Carbapenamase-Producing Acinetobacter baumannii in China, Latin America and the Caribbean: A Systematic Review and Meta-Analysis

INTRODUCTION Carbapenem-resistant Acinetobacter baumannii is a complex health problem, causing diﬃ culties in clinical–therapeutic management worldwide. It is of particular concern in Latin America, the Caribbean and China, where it is an emerging health problem. Carbapenemases produced by these organisms inactivate carbapenem antibiotics. Monitoring circulating genotypes’ geographic dispersion contributes to more eﬀ ective control measures. However, exhaustive studies on carbapenem-resistant A. baumannii are scarce. OBJECTIVES Study the production of carbapenemases in clinical isolates of A. baumannii resistant to carbapenem antibiotics and the geographic distribution of the sequences circulating in China, Latin America and the Caribbean. DATA ACQUISITION We followed PRISMA indications. We carried out a systematic search in Pubmed, BVS and CKNI on papers on A. baumannii and carbapenemases published during 2015–2020 in English, Spanish and Chinese, and selected 29 cross-sectional studies that met the search criteria. Studies were evaluated using JBI Critical Appraisal tools, and quantitative data were collated for meta-analysis using the Metaprop library in Stata15. Colombia. In total, 41 sequence types were identiﬁ ed; in Latin America and the Caribbean the most common types were: ST79, ST25, ST1 and ST15; in China, the sequences ST195, ST208, ST191, ST368 and ST369 were the most prevalent. ST2 was found in both regions. CONCLUSIONS The most prevalent carbapenemases and sequence types vary by region, indicating diﬀ erent ancestral strains. Microbiological surveillance, antibiotic use optimiza-tion, adequate infection treatment and timely control strategies are essential for carbapenem-resistant A. baumannii prevention and control in geographies such as Latin America, the Caribbean and China where such resistance is an emerging health problem.


INTRODUCTION
The genus Acinetobacter (Acinobacter spp.) is made up of several species. These gram-negative bacilli are among the most common nosocomial pathogens worldwide. Acinetobacter baumannii is the most clinically relevant species, due to its ability to develop various mechanisms that lend themselves to antibiotic resistance. [1] A member of the beta-lactam class (the same class of antibiotics as penicillins and cephalosporins), carbapenemic antibiotics are the last resort in treating A. baumannii infections.
There has been a worldwide increase in carbapenem resistance observed in clinical isolates of A. baumannii. [2] The Latin American Antimicrobial Resistance Surveillance Network found A. baumannii to have high resistance to carbapenems in 15 countries in the region during [2014][2015][2016]. The percentage of resistant isolates varied from 8% to 89%. [3] In China in 2016, 71.4% of A. baumannii isolates were resistant to carbapenems. [4] WHO published a list of priority pathogens in 2017, including A. baumannii, Pseudomonas aeruginosa, and carbapenem-resistant Enterobacteriacae spp. as critical priorities. [5] The main mechanism of carbapenem resistance in A. baumannii (CRAB) strains is carbapenemases production. The most common of these are Ambler's class D oxacillinases (OXAs). [2] Six subgroups of OXAs have been identifi ed in A. baumannii: the species' intrinsic carbapenemase OXA-51-like, and the acquired carbapenemases OXA-23-like, OXA-24-like, OXA-58-like, OXA-143-like, and OXA-235-like. [6] Class B metallobetalactamases (MBLs) are also a major threat because they are often located in mobile genetic elements, easily transferable between bacteria. Four types of MBLs are frequently detected in A. baumannii: imipenemase (IMP), Verona imipenemase (VIM), Seoul imipenemase (SIM), and New Delhi betalactamase (NDM). [7] Molecular characterization of A. baumannii isolates is very useful in identifying the source of an outbreak and in helping to control its spread. Multilocus sequence typing (MLST) is highly discriminative and has been applied successfully to several bacterial patho-

Review Article
Peer Reviewed gens, including A. baumannii. Additionally, this typing allows for comparisons between laboratories and provides a powerful tool for conducting epidemiological studies worldwide. [8] The emergence of carbapenem-resistant forms of A. baumannii is a complex global problem, diffi cult to manage both clinically and therapeutically. Monitoring carbapenemase genotypes and molecular epidemiology studies contribute to more eff ective control measures. However, comparative analyses of circulating forms in diff erent geographies are scarce. This work is a systematic review of published information on carbapenemase production and sequence types characterized in A. baumannii isolates in China, Latin America and the Caribbean (LAC).

