Helicobacter pylori (H. pylori) is a major pathogenic factor for chronic gastritis, duodenal ulcer, gastric mucosa-associated lymphoid tissue lymphoma, gastric cancer, and other types of gastric and extragastric diseases (Fischbach and Malfertheiner, 2018). Since its discovery in 1982, the treatment of H. pylori infection has continued to evolve and remains a global research topic (O’connor et al., 2017). Globally, triple therapy containing pump proton inhibitors (PPIs), amoxicillin, and clarithromycin used to be the most frequently recommended primary eradication therapy for H. pylori. However, due to an increase in antibiotic resistance, the efficacy of triple therapy has drastically declined (Mégraud, 2004; Graham and Shiotani, 2008; Li et al., 2015; Lee and Park, 2016; Yeo et al., 2018). Thus, tailored susceptibility-guided treatment has been suggested to overcome antibiotic resistance (Fallone et al., 2016; Malfertheiner et al., 2017; Liou et al., 2018a). In tailored therapy, suitable medications are selected according to the results of susceptibility testing to achieve better eradication efficacy. Currently, the terms “tailored therapy,” “susceptibility-guided treatment,” “personalized eradication therapy,” and “customized H. pylori therapy” are often used interchangeably. Additionally, bismuth quadruple therapy is gradually being recommended as a first-line treatment (the initial therapy) (Fallone et al., 2016; Chey et al., 2017; Malfertheiner et al., 2017; Nyssen et al., 2021).

Antibiotic sensitivity was mainly detected by the following two methods: one was phenotypic identification, that is, H. pylori was cultured by gastroscopic biopsy, and antibiotic sensitivity was detected by agar dilution, disk diffusion, or E-test. The agar dilution method is the gold standard, but it is time-consuming and laborious (Ogata et al., 2014). The disk diffusion method and E-test are easy to apply, but the disk diffusion method may be difficult to explain (Lang and García, 2004; Miftahussurur et al., 2020; Tang et al., 2020). Moreover, the Clinical and Laboratory Standards Institute (CLSI) does not prescribe a standardized break point for antibiotics except for clarithromycin (Gingold-Belfer et al., 2021). On the other hand, the European Committee for Antimicrobial Sensitivity Testing (EUCAST) recommends an E-test and provides minimum inhibitory concentration (MIC) thresholds for six antibiotics (Alarcón et al., 2017). Another method is genotyping, by molecular detection (real-time PCR and fluorescence in situ hybridization) from stool and stomach biopsy specimens. Point mutations associated with resistance to specific antibiotics are usually detected through kits (Gingold-Belfer et al., 2021). In addition, high-throughput whole-genome sequencing techniques have been used to identify drug-resistant mutations (Chen et al., 2018; Nezami et al., 2019).

Despite the commercial availability and guideline recommendations to perform a susceptibility test, data to support this practice are scarce (Fallone et al., 2016; Chey et al., 2017; Malfertheiner et al., 2017). At present, there are three meta-analyses evaluating the effects of tailored therapy versus empirical therapy for H. pylori eradication (Lopez-Gongora et al., 2015; Chen et al., 2016; Gingold-Belfer et al., 2021). Lopez-Gongora et al. (2015) published the first meta-analysis in June 2015, which included 12 randomized controlled trials (RCTs) and quasi-RCTs published before February 2015, and their results showed that the evidence supporting the widespread use of customized H. pylori therapy, either as a first-line treatment or as a remedial treatment, was too limited. Another meta-analysis published by Chen et al. (2016) included 13 RCTs and controlled clinical trials (CCTs) published before October 2015, and their conclusions support tailored therapy as a better alternative to H pylori eradication. The latest study, a meta-analysis of 16 RCTs published before May 2020 by Boltin et al., found that tailored therapy was slightly better than empirical first-line triple therapy but not better than empirical first-line quadruple therapy or empirical rescue therapy (Gingold-Belfer et al., 2021). However, this study missed some studies published in the past 5 years.

Thus, we conducted a systematic review and meta-analysis without language restriction to systematically review RCTs (excluding quasi-RCTs and CCTs) comparing tailored H. pylori eradication regimens with empirical therapies and evaluate the effects of the two approaches for H. pylori eradication.

This meta-analysis was conducted in accordance with the PRISMA 2020 guidelines (Supplementary Table S1) (Page et al., 2021) and the standard methodology recommended by the Cochrane Collaboration (Cumpston et al., 2019; Higgins et al., 2019). The protocol of this meta-analysis has been registered at the International Platform of Registered Systematic Review and Meta-analysis Protocols (INPLASY) (Registration number: INPLASY202230166).

The main result of our meta-analysis, supported by both the ITT and PP analyses, is that tailored therapies might be superior to empirical therapies in the first-line treatment. However, there were no significant differences between tailored and empirical therapies in second-line therapy and mixed second- and third-line therapy. Notably, the substantial heterogeneity in ITT and PP cure rates indicated that the results should be interpreted with caution. Because of the diversity in antibiotic resistance, diagnosis methods, and treatment regimens of H. pylori, the evidence included in our research was highly heterogeneous. The high heterogeneity in both first-line treatment and rescue treatments prevented us from drawing a valid conclusion.

