Prevalence and trends of aminoglycoside resistance in Shigella worldwide, 1999-2010

Shigellosis causes diarrheal disease in humans in both developed and developing countries, and multi-drug resistance in Shigella is an emerging problem. Understanding changing resistance patterns is important in determining appropriate antibiotic treatments. This meta-analysis systematically evaluated aminoglycoside resistance in Shigella. A systematic review was constructed based on MEDLINE and EMBASE databases. Random-effect models or fixed-effect models were used based on P value considering the possibility of heterogeneity between studies for meta-analysis. Data manipulation and statistical analyses were performed using software STATA 11.0. By means of meta-analysis, we found a lower resistance to three kinds of aminoglycosides in the Europe-America areas during the 12 year study period than that of the Asia-Africa areas. Kanamycin resistance was observed to be the most common drug resistance among Shigella isolates with a prevalence of 6.88% (95%CI: 6.36%-7.43%). Comparison of data from Europe-America and Asia-Africa areas revealed that Shigella flexneri resistance was greater than the resistance calculated for Shigella sonnei. Importantly, Shigella sonnei has played a significant role in aminoglycoside-resistance in recent years. Similarly, data showed that resistance to these drugs in children was higher than the corresponding data of adults. In conclusion, aminoglycoside-resistant Shigella is not an unusual phenomenon worldwide. Distribution in Shigella resistance differs sharply based on geographic areas, periods of time and subtypes. The results from the present study highlight the need for continuous surveillance of resistance and control of antibiotic usage.


INTRODUCTION
Acute gastroenteritis and diarrheal diseases continue to be a health problem worldwide, especially in developing countries. They account for approximately 2.5 million deaths per year in children < 5 years of age [1,2] . Worldwide, the most common bacterial pathogens causing these diseases are: Salmonella spp, Shigella (S.) spp, Campylobacter spp, Escherichia coli O157:H7, Listeria monocytogenes, Vibrio cholerae, and Yersinia enterocolitica [3,4] . The common route of infection by these pathogens is the ingestion of contaminated food and drinks [5] . Infection by Shigella species is an important global public health problem [6] . Shigella infections, especially S. flexneri and S. sonnei infections, can lead to illness ranging from mild, self-limited diarrhea to severe dysentery with frequent passages of blood and mucus, high fever, cramps, tenesmus, and in rare cases, bacteremia. Complications of shigellosis are seen most frequently in children, the elderly, and the immunocompromised. Therefore, shigellosis is recognized by the World Health Organization (WHO) as major global public health concern [7,8] .
Prompt treatment with effective antimicrobial agents shortens the duration of symptoms and carriage, and reduces the spread of infection. However, antimicrobial resistance has complicated the selection of empirical agents for the treatment of shigellosis, particularly in children. Shigella isolates often showed resistance to commonly used, inexpensive antimicrobials, including ampicillin, piperacillin, trimethoprim-sulfamethoxazole, thereby drastically reducing therapeutic possibilities. Thus, the use of sulfonamide or β-lactam antibiotics would not be appropriate for empirical treatment of shigellosis. Shigella strains have become progressively resistant to multiple antimicrobial agents, initially to sulfonamides [9,10] , shortly after they became commercially available; resistance to tetracycline, chloramphenicol, and streptomycin was seen less than 10 years after each was introduced, with subsequent resistance to ampicillin, kanamycin, and trimethoprim-sulfamethoxazole [11,12] . In certain eastern Africa populations and in a study from China, aminoglycoside resistance of Shigella is a common finding [13,14] .
The present study aimed to identify the worldwide prevalence and distribution of aminoglycosideresistant Shigella using meta-analysis based on data gathered from a systematic review of articles reported between January 1999 and July 2012. The relevant estimates were evaluated for new cases and previously treated cases, respectively, which could provide a clear profile for the status of aminoglycoside-resistant Shigella globally.

Literature identification
We conducted a computerized search of MEDLINE (January 1999--July 2012) and EMBASE (January 1999--July 2012) to identify all reports on aminoglycosides resistance associated with Shigella infections. The following keywords were used in searches: "bacterial surveillance" or "antimicrobial resistance" or "bacterial resistance" and "Shigella" [15] . We also attempted to identify potentially relevant articles by checking the references of the germane articles and through personal communications with colleagues.

