Primary antibiotic resistance in Helicobacter pylori in the Asia-Pacific region: a systematic review and meta-analysis

Articles Primary antibiotic resistance in Helicobacter pylori in the Asia-Pacific region: a systematic review and meta-analysis Yu-Ting Kuo*, Jyh-Ming Liou*, Emad M El-Omar, Jeng-Yih Wu, Alex Hwong Ruey Leow, Khean Lee Goh, Rajashree Das, Hong Lu, Jaw-Town Lin, Yu-Kang Tu, Yoshio Yamaoka, Ming-Shiang Wu, for the Asian Pacific Alliance on Helicobacter and Microbiota Summary Lancet Gastroenterol Hepatol 2017; 2: 707–15 Background So far, a comprehensive systematic review and meta-analysis has not been done of the prevalence of primary antibiotic resistance in Helicobacter pylori in the Asia-Pacific region. We aimed to assess the trends and Published Online regional differences in primary antibiotic resistance to H pylori in the Asia-Pacific region and to examine the relation August 3, 2017 between resistance and first-line eradication. http://dx.doi.org/10.1016/ S2468-1253(17)30219-4 Methods We did a systematic review and meta-analysis of primary antibiotic resistance to H pylori and the efficacy of first-line regimens in the Asia-Pacific region. We searched PubMed, Embase, and the Cochrane Library for articles See Comment page 692 published between Jan 1, 1990, and Sept 30, 2016; we also searched abstracts from international conferences. Both observational studies and randomised controlled trials were eligible for inclusion in the analysis of primary antibiotic *These authors contributed resistance, but only randomised controlled trials were eligible for inclusion in the analysis of efficacy of first-line equally therapies. Meta-analysis was by the random-effects model to account for the substantial variations in resistance across the region. We did subgroup analyses by country and study period (ie, before 2000, 2001–05, 2006–10, and 2011–15) to Department of Internal establish country-specific prevalences of primary antibiotic resistance and first-line eradication rates. This study is Medicine, National Taiwan registered with PROSPERO, number CRD42017057905. University Hospital Bei-Hu Branch, National Taiwan Findings 176 articles from 24 countries were included in our analysis of antibiotic resistance. The overall mean University College of Medicine, prevalences of primary H pylori resistance were 17% (95% CI 15–18) for clarithromycin, 44% (95% CI 39–48) for Taipei, Taiwan (Y-T Kuo MD); metronidazole, 18% (95% CI 15–22) for levofloxacin, 3% (95% CI 2–5) for amoxicillin, and 4% (95% CI 2–5) for Department of Internal tetracycline. Prevalence of resistance to clarithromycin and levofloxacin rose significantly over time during the period Medicine, National Taiwan investigated, whereas resistance to other antibiotics remained stable. 170 articles from 16 countries were included in University Hospital, National analysis of efficacy of first-line therapies. We noted unsatisfactory efficacy (ie, <80%) with clarithromycin-containing Taiwan University College of regimens in countries where the clarithromycin resistance rates were higher than 20%. Medicine, Taipei, Taiwan (J-M Liou MD, Prof J-T Lin MD, Interpretation The prevalence of primary antibiotic resistance varied greatly among countries in the Asia-Pacific Prof M-S Wu MD); Department region, and thus treatment strategy should be adapted relative to country-specific resistance patterns. of Medicine, St George & Clarithromycin-containing regimens should be avoided in countries where the prevalence of clarithromycin Sutherland Clinical School, resistance is higher than 20%. University of New South Wales, Sydney, NSW, Australia Funding Ministry of Health and Welfare of Taiwan, Ministry of Science and Technology of Taiwan, and Amity (Prof E M El-Omar MD); University. Department of Internal Medicine, Kaohsiung Municipal Introduction has been reported in some reviews, but these reviews Hsiao-Kang Hospital, were not systematic and therefore were prone to selection Kaohsiung Medical University, Gastric cancer is the third leading cause of cancer-related bias.12 Therefore, we did a systematic review to provide an Kaohsiung, Taiwan mortality worldwide, and the Asia-Pacific region accounted overview of trends in antibiotic resistance in the past (J-Y Wu MD); Department of for about two-third of deaths from gastric cancer in 2012.1 25 years in different countries in the Asia-Pacific region, Medicine, Faculty of Medicine, Epidemiological studies2–4 suggest that up to 90% of and to assess the relation between resistance and University of Malaya, Kuala gastric cancer could be attributed to Helicobacter pylori eradication rates with first-line antibiotic therapies. Lumpur, Malaysia infection, and that eradication of this pathogen could (A H R Leow MD, reduce the incidence of gastric cancer. Screening for, and Methods Prof K L Goh MD); Amity eradication of, H pylori is thus a promising strategy to Institute of Biotechnology, eliminate gastric cancer in the region. However, the Search strategy and study selection Amity University, Noida, Uttar efficacy of standard clarithromycin-based triple therapy in In the first part of the systematic review, we focused on Pradesh, India (R Das PhD); eradicating infection has fallen below 80% in many primary H pylori resistance to antibiotics. We searched Gastrointestinal Division, Renji Asian countries because of the rising prevalence of PubMed, Embase, and the Cochrane Library for Hospital, School of Medicine, clarithromycin resistance.5–11 observational studies and randomised controlled trials Shanghai Jiao-Tong University, published in any language from Jan 1, 1990, to Sept 30, 2016 Shanghai Institution of The prevalence of resistance changes with time, and (appendix pp 85–86). In the second part of the systematic Digestive Disease, Shanghai, should be monitored to establish the optimal first-line review, we focused on the efficacy of first-line H pylori China (Prof H Lu MD); School of eradication regimen in each country. An overview of eradication therapies. We searched PubMed and the Medicine, Fu Jen Catholic antibiotic resistance in H pylori in the Asia-Pacific region University, New Taipei City, Taiwan (Prof J-T Lin); Institute of Epidemiology and Preventive Medicine, College of Public Health, National Taiwan University, Taipei, Taiwan (Y-T Kuo, Prof Y-K Tu PhD); Department of Environmental and Preventive Medicine, Oita University Faculty of Medicine, Yufu, Japan (Prof Y Yamaoka MD); and www.thelancet.com/gastrohep Vol 2 October 2017 707

