Antimicrobial Resistance in Streptococcus pneumoniae before and after the Introduction of Pneumococcal Conjugate Vaccines in Brazil: A Systematic Review

Streptococcus pneumoniae causes serious illnesses, such as pneumonia, bacteremia, and meningitis, mainly in immunocompromised individuals and those of extreme ages. Currently, pneumococcal conjugate vaccines (PCVs) are the best allies against pneumococcal diseases. In Brazil, the 10-valent and 13-valent PCVs have been available since 2010, but the threat of antimicrobial resistance persists and has been changing over time. We conducted a systematic review of the literature with works published since 2000, generating a parallel between susceptibility data on isolates recovered from colonization and invasive diseases before and after the implementation of PCVs for routine childhood use in Brazil. This systematic review was based on the Cochrane Handbook for Systematic Reviews of Interventions and Preferred Reporting Items for Systematic Literature Reviews and Meta-Analyses (PRISMA) guidelines. Despite the inclusion of PCVs at a large scale in the national territory, high frequencies of non-susceptibility to important drugs used in pneumococcal diseases are still observed, especially penicillin, as well as increasing resistance to macrolides. However, there are still drugs for which pneumococci have a comprehensive sensitivity profile.


Introduction
Streptococcus pneumoniae is a common colonizer of the human upper respiratory tract.However, pneumococci can cause milder diseases, such as acute otitis media (AOM) and sinusitis, as well as severe diseases, including community-acquired pneumonia (CAP), bacteremia, and meningitis, affecting individuals of all age groups, especially those of extremes ages and immunocompromised people [1][2][3][4].
The main prevention strategy against pneumococcal diseases is pneumococcal conjugate vaccines (PCVs).They confer a high degree of protection against specific serotypes, interfering in the fluctuation of their distribution and in the prevalence of resistance to antimicrobial agents [5,6].There are several PCVs approved for use in children and adults in different countries [7][8][9][10][11][12][13]. In Brazil, the 7-valent PCV (PCV7; serotypes 4, 6B, 9V, 14,18C, 19F, and 23F) was initially made available, in 2001, in private immunization clinics for children and in the Brazilian public health system (Sistema Único de Saúde, SUS) for children < 5 years old who were at high risk of invasive pneumococcal diseases (IPD).In 2010, the 10-valent PCV (PCV10; PCV7 serotypes + 1, 5, and 7F) was introduced into the Brazilian National Immunization Program (NIP) for free-of-charge immunization of all children < 5 years old.Initially, the PCV10 schedule comprised three primary doses at 2, 4, and 6 months of age and a booster dose at 12-15 months of age (3p + 1), but, since 2016, the PCV10 dosing regimen in the Brazilian NIP changed to 2p + 1 (at 2 and 4 months of age and a booster dose at 12 months of age).In 2010, the 13-valent PCV (PCV13; PCV10 serotypes + 3, 6A, and 19A) replaced PCV7 in private clinics, and it was made available via SUS in 2019 for individuals aged 5 years or older who are at the highest risk for IPD, including patients living with HIV/AIDS, patients with cancer, and those who underwent solid organ or bone marrow transplantations.In 2023, the PCV15 (PCV13 serotypes + 22F and 33F) was approved for use in Brazil [13][14][15][16][17][18][19].