Inclusion criteria
We included studies with the following characteristics: 1) observational studies with a cross-sectional design; 2) analysis of A. baumannii clinical isolates in adult populations; 3) reports on CRAB; 4) detection of class A, B and D carbapenemases by molecular methods such as reverse-transcriptase polymerase chain reaction testing (RT-PCR) or MLST; 5) carbapenemase genotype analysis in LAC or China published in 2015-2020 in English, Spanish or Chinese.
Exclusion criteria Studies with one or more of the following characteristics were excluded: 1) contained no information on CRAB; 2) studied isolation in pediatric populations; 3) did not report on class A, B and D carbapenemase classes; 4) duplicated other research; and 5) studies that reported experiments in nonhuman subjects or were review articles, conference abstracts, meta-analyses or systemic reviews.
Review articles and meta-analyses were only considered in this review's discussion section. Screening for inclusion was carried out individually by two team members. When there were diff erences of opinion, these were discussed with the principal investigator. Endnote X9 and Excel were used to manage references.
Study quality evaluation Study quality was evaluated by two researchers using the JBI Critical Appraisal Tools for Prevalence Studies. [11] This tool includes nine items, each of which was scored as either 'Yes' (when the requirement was met), as 'No' (when the requirement was not met) or as 'Not Clear' (if it was unknown whether the requirement was met) or as 'Not Applicable'. [11] Studies were considered high quality when their score was ≥80% of the maximum possible score (8 or more items scored "Yes"), average quality when their score was 70%-79% (6-7 items scored 'Yes'), and low quality when their score was <70% (6 or fewer items scored 'Yes').
Data extraction Two team members carried out data selection and extraction individually, as well as bias risk analysis. We organized data on carbapenemase genotypes in China, Latin America and the Caribbean into a matrix (Table 1).

Data analysis and statistics
We carried out a quantitative study (meta-analysis) on A. baumannii carbapenemase genotypes and a qualitative study (qualitative descriptive synthesis) for sequence types, due to the great diversity of sequence types and diff erences between geographical areas. Two investigators undertook data analysis for the quantitative study. Statistical analysis was performed using the Metaprop module of Stata version 15 (Stata-Corp LLC U.S.A), [12] and obtained estimates of the combined prevalence of predominant A. baumannii carbapenemases in different regions, as well as their 95% confi dence intervals (95% CI), which are represented in corresponding forest graphs ( Figure 2).
We used either fi xed-eff ect or random-eff ect models according to statistical heterogeneity between studies, which was assessed using the Cochran I2 statistic (a value of 0% indicates no heterogeneity; 25%, 50% and 75% are considered to have low, medium and high heterogeneity, respectively). Egger's weighted linear regression test, combined with a funnel plot, was used to assess publication bias. An assessment of 'no publication bias' was made when the regression line started from the origin of the ordinate axis (Y) (publication bias increases as the line moves away from the Y coordinate's origin). Statistical signifi cance was assessed at 0.1 and not 0.05. [13] DEVELOPMENT Literature search, selection and validation We initially selected 334 articles, which were reduced to 318 after eliminating duplicates. Of these, 261 were excluded for failure to meet inclusion criteria or because they were outside the scope of this review. Finally, we fully reviewed a total of 57 articles, and 29 were selected that met all established criteria, which were then included in the meta-analysis ( Figure 1).
Medium-high scores were obtained upon evaluation of the crosssectional observational studies (Table 2). In this study, 95% is According to the I2 values, high heterogeneity was observed among studies and we consequently used a random-eff ects model for the meta-analysis. Egger tests (p >0.1) and funnel plots show the characteristic shape of the asymmetric dispersion (Table  3 and Figure 2).

Molecular typing characteristics
In 17 of the 29 studies, MLST was performed on CRAB isolates. In total, 41 sequence types (STs) were identifi ed: 16 in LAC and 26 in China. Clear geographical diff erences were observed in predominant ST frequencies: ST79, ST25, ST1 and ST15 were more common in LAC; while ST195, ST208, ST191, ST368 and ST369 were found more frequently in China. ST2 was found in both (Table 2 and Figure 3).

DISCUSSION
Carbapenemase types Antibiotic overuse has led to an increase in multi-drug-resistant Acinetobacter. More than 50% of Acinetobacter spp. isolates in the United States, South America, India and China are resistant to carbapenem antiobiotics. [ [47] OXA-24-like carbapenemases (including OXA-24, 25, 26, 40 and 72) have been found in both plasmid and chromosomal structures; OXA-58-like and OXA-23-like carbapenemases are encoded by plasmids, [48] which increases the probability of horizontal transmission. The plasmid-encoded carbapenemases OXA-23, OXA-24 and OXA-58 were the most frequently isolated carbapenemases in the two regions analyzed in this study. These results justify the noninclusion of OXA-51 type carbapenemases, as their resistance to CRAB is intrinsic, and thus has little impact on amplifying carbapenem resistance in this species.
Carbapenemase OXA-72 was fi rst identifi ed in 2004 in an A. baumannii isolate from Thailand, [49] and subsequently detected in Brazil, Mexico, Ecuador, Peru, China and Europe. [27,43,45,50,51] PXA-231 and OXA-253 were identifi ed in Brazil and Peru, respectively. Both belong to the OXA-143-like group. OXA-231 and OXA-253 were reported the fi rst time in A. baumannii isolates from Brazil (in 2007 and 2014, respectively) and still appear mainly in that country. [52,53] The dissemination of this subgroup requires monitoring, as it has been described more recently in Iran (2017), Colombia (2017) and Peru (2018). [45,54,55] Peer Reviewed used as expected prevalence in calculating sample size, as carbapenemase production is the CRAB's dominant cause. Considering three studies on molecular isolate typing in hospitals, [14][15][16] the confi dence level equal to 95% and precision equal to 5%, the estimated minimum CRAB sample size was 73. Consequently, we excluded 14 articles due to small sample size. Another 14 studies on carbapenemase genoptyping were excluded for only reporting detection of Class D carbapenemases.  Table 3). Only one clinical isolate Class A carbapenemase producer (KPC) was found, and it was isolated in Colombia.