Articles Department of Medicine, Research in context Added value of this study Gastroenterology and To our knowledge, ours is the first systematic review and Evidence before this study meta-analysis that provides an overview of the prevalence of Hepatology Section, Baylor We searched PROSPERO, PubMed, Embase, and the Cochrane primary antibiotic resistance in H pylori in the past 25 years in College of Medicine, Houston, Library with the terms (“Helicobacter pylori” [medical subject the Asia-Pacific region. We reported country-by-country heading] OR “Helicobacter pylori”) and (“drug resistance, microbial” variability in prevalence of resistance and assessed the relation TX, USA (Prof Y Yamaoka) [medical subject heading] OR “antimicrobial resistance” OR between resistance patterns and eradication rates with first-line “antibiotic resistance”) and “Asia” without any language therapies. We identified significant increases in primary Correspondence to: restrictions for articles published on or before Jan 31, 2017. We did clarithromycin and levofloxacin resistance from before 2000 Prof Yoshio Yamaoka, not find any published, ongoing, or planned systematic reviews of to 2011–15 in the Asia-Pacific region. According to data from Department of Environmental the prevalence of primary Helicobacter pylori antibiotic resistance in 2006–15, the overall prevalences of clarithromycin (20%) and and Preventive Medicine, Oita the Asia-Pacific region. After completion of our study, we searched levofloxacin resistance (21%) was higher than those in Europe University Faculty of Medicine, PROSPERO, PubMed, Embase, and the Cochrane Library again and Latin America. The efficacies of clarithromycin-based 1-1 Idaigaoka, Hasama-machi, with the same search terms for articles published on or before therapies were unsatisfactory (ie, <80%) in countries where the Yufu-City, Oita 879-5593, Japan March 31, 2017. Reviews that focused on patterns of primary prevalence of clarithromycin resistance was higher than 20%, [email protected] H pylori antibiotic resistance in the Asia-Pacific region were eligible; and thus such regimens should be avoided in these regions. those of secondary or tertiary H pylori resistance were excluded. We or identified one review of the global emergence of H pylori resistance Implications of all the available evidence (which included details of antibiotic resistance in four Asian Our findings provide policy makers with the necessary evidence Prof Ming-Shiang Wu, countries: Japan [two articles], China [two articles], Taiwan [one to decide optimal first-line eradication regimens according to Department of Internal article], and South Korea [three articles]), and three county-specific local prevalence of primary antibiotic resistance and develop Medicine, College of Medicine, reviews, one each about resistance in Saudi Arabia, Turkey, effective strategies to control the rising antibiotic resistance in National Taiwan University, No 7, and Vietnam. However, none of these papers were systematic their countries. Chung-Shan S Road, Taipei, reviews of primary antibiotic resistance in H pylori in the Asia-Pacific region. Taiwan [email protected] See Online for appendix Cochrane Library of randomised controlled trials for H pylori, the efficacy of second-line or third-line treatments, randomised controlled trials published in any language or the efficacy of regimens other than first-line therapy; or from Jan 1, 1990, to Sept 30, 2016, filtering with clinical if they were reviews or letters to the editor. The definition trial (appendix pp 87–88). We also searched the reference of the Asia-Pacific region was based on the UN geoscheme list of papers identified by our searches in both parts and devised by the United Nations UN Statistics Division. We related reviews for additional references. In both focused on countriesin this region with populations of systematic reviews, we identified relevant abstracts that more than 1 000 000 people. When important information were presented at Digestive Disease Week, the United was absent from articles, we attempted to contact the European Gastroenterology Week meeting, and the Asian corresponding author to clarify eligibility. Two reviewers Pacific Digestive Week between Jan 1, 2010, and Dec 31, (Y-TK and J-ML) screened all titles and abstracts 2015. The preplanned protocol of our study is available independently and excluded irrelevant or duplicate articles online (PROSPERO registry CRD42017057905). We first. Both reviewers then independently assessed the followed the PRISMA guidelines for systematic reviews.13 remaining articles for inclusion. Discrepancies were resolved by discussion. We established the inclusion and exclusion criteria before our search to minimise selection bias. Eligibility Data extraction and quality of assessment criteria for both systematic reviews were diagnosis of Two reviewers (Y-TK and J-ML) used a standardised data H pylori based on at least one of the standard diagnostic extraction form to independently collect information methods (rapid urease test, histology, culture, ³C urea from each eligible study—specifically, publication year, breath test or stool H pylori antigen test), and patients start and end dates of the study, study country, without any history of eradication treatment for H pylori characteristics of participants, diagnostic method of and no use of any antibiotics within the 2 weeks before susceptibility test, cutoffs for antibiotic resistance, they were recruited. Additional eligibility criteria for the details of H pylori treatment, number of included analysis of efficacy were patients treated with first-line patients, number of patients with resistance to different therapies recom­mended by the Maastricht and the Toronto antibiotics, and number of patients in whom H pylori consensus reports (appendix p 88).4,14 Articles were was successfully eradicated in intention-to-treat analyses excluded if participants’ mean or median age was younger (appendix pp 1–57). When different antibiotics of the than 18 years; if the trial focused on a special population same class were assessed (eg, levofloxacin, ciprofloxacin, (eg, children, patients with diabetes mellitus, patients moxifloxacin), they were combined in analyses of anti­ undergoing haemodialysis); if the aim of the study was to biotic resistance or eradication rates. investigate secondary or tertiary antibiotic resistance in 708 www.thelancet.com/gastrohep Vol 2 October 2017

Articles A B 3757 potentially eligible articles identified 2028 potentially eligible articles identified by database search by database search 1014 in PubMed 1028 in PubMed 1772 in Embase 1000 additional relevant abstracts 971 additional relevant abstracts identified from international identified from international conferences conferences 3016 articles judged to be duplicates or not relevant 1262 articles judged to be duplicates or not relevant after title and abstract review after title and abstract review 741 articles identified for full-text screening 766 articles identified for full-text screening 565 articles excluded 596 articles excluded 52 studies were in children 53 were done in ineligible participants (eg, children, patients with diabetes mellitus, 204 not about primary Helicobacter pylori patients undergoing haemodialysis) resistance 319 not of first-line therapy 61 had no information on the participants 118 assessed other regimens of first-line 180 did not originate in the Asia-Pacific region therapy 54 were review articles, meta-analyses, or 26 not randomised controlled trials letters to the editor 63 did not originate in the Asia-Pacific region 17 were review articles, meta-analyses, or 14 duplicates letters to the editor 176 full-text articles extracted for detailed 170 full-text articles extracted for detailed assessment and data synthesis assessment and data synthesis Figure 1: Review of literature for prevalence of primary antibiotic resistance in (A), and efficacy of first-line eradication therapies for (B), Helicobacter pylori infection in the Asia-Pacific region To increase robustness of our systematic review of antibiotic use with national populations in the same resistance, we used a risk-of-bias tool modified from the period, we adjusted population data from the World Newcastle–Ottawa quality-assessment scale.15 We used the Bank on the basis of the percentage of market coverage tool, which comprised five items, to assess adequacy in (appendix p 83).17,18 selection of participants, measurement of exposure (ie, H pylori infection), and outcomes (ie, H pylori resistance Statistical analysis to antibiotics). To investigate the risk of bias in randomised We did meta-analyses of the prevalence of H pylori controlled trials of first-line therapy, we used the Cochrane resistance to antibiotics and the eradication rate of first- Collaboration tool,16 which consists of six items: sequence line therapy. In view of the substantial variations in generation, allocation concealment, blinding of partici­ antibiotic resistance across the Asia-Pacific region, pants and personnel, blinding of outcome assessors, a random-effects model was used to adjust for incomplete outcome data, and selective outcome reporting. heterogeneity.19 Because the proportion of H pylori For each item, studies were recorded as “+” if risk of bias resistance in some studies was close to or at 0%, we was low, “−” if risk of bias was high, or “?” if risk was used the DerSimonian and Laird method to calculate unclear. All included studies were assessed by two authors pooled estimates and 95% CIs after Freeman-Tukey (Y-TK and J-ML) independently, and any discrepancies double arcsine transformation to avoid exclusion of were resolved through discussion. studies with estimated proportions of 0%.20,21 Heterogeneity among included studies was assessed Use of antibiotics in the Asia-Pacific region with Cochran’s Q test and the I² statistic (I² of <25%, We used sales data for Jan 1, 2006, to Dec 31, 2015, from 25–75%, and >75% were considered to represent low, the IMS Health MIDAS database (an analysis tool and moderate, and high degrees of between-study online system that brings together health-care data from heterogeneity, respectively).22 We used funnel plots and more than 70 countries) to estimate use of macrolides Egger’s test to assess potential small study bias.23 The and fluoroquinolones in the 18 countries for which sales Cochran-Armitage test was used in the trend analysis of data were available. Antibiotic consumption was given in national antibiotic consumption per person from standard units—ie, number of doses sold (the IMS Jan 1, 2006, to Dec 31, 2015.24,25 All tests were two-sided, identifies a dose as a pill, capsule, or ampoule). To link and the significance level was set at 0·05. www.thelancet.com/gastrohep Vol 2 October 2017 709