Although not compulsory, vaccination in Brazil is strongly recommended.The most used antipneumococcal vaccine in Brazil is PCV10, and between 2011 and 2022, considering the different geographical regions, the average PCV10 vaccination coverage with primary doses and the booster dose was 88.5% and 80.6%, respectively.Data on PCV13 coverage are limited, but a few studies report a low coverage (<8%) among children < 5 years old [20][21][22].Some post-PCV10 introduction studies in Brazil indicate a reduction in the average mortality rate of pneumonia (11%; from 29.69 to 23.40 per 100,000) in children younger than 1 year after four years of vaccination [23] and a significant reduction, between 13.9% and 17.6%, in hospitalizations for pneumonia in the target groups of vaccination over five years after PCV10 implementation [24].On the other hand, the incidence of pneumococcal meningitis remains high in Brazil, with approximately 1000 cases/year [25].
Beta-lactams, especially penicillin and amoxicillin, are the main, but not exclusive, choices to treat pneumococcal diseases.Other antimicrobial agents frequently used against pneumococcal diseases include macrolides, fluoroquinolones, and lincosamides [26][27][28].The first choice for AOM is amoxicillin, which may be combined with clavulanate in cases of recurrence within 30 days or when associated with other symptoms.For allergic people, cefdinir or azithromycin has been frequently prescribed [29].For CAP patients without comorbidities, the most indicated treatment includes amoxicillin, doxycycline, or a macrolide, and for those with comorbidity, a fluoroquinolone or a combination of amoxicillin with clavulanate or cephalosporin plus a macrolide or doxycycline.Furthermore, for patients admitted to a hospital, a fluoroquinolone in monotherapy or the combination of a macrolide with a beta-lactam is recommended, with a difference in treatment for patients in intensive care, who require the combination of a beta-lactam with a macrolide or a fluoroquinolone [30].
Antimicrobial resistance, however, is a concern among S. pneumoniae.Penicillin nonsusceptible pneumococci (PNSP) are considered a medium priority risk to human health by the World Health Organization [31].Drug-resistant S. pneumoniae is also classified as a serious threat in the USA [32].The growing report of resistance to different antimicrobial agents has been a cause for concern in public health and demands strategies in public policies, as well as therapeutic alternatives [27].
This systematic literature review aims to verify the Brazilian scenario pre-and post-PCV10 regarding antimicrobial resistance among S. pneumoniae isolates associated with colonization and diseases recovered from individuals from all age groups.For this reason, we selected the year 2000 as a starting point, considering that it corresponds to 10 years before the introduction of PCV10 in Brazil.

