INTRODUCTION

Gastroesophageal reflux disease (GERD) is a worldwide public health issue affecting 13% of the global population (1). Based on visible signs of mucosal damage under endoscopy, GERD could be subclassified as erosive esophagitis (EE), nonerosive reflux disease, and Barrett's esophagus. EE accounts for 25%–50% of patients with GERD (2), and it is a known risk factor of advanced GERD complications, such as Barrett's esophagus, esophageal adenocarcinoma, and esophageal stricture (3). Exposure of excessive gastric content, basically acidic, is the major cause of esophageal mucosal injury in EE. Healing of EE is directly related to the duration of maintained neutralization of intragastric content (4). As the severity of esophagitis increases, more potent and long-lasting acid inhibition therapy is required.

Proton-pump inhibitors (PPI) have been recommended as first-line acid-suppressive agents for EE treatment. Although PPI are effective in healing EE, they fail to work in some patients, especially those with high-grade esophagitis. Approximately 10%–30% of patients with Los Angeles (LA) grade C/D esophagitis failed to achieve endoscopic remission with 8 weeks of PPI treatment (5–7), meanwhile 20%–30% of them had persistent or relapsed esophagitis despite maintained use of PPI (8,9).

Potassium competitive acid blocker (P-CAB) is a novel antisecretory agent that binds competitively to the potassium binding site of proton pump. Because of its pharmaceutical property, it provides more potent and prolonged acid-inhibitory effect than PPI (10). Vonoprazan, tegoprazan, fexuprazan, revaprazan, and keverprazan are P-CAB that are currently put into clinical use. A latest trial conducted among Western population indicated that vonoprazan is superior in initial and maintained healing of high-grade esophagitis (11), while randomized controlled trials in Asia showed that healing rate of vonoprazan was only numerically higher than lansoprazole in LA grade C/D EE, probably due to limited number of subjects with high-grade EE (12–14).

Grade C/D esophagitis represents a severe manifestation of GERD that requires active treatment and close follow-up. A great proportion of patients with grade C/D esophagitis could not achieve mucosal healing despite PPI use. P-CAB are promising alternatives to PPI; however, whether P-CAB are superior to PPI in treating high-grade esophagitis is uncertain. At present, comparative studies regarding P-CAB and lansoprazole only included limited patients with grade C/D esophagitis (11–15), while head-to-head comparisons between P-CAB and PPI other than lansoprazole are hardly obtainable. This systematic review and network meta-analysis (NMA) aimed to combine both direct and indirect evidence to compare the efficacy and safety between P-CAB and various PPI for LA grade C/D esophagitis to aid clinical decision-making.

METHODS

The protocol of this analysis was registered in PROSPERO (CRD42023418273) (https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42023418273).

Search strategy and selection criteria

Literature search was conducted in 3 databases, including PubMed, EMBASE, and Cochrane Central Register of Controlled Trials, from inception up to March 15, 2023.

Randomized controlled trials (RCT) were eligible for inclusion if: (i) the efficacy of P-CAB and PPI vs each other or vs a placebo in adult patients with EE was reported; (ii) written in English; (iii) EE was graded according to LA classification and the number of patients with LA grade C/D EE at baseline was reported; (iv) the minimum duration of therapy was 8 weeks. (v) remission rate of LA grade C/D EE was reported; and (vi) more than 5 cases of LA grade C/D EE were included. Reviews and conference abstracts were excluded.

Key search terms including EE, reflux esophagitis, peptic esophagitis, gastroesophageal reflux disease or GERD in both MeSH term and free text were used to identify candidate RCT. Identified studies were then combined with the following terms with set operator AND for further search: PPI, lansoprazole, omeprazole, esomeprazole, dexlansoprazole, rabeprazole, pantoprazole, potassium competitive acid blocker, P-CAB, vonoprazan, tegoprazan, fexuprazan, revaprazan, or keverprazan.

After duplicates removed, all citations were first screened through titles and abstracts and then eligibility of studies was thoroughly assessed by full-text reading. In the case of unclear reported data, original data were asked to be provided from the original authors.

Outcome measures

The primary outcome was failure in healing LA grade C/D EE at initial treatment phase (8 weeks). Secondary outcomes included failure in healing at 24 weeks, occurrence of adverse events (AE), severe AEs (SAE), and withdrawal related to drug in short-term (8 weeks) and long-term (24–52 weeks).

Data extraction

Data were extracted into a prespecified Microsoft Excel spreadsheet as dichotomous outcome by 2 authors independently. Results of data extraction would be compared and discussed to solve disagreements by 2 authors. We extracted the following data of each trial: author, year of publication, study location, number of centers, type of included subjects, age, sex ratio, CYP2C19 status, number of patients with LA grade C/D esophagitis, end point(s) of study, treatment arm, treatment dose, PPI administration time (in relation to meal), and number of subjects who failed to achieve endoscopic remission. In multiple-arms study, the included subjects in control group would be divided equally to pair with intervention groups. Data were extracted as intention-to-treat analyses, with all dropouts considered to be treatment failure.