Articles Clarithromycin Metronidazole Levofloxacin Amoxicillin Tetracycline Articles Patients Articles Patients Articles Patients Prevalence Articles Patients Prevalence (n) (n) Prevalence (n) (n) Prevalence Articles Patients Prevalence (n) (n) (95% CI) (n) (n) (95% CI) (95% CI) (95% CI) (n) (n) (95% CI) ·· ·· 2 294 Australia 4 404 5% 5 657 57% 1 56 ·· 2 176 0% 1 137 1% 176 (2–9) (53–60) 1 111 1 111 (0–1) (0–4) 111 13 3266 66% 17 5307 Bangladesh 2 6463 18% 2 176 84% 1 127 (53–77) 3 571 6% 2 176 8% 926 (12–24) (78–89) 3 370 5 505 (2–10) (4–12) 505 1 77 3% 1 77 Bhutan 1 0% 1 111 83% 5 525 (1–8) 9 926 0% 1 111 0% 77 (0–3) (75–89) ·· ·· 36% 1 110 (0–3) (0–3) 1026 4 857 (28–45) 15 7400 China 23 26% 19 5717 61% 1 119 ·· ·· 2% 6 2205 2% 111 (21–32) (54–67) 4 333 17% (1–3) (1–4) 13 138 1 52 (12–25) Hong Kong 6 10% 8 1052 53% 1 42 0% 1 193 0% 119 (5–17) (39–66) 1 63 6% (0–1) (0–2) 523 1 178 (0–15) 18% India 5 52 17% 5 505 70% 1 26 (1–47) 5 505 11% 42 (1–45) (54–83) 1 27 31% (0–34) 330 2 806 (22–42) 5% Indonesia 1 289 9% 1 77 47% 12 2410 (2–13) 1 77 3% 26 (4–18) (36–58) 10 4672 8% 10% (1–9) 433 4 446 (3–16) (5–15) Iran 10 1152 19% 9 926 58% 2 109 8 848 3% 3257 (14–25) (50–66) 2 176 ·· 1% (1–6) 5809 72 14 848 (0–5) Israel 1 1299 8% 1 110 38% 36% 1 110 0% 775 (4–15) (30–48) (15–60) 3% (0–3) 176 (0–7) Japan 26 37 219 19% 16 7568 10% 13% 3 533 2% (16–22) (7–13) (8–21) ·· (0–7) Laos 1 13% ·· ·· ·· 2% ·· ·· ·· (8–20) (0–7) Malaysia 7 9 655 41% 5 385 0% 4 295 0% 2% (29–55) 6% (0–0) (0–0) (1–5) (2–16) Myanmar 1 1 52 37% 43% 1 52 0% 1 52 0% 0% (25–50) (29–58) (0–7) (0–7) (0–7) 10% Nepal 1 1 42 88% (4–19) 1 42 0% 1 42 0% 21% (75–95) 19% (0–8) (0–8) (12–36) (14–25) New 2 2 330 36% 42% 1 73 5% 1 73 0% Zealand 2 9% (31–42) (26–61) (2–13) (0–5) Pakistan (6–12) 1 178 84% 11% 1 178 37% 1 178 12% 37% (78–88) (4–28) (30–44) (8–17) (32–43) 10% Russia* 1 1 26 69% (8–13) 1 26 0% ·· ·· ·· 8% (50–83) 19% (0–13) (2–24) (14–24) Saudi Arabia 3 3 428 60% 3 387 8% 3 403 3% 16% (35–82) 11% (0–24) (0–12) (3–38) (7–14) Singapore 4 4 1152 39% 3 870 3% 3 870 4% 10% (33–45) 15% (1–5) (2–8) (6–15) (8–24) South Korea 17 14 2778 37% 28% 13 2554 7% 11 2401 9% 17% (33–42) (19–37) (5–10) (5–13) (13–21) 25% Taiwan 21 21 5784 32% (19–32) 20 5774 1% 5 2202 1% 11% (29–35) 18% (0–3) (0–2) (10–13) (15–22) Thailand 10 8 894 43% 2 524 4% 2 524 1% 8% (32–54) (2–6) (0–2) (4–14) Turkey 11 2 109 35% 3 155 0% 2 109 0% 27% (26–45) (0–1) (0–1) (19–37) Vietnam 2 2 176 72% 2 176 0% 1 103 6% 34% (65–79) (0–1) (3–12) (27–41) Total 162 136 29 503 44% 112 26 673 3% 64 12 156 4% 17% (39–48) (2–5) (2–5) (15–18) A detailed geographical view of antibiotic resistance is in the appendix (pp 222–26). *Data are for eastern Russia only. Table 1: Clarithromycin, metronidazole, levofloxacin, amoxicillin, and tetracycline resistance in the Asia-Pacific region To investigate potential sources of heterogeneity, we resistance as factors of interest. We grouped study did subgroup analyses and meta-regression with geo­ countries into regions on the basis of the UN geoscheme graphical region, period of sample collection, cutoffs for (devised by the UN Statistics Division).26 Sample antibiotic resistance, and assessment method for H pylori collection periods were divided into four periods: 710 www.thelancet.com/gastrohep Vol 2 October 2017

Articles before 2000, 2000–05, 2006–10, and 2011–15 for subgroup Resistance rate (%) 100 Clarithromycin 40% 50% 45% analysis. Articles in which the collection time spanned Metronidazole 27% two periods were grouped into the period in which most 16% 19% 21% of the study took place. If the sample collection period 90 Levofloxacin 9% 17% was not provided, we defined 2 years before the 80 2006–10 2011–15 publication year as the study period (31 studies included 70 2000–05 in the first part of our analysis and 37 included in the 60 Year second part did not provide a sample collection period). 50 For articles with two or more non-consecutive collections, 40 36% each con­secutive collection period was considered as an 30 individual study in the meta-analysis. Statistical analyses 20 were done in Stata (version 13.1). 10 7% Role of the funding source 0 2% The funders of the study had no role in study design; Before 2000 data collection, analysis, or interpretation; or writing of the report. All authors had full access to all study data Figure 2: Prevalences of primary clarithromycin, metronidazole, and levofloxacin resistance in the Asia-Pacific and had final responsibility for the decision to submit for region publication. Error bars represent 95% CIs. Results Mean overall prevalence of resistance to levofloxacin was 18% (95% CI 15–22; table 1; appendix p 224), ranging We screened 3757 articles, 176 of which were eligible for from 2–3% (Bhutan and Malaysia) to 66% in Bangladesh. inclusion in the analysis of prevalence of primary Between-study heterogeneity was high (I²=95·9%, antibiotic resistance (figure 1A). These 176 articles were p<0·0001; appendix pp 59–60). Subgroup analysis by from 24 countries. 