Assessment of the Methodological Quality of the Articles
Regarding the description and case definition of the population of the studies, only two (11.8%) of seventeen articles were negatively classified.Seven (41.2%) of the seventeen articles described the representativeness of the sample and its sampling in a clear way.All articles described the type of test used and mentioned or referenced the evaluative standard used.However, only five (29.4%) articles described the use of internal quality control.Detailed data can be found in Table 1.

Assessment of the Methodological Quality of the Articles
Regarding the description and case definition of the population of the studies, only two (11.8%) of seventeen articles were negatively classified.Seven (41.2%) of the seventeen articles described the representativeness of the sample and its sampling in a clear way.All articles described the type of test used and mentioned or referenced the evaluative standard used.However, only five (29.4%) articles described the use of internal quality control.Detailed data can be found in Table 1.
[     Considering all the references included in this study, we obtained data on 18,273 isolates; data on 15,437 (84.5%) isolates were provided by SIREVA II (invasive isolates) and data on 2839 (15.5%) isolates were obtained through the included articles.Of 18,273 isolates, 2683 (14.7%) isolates were associated with colonization, 117 (0.6%) isolates with non-invasive diseases, and 39 (0.2%) isolates were associated with invasive diseases, but not presented by SIREVA II.In total, 8991 (49.2%) isolates were from the pre-PCV10 period and 9285 (50.8%) were from the post-PCV10 period.
Invasive isolates included those from sterile sites, such as blood, pleural fluid, and cerebrospinal fluid (CSF).The colonization isolates, obtained through the articles, were mainly collected through sterile swabs in contact with the nasopharynx and oropharynx.Other types of isolates were included in non-invasive pneumococcal diseases, such as ear abscess, cervical abscess, buttock abscess, nasal/eye abscess, bronchial aspirate, corneal aspirate, sinus aspirate, pulmonary aspirate, tracheal aspirate, sputum, bronchoalveolar lavage, auricular secretion, bronchial secretion, tear duct secretion, conjunctival secretion, ocular secretion, wound secretion, skin secretion, pulmonary secretion, postauricular secretion, tracheal secretion, rectal swab, corneal ulcer, and urine.Antimicrobial resistance data were compiled and organized into tables separated by pre-and post-PCV10 introduction periods (Tables 3 and 4).Higher frequencies of resistance to sulfamethoxazole-trimethoprim were observed in invasive isolates in the pre-PCV10 period (60.1%; 4815/8016).No case of non-susceptibility (intermediate + resistant) in the pre-PCV10 period was observed for vancomycin, linezolid, trovafloxacin, telithromycin, and quinupristin-dalfopristin, as well as resistance to amoxicillin.In the post-PCV10 introduction period, no resistance was observed to vancomycin, linezolid, telithromycin, and quinupristin-dalfopristin.
Data on susceptibility to penicillin and ceftriaxone were separated into meningitis and non-meningitis and by period, respectively, in  For ceftriaxone, we observed a higher proportion of resistance to general in the post-PCV10 introduction period (6.5%, 25/384), similar to penicillin, which showed a higher proportion (44.6%; 499/1118).
Regarding macrolide resistance, a greater volume of data were obtained for erythromycin.There is a high susceptibility for colonization isolates in the pre-PCV10 period (95.2%; 719/755), with a decline in susceptibility in the post-PCV10 period (82%; 596/727).These findings were similar to invasive isolates, which in the pre-PCV10 period were 94.5% susceptible (6795/7186) and 81.1% (6397/7884) in the post-PCV10 period.Finally, a small number of isolates was tested against fluoroquinolones and, as a result, there are data on ofloxacin and trovafloxacin susceptibility only for the pre-PCV10 period.All ninety-two (100%) carriage isolates tested against ofloxacin and the two (100%) carriage isolates tested against trovafloxacin were susceptible.For invasive (n = 1) and non-invasive (n = 2) disease isolates, the susceptibility against trovafloxacin was also 100%.Levofloxacin had a higher number of susceptible isolates, with a proportion of 98.8% (557/564) in the pre-PCV10 period and 100% (565/565) in the post-PCV10 period for colonization isolates.All the 20 invasive isolates from the pre-PCV10 period were susceptible to levofloxacin.All non-invasive disease isolates from the pre-PCV10 (48/48) and post-PCV10 (22/22) periods were also susceptible to levofloxacin.

Statistical Analysis
The proportion of erythromycin non-susceptible isolates was higher among carriage (p < 0.01) and invasive (p < 0.01) isolates of the post-PCV10 period.The proportion of sulfamethoxazole-trimethoprim susceptibility (p < 0.01) was higher among isolates of the post-PCV10 period, regardless the isolation source.Although a limited number of isolates has been tested against meropenem, susceptibility to this drug was higher among non-invasive isolates (p < 0.01) of the post-PCV10 period.Among carriage isolates, the frequencies of susceptibility to chloramphenicol (p = 0.01), as well as non-susceptibility to clindamycin (p < 0.01) and tetracycline (p < 0.01), were higher after PCV10 introduction.Among invasive isolates (meningitis and non-meningitis), the proportion of susceptibility to penicillin (p ≤ 0.01) and ceftriaxone (p ≤ 0.02) was higher after PCV10 introduction.On the other hand, the frequency of penicillin non-susceptible pneumococci was higher among carriage isolates (p < 0.01) in the general parameter in the post-PCV10 period.Figure 2 shows the main results of proportion tests when statistically significant differences in the antimicrobial susceptibility profile were detected between isolates of the pre-and post-PCV10 periods.