Besides, we extracted number of subjects included in the safety set of each trial. Number of subjects incurring AE, SAE, withdrawal related to drug, increased gastrin, and abnormal liver function were also documented. As the incidences of AE were similar in all grades of EE, patients who received at least 1 dose of treatment regardless of their baseline endoscopic grading were included in our safety analysis.

Quality assessment and risk of bias

The Cochrane Collaboration Risk of Bias tool in RevMan 5.4 software was used to summarize the risk of bias of included trials. Two authors assessed study quality independently and solved disagreements by discussion. Seven specific individual domains were recorded, including sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcomes assessment, incomplete outcome data, selective outcome reporting, and specific other issues. Each domain could be graded as low risk of bias, high risk of bias, and unclear risk (16). The overall risk-of-bias judgment of included studies was assessed according to the following criteria: (i) if the trial was judged to be at high risk of bias in at least 1 domain, it was regarded as low quality; (ii) if the trial was judged to be at low risk of bias for all domain, it was regarded as high quality; and (iii) if the study did not meet the criteria of high or low quality, it was considered as of moderate quality. Sensitivity analysis was conducted by excluding studies of low quality.

Data synthesis and statistical analysis

This study was reported according to the Preferred reporting Items for Systematic Reviews and Meta-Analyses extension statement for network meta-analyses.

Direct comparison of efficacy and AE.

For treatment efficacy between different P-CAB and PPI, risk ratios (RR) of failure to achieve endoscopic remission were pooled using random-effect model. This approach is more stable, when compared with RR of improvement or using the odd ratios (17). As for AEs between P-CAB and PPI, dichotomous data were summarized as RR as well. All data were reported with 95% confidence interval (CI). Direct comparisons were conducted with Revman 5.4 software.

Geometry of network.

A network plot was produced to examine the symmetry and geometry of the direct and network comparison between P-CAB and PPI using pcnetmeta package in R software (version 4.2.3, The R Foundation, https://www.r-project.org).

Network meta-analysis.

Network meta-analysis (NMA) using Bayesian random-effects model was conducted to compare the efficacy and safety of P-CAB with various types of PPI. NMA facilitates simultaneous comparisons of several interventions, creates treatments rankings, allows the estimation of the effect size of interventions that have not been compared in head-to-head RCT, and can potentially improve the precision of existing direct evidence. NMA was performed using the rjags, gemtc, and pcnetmeta packages in R software. A mixed treatments comparison was conducted using Bayesian framework with Monte Carlo Markov Chain. Model diagnosis, consistency analysis, and heterogeneity analysis are shown in Supplementary Digital Content (see Appendix, https://links.lww.com/AJG/D203). The relative differences between treatments were presented as RR of treatment failure with 95% credible intervals (CrI). The absolute differences between treatments were summarized as risk difference (RD) with 95% CrI. Differences were considered significant if the 95% CrI did not cross 1 for RR and did not cross 0 for RD.

Treatments rankings.

The surface under the cumulative probability ranking (SUCRA) was used to comprehensively rank the treatments to be the best, the second best, etc. The SUCRA score was calculated using gemtc package in R. When interpreting results, it is important to take RR and 95% CrI into account rather than relying on rankings alone.

Pooled rates of treatment failure.

Absolute rates of treatment failure were pooled using random-effects models. No transformation was applied to the original data because the data were normally distributed. The absolute rates of treatment failure were summarized as proportion (%) and 95% CI. The calculation was performed using meta package in R.

Assessment of heterogeneity and publication bias

Global statistical heterogeneity of the direct and indirect comparisons was assessed using the I2 measure with Revman 5.4 software and anohe function in R, respectively. The I2 measure ranges from 0% to 100%, with value of 25%–49% indicating low study heterogeneity, 50%–74% moderate heterogeneity, and ≥75% high heterogeneity. If there were sufficient studies (≥10 studies), publication bias and small study effects was assessed using comparison-adjusted funnel plots with network statistical packages in Stata V.16.