170 articles from 16 countries were collection period showed that overall levofloxacin eligible for inclusion in the analysis of efficacy of first-line resistance increased from 2% (95% CI 0–13) before 2000 eradication therapies (figure 1B). Detailed base­line to 27% (95% CI 21–34) during in 2011–15 (figure 2; characteristics for each study are in the appendix appendix pp 66–69), with significant between-group (pp 1–57). For studies included in the meta-analysis heterogeneity (p<0·0001). Resistance to levofloxacin of primary H pylori antibiotic resistance, inadequate increased over time in all included countries for which description of participant characteristics and unclear data were available, except Iran (appendix pp 66–69, 229). methods of participant selection were the main sources of potential bias (appendix pp 89–90). For studies included Mean overall prevalence of resistance to amoxicillin in the meta-analysis of first-line antibiotic efficacy, in­ was 3% (95% CI 2–5; table 1; appendix p 225), with high adequate allocation concealment and deficiencies of between-study heterogeneity (I²=95·4%, p<0·0001; outcome assessment were sources of potential bias appendix pp 60–61). Country-specific data showed no (appendix pp 91–92). remarkable changes in resistance over time (appendix pp 66–69). Overall resistance to tetracycline was 4% Mean overall prevalence of resistance to clarithromycin (95% CI 2–5; table 1; appendix p 226). Between-study was 17% (95% CI 15–18; table 1; appendix p 222), ranging heterogeneity was high (I²=91·8%, p<0·0001; appendix from 0% in Bhutan and Myanmar to 34% in Vietnam p 61). Country-specific resistance did not remarkably and 37% in Pakistan. Between-study heterogeneity was change between classified periods (appendix pp 70–72). high (I²=95·2%, p<0·0001; appendix p 58). Subgroup analysis by collection period showed that overall According to data for 2006–15 (appendix pp 62–65, 230), clarithromycin resistance increased from 7% (95% CI clarithromycin resistance was higher than 15% in 5–8) before 2000 to 21% (95% CI 18–25) in 2011–15 13 countries: Bangladesh, China, India, Iran, Japan, Nepal, (figure 2). Clarithromycin resistance increased in most New Zealand, Pakistan, Saudi Arabia, Singapore, South countries during the periods we assessed (appendix Korea, Turkey, and Vietnam. By contrast, frequency of pp 62–65, 227). resistance was less than 15% in eight countries: Bhutan, Indonesia, Laos, Malaysia, Myanmar, Russia (data were Mean overall prevalence of resistance to metronidazole specifically from eastern Russia), Taiwan, and Thailand. was 44% (95% CI 39–48; table 1; appendix p 223) ranging During the same period of time, metronidazole resistance from 10% in Japan to 84% in Bangladesh and 88% in was higher than 40% in most included countries, except Nepal. Between-study heterogeneity was high (I²=98·3%, Japan, Myanmar, South Korea, Taiwan, and Turkey. p<0·0001; appendix pp 58–59). Subgroup analysis Levofloxacin resistance in the most recent collection period showed that metronidazole resistance increased from was higher than 20% in nine countries: Bangladesh, 36% (95% CI 27–45) before 2000 to 45% (95% CI 37–52) China, Indonesia, Japan, Nepal, Russia, South Korea, in 2011–15 (figure 2; appendix p 228), without significant Turkey, and Vietnam (appendix pp 66–69). Mean overall between-group heterogeneity (p=0·085). www.thelancet.com/gastrohep Vol 2 October 2017 711

Articles resistance was high (ie, >10%) to both amoxicillin and period (tables 2, 3). Levofloxacin resistance was signifi­ tetracycline in India and Pakistan (table 1; appendix cantly higher in studies done during 2011 to 2015 pp 66–72, 225–26). compared with those done before 2000 (difference 26% [95% CI 7–46]; p=0·008; tables 2, 3). Southeastern Asia During 2006–15, in the Asia-Pacific region, had the lowest risk of clarithromycin resistance. clarithromycin resistance (20%) was higher than that Southern Asia had the highest risk of both metronidazole reported for Latin America (13% in 2006–11) and Europe and amoxicillin resistance (tables 2, 3). Eastern Asia had (18% in 2008 and 2009; appendix p 232), although we did significantly higher levofloxacin resistance than western not do significance testing.27,28 Prevalence of metroni­ and southeastern Asia (tables 2, 3). Different suscepti­ dazole resistance in the Asia-Pacific region in 2006– bility tests might contribute to the between-study hetero­ 15 (47%) was between the reported 35% in Europe and geneity of tetracycline resistance. Multivariable analyses the 50% in Latin America (appendix p 232), and that of (table 3) showed that the prevalence of tetracycline levofloxacin resistance (21%) was higher than that resistance was lower when E-test rather than agar reported for Europe (14%) and Latin America (19%; dilution tests (p=0·012) or disc diffusion tests (p=0·023) appendix p 232).27,28 Prevalences of amoxicillin and wer used. Additionally, the prevalence of metronidazole tetracycline resistance (both 3%) were similar to those resistance was higher when E-tests rather than agar reported for Europe (both 1%) and Latin America dilution tests were used (p=0·002; table 3). (4% and 5%; appendix p 232).27,28 When classified by clarithromycin resistance according Meta-regression analysis showed no significant to our meta-analysis, efficacy of clarithromycin-based correlation between cutoff values of minimum inhibitory triple therapy was lower than 80% in countries concentration and antibiotic resistance (appendix p 84). where clarithromycin resistance was higher than 20% Univariable and multivariable analyses showed positive (appendix pp 73–75, 233–34). The efficaces of sequential correlation between clarithromycin resistance and study Clarithromycin Levofloxacin Metronidazole Amoxicillin Tetracycline Difference Difference Difference Difference Difference p value (95% CI) p value (95% CI) p value (95% CI) p value (95% CI) p value (95% CI) ·· Period Reference <0·0001 Reference ·· Reference ·· Reference ·· Reference ·· Before 2000 0·10 <0·0001 0·05 0·625 0·04 0·504 0·09 0·001 0·01 0·764 2000–05 <0·0001 0·215 (0·05 to 0·15) (–0·16 to 0·26) 0·029 (–0·07 to 0·14) 0·013 (0·04 to 0·15) 0·032 (–0·07 to 0·09) 0·400 2006–10 0·14 ·· 0·13 0·13 0·06 0·03 0·798 ·· 0·106 0·406 0·993 2011–15 (0·09 to 0·19) 0·437 (–0·07 to 0·32) 0·874 (0·03 to 0·23) (0·004 to 0·11) (–0·04 to 0·10) 0·16 0·008 0·22 0·952 0·08 0·02 –0·0003 Susceptibility test 0·396 0·577 (–0·07 to 0·07) E-test (0·11 to 0·21) (0·02 to 0·42) 0·712 (–0·02 to 0·18) (–0·03 