Inclusion and exclusion criteria
Two investigators (BG and XK) reviewed potentially appropriate studies independently, to determine whether they met predetermined eligibility criteria. Disagreements between the reviewers were resolved by consensus. Studies obtained from the literature search were checked by title and citation. If an article appeared relevant, the abstract was reviewed. Relevant abstracts were examined in full text. The inclusion and exclusion criteria were established by the investigators prior to review of the literature. The inclusion criteria were as follows: original article, short communication, correspondence or letter which provided sufficient original data, and all strains isolated from stool. Studies were excluded if they met the following conditions: (1) review or case report; (2) not 1999--2010 data; (3) not separated by country/region; (4) non-human bacterial source; (5) did not include study drugs; (6) did not include resistance results for study pathogens; (7) inclusion/exclusion criteria were not presented; (8) non-recommended regimens/dosing; (9) susceptibility results were not presented. Before we excluded the studies, authors of such studies were contacted in an effort to obtain missing data.

Validity assessment
Studies were assessed for quality and only high quality studies were included for analysis. Characteristics of high quality studies were: prospective cohort, retrospective consecutive cohort; provided basic data including study period and area, total tested numbers and resistant numbers; susceptibility test was performed in accordance with guidelines established by the Clinical and Laboratory Standard Institute (CLSI) [16] ; reported at least one of three antimicrobials (gentamicin, kanamycin and amikacin) with quality control; in-dividuals included in studies had no infections other than bacillary dysentery. Only one representative case for each outbreak was included, unless the isolates had different antibiotic susceptibility patterns. When study strains overlapped, we included strains from the more recent and larger study in the analysis. If the strains from the smaller study provided data that was not reported in the larger study, results were included for that specific variable.

Data extraction and statistical analysis
Data extraction was performed by two reviewers (BG and XK) using a standardized extraction form. When there was disagreement, the relevant paper was reviewed and differences were resolved by consensus. Microsoft Excel (version 12.0) software was used for data entry and analysis. In our review, considering the possibility of significant heterogeneity between studies which were tested with the Q test (P < 0.10 was considered indicative of statistically significant heterogeneity), random effect models or fix effect models were chosen by P value for meta-analysis. Freeman-Tukey arcsin transform to stabilize variances, and after the meta-analysis, investigators can transform the summary estimate and the CI boundaries back to proportions using sin function. Specific conversion details were previously described [17] . Data manipulation and statistical analyses were undertaken using the Statistical Software Package (STATA) 11.0 (STATA Corporation, College Station, TX, USA).

Studies and endpoints
We reviewed 3,176 publications from MEDLINE and EMBASE reported from 1999 to 2012. Candidate articles are shown in Fig. 1. After exclusion based on title and abstract evaluation, 580 articles were retrieved for detailed, full-text evaluation. As shown in Fig. 1, among the included articles, 46 studies were reviews or case reports. Ninety-one articles did not use data that was within the 12-year study period. Findings in 18 articles were not separated by country/ region. Human or resistance results for study pathogens were not presented in 28 and 69 studies, respectively. Detailed results of drug susceptibility testing (DST) with respect to study drugs were not provided in 181 studies. Recommend regimens, recommended dosing or data on minimal inhibitory concentrations (MIC) was not included in 9, 42, and 28 studies, respectively. Finally, 68 studies, addressing the prevalence of aminoglycoside-resistant Shigella in new cases or in previously treated cases, were identified.  3.59%-4.22%) (n/N= 937/14,059), 6.88% (6.36%-7.43%) (n/N=1,106/8,647) and 1.29% (0.97%-1.68%) (n/N=432/8,614), respectively. Importantly, evident heterogeneity was observed (P < 0.001). In the stratified analyses, the prevalence of any drug resistance was observed to vary by geographic areas, study years and subtypes. Lower rates were observed for studies from Europe-America and the period of 1999 to 2004, while the rates from Asia-Africa and using the subtypes of S. flexneri were higher. The end time for enrollment of the cases (after 2008) did not significantly change the results.

Status of aminoglycoside-resistant Shigella
The most common drug resistance was observed for kanamycin. Among kanamycin resistance, the highest drug resistance rate by geographic areas was found in Asia with a prevalence of 16.78% (7.58%-28.71%). Similarly, the most common resistance was observed for 2005-2007 and S. flexneri with a summarized combined prevalence of 12.05% (11.18%-14.21%) and 9.25% (7.69%-10.96%), respectively.

