137-ST3982 (n = 1).

During the SARS-CoV-2 pandemic (March 2020 to December 2023), 66 COVID-19 patients admitted to the hospital had concomitant pneumococcal infection ( Supplementary Table S3 ). A total of 52 episodes were classified as non-invasive, while 14 were considered IPD. The most common serotypes (57/66 serotypes available) were 3 (17.5%), 11A (15.8%), 6C (8.8%), 23A (8.8%), and 19F (5.3%). Despite the relatively low numbers of episodes per year, notable changes in the prevalence of certain serotypes coinfecting SARS-CoV-2 patients were observed. For example, serotype 3 was consistently present in 2020 (10.0%), 2021 (18.2%), and 2022 (30.4%), but it was not detected in 2023 ( Supplementary Figure S3 ). In contrast, serotypes 11A and 6C were detected in all four years.

Given the high use of antibiotics during the SARS-CoV-2 pandemic, especially in the early stages, we analyzed resistance rates among the S. pneumoniae isolates from COVID-19 patients ( Figure 3 ). All 57 sequenced isolates had penicillin MICs ≤ 2 mg/L (range ≤0.06–2 mg/L). Fifteen isolates (26.3%) were macrolide-resistant, harboring erm(B) and/or mef(E) genes. Tetracycline-resistance (n = 13, 22.8%) was associated with the presence of the tet(M) gene. The resistance rates for co-trimoxazole and levofloxacin were 15.8% and 5.3%, respectively. In both cases, the resistance was linked to point mutations in target genes. Figure 3 displays the phylogenetic tree of pneumococci isolated from COVID-19 patients.

The re-emergence of serotype 12F in 2023 was associated with a new ST, the ST3377, of the major GPSC26 lineage. This sub-lineage exhibited MDR phenotype, including non-susceptibility to penicillin, macrolides, clindamycin, and tetracycline. After a deeper analysis ( Figure 4 ; Supplementary Table S4 ), a total of 17 recombination blocks (RB) between ST989 and ST3377 were detected. Some of these RB included important genes such as pbp2x in RB5 that changed from allele 3 in ST989 to 20 in ST3377, pbp1a in RB6 which changed from allele 12 to 17 or pbp2b (allele 4 to new) in RB15. This new PBP type in ST3377 results in an increase of penicillin (1 mg/L) and amoxicillin (2 mg/L) MICs. RB15 also includes ddl gene explaining the ST change in this new sub-lineage. In addition, the different rearrangements of the Tn5252 present in the reference genome include the loss of the cat gene (resistance to chloramphenicol) and the acquisition of the erm(B) which confers resistance to erythromycin and clindamycin. There were no differences in the structure of the capsular operon between strains of these two sub-lineages. This acquired resistance represents a new hallmark of serotype 12F and warrants close surveillance.

A total of 461 nasopharyngeal swabs from patients with SARS-CoV-2 infection, who attended primary care centers (n = 133) or the hospital (n = 328), were analyzed ( Table 1 ). Hospitalized patients were also tested for influenza viruses (A and B) and respiratory syncytial virus (RSV), but none tested positive, indicating no viral co-detection. Pneumococcal colonization was screened through a dual-targeted PCR (lytA plus piaB). Of the 461 samples studied, 181 (39.3%) were from men and 273 (59.2%) from patients over 64 years of age. Pneumococcal colonization was detected in 13 patients (2.8%), 10 of whom presented mild COVID symptoms. In concordance with the circulating SARS-CoV-2 variants in the period studied, the majority of infections, both in colonized and non-colonized patients, involved recombinant XBB variants. Table 1 presents the analysis of pneumococcal colonization in patients with SARS-CoV-2 infection. The colonization rate was similar in primary care (2.3%) and hospitalized (3.0%) patients. In terms of disease severity, colonization was more common in patients with severe COVID-19 (4.7%) compared to those with mild disease (2.5%) (p = 0.404; IC 95% 0.13-3.05). By age group, the colonization rate was higher in young adults (18–64 years old, 4.8%) than in older people (>64 years old, 1.5%) (p = 0.04; IC 95% 0.92-15.21). Regarding differences in colonization among Omicron lineages, some slight variations were observed although they did not reach the statistical significance. Further statistical analysis adjusting for confounding factors was not possible due to low colonization rates.

We compared the clinical characteristics of SARS-CoV-2 patients between two groups: patients with SARS-CoV-2 and pneumococcal colonization (n = 10, 3 clinical charts unavailable) versus patients with SARS-CoV-2 and pneumococcal infection (n = 66) ( Table 2 ). We also selected a control group of patients with only SARS-CoV-2 infection (n = 26). In general, patients without pneumococcal colonization or infection had fewer underlying conditions and comorbidities compared to those with pneumococcal colonization. The clinical presentation of patients with pneumococcal infection was more severe, with a higher rate of pneumonia (65.2%) and respiratory failure (69.7%). These patients also required hospital admission more frequently (84.8%).

The rate of pneumococcal vaccination was low in patients with pneumococcal infection. Almost all patients were appropriately vaccinated against SARS-CoV-2. However, as some pneumococcal co-infection episodes occurred before SARS-CoV-2 vaccines were available, the vaccination rate in this group was lower. To address this, we analyzed this group in more detail, differentiating between episodes occurring before SARS-CoV-2 vaccines were available (prior to February 2021) and those after vaccines became available (from February 2021 onward). Patients prior to February 2021 had more risk factors and comorbidities, with half of them having more than two comorbidities, compared to 15 out of 55 (27.3%) in the other group. Their clinical presentation was also more severe, and their 30-day mortality rate was higher (27.3% vs 6.5%). Pneumococcal serotypes also varied between the groups: serotype 3 was nearly absent in the pre-vaccine period, whereas it was the most prevalent after February 2021. There were only two cases of pneumococcal infection in PCV13-vaccinated patients, with strains belonging to serotype 3 and 19F.