to 0·07) Agar test ·· Reference 0·395 Reference ·· Reference ·· Reference ·· Reference ·· Disc diffusion test 0·01 0·143 –0·01 0·080 –0·11 0·004 0·04 0·039 0·06 0·006 0·001 (–0·08 to 0·07) 0·385 (–0·19 to –0·04) PCR (–0·04 to 0·05) 0·026 –0·004 0·006 0·06 0·324 (0·002 to 0·07) 0·017 (0·02 to 0·10) 0·241 –0·02 (–0·17 to 0·16) 0·837 (–0·06 to 0·17) 0·09 0·04 Two methods (–0·08 to 0·03) 0·05 –0·33 0·158 ·· ·· 0·10 (–0·13 to 0·23) (–0·78 to 0·13) (0·02 to 0·16) (–0·03 to 0·12) Region (0·02 to 0·17) –0·03 –0·06 0·358 N/A 0·523 N/A 0·440 Eastern Asia* –0·03 (–0·21 to 0·14) (–0·19 to 0·07) Western Asia† (–0·11 to 0·04) 0·03 0·06 Reference (–0·07 to 0·14) (–0·10 to 0·23) Southern Asia‡ Reference –0·11 0·03 (–0·24 to 0·01) Reference ·· Reference ·· Reference ·· Southeastern Asia§ –0·06 0·17 0·003 0·03 0·339 –0·03 0·284 (–0·03 to 0·09) (–0·18 to 0·07) (–0·09 to 0·03) Russia or Oceania¶ 0·06 –0·12 (0·06 to 0·28) <0·0001 (–0·03 to 0·09) <0·0001 0·04 0·288 (–0·20 to –0·03) 0·39 0·16 (–0·03 to 0·11) (–0·02 to 0·15) –0·03 0·213 0·305 –0·03 0·248 –0·09 (–0·30 to 0·25) (0·26 to 0·53) (0·09 to 0·24) (–0·09 to 0·02) (–0·15 to –0·04) 0·05 0·053 –0·03 0·477 –0·06 0·407 –0·12 (–0·08 to 0·03) (–0·20 to 0·08) (–0·23 to –0·01) (–0·03 to 0·14) –0·04 0·15 (–0·16 to 0·07) (–0·002 to 0·29) *Eastern Asia=China, Hong Kong, Japan, South Korea, and Taiwan. †Western Asia=Israel, Saudi Arabia, and Turkey. ‡Southern Asia=Bangladesh, Bhutan, India, Iran, Nepal, and Pakistan. §Southeastern Asia=Indonesia, Laos, Malaysia, Myanmar, Singapore, Thailand, and Vietnam. ¶Russia or Oceania=Russia (specifically eastern Russia) , Australia, and New Zealand. Table 2: Meta-regression univariable analysis of the prevalence of antibiotic resistance in Asia-Pacific region 712 www.thelancet.com/gastrohep Vol 2 October 2017

Articles Clarithromycin Levofloxacin Metronidazole Amoxicillin Tetracycline Difference Difference Difference Difference Difference p value (95% CI) p value (95% CI) p value (95% CI) p value (95% CI) p value (95% CI) Reference ·· Period Reference ·· Reference ·· Reference ·· Reference ·· 0·04 0·359 Before 2000 0·09 0·001 0·07 0·471 0·03 0·599 0·09 0·001 0·070 2000–05 0·078 (–0·04 to 0·12) 0·320 (0·04 to 0·14) <0·0001 (–0·13 to 0·27) 0·008 (–0·07 to 0·13) 0·016 (0·04 to 0·15) 0·014 0·07 2006–10 0·15 0·17 0·12 0·07 ·· <0·0001 ·· 0·054 0·137 (–0·01 to 0·15) 0·012 2011–15 (0·09 to 0·20) (–0·02 to 0·37) ·· (0·02 to 0·21) (0·01 to 0·12) 0·04 0·023 0·16 0·26 ·· 0·09 0·04 Susceptibility test ·· (–0·04 to 0·12) ·· E-test (0·11 to 0·21) (0·07 to 0·46) ·· (–0·001 to 0·18) (–0·01 to 0·09) 0·207 Agar test Reference Reference ·· ·· Reference ·· Reference ·· 0·06 ·· Disc diffusion test –0·01 0·502 ·· –0·12 0·002 0·03 0·054 0·076 (–0·06 to 0·03) (–0·19 to –0·05) (0·01 to 0·11) 0·434 PCR 0·01 0·809 ·· 0·04 0·538 (–0·001 to 0·07) 0·020 0·12 0·760 (–0·05 to 0·06) (–0·08 to 0·15) 0·09 0·829 Two methods 0·08 0·012 ·· –0·25 0·230 ·· (0·02 to 0·22) (0·02 to 0·15) (–0·67 to 0·16) (0·01 to 0·17) N/A Region –0·03 0·408 ·· –0·04 0·502 N/A 0·524 Eastern Asia* (–0·10 to 0·04) (–0·16 to 0·08) 0·12 Western Asia† 0·03 (–0·07 to 0·30) Reference (–0·07 to 0·14) Southern Asia‡ 0·002 Reference ·· Reference ·· Reference ·· Reference ·· –0·08 Southeastern Asia§ (–0·06 to 0·06) 0·956 –0·14 0·023 0·10 0·099 –0·01 0·740 (–0·16 to 0·01) 0·04 (–0·26 to –0·02) (–0·08 to 0·06) 0·03 Russia or Oceania¶ 0·288 –0·04 0·514 (–0·02 to 0·22) <0·0001 0·14 <0·0001 (–0·04 to 0·10) (–0·04 to 0·12) (–0·16 to 0·08) 0·38 (0·07 to 0·21) –0·01 –0·13 <0·0001 –0·13 0·001 0·634 –0·02 0·540 (–0·07 to 0·05) (–0·18 to –0·08) (–0·21 to –0·06) (0·25 to 0·51) (–0·07 to 0·04) –0·02 –0·06 0·221 –0·08 0·545 –0·02 0·119 –0·02 0·700 (–0·17 to 0·13) (–0·16 to 0·04) (–0·34 to 0·18) (–0·11 to 0·07) (–0·13 to 0·09) 0·12 (–0·03 to 0·27) *Eastern Asia=China, Hong Kong, Japan, South Korea, and Taiwan. †Western Asia=Israel, Saudi Arabia, and Turkey. ‡Southern Asia=Bangladesh, Bhutan, India, Iran, Nepal, and Pakistan. §Southeastern Asia=Indonesia, Laos, Malaysia, Myanmar, Singapore, Thailand, and Vietnam. ¶Russia or Oceania=Russia (specifically eastern Russia), Australia, and New Zealand. Table 3: Meta-regression multivariable analysis of the prevalence of antibiotic resistance in Asia-Pacific region therapy and concomitant therapy were also less than The significant increase in clarithromycin resistance 80% in most countries with clarithromycin resistance from 7% before 2000 to 21% in 2011–15 in the Asia-Pacific higher than 20% (appendix pp 78–79, 82, 235–36). region might be attributed to the increased consumption Efficacy of bismuth quadruple therapy and hybrid of macrolides. Metronidazole resistance tended to therapy was reported in few studies in the Asia-Pacific be higher in developing countries, such as Nepal, region in 2016–15 (appendix pp 78–81, 237–38). Overall, Bangladesh, Pakistan, Bhutan, Vietnam, and India, but use of macrolides and fluoroquinolones increased seemed lower in countries with higher socioeconomic steadily from 2006 to 2015 in the Asia-Pacific region development indices, such as Japan. The use of this (p<0·0001 for both; appendix pp 93–94). inexpensive antibiotic for parasite infestation and pelvic inflammatory disease or dental infections in developing Discussion countries might contribute to these differences.29,30 We also noted a remarkable rise in levofloxacin resistance, To our knowledge, this study is the first comprehensive from 2% before 2000 to 27% in 2011–15. Megraud and systematic review and meta-analysis of primary antibiotic colleagues28 and Liou and colleagues31 showed that resistance in H pylori in the past 25 years in the Asia-Pacific fluoroquinolone resistance correlated with con­sumption region. We showed that levels of metronidazole, amoxicillin, of fluoroquinolones in Europe and Taiwan, respectively. and tetracycline resistance were stable over time, but that Van Boeckel and colleagues17 reported that global clarithromycin and levofloxacin resistance are increasing. consumption of fluoroquinolones has increased by 64%, We also showed that the efficacy of clarithromycin- and of macrolides by 19%, between 2000 and 2010.We containing regimens, including triple therapy, sequential also noted increased consumption of fluoroquinolones therapy, and concomitant therapy, was lower than 80% in and macrolides in this study (appendix pp 93–94), which countries where clarithromycin resista­nce was higher might be explained by the recommendation in 2004 than 20%. www.thelancet.com/gastrohep Vol 2 October 2017 713