Studies considered for primary analysis
A total of 68 reports were included in the metaanalysis. As specified a priori in our analysis plan, meta-analyses were performed for outcomes in which there were one or more observations of pediatric or adult group resistance that could be aggregated in the three drugs. Analyses were conducted across the geographic areas, study years and different subtypes for selected primary endpoints, including the resistance of Shigella to gentamicin, kanamycin and amikacin; S. flexneri to gentamicin, kanamycin and amikacin; S. sonnei to gentamicin, kanamycin and amikacin (Supplementary Table 1 available online).

Resistance patterns in European-American
and Asian-African countries Gentamicin Table 2 shows the resistance rates of total Shigella isolates in European-American and Asian-African countries. A lower prevalence of gentamicin resistance was found in European-American countries at 0.68% (0.39%-1.05%). After analyzing the study data on years, we observed a minimal change in the resistance prevalence of gentamicin, from 0.25% (0.04%-0.64%) to 0.84% (0.08%-2.40%) in Europea-American countries, in contrast to data in Asian-African countries, which fluctuated from 6.05% (1.18%-14.28%) to 20.83% (12.67%-30.40%). It is worth noting that the resistance prevalence of gentamicin increased annually in Asian-African countries, while the resistance prevalence decreased year by year in European-American countries.

Kanamycin
The kanamycin resistance calculations of Shigella isolates among different areas are shown in Table 2. The prevalence of gentamicin resistance in Asian-African countries increased sharply from 14.00% Amikacin Table 2 compares the amikacin resistance of Shigella isolates between European-American and Asian-African countries. In European-American regions, a lower amikacin resistance was also found during the 12-year study period. In fact, amikacin resistance decreased from 0.28% (0.00-1.08) to 0.05% (0.04-0.40). The highest resistance of Shigella isolates to amikacin was only 0.28% (0.00%-1.08%). We observed that the prevalence of amikacin resistance remarkably increased from 6.39% (1.40%-14.63%) to 48.06% (34.57%-61.65%) in Asian-African countries.