We further conducted first-line subgroup analyses in terms of the antibiotic resistance detection method, the region, and the empirical regimen. Tailored therapies took advantage of the empirical therapies found in both the culture and molecular methods. It should be noted that, in clinical practice, it is troublesome to isolate and culture H. pylori successfully from gastric biopsy specimens. The results are largely dependent on the quality of clinical specimens, time interval between sampling and culture, and transportation conditions (Pohl et al., 2019). Moreover, H. pylori culture requires highly trained staff to take 7 days until samples can be reported as negative and 2 weeks to reveal antibiotic susceptibility results (Pohl et al., 2019). Molecular testing contrasts with traditional culture-based testing because it can use clinical isolates, fresh or formalin-fixed gastric biopsies, or stool samples and could rapidly provide data on multiple antibiotics (Graham, 2021). To effectively use these data, we need studies to identify the advantages and limitations to molecular methods using different types of samples and more comparisons between molecular- and culture-based methods concerning treatment outcomes (Graham, 2021).

Regarding regionalism, tailored treatment was significantly more effective than first-line empirical therapy in both Asian and non-Asian areas. The effect was slightly more pronounced in Asia. It is noticeable that the literature from different regions showed discrepant geographical resistance rates to antibiotics (Ierardi et al., 2013). Since the gene mutation site increasing clarithromycin resistance might differ from Asia to other continents, epidemiological data about resistance vary across regions (Oleastro et al., 2003). In China, the resistance rate ranges between 21.5% and 23.8% (Su et al., 2013). Interestingly, in Japan, the rate was at least 15%, and a much higher resistance rate was observed in another study (86.4%), suggesting that even in the same geographical region, relevant differences may occur. In Northern Europe, the trend is relatively inconspicuous (Koivisto et al., 2004; Storskrubb et al., 2006; Selgrad et al., 2013), while in Italy, the rate is not only high (24.1%) but also increases rapidly (De Francesco et al., 2007). In contrast, the percentages of amoxicillin resistance are almost negligible worldwide except in a few regions (Ierardi et al., 2013). Iran and Japan have reported resistance rates of 28.6% (Milani et al., 2012) and 8.2%–15.2% (Murakami et al., 2013), respectively. In Cameroon, the rate is specifically high (85.6%) (Murakami et al., 2013). Over the years, levofloxacin-resistant strains have increased, and an unfavorable tendency has been revealed in Asia (Ierardi et al., 2013). Resistance was detected at approximately 18.4% in Vietnam and 20.6% in China (Ierardi et al., 2013). In Europe, a multicentric epidemiologic study reported an overall percentage of 14.1% (Megraud et al., 2013), indicating that levofloxacin may be an ineffective option in the future.

In the subgroup analysis of diverse empirical methods, tailored therapy showed the best advantage over the standard triple therapy. However, current guidelines recommend the use of triple therapy as first-line treatment only in areas with clarithromycin resistance rates <15%. However, the majority of studies using triple therapy as a control group had clarithromycin resistance rates>20%. In addition, tailored therapy was superior to bismuth-quadruple therapy in terms of both ITT and PP cure rates. Nevertheless, it should be noted that in the PP analysis, the heterogeneity was high. This might be explained by the very limited evidence—only five studies were suitable for this analysis. Among them, the PP analysis of two studies from China (Chen Q et al., Zhou L et al.), both found very similar cure rates between bismuth quadruple therapy and the empirical therapy, which were inconsistent with the results of other therapies. This suggests that the relationship between the eradication rate and race can be further explored. Finally, only three studies in the control group used non-bismuth quadruple therapy, and no significant difference was found in either ITT or PP analyses. In fact, the number of studies using non-bismuth quadruple therapy or other therapies (such as sequential therapy) was too limited to conclude.

Regarding second-line or mixed second- and third-line treatments, research has been relatively limited. In total, nine RCTs met the inclusion criteria. The cure rate did not show significant differences between the two treatments. Above all, the effectiveness of tailored therapy following previous treatment failures is uncertain.

Concerning safety, we found a trend in favor of tailored therapy in terms of AEs. However, this trend was not statistically significant, which might be caused by the small sample size.

Our study has its own strengths and limitations. This meta-analysis has the following advantages. First, all the included studies were randomized controlled trials, which ensured the validity of the overall results and reduced the possibility of bias in individual studies. Second, the study selection process was rigorous, with two independent reviewers screening eligibility and reviewing to ensure the accuracy and completeness of the included literature. In addition, our data retrieval results were complete. Under strict screening conditions, our study included 22 randomized controlled trials, the largest meta-analysis of qualified studies ever conducted. Finally, we analyzed the side effects to further investigate the feasibility of tailored regimens.

Additionally, our study was also limited in that we did not conduct a health economic analysis of the two treatments to study the difference in economic cost between tailored and empirical treatments. Although two trials have shown that the tailored treatment saves more money than the standard triple therapy (saving an average of $5 and $12, respectively) (Chen et al., 2016), there are still not enough data to determine whether the custom treatment saves more money than other popular empirical regimens. In addition, we attempted to conduct a subgroup analysis of the studies with first-line treatment in terms of the clarithromycin resistance rate, but the clarithromycin resistance rate reported in most studies was > 20%, thereby preventing us from conducting the analysis. Furthermore, we did not search the Web of Science and Scopus databases, which might have caused us to miss the potentially qualified literature. Finally, the high heterogeneity in the current study prevented us from drawing a valid conclusion.

In conclusion, in terms of the eradication rate, tailored therapy might be a better choice than empirical therapy in first-line treatment. The safety profiles of tailored therapy and empirical treatment might be comparable. However, evidence is too limited to support the generalized use of tailored therapy for H. pylori treatment, either as first-line or as rescue treatment; more studies are needed to reach an evidence-based conclusion.