RESULTS Study selection

A total of 5,876 citations were retrieved from the search strategy. Following removal of duplicate studies, 2,889 citations were screened through titles and abstracts and 286 full texts were evaluated. After that, a total of 24 RCT fulfilled the inclusion criteria (Figure 1). Seven trials compared the efficacy of P-CAB with PPI, 5 of which compared 20 mg of vonoprazan q.d. with lansoprazole (740 patients) (11–15), 1 compared 20 mg of tegoprazan q.d. with 15 mg of lansoprazole q.d. (18 patients) (18) and 1 compared 20 mg of keverprazan q.d. with 30 mg of lansoprazole q.d. (49 patients) (19). One trial compared the efficacy between 20 mg of vonoprazan q.d. and 10 mg of vonoprazan q.d. (16 patients) (20), and 15 trials the efficacy between different types of PPI, including lansoprazole, esomeprazole, omeprazole, pantoprazole, dexlansoprazole, and rabeprazole ER (extended release) (8,714 patients) (6,7,9,21–33). We incorporated only P-CAB and PPI at licensed dose in further analysis. Detailed characteristics of included studies are shown in Supplementary Digital Content (see Table S1, https://links.lww.com/AJG/D202). Detailed safety outcomes of included studies are shown in Supplementary Digital Content (see Table S2, https://links.lww.com/AJG/D202). Of note, 2 studies by Kinoshita et al compared only the efficacy of rabeprazole in different dosages (25,26); thus these studies were unable to be included in the network analysis.

Figure 1.:

Flowchart of literature screening. EE, erosive esophagitis; LA, Los Angeles; P-CAB, potassium competitive acid blocker; PPI, proton pump inhibitor; RCT, randomized controlled trial.

Risk of bias

Of the 24 included RCT, 14 clearly stated the method of generation of randomization sequence, 20 the method of allocation concealment, 13 the details of blinding of patients or personnel, and 6 the details of blinding of outcome assessment. None of the trials had evidence of selective reporting outcomes (see Supplementary Figure S1, https://links.lww.com/AJG/D202). A total of 15 studies were of high quality, 7 of moderate quality, and 2 of low quality.

Network synthesis and model diagnosis

Network evidence plots for efficacy and safety analyses are shown in Figure 2 and Supplementary Digital Content (see Figure S2, https://links.lww.com/AJG/D202). A total of 9,267 subjects with grade C/D esophagitis were included for initial efficacy analysis (7,11–14,21,23,24,28,29,31–33), 1,834 subjects for maintenance efficacy analysis (9,11,15,27), 18,531 subjects for short-term overall AE analysis (11–14,21,23,24,28,29,31–33), and 5,026 subjects for long-term overall AE analysis (9,11,15,20,27). Results of model diagnosis, consistency validation, and heterogeneity analysis are shown in Supplementary Digital Content (see Appendix, https://links.lww.com/AJG/D203).

Figure 2.:

Network evidence plot for efficacy and safety analyses between P-CAB and PPI at initial and maintenance treatment phase. Node size is proportional to the number of study participants assigned to receive each intervention. The line width corresponds to the number of studies comparing the individual treatments. Treatment dose: the drug was administered once daily. DEX, dexlansoprazole; ESO, esomeprazole; KPZ, keverprazan; LPZ, lansoprazole; OME, omeprazole; PAN, pantoprazole; RAB, rabeprazole extended release; TGZ, tegoprazan; VPZ, vonoprazan.

Because most evidence regarding P-CAB were about vonoprazan, we would focus on the comparative efficacy and safety between the most studied P-CAB, vonoprazan, and PPI.

Efficacy and safety of vonoprazan vs PPI Initial efficacy.

Direct comparison showed that 20 mg of vonoprazan q.d. was superior over 30 mg of lansoprazole q.d. in healing grade C/D esophagitis at 8 weeks (risk of treatment failure vs 20 mg of vonoprazan q.d.: RR = 2.72, 95% CI [1.10–6.73], 4 studies, 740 patients, P = 0.03, I2 = 61%) (11–14). Pooled analysis showed that 20 mg of vonoprazan q.d. was significantly superior over 30 mg of lansoprazole q.d. and 20 mg of omeprazole q.d. (risk of treatment failure: 30 mg of lansoprazole q.d. vs 20 mg of vonoprazan q.d.: RR = 2.97, 95% CrI [1.65–6.8], RD = 0.16, 95% CrI [0.00–0.28]; 20 mg of omeprazole q.d. vs 20 mg of vonoprazan q.d.: RR = 3.51, 95% CrI [1.31–12.14], RD = 0.21, 95% CrI [0.01–0.38]) (Table 1). Significant difference was not found in the efficacy between 20 mg of vonoprazan q.d., 40 mg of esomeprazole q.d., 20 mg of esomeprazole q.d., 40 mg of pantoprazole q.d., 50 mg of rabeprazole ER q.d., and 60 mg of dexlansoprazole q.d. Nevertheless, we found that 20 mg of vonoprazan q.d. reduced the absolute risk of treatment failure by 11%–21% when compared with other PPI (Table 1). Estimated rates of treatment failure were 6% (95% CI [0%–13%]) for 20 mg of vonoprazan q.d. and 21% (95% CI [18%–23%]) for overall PPI (Figure 3a).