Discussion
Based on the 17 articles selected through this systematic literature review, a high number of articles (88.2%; 15/17) were positively classified within the tool used (modified Newcastle-OQawa assessment scale) [63][64][65], offering greater reliability in the use of the data obtained.Notably, a considerably high number of invasive isolates originated from SIREVA II (84.4%, 15,437/18,276), considered an important epidemiological surveillance tool for S. pneumoniae and other microorganisms in Latin America.
The susceptibility to sulfamethoxazole-trimethoprim (SXT) was higher after PCV10 introduction for routine use in Brazil (p = 0.01).In the pre-PCV10 period among invasive isolates, the proportion of SXT susceptibility and non-susceptibility was 40.3% (2907/7218)

Discussion
Based on the 17 articles selected through this systematic literature review, a high number of articles (88.2%; 15/17) were positively classified within the tool used (modified Newcastle-Ottawa assessment scale) [63][64][65], offering greater reliability in the use of the data obtained.Notably, a considerably high number of invasive isolates originated from SIREVA II (84.4%, 15,437/18,276), considered an important epidemiological surveillance tool for S. pneumoniae and other microorganisms in Latin America.
The susceptibility to sulfamethoxazole-trimethoprim (SXT) was higher after PCV10 introduction for routine use in Brazil (p = 0.01).In the pre-PCV10 period among invasive isolates, the proportion of SXT susceptibility and non-susceptibility was 40.3% (2907/7218) and 59.7% (4311/7218), respectively.In the post-PCV10 introduction period, there was a drop in the number of non-susceptible isolates (37.7%; 2957/7839) compared to the susceptible ones (62.3%; 4882/7839).This comparison is interesting because it presents a change in the general panorama of antimicrobial resistance of this drug, tending to a drop in resistance levels.However, it is noteworthy that this phenomenon is not uniformly observed in other countries; for example, a recent study carried out in Malawi (southeast Africa) with colonization and invasive isolates verified a high frequency of resistance to SXT (96%; 137/143), with similar resistance profiles worldwide [66].
For penicillin, there was a statistically significant difference in the percentage of nonsusceptibility between the pre-and post-PCV10 introduction periods among invasive isolates, with lower frequencies for both meningitis (31.9% to 28.7%; p = 0.01) and nonmeningitis (17.7% to 6.3%; p < 0.01) after PCV10 use.This finding is very important since in Latin America most countries usually report a prevalence of penicillin resistance among meningitis isolates over 30% [62].On the other hand, regarding the general parameter, there was an increase in non-susceptibility between the same periods from 25.9% (387/1495) to 44.1% (461/1000) for colonization isolates (p < 0.01), respectively.This may be explained mainly by the impact of childhood vaccination with PCV10 in Brazil since before PCV10 introduction, resistance to beta-lactams was mostly associated with serotypes included in the vaccine formulation, especially 6B, 14, 19F, and 23F [38,40,67].After PCV10 introduction, these serotypes were nearly eliminated from both colonization and diseases [21,22,67].
Due to the serotype replacement phenomenon, some of the main serotypes circulating in Brazil are currently 19A in invasive diseases with high resistance to different classes of antimicrobial agents and 6C in colonization isolates [21,36,40,44].In this context, a replacement by PCV13, PCV15, PCV20, and even Pneumosil ® , which also protects against 10 vaccine serotypes, would be appropriate to replace PCV10 in the Brazilian National Immunization Program [6,[9][10][11]19].However, this phenomenon may continue due to the varied range of capsular serotypes and their distribution among populations.
For ceftriaxone, the general parameter shows a higher frequency of non-susceptible isolates (4.6%; 25/542) in isolates associated with colonization in the post-PCV10 introduction period.Although not statistically significant (p = 0.43), this is of paramount importance since third-generation cephalosporins are frequently used to treat pneumococcal meningitis [26], and isolates with this resistance profile circulating within a population represent a high risk of transmission and development of severe diseases.In turn, susceptibility to ceftriaxone was significantly higher (p ≤ 0.02) among invasive isolates recovered in the post-PCV10 period.
Frequencies of susceptibility to macrolides, namely erythromycin and clarithromycin, exceeded 70% across all periods evaluated.A similar profile between invasive and colonization isolates was observed, with an important decline in susceptibility in the post-PCV10 period.Resistance in the pre-PCV10 period was around 5% for both colonization and invasive isolates.However, the proportion of macrolide-resistant isolates almost reached 20% in the post-PCV10 period.Macrolide resistance has been increasing worldwide.A nationwide surveillance in the USA between 2018 and 2019, with isolates recovered from blood and respiratory specimens from adults, revealed a high burden of macrolide resistance among S. pneumoniae, reaching almost 40% [68].
Levofloxacin is the fluoroquinolone with the greatest amount of data available for analysis, and the authors observed a high proportion of susceptibility among colonization isolates in the pre-PCV10 (98.8%; 557/564) and the post-PCV10 (100%; 565/565) periods.A similar scenario was observed for invasive isolates, in which all isolates (24 isolates from pre-PCV10 and 50 isolates from post-PCV10 periods) were susceptible to fluoroquinolones.
Despite the increasing and worrying resistance to beta-lactams and macrolides, all isolates were susceptible to vancomycin, linezolid, telithromycin, and quinupristin-dalfopristin in both the pre-and post-PCV10 introduction periods.
The main limitation of this work was the high variation of data presentation in the articles included in this review, making it difficult to group them.Also, 22 articles with important data were not made available in time by the authors, despite attempts to contact them.Still, we retrieved data on an extensive collection of isolates recovered from various clinical sources, mainly associated with IPD, and from different geographical regions of Brazil, providing a comprehensive scenario of antimicrobial resistance in pneumococci before and after PCV introduction for routine use in Brazil.