Articles guidelines to use fluoroquinolone monotherapy as an Contributors alternative first-line therapy for community-acquired J-ML conceived the study, with input from M-SW, YY, and EME-O. J-ML pneumonia.32 Strategies to restrict use of fluoroquinolones designed the study and wrote the protocol, with input from all authors. should be considered in future guidelines. Y-TK and J-ML did the systematic review, and Y-TK did the statistical analysis. Y-TK and J-ML wrote the Article, which was critically revised by Although amoxicillin (3%) and tetracycline resist­ M-SW, YY, Y-KT, and EME-O. All authors commented on the drafts and ance (4%) were uncommon in the Asia-Pacific region, approved the final version. J-ML, EME-O, J-YW, AHRL, HL, YY, M-SW resistance to both drugs was higher than 10% in Pakistan were on the steering committee of the Asia-Pacific Alliance of and India. Easy access to these antibiotics without Helicobacter and Microbiota, of which all other investigators except for prescription, and their wide use for treatment of respiratory Y-KT were members. and skin infections in these countries might explain this finding.33,34 However, data should be interpreted with Declaration of interests caution because the study numbers and case numbers We declare no competing interests. were small; further validation studies are warranted. Acknowledgments Efficacy of clarithromycin triple therapy was higher The study was funded by the Ministry of Science and Technology, than 80% in countries with low clarithromycin resist­ Executive Yuan, China, Taiwan (103-2314-B-002-173-MY3, ance (<15%), such as Thailand, Malaysia, Hong Kong, 105-2325-B-002-034, 102-2628-B-002-034-MY3, 105-2325-B-002-042), and Taiwan. By contrast, the studies from countries with Ministry of Health and Welfare of Taiwan (MOHW104-TDU-B-211-113001, high clarithromycin resistance, including Turkey, South MOHW105-TDU-B-211-133015), and intramural funding from Amity Korea, China, and Japan, have shown unacceptable University (AUUP/projects/2015). The database of IMS Health was eradication results with clarithromycin-based triple sponsored by Abbott. We thank Yun-Chun Wu (Institute of Epidemiology therapy (appendix p 82). These findings collectively and Preventive Medicine, College of Public Health, National Taiwan suggest that, although clarithromycin-based triple University) for her statistical expertise and assistance, and the Eighth therapy can still be used as the standard first-line Core Lab (Department of Medical Research, National Taiwan University treatment in countries where clarithromycin resistance Hospital) for technological support. is lower than 15%, alternative first-line regimens, such as bismuth quadruple therapy or non-bismuth quadruple References therapies, should be considered in countries with high 1 Ferlay J, Soerjomataram I, Dikshit R, et al. Cancer incidence and clarithromycin resistance.35–37 mortality worldwide: sources, methods and major patterns in Our study has limitations. First, between-study hetero­ GLOBOCAN 2012. Int J Cancer 2015; 136: E359–86. geneity among enrolled studies was remarkable. We have 2 Gonzalez CA, Megraud F, Buissonniere A, et al. Helicobacter pylori identified that the study periods, the study countries or infection assessed by ELISA and by immunoblot and noncardia regions, and the methods used to measure antibiotic gastric cancer risk in a prospective study: the Eurgast-EPIC project. resistance contributed to the heterogeneity. Another Ann Oncol 2012; 23: 1320–24. explanation for heterogeneity is intrinsic geo­graphical 3 Lee YC, Chiang TH, Chou CK, et al. Association between differences, which could have been caused by various Helicobacter pylori eradication and gastric cancer incidence: local infectious diseases contributing to different patterns a systematic review and meta-analysis. Gastroenterology 2016; of antibiotic consumption. Furthermore, participants in 150: 1113–24. most published trials were enrolled from medical 4 De Martel C, Ferlay J, Franceschi S, et al. Global burden of cancers centres, and thus might not be representative of the attributable to infections in 2008: a review and synthetic analysis. general population in a region. Second, there was Lancet Oncol 2012; 13: 607–15. potential publication bias in this systematic review. 