Table 1.:

Summarized efficacy of vonoprazan and PPI for failure to achieve mucosal healing in initial and maintenance treatment of grade C/D esophagitis (compared with 20 mg of vonoprazan q.d. as reference)

Figure 3.:

Pooled rates of failure to achieve mucosal healing in initial (a) and maintenance (b) treatment. DEX, dexlansoprazole; ESO, esomeprazole; Het.I2, heterogeneity; KPZ, keverprazan; LPZ, lansoprazole; N, number of included subjects; n, number of patients failed to achieve mucosal healing; OME, omeprazole; PAN, pantoprazole; RAB, rabeprazole extended release; TGZ, tegoprazan; VPZ, vonoprazan.

According to the SUCRA score, 20 mg of vonoprazan q.d. ranked the best in healing grade C/D esophagitis (SUCRA = 0.89), followed by 60 mg of dexlansoprazole q.d. (0.66) and 40 mg of esomeprazole q.d. (0.61) (Table 1).

Maintenance efficacy.

Direct comparisons showed only the superiority of 20 mg of vonoprazan q.d. over 15 mg of lansoprazole q.d. at maintenance healing phase (risk of treatment failure: 15 mg of lansoprazole q.d. vs 20 mg of vonoprazan q.d.: RR = 8.39, 95% CI [2.06–34.24], 1 study, 84 patients, P = 0.001, I2 = NA) (15), while 10 mg of vonoprazan q.d. was comparable with 15 mg of lansoprazole q.d. and 30 mg of lansoprazole q.d. in maintained healing (11,15). Pooled analysis showed no significant difference between vonoprazan and PPI in maintained healing of grade C/D esophagitis (Table 2). Twenty milligrams of vonoprazan q.d. reduced the absolute risk of treatment failure by 4%–18% when compared with PPI and 10 mg of vonoprazan q.d. by −3%–10% (Table 2). Estimated rates of treatment failure were 16% (95% CI [0%–39%]) for 20 mg of vonoprazan q.d., 25% (95% CI [18%–33%]) for 10 mg of vonoprazan q.d., and 30% (95% CI [23%–37%]) for other PPI (Figure 3b).

Table 2.:

Summarized safety profile of vonoprazan and PPI in initial and maintained healing of esophagitis (compared with 20 mg of vonoprazan q.d. as reference)

SUCRA scores showed that 20 mg of vonoprazan q.d. ranked the best in maintaining remission of grade C/D esophagitis (SUCRA = 0.87), followed by 30 mg of lansoprazole q.d. (0.73) and 10 mg of vonoprazan q.d. (0.50) (Table 1).

Safety profile.

For the short-term safety, direct comparisons showed that 20 mg of vonoprazan q.d. and 30 mg of lansoprazole q.d. had similar risk of AE occurrence (30 mg of lansoprazole q.d. vs 20 mg of vonoprazan q.d.: RR = 0.97, 95% CI [0.85–1.10], 4 studies (11–14), 2,204 patients, P = 0.62, I2 = 0%), SAE occurrence (RR = 1.33, 95% CI [0.46–3.82], 3 studies (11–13), 1,720 patients, P = 0.60, I2 = 0%), and drug withdrawal (RR = 1.51, 95% CI [0.72–3.18], 3 studies (11,12,14), 1,912 patients, P = 0.28, I2 = 0%). NMA did not detect significant difference between vonoprazan and PPI in short-term incidence rates of overall AE, SAE, and drug withdrawal (Table 2). SUCRA scores showed that 30 mg of lansoprazole q.d. (SUCRA = 0.73) ranked first for the short-term safety, followed by 20 mg of omeprazole q.d. (0.64), 40 mg of pantoprazole q.d. (0.64), and 20 mg of vonoprazan q.d. (0.59) (Table 2).

For the long-term safety, direct comparisons showed that incidences of overall AE, SAE, and drug withdrawal during maintenance treatment were similar between 20 mg of vonoprazan q.d. group and 30 mg of lansoprazole q.d. group (11,15). Pooled analysis did not detect significant difference between vonoprazan and PPI in incidence rates of long-term AE, SAE, and drug withdrawal (Table 2). According to SUCRA scores, 20 mg of tegoprazan q.d. (SUCRA = 0.80) ranked first for long-term safety, followed by 30 mg of lansoprazole q.d. (0.71), 10 mg of vonoprazan q.d. (0.58), and 20 mg of vonoprazan q.d. (0.49) (Table 2).