Yes --High
All were cross-sectional studies. & : Quality criteria: 1. The sample was appropriate to address the target population, 2. The sample was obtained using an adequate method, 3. The sample size was adequate 4. Participants and the context are described in detail 5. Data analysis was carried out with suffi cient coverage of the identifi ed sample 6. Eff ective methods were used to identify diseases or health problems 7. The sample was measured using standard and reliable methods for all participants 8. The statistical analysis was appropriate 9. Response rate was adequate or low response rate was adequately managed. n: number of times the sequence was found; N: total isolates studied; NA: Not applicable (detection not performed); ND: Not detected; X: did not meet the requirement; -: met the requirement.

64
MEDICC Review, January 2022, Vol 24, No 1 Peer Reviewed   The diff erences in OXA and MBL prevalence between countries is likely due to pressures of antimicrobial selection, horizontal transfer of carbapenemase genes by mobile genetic elements (plasmids) between species, propagation of clones carrying these genes or to a combination of all these factors.

Review Article
Detected sequence types MLST is considered the gold standard for detecting bacterial sequence types and is highly useful in epidemiology. This review includes 17 studies based on MLST.
Due to the great diversity in sequence types and the diff erences between countries and provinces, we only carried out a descriptive qualitative study on these articles (Figure 3). In the case of LAC, these studies mainly originated in South America (namely in Brazil, Peru, Argentina, Colombia, Chile, Bolivia, Ecuador, Paraguay and Uruguay). The most common sequence types in the region were ST79, ST25, ST1 and ST15. ST79 had the widest geographic spread, and was detected in Ecuador, Peru, Brazil, Paraguay, Uruguay and Argentina, [34,35,39,43,45]  baumannii clones ST191 and ST369 has also been reported in China and South Korea. [67,68] The STs that are the most common in China are rarely reported in LAC; only ST369 has been reported in Mexico in 2020. [69] This may be the result of diff ering clonal ancestries in both regions and selection pressure from diff erent antimicrobial use habits.
ST2 was found in both regions (Peru in LAC and Liao Ning in China). ST2 is prevalent in some countries in Pacifi c Asia and Europe, [70][71][72][73][74] but has not been reported with any frequency in LAC in the last fi ve years. Although this study found ST2 to be associated with OXA-72, other studies have described ST2 A.
This study has provided estimates of the prevalence of diff erent types of carbapenemases in A. baumannii in LAC and China, and the geographical distribution of diff erent circulating CRAB STs and supports adjustments to resistance surveillance programming and antimicrobial management.
Study limitations There were some Latin American and Caribbean countries and Chinese provinces that produced no studies that met our selection criteria. Consequently, there is no information on carbapenemase-producing CRAB strain genotypes in these regions. Additionally, most studies do not detect Class A carbapenemases. This exclusion may have led to an overrepresentation of other carbapenemase classes and this preferential study of certain classes may have introduced bias. Finally, while the Egger test in not statistically signifi cant and publication bias is unlikely, there is considerable heterogeneity between studies, as suggested by the asymmetric and sparse shape of the funnel plots. This could be due to insuffi cient sample size in the included studies, or to some variation (either geographical or temporal) in the strains or other methodological aspects such as diff erences in sample inclusion and exclusion criteria, among other factors.

CONCLUSION
A. baumannii resistance to carbapenem antibiotics is a global threat, and there is increasing diff usion of carbapenem-producing A. baumannii isolates and increasing geographic ST variation. Enzymes produced by regional isolates are generally very simi- America and the Caribbean. Diff erent regions have diff erent epidemic CRAB strains, and it is important to consider local epidemiology in order to best tailor patient treatment. Studying circulating enzymes, clones and genetic lines facilitates understanding of region-specifi c resistance mechanisms.
Multidisciplinary collaboration is necessary (microbiologists, clinicians, epidemiologists and specialists in preventive medicine with experience in infection control), and will allow for early detection and study of resistance using molecular methods; adequate treatment, taking into consideration the patient's clinical status, common circulating strains, and infection characteristics; and adoption of epidemiological measures aimed at multi-drug-resistant infection prevention and control, reducing transmission at community-and hospital levels. A multidisciplinary approach can provide better results in managing a complex, multifactorial problem with major implications for public health.