Search Strategy
This systematic review was structured between May 2022 and July 2023 with the search date on 23 May 2023.The following databases were consulted: Lilacs (Latin American & Caribbean Health Sciences Literature), Embase, Pubmed, Scopus, and Web of Science.In addition to these, a manual search was carried out in the bibliographic references of the selected articles and data extraction from the documents was produced by the System of Surveillance Networks of Responsible Agents for Bacterial Pneumonia and Meningitis (SIREVA II; Electronic page: https://www3.paho.org/hq/index.php?option= com_docman&view=list&slug=sireva-ii-8059&Itemid=270&lang=pt#gsc.tab=0;accessed on 23 May 2023).
The files referring to the search strategies according to the base can be found in the Supplementary Material as Table S1 and the manual search as Table S2.
This review was based on the question: "How is the resistance profile to antimicrobial agents of Streptococcus pneumoniae isolates before and after the introduction of pneumococcal conjugate vaccines in Brazil?".It is noteworthy that this research was submitted to the Prospero platform [PROSPERO acknowledgment of receipt (364743)].

Article Selection and Data Extraction
All articles found were initially evaluated based on titles and abstracts.After this step, some articles were selected for full reading based on the inclusion and exclusion criteria listed in Table 7. Two authors performed these steps and a third author was consulted in case of doubt.Then, data were extracted using the Microsoft Excel ® program.

Quality Assessment
Individual quality control of each academic work was evaluated according to the Newcastle-Ottawa Quality Tool Assessment Scale with modifications according to models by Sugianli et al., 2021 andMancini et al., 2017 for cross-sectional studies and according to data demand [63][64][65].

Data Compilation
The data obtained were compiled and analyzed using Excel ® , allowing the division of data according to the vaccination period: pre-or post-PCV introduction in the Brazilian immunization program.
In the case of penicillin and ceftriaxone (beta-lactams), from 2007 onwards, the evaluation parameters were divided into two groups: meningitis and non-meningitis [69].Data from articles with these definitions were added to their respective classifications (meningitis and non-meningitis).Articles that did not use the parameters listed above for beta-lactams were assigned to the general parameter column for better organization and analysis of the data.Furthermore, when the data were provided by the authors (raw data), in the case of penicillin specifically, originating from sources of colonization (non-invasive), the parameters of oral penicillin were used and the data were added in the general column; when invasive, meningitis and non-meningitis parameters were used.In the case of ceftriaxone, meningitis criteria were applied for invasive isolates and non-meningitis for colonization isolates.
Articles presenting data from a long time covering both periods (pre-and post-PCV10 introduction) had their data organized separately.
It is also noteworthy that among the works that required data supplementation, we received only the raw data from the article by Pinto et al. 2019 [33] on time for this study.In this context, the data were separated into three distinct groups according to the scope of this systematic literature review.