Among the AE, increased blood gastrin and abnormal liver function are most worrisome ones for P-CAB use. Subgroup analyses found that incidence rates of increased blood gastrin and abnormal liver test were similar between vonoprazan and lansoprazole in short-term, while increased gastrin was found to be more frequent in the 20 mg of vonoprazan q.d. group (RR = 11.79 95% CI [4.30–32.31], 1 study (11), 593 patients, P < 0.00001, I2 = NA) and 10 mg of vonoprazan q.d. group than in lansoprazole (RR = 8.28 95% CI [2.97–23.07], 1 study (11), 593 patients, P < 0.00001, I2 = NA).

Efficacy and safety of other P-CAB vs PPI Keverprazan.

A total of 24 subjects with grade C/D esophagitis were included in 20 mg of keverprazan q.d. group in initial efficacy analysis and 119 subjects in safety analysis. Twenty milligrams of keverprazan q.d. was shown to have increased risk of AE occurrence when compared with 30 mg of lansoprazole q.d. in direct comparisons (30 mg of lansoprazole q.d. vs 20 mg of keverprazan q.d.: RR = 0.78, 95% CI [0.63–0.97], 1 study (19), 238 patients, P = 0.03, I2 = NA). Both direct and indirect comparisons could not detect significant difference between 20 mg of keverprazan q.d. and PPI in initial efficacy, incidence rates of short-term AE, and incidence rates of withdrawal related to drug, except that 20 mg of keverprazan q.d. had significantly lower risk for incurring SAE when compared with PPI (see Supplementary Table S3, https://links.lww.com/AJG/D202). Of note, the 95% CrI for RR in SAE analysis were extremely wide, which suggested that the pooled results of SAE might be unreliable because limited number of subjects had incurred SAE. Twenty milligrams of keverprazan q.d. reduced the absolute risk of treatment failure by 3%–13% when compared with PPI (see Supplementary Table S3, https://links.lww.com/AJG/D202). Pooled rates of treatment failure were 8% (95% CI [1%–27%]) for 20 mg of keverprazan q.d., when compared with 21% (95% CI [18%–23%]) for overall PPI (Figure 3a). According to ranking analysis, keverprazan ranked higher than all PPI in efficacy ranking analysis (SUCRA = 0.87), but ranked the last in short-term safety (SUCRA = 0.13) (see Supplementary Table S3, https://links.lww.com/AJG/D202).

Tegoprazan.

Eight subjects with grade C/D esophagitis were included in the 20 mg of tegoprazan q.d. group in maintenance efficacy analysis and 173 subjects in safety analysis. Direct comparisons showed only 20 mg of tegoprazan q.d. had lower risk of incurring AE when compared with 15 mg of lansoprazole q.d. (lansoprazole v.s. tegoprazan: RR = 1.67 95% CI [1.03–2.71], 1 study (18), 247 patients, P = 0.03). Both direct and indirect comparisons did not detect significant difference between 20 mg of tegoprazan q.d. and PPI in maintained efficacy and incidence rates of long-term AE, but there were higher risk of SAE occurrence in tegoprazan groups (see Supplementary Table S4, https://links.lww.com/AJG/D202). Twenty milligrams of tegoprazan q.d. reduced the absolute risk of maintenance treatment failure by −9%–9% when compared with PPI (see Supplementary Table S4, https://links.lww.com/AJG/D202). Pooled rates of maintenance treatment failure were 25% (95% CI [3%–65%]) for 20 mg of tegoprazan q.d., when compared with 30% (95% CI [23%–37%]) for overall PPI (Figure 3b). According to SUCRA scores, 20 mg of tegoprazan q.d. ranked behind vonoprazan and 30 mg of lansoprazole q.d. for long-term efficacy, but ranked first for long-term safety (SUCRA = 0.80), followed by 30 mg of lansoprazole q.d. (0.71), 10 mg of vonoprazan q.d. (0.58), and 20 mg of vonoprazan q.d. (0.49) (see Supplementary Table S4, https://links.lww.com/AJG/D202).

Additional analysis

The recommended PPI administration time is 30–60 minutes before a meal, while inappropriate PPI dosing time could bias the results against PPI. A subgroup analysis was performed to include only studies that had given appropriate instruction on PPI administration (4,11,19,23,28,29,33). For initial efficacy, 7 RCT that contained vonoprazan, keverprazan, and 5 PPI were included in NMA (see Supplementary Figure S4, https://links.lww.com/AJG/D202). Pooled analysis showed that 20 mg of vonoprazan q.d. was still superior over 30 mg of lansoprazole q.d. in healing grade C/D esophagitis (risk of treatment failure: 30 mg of lansoprazole q.d. vs 20 mg of vonoprazan q.d.: RR = 2.97, 95% CrI [1.00, 8.98], RD = 0.13, 95% CrI [−0.30 to 0.28]); besides 20 mg of vonoprazan q.d. remained the first in ranking analysis, followed by 20 mg of keverprazan q.d. (see Supplementary Table S5, https://links.lww.com/AJG/D202). However, there were insufficient studies to perform additional analysis for maintenance efficacy.