5 Malfertheiner P, Megraud F, O’Morain CA, et al. Management of Information about the prevalence of antibiotic resistance Helicobacter pylori infection—the Maastricht V/Florence Consensus was scarce in some countries, including Banglandesh, Report. Gut 2017; 66: 6–30. Nepal, Russia, Indonesia, Bhutan, Myanmar, Malaysia, 6 Chey WD, Wong BC, Practice Parameters Committee of the and various Middle East countries. More data are needed American College of Gastroenterology. American College of from these countries. Gastroenterology guideline on the management of Helicobacter pylori infection. Am J Gastroenterol 2007; 102: 1808–25. The rise in clarithromycin and levofloxacin resistance in 7 Gungor G, Baglicakoglu M, Kayacetin E, et al. Current status of the past 25 years in most countries in the Asia-Pacific five different regimens for empiric first-line Helicobacter pylori region is concerning, and this should prompt the eradication in Turkey. Digestion 2015; 92: 55–59. development of effective strategies to control antibiotic 8 Liou JM, Lin JT, Chang CY, et al. Levofloxacin-based and resistance. First-line treatment strategies should be clarithromycin-based triple therapies as first-line and second-line adapted to resistance patterns on a country-by-country treatments for Helicobacter pylori infection: a randomised basis, and clarithromycin-containing regimens should be comparative trial with crossover design. Gut 2010; 59: 572–78. avoided in countries where clarithromycin resistance is 9 Chung JW, Han JP, Kim KO, et al. Ten-day empirical sequential or higher than 20%. Prospective studies are warranted concomitant therapy is more effective than triple therapy for to con­tinuously monitor the prevalence of antibiotic Helicobacter pylori eradication: a multicenter, prospective study. resistance in H pylori in the Asia-Pacific region. Dig Liver Dis 2016; 48: 888–92. 10 Zhou L, Zhang J, Chen M, et al. A comparative study of sequential therapy and standard triple therapy for Helicobacter pylori infection: a randomized multicenter trial. Am J Gastroenterol 2014; 109: 535–41. 11 Nishizawa T, Maekawa T, Watanabe N, et al. Clarithromycin versus metronidazole as first-line Helicobacter pylori eradication: a multicenter, prospective, randomized controlled study in Japan. J Clin Gastroenterol 2015; 49: 468–71. 12 Thung I, Aramin H, Vavinskaya V, et al. Review article: the global emergence of Helicobacter pylori antibiotic resistance. Aliment Pharmacol Ther 2016; 43: 514–33. 13 Shamseer L, Moher D, Clarke M, et al. Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015: elaboration and explanation. BMJ 2015; 349: g7647. 714 www.thelancet.com/gastrohep Vol 2 October 2017

Articles 14 Fallone CA, Chiba N, van Zanten SV, et al. The Toronto Consensus 27 Camargo MC, Garcia A, Riquelme A, et al. The problem of for the treatment of Helicobacter pylori infection in adults. Helicobacter pylori resistance to antibiotics: a systematic review in Gastroenterology 2016; 151: 51–69. Latin America. Am J Gastroenterol 2014; 109: 485–95. 15 Stang A. Critical evaluation of the Newcastle-Ottawa scale for the 28 Megraud F, Coenen S, Versporten A, et al. Helicobacter pylori assessment of the quality of nonrandomized studies in resistance to antibiotics in Europe and its relationship to antibiotic meta-analyses. Eur J Epidemiol 2010; 25: 603–05. consumption. Gut 2013; 62: 34–42. 16 Higgins JP, Altman DG, Gotzsche PC, et al. The Cochrane 29 Megraud F. H pylori antibiotic resistance: prevalence, importance, Collaboration’s tool for assessing risk of bias in randomised trials. and advances in testing. Gut 2004; 53: 1374–84. BMJ 2011; 343: d5928. 30 Megraud F. Epidemiology and mechanism of antibiotic resistance 17 Van Boeckel TP, Gandra S, Ashok A, et al. Global antibiotic in Helicobacter pylori. Gastroenterology 1998; 115: 1278–82. consumption 2000 to 2010: an analysis of national pharmaceutical sales data. Lancet Infect Dis 2014; 14: 742–50. 31 Liou JM, Chang CY, Chen MJ, et al. The primary resistance of Helicobacter pylori in Taiwan after the national policy to 18 World Bank. Population estimates and projections. restrict antibiotic consumption and its relation to virulence http://data.worldbank.org/data-catalog/population-projection-tables factors—a nationwide study. PLoS One 2015; 10: e0124199. (accessed July 6, 2016). 32 Niederman MS. Review of treatment guidelines for 19 DerSimonian R, Laird N. Meta-analysis in clinical trials. community-acquired pneumonia. Am J Med 2004; Control Clin Trials 1986; 7: 177–88. 117 (suppl 3A): 51s–57s. 20 Nyaga VN, Arbyn M, Aerts M. Metaprop: a Stata command to perform 33 Gehlot V, Mahant S, Mukhopadhyay AK, et al. meta-analysis of binomial data. Arch Public Health 2014; 72: 39. Antimicrobial susceptibility profiles of Helicobacter pylori isolated from patients in north India. J Glob Antimicrob Res 2016; 21 Freeman MF, Tukey JW. Transformations related to the angular and 5: 51–56. the square root. Ann Math Stat 1950; 21: 607–11. 34 Pandya HB, Agravat HH, Patel JS, Sodagar NR. 22 Higgins JP, Thompson SG, Deeks JJ, Altman DG. Emerging antimicrobial resistance pattern of Helicobacter pylori in Measuring inconsistency in meta-analyses. BMJ 2003; central Gujarat. Indian J Med Microbiol 2014; 32: 408–13. 327: 557–60. 35 Liou J-M, Chen C-C, Chen M-J, et al. Sequential versus triple therapy 23 Egger M, Davey Smith G, Schneider M, Minder C. Bias in for the first-line treatment of Helicobacter pylori: a multicentre, meta-analysis detected by a simple, graphical test. BMJ 1997; open-label, randomised trial. Lancet 2013; 381: 205–13. 315: 629–34. 36 Liou JM, Chen CC, Chang CY, et al. Sequential therapy for 10 days 24 Armitage P. Tests for linear trends in proportions and frequencies. versus triple therapy for 14 days in the eradication of Biometrics 1955; 11: 375–86. Helicobacter pylori in the community and hospital populations: a randomised trial. Gut 2016; 65: 1784–92. 25 Cochran WG. Some methods for strengthening the common x2 tests. Biometrics 1954; 10: 417–51. 37 Liou J-M, Fang Y-J, Chen C-C, et al. Concomitant, bismuth quadruple, and 14-day triple therapy in the first-line treatment of 26 UN Statistics Division. Composition of macro geographical Helicobacter pylori: a multicenter, open-label, randomised trial. (continental) regions, geographical sub-regions, and selected Lancet 2016; 388: 2355–65. economic and other groupings. https://unstats.un.org/unsd/ methodology/m49/ (accessed June 15, 2017). www.thelancet.com/gastrohep Vol 2 October 2017 715