Statistical Analysis
We used a two-proportion Z-test to compare independent samplings, at a confidence level of 95%, and verify if the proportion of pneumococci non-susceptible to antimicrobial agents has significantly changed in the post-PCV10 period.

Ethical Aspects
All included studies were approved by their respective Ethics Committees.Other data were retrieved from a public database.

Conclusions
There is evidence that the proportion of isolates that are susceptible to chloramphenicol and sulfamethoxazole-trimethoprim is higher after PCV10 implementation for routine use in Brazil.More importantly, the same scenario was observed for penicillin and ceftriaxone among isolates associated with IPD.However, it is important to highlight the higher frequency of penicillin non-susceptible pneumococci associated with colonization in the post-PCV10 introduction period due to the emphasis on its use in the treatment of pneumococcal diseases.The emergence of macrolide-resistant isolates, associated with both colonization and diseases, is also a concern.Similarly, resistance to clindamycin and tetracycline is significantly higher among carriage isolates of the post-PCV10 period.On the other hand, susceptibility to other antimicrobial agents, such as ansamycins, fluoro-

Figure 1 .
Figure 1.Detailed flowchart for obtaining and selecting eligible articles for this systematic review.

Figure 1 .
Figure 1.Detailed flowchart for obtaining and selecting eligible articles for this systematic review.

Figure 2 .
Figure 2. The proportion of isolates susceptible and non-susceptible to antimicrobial agents according to isolation source ((a).carriage isolates; (b).invasive isolates) before and after the introduction of the 10-valent pneumococcal conjugate vaccine (PCV10) for universal use in Brazil (p-value was calculated using a two-proportion Z-test to compare independent samplings).

Figure 2 .
Figure 2. The proportion of isolates susceptible and non-susceptible to antimicrobial agents according to isolation source ((a).carriage isolates; (b).invasive isolates) before and after the introduction of the 10-valent pneumococcal conjugate vaccine (PCV10) for universal use in Brazil (p-value was calculated using a two-proportion Z-test to compare independent samplings).

Table 2 .
(a) Main results retrieved from articles with data of the pre-PCV10 period.(b) Main results retrieved from articles with data of the post-PCV10 period.(c) Main results retrieved from article with data of the extended period (covering pre-and post-PCV10 period) with raw data provision.

Table 3 .
Data related to antimicrobial resistance evaluated in the pre-PCV10 period divided into colonizing, non-invasive, and invasive isolates.

Table 4 .
Data related to antimicrobial resistance evaluated in the post-PCV10 introduction period divided into colonizing, non-invasive, and invasive isolates.

Table 5 .
(a) Data related to penicillin resistance in the pre-PCV10 period divided into meningitis, non-meningitis, and general parameters.(b) Data related to penicillin resistance in the post-PCV10 period divided into meningitis, non-meningitis, and general parameters.

Table 6 .
(a) Data related to ceftriaxone resistance in the pre-PCV10 period divided into meningitis, non-meningitis, and general parameters.(b) Data related to ceftriaxone resistance in the post-PCV10 introduction period divided into meningitis, non-meningitis, and general parameters.
N = number of isolates; I = intermediate; R = resistant; S = susceptible.The "general parameter" column refers to data prior to the change in interpretation criteria for beta-lactams or not specified in academic productions.

Table 7 .
Criteria used for article selection.Regional Surveillance System) is a compilation of data on Haemophilus influenzae, Neisseria meningitidis, and Streptococcus pneumoniae from Latin American countries since 2000.