Sensitivity analysis

Two studies with high risk of bias were excluded in NMA of 8-week efficacy in sensitivity analysis (12,32). Consistency hypothesis was not rejected for sensitivity analyses. Relative effects in NMA did not change significantly after removal of studies with high risk of bias. Ranking probability analysis and SUCRA score remained the same in sensitivity analysis as well.

Publication bias

Only outcome measures at 8 weeks had sufficient studies (≥10 studies) to assess publication bias and small study effects. Assessment of publication bias demonstrated symmetrical distribution, indicating no small sample effect or publication bias in NMA (see Supplementary Figure S5, https://links.lww.com/AJG/D202).

DISCUSSION

In this NMA, we aimed to compare the efficacy and safety of P-CAB with various types of PPI for initial and maintenance treatment of LA grade C/D esophagitis. Because most evidence regarding P-CAB were about vonoprazan, we have put emphasis on the comparative efficacy and safety between vonoprazan and PPI. At initial treatment phase, 20 mg of vonoprazan q.d. was significantly superior to 30 mg of lansoprazole q.d. and 20 mg of omeprazole q.d. in preventing treatment failure, while significant difference was not found in mixed comparisons between 20 mg of vonoprazan q.d. and other PPI. Nevertheless, 20 mg of vonoprazan q.d. reduced the absolute risk of treatment failure by 11%–21% when compared with other PPI, and the estimated rates of treatment failure were 6% (95% CI [0%–13%]) for 20 mg of vonoprazan q.d. and 21% (95% CI [18%–23%]) for overall PPI. According to ranking analysis, 20 mg of vonoprazan q.d. (SUCRA = 0.89) was the best treatment for grade C/D esophagitis, followed by 20 mg of keverprazan q.d. (SUCRA = 0.87), 60 mg of dexlansoprazole q.d. (SUCRA = 0.66), and 40 mg of esomeprazole q.d. (SUCRA = 0.61). Of note is that 40 mg of esomeprazole q.d. and 60 mg of dexlansoprazole q.d. are PPI treatments at max recommended dosing. The results of ranking analysis suggested that the initial starting dose of PPI for patients with grade C/D esophagitis should be maximized to ensure mucosal healing, and more importantly, 20 mg of vonoprazan q.d. was comparable or even superior to maximized-dose PPI in treating grade C/D esophagitis. Of short-term safety profile, vonoprazan had similar rates of AE, SAE, and drug withdrawal to PPI. Twenty milligrams of vonoprazan q.d. (SUCRA = 0.59) ranked higher than 40 mg of esomeprazole q.d. (SUCRA = 0.42) and 60 mg of dexlansoprazole q.d. (SUCRA = 0.52) in short-term safety analysis. Combined the results from efficacy and safety analyses, 20 mg of vonoprazan q.d. had superior efficacy and was relatively safe in initial healing of grade C/D esophagitis. As for keverprazan, a newly developed P-CAB in China, it ranked second in efficacy ranking analysis (SUCRA = 0.87), but the last in short-term safety analysis (SUCRA = 0.13), which largely compromised its use in healing grade C/D esophagitis.

As for the maintenance efficacy, our study showed that 30 mg of lansoprazole q.d. and 20 mg of vonoprazan q.d. ranked as the best treatments in maintained healing at 24 weeks. Twenty milligrams of vonoprazan q.d. reduced the absolute risk of treatment failure by 4%–18% when compared with PPI, and the absolute rates of treatment failure were 16% (95% CI [0%–39%]) for 20 mg of vonoprazan q.d. and 30% (95% CI [23%–37%]) for other PPI. An earlier NMA had assessed the efficacy of vonoprazan and various PPI for maintenance treatment of EE in all grades (34). The maintenance effect with 20 mg of vonoprazan was significantly better than that with10 mg of rabeprazole, 15 mg of lansoprazole, 10 mg of esomeprazole, and 10 mg of omeprazole, while that with 10 mg of vonoprazan was only better than that with 10 mg of esomeprazole and 10 mg of omeprazole (34). Same as this study, we also found that standard dose of vonoprazan or lansoprazole are more effective in maintained healing of EE, while standard dose of tegoprazan had inferior maintenance effect and higher risk of SAE occurrence. Given that recurrence occurs in nearly 100% of grade C/D esophagitis without therapy, indefinite PPI maintenance treatment is recommended for such patients (35–37). The medication is recommended to be administered in the lowest dose that could effectively maintain healing of EE because chronic acid-suppressive therapy might lead to SAE (35). For those who attempted a step-down therapy, 10 mg of vonoprazan q.d. (SUCRA = 0.50) ranked higher than 20 mg of esomeprazole q.d. (SUCRA = 0.44), 15 mg of lansoprazole q.d. (SUCRA = 0.24), and 20 mg of pantoprazole q.d. (SUCRA = 0.22) in maintained healing analysis and with moderate safety as indicated by safety ranking analysis. Of note, most of the end points in the included studies were 24 weeks, and long-term follow-up is required to assess the efficacy and safety of P-CAB.