Comparison between S. flexneri and S. sonnei
Comparison of the data from Europe-America or Asia-Africa revealed that S. flexneri resistance to gentamicin [0.65% (0.30%-1.14%); 9.66% (5.51%-14.81%)] was greater than the resistance calculated for S. sonnei [0.39% (0.21%-0.61%); 13.66% (7.72%-20.96%)] ( Tables 3, 4  Comparison between children and adults Table 6 shows the clear relationship between the resistance rate and the age of patients with diarrhea. Strains explicitly isolated from children were naturally classified into the pediatric group, whereas the remaining strains were viewed as isolates from adults. For Shigella, an additional analysis was conducted for a pediatric group population (34 studies).
In this pediatric group, the resistance of Shigella to gentamicin was higher than that among the adult group population [5.93% (3.97%-8.23%) and 18.34% (9.81%-28.76%)]. Kanamycin resistance in the pediatric group was significantly higher than that in the adult group, which showed 70.72% (33.95%-96.25%) versus 5.40% (1.87%-10.62%) for kanamycin. Similarly, greater resistance to amikacin was shown in the pediatric group than in the adults group [ Table 4 Rates of resistance to kanamycin reported for S. flexneri and S. sonnei isolated from Europe-America and Asia-Africa during 1999-2010 1.7 million episodes of bacillary dysentery annually, with up to 200,000 patients admitted to hospitals [18] . Any of four subtypes of Shigella (S. dysenteriae, S. flexneri, S. boydii, and S. sonnei) can cause shigellosis. Children are at a higher risk of being affected by the disease, which might be a reflection of secondary infection from the adults as well as poor personal hygiene [19] . Because of increasing antimicrobial resistance to Shigella, empiric treatment options are dwindling. In a recent study, aminoglycosides showed higher in vitro activity against members of the family Enterobacteriaceae. The function of aminoglycoside antibiotics for empirical treatment of patients with serious infections caused by Shigella merits our attention.
As with other classes of antibiotics, significant differences in the spectrum of antimicrobial activity exist among various aminoglycosides [20] . Aminoglycosides bind to the bacterial ribosome and inhibit protein synthesis. Generally, newer aminoglycosides, such as gentamicin, tobramycin, amikacin, netilmicin, isepamicin, dibekacin, and arbekacin, have broader spectra of activity than older compounds like streptomycin and kanamycin. Aminoglycosides are often administered in combination with other antibacterial agents. Despite their potential nephrotoxicity, ototoxicity and problems associated with aminoglycoside-resistant organisms, aminoglycoside antibiotics remain valuable and sometimes indispensable for the treatment of various infections and prophylaxis in special situations [21] . Several mechanisms have been proposed for bacterial resistance to aminoglycoside antibiotics, including decreased antibiotic uptake and accumulation, modification of the ribosomal target, efflux of antibiotic, and enzymatic modification of aminoglycosides.
The resistance of clinical isolates to aminoglycoside antibiotics varies with the specific drug, the microorganism, mechanism of resistance, geographic area, and many other factors. Of the 580 articles and abstracts fully evaluated for this meta-analysis, only 68 published reports met our strict inclusion criteria. Several studies which we reviewed were not included because they did not report on the outcomes of the study drugs in the study period or did not meet other inclusion criteria. However, use of a statistical tool to assess potential heterogeneity among included studies (STATA) gave us further confidence in the metaanalysis results. This meta-analysis demonstrated that the resistance to aminoglycoside is a significant issue in Asia and Africa, as well as in Europe and America. Reported aminoglycoside resistance in Shigella varies greatly from country to country and is likely to be an important problem in certain regions. The goals of our study were to analyze the distribution of antimicrobial resistance associated aminoglycoside-resistant Shigella based on articles reported between January 1999 and December 2010 and to examine issues related to N/A: results not available. reasonable treatment about shigellosis. The first noteworthy finding of this study was that all of these three drugs, in the European-American regions, had an obviously lower resistance rate during the 12 years than in the Asian-African regions. Regardless of its origins and mechanisms, the widespread resistance that we found among S. flexneri and S. sonnei suggests that infections due to drug-resistant Shigella are now endemic around the world. Although mild illness can resolve without antimicrobial treatment, current recommendations guide clinicians to treat Shigella infections with antimicrobial therapy to reduce the duration and severity of clinical symptoms and decrease the shedding period of Shigella. Because isolation and antimicrobial susceptibility testing for Shigella take only several days, antimicrobial agents were generally selected empirically. Combined with the characteristics of antibiotics use in different areas, we found that the use trends of aminoglycoside for shigellosis were not very serious. Selected literature reported from America [22,23] showed that trimethoprimsulfamethoxazole and ampicillin can no longer be considered appropriate empirical therapies. A Chinese article found that [18,24] , for the treatment of shigellosis, 35 to 56 different antibiotics were used, with penicillin, cephalosporin, and macrolides accounting for the largest total volume. In some cases, aminoglycoside is effective and approved for use as treatment of Enterobacteriaceae. Stronger antibiotics like aminoglycoside may also present notable side effects. Rising resistance rates of these stronger antibiotics have a major impact on the ability of physicians to treat common infections. Patients, especially from Asian-African regions, faced more severe infections with increased duration as resistance increases. They may also experience heightened toxicity associated with the use of stronger antibiotics. Clinicians eventually encountered infections caused by highly resistant pathogens for which no effective antibiotics are available.