LA grade C/D esophagitis remained a great challenge in GERD management. PPI are recommended as first-line treatment for EE; however, grade C/D esophagitis have lower rates of mucosal healing than those with milder esophagitis when given the same dose of PPI, with differences up to 25% (29). Multiple factors contribute to treatment failure of PPI, including incomplete acid control, inappropriate administration time, and CYP2C19 metabolism (36). The recommended PPI administration time is 30–60 minutes before a meal, while inappropriate dosing time could diminish the antisecretory effect of PPI (35). In our case, this might exaggerate the healing effect of P-CAB and underestimate the one of PPI. We then performed a subgroup analysis that included only studies that had given appropriate instruction on PPI administration time. In the subgroup analysis, 20 mg of vonoprazan q.d. still remained the first rank in SUCRA. In clinical practice, only 65% of patients received correct direction of mealtime-related dosing and 10% of whom failed to comply proper PPI administration (37). Regarding this, vonoprazan might be a preferred and convenient choice for patients with EE. CYP2C19 genotype is another factor affecting PPI response. CYP2C19 status was reported in 8 of 24 included RCT in our network analysis, which contained vonoprazan, keverprazan, tegoprazan, lansoprazole, and rabeprazole. The percentage of extensive metabolizer ranged from 1.7% to 100% among these studies. Studies included high proportion of extensive metabolizers might also exaggerate the healing effects of P-CAB. However, we were unable to perform stratified analysis or meta regression analysis to assess the role of CYP2C19 status in treatment efficacy because there were insufficient studies (at least 10 studies were required). Anyhow, genetic testing of CYP2C19 has no established role in clinical practice, and it is not practical to perform genetic testing in every patient before starting acid-suppressive treatment. We assumed that population with a higher proportion of extensive metabolizers (e.g., White individuals) (38) might benefit more from P-CAB treatment.

Grade C/D esophagitis is challenging in GERD management; however, head-to-head RCT to compare the efficacy between P-CAB and various PPI in LA grade C/D esophagitis were unobtainable. The strength of our study was that we simultaneously compared the efficacy of P-CAB and commonly prescribed PPI using Bayesian NMA to aid clinical decision-making. Limitations of this study include the fact that our data were extracted mostly from subgroup analysis of eligible included trials. Baseline characteristics of patients with LA grade C/D esophagitis including sex ratio, age, and CYP2C19 status were missing in a great proportion of included studies. To minimize heterogeneity, we included only studies using LA classification to identify EE severity. The heterogeneity across the comparisons in our study was low to moderate, which was within the acceptable range. Second, none of the included studies reported the symptom response rate in patients with grade C/D esophagitis; thus, the effects of P-CAB and PPI on symptom improvement could not be determined. Third, we included only studies regarding daily-dose PPI because data of twice-daily or once-daily double-dose PPI treating grade C/D esophagitis were scarce. As a result, our study could not conclude that P-CAB were superior over optimized PPI therapy. Of note, there was no evidence that patients with grade C/D esophagitis should start with twice-daily or double-dose PPI at initial healing phase, unless patients were with refractory symptoms or PPI-persistent esophagitis (35). Besides, although our study was the first meta-analysis to include P-CAB other than vonoprazan, the pooled efficacy of other P-CAB should be carefully interpreted. Only 2 RCT regarding tegoprazan and keverprazan fulfilled our inclusion criteria and were both with limited number of included subjects. Our results of other P-CAB might be modified if updated RCT present in the future.

Summarized SUCRA rankings for overall efficacy and safety of each treatment are shown in Figure 4. Our NMA showed that the efficacy of vonoprazan ranked the best at initial healing phase when compared with other PPI, with moderate safety as indicated by short-term safety analysis. Twenty milligrams of vonoprazan q.d. might be an efficacious and convenient option for start-up treatment of grade C/D esophagitis. In maintained healing of grade C/D esophagitis, standard dose of vonoprazan and lansoprazole ranked first in efficacy, while 10 mg of vonoprazan q.d. was a reasonable choice for those who attempted to lower maintenance dose, considering its moderate efficacy and safety. Combined the results of efficacy and safety analyses, vonoprazan is a preferred choice for treating grade C/D esophagitis in short-term and long-term. Whether keverprazan and tegoprazan were superior over PPI was doubted because of their compromised performance in safety and efficacy ranking analyses, respectively. In conclusion, vonoprazan has considerable efficacy in initial and maintained healing of grade C/D esophagitis compared with PPI, with moderate short-term and long-term safety, while other P-CAB warranted further evidence to prove their advantages.