To compare the resistance difference among Shigella species, additional clinical data from large-scale observational studies were needed to evaluate the link between in vitro aminoglycoside resistance and different subtypes. We found that S. flexneri resistance to these antibiotics was greater than the resistance calculated for S. sonnei, and this effect was not influenced by district. In some hospital-based surveillance studies, S. flexneri infections have been more frequently detected [25] . One explanation for this could be that S. flexneri infections result more frequently in hospitalizations than S. sonnei infections. The notion that S. sonnei is less virulent than S. flexneri is supported by the possibly shorter duration of diarrhea occurring among patients infected with S. sonnei than among those with S. flexneri. Moreover, the molecular mechanisms of different antimicrobial resistance rate in Shigella may be associated with the function of integrons. In Shigella species, antimicrobial resistance is often associated with the presence of class 1 and class 2 integrons that contain resistance gene cassettes. Multiple and complex expression regulation mechanisms involving mobile genetic elements in integrons have been developed in the evolution of Shigella strains. S. sonnei and S. boydii strains often contain a single integron of class 2, whereas S. flexneri and S. dysenteriae strains carry a class 1 integron, either alone or associated with a class 2 integron [26,27] . Finally, S. flexneri could be linked with more complex drug resistant genes. This may be one of the explanations why S. flexneri resistance was greater than S. sonnei. Information about the rates of Shigella-resistance to kanamycin and amikacin in Asian-African regions was not reported, which stresses the need for continuous surveillance of resistance in those countries.
The third significant finding is that, in some countries, the prevalence of S. sonnei resistance was greater in 2008-2010. Obviously, the resistance rates among S. sonnei have an increasing trend. In some areas, S. sonnei has become the primary cause of shigellosis. Some of the S. sonnei isolates recovered showed resistance to several kinds of antimicrobial agents. There are distinct phenotypic and genotypic differences in terms of biotypes, antimicrobial susceptibilities and PFGE profiles, and antimicrobial susceptibilities. [70,72,73] . An increasing trend in the use of strong antibiotics, such as gentamicin and kanamycin, for shigellosis might be responsible for the acquisition of resistance to these antibiotics in S. sonnei isolates. The increased prevalence of S. sonnei resistance prompted us to suspect that S. sonnei may play a significant role in aminoglycosides-resistance in the future. It is essential to curb the spread of antibiotic resistance and diffusion of S. sonnei should be prevented. Overall, the surveillance of antimicrobial resistance of S. sonnei isolates should be continued, particularly to monitor the emergence of strains fully resistant to aminoglycosides. Thus, analyses of subtypes-based experiences in Shigella resistance can provide an important contribution to the understanding of real-world resistance issues from the perspective of day-to-day medical practice. This analysis also tells us that we should protect against S. flexneri, which caused more hospitalizations in the past, as well as S. sonnei, which caused the majority of shigellosis cases in this study.
Our fourth finding was that, in the pediatric group, the resistance to aminoglycosides was higher than in the adult group population. For children and adults with acute infectious gastroenteritis, the use of specific antimicrobial therapy should be limited to welldefined bacterial and protozoal agents. However, the use of antimicrobial agents in humans for many conditions, including therapy for children with diarrheal disease, is widespread. Although aminoglycosides showed higher in vitro activity against members of the family Enterobacteriaceae, antimicrobial therapy should be considered for specific clinical circumstances including the safety and tolerability of antimicrobial agents, particularly in young children. Considering potential side effects associated with aminoglycoside antibiotics to children, aminoglycoside would not be the best therapeutic choice for gastrointestinal diseases of children. For empiric treatment of diarrheal infections among children, in the report of Abu Elamreen et al. [28,29] , ampicillin and trimethoprim-sulfamethoxazole are most often used. In the meantime, the greater aminoglycoside resistance rates to Shigella in the pediatric group cannot be ignored. This meta-analysis provides an important synthesis of the reported aminoglycoside resistance rates for S. flexneri and S. sonnei. Based on these findings, aminoglycoside resistance is consistently present in a variable proportion of multiple populations. We found that Shigella showed greater resistance to gentamicin, kanamycin or amikacin in the pediatric group than it did in the adult group. As interest evolves in the resistance patterns and rates of Shigella to aminoglycoside antibiotics, these results can be used to guide treatment decisions and to formulate consensual recommendations for appropriate treatment paradigms, especially for children. The meta-analysis technique used here can help to develop appropriate guidelines governing antibiotic use and to monitor drug resistance trends in different population groups.
Analysis of data on the use of various aminoglycoside antibiotics in different countries and regions of the world indicates that a correlation exists between the selective pressure of antibiotics and the patterns of combinations of aminoglycoside resistance mechanisms. For example, gentamicin has been most frequently used in the USA, while amikacin was used more extensively in Japan. In that time, the significant mechanisms of aminoglycoside resistance in the USA were production of ANT(2")-I (resistance to gentamicin, tobramycin, dibekacin, and kanamycin), and AAC(3)-I (resistance to gentamicin), whereas in Japan, Europe and Latin America, in addition to ANT(2")-I, AAC(6')-I (resistance to amikacin, netilmicin, tobramycin, dibekacin, and kanamycin but not to gentamicin) was identified [20,30] . The epidemiology of aminoglycoside resistance is becoming more complex, in part because of the multitude of aminoglycosidemodifying enzymes that exist for these antibiotics and also from the presence of disparate additional mechanisms for antibiotic resistance other than enzymatic resistance determinants. Because the genes for the aminoglycoside-modifying enzymes are often located on plasmids or transposons, together with the genes encoding resistance to other classes of antibacterials, the total consumption of non-aminoglycosides can also significantly influence the epidemiological features of aminoglycoside resistance [20] .
In summary, this meta-analysis has provided important information on resistance by S. flexneri and S. sonnei to aminoglycosides in European-American and Asian-African countries. The use of the metaanalysis technique has allowed us to summarize data from individual studies and to determine robust values for both overall resistance and resistance among subgroups. Because identifying resistance patterns can be informative for empiric treatment recommendations, these results will be helpful in developing future guidelines and treatment paradigms for S. flexneri and S. sonnei, as well as in helping to direct future research on the impact of bacterial resistance and appropriate antimicrobial use.