Figure 4.:

Summarized 2-dimensional SUCRA ranking plot for efficacy and safety at initial (a) and maintenance (b) treatment phase. Treatments lying in the upper right corner are more effective and safe. AE, adverse event; DEX, dexlansoprazole; ESO, esomeprazole; KPZ, keverprazan; LPZ, lansoprazole; OME, omeprazole; PAN, pantoprazole; TGZ, tegoprazan; VPZ, vonoprazan.

CONFLICTS OF INTEREST

Guarantor of the article: Yinglian Xiao, MD, PhD.

Specific author contributions: Y.X.: planned the study. Q.Z. and S.C.: conducted the study. Q.Z., X.J., M.Z., and J.H.: collected data. Q.Z., S.C., X.Z., and M.Z.: analyzed the data; N.T., F.C., and Z.Z.: interpreted the data. Q.Z. and Y.X.: in drafting the manuscript. Y.X. and M.Z.: finalized the manuscript.

Financial support: This systematic review and network meta-analysis was supported by Provincial Natural Science Foundation of Guangdong (2023A1515010187), Provincial and municipal joint Fund Youth Project of Guangdong (2022A1515111159), and National Natural Science Foundation of China (81970479). This study did not receive any funding from manufacturers of treatments in the network.

Potential competing interests: None to report.

REFERENCES 1. Eusebi LH, Ratnakumaran R, Yuan Y, et al. Global prevalence of, and risk factors for, gastro-oesophageal reflux symptoms: A meta-analysis. Gut 2018;67(3):430–40. 2. Ronkainen J, Aro P, Storskrubb T, et al. High prevalence of gastroesophageal reflux symptoms and esophagitis with or without symptoms in the general adult Swedish population: A Kalixanda study report. Scand J Gastroenterol 2005;40(3):275–85. 3. Shaheen N, Ransohoff DF. Gastroesophageal reflux, Barrett esophagus, and esophageal cancer: Scientific review. JAMA 2002;287(15):1972–81. 4. Katz PO, Ginsberg GG, Hoyle PE, et al. Relationship between intragastric acid control and healing status in the treatment of moderate to severe erosive oesophagitis. Aliment Pharmacol Ther 2007;25(5):617–28. 5. Higuchi K, Joh T, Nakada K, et al. Is proton pump inhibitor therapy for reflux esophagitis sufficient? A large real-world survey of Japanese patients. Intern Med 2013;52(13):1447–54. 6. Lightdale CJ, Schmitt C, Hwang C, et al. A multicenter, randomized, double-blind, 8-week comparative trial of low-dose esomeprazole (20 mg) and standard-dose omeprazole (20 mg) in patients with erosive esophagitis. Dig Dis Sci 2006;51(5):852–7. 7. Richter JE, Kahrilas PJ, Johanson J, et al. Efficacy and safety of esomeprazole compared with omeprazole in GERD patients with erosive esophagitis: A randomized controlled trial. Am J Gastroenterol 2001;96(3):656–65. 8. Sjostedt S, Befrits R, Sylvan A, et al. Daily treatment with esomeprazole is superior to that taken on-demand for maintenance of healed erosive oesophagitis. Aliment Pharmacol Ther 2005;22(3):183–91. 9. Devault KR, Johanson JF, Johnson DA, et al. Maintenance of healed erosive esophagitis: A randomized six-month comparison of esomeprazole twenty milligrams with lansoprazole fifteen milligrams. Clin Gastroenterol Hepatol 2006;4(7):852–9. 10. Zhang M, Xiao Y, Chen M. The role of vonoprazan in patients with erosive esophagitis. Therap Adv Gastroenterol 2022;15:17562848221122623. 11. Laine L, DeVault K, Katz P, et al. Vonoprazan versus lansoprazole for healing and maintenance of healing of erosive esophagitis: A randomized trial. Gastroenterology 2023;164(1):61–71. 12. Ashida K, Sakurai Y, Hori T, et al. Randomised clinical trial: Vonoprazan, a novel potassium-competitive acid blocker, vs. lansoprazole for the healing of erosive oesophagitis. Aliment Pharmacol Ther 2016;43(2):240–51. 13. Ashida K, Sakurai Y, Nishimura A, et al. Randomised clinical trial: A dose-ranging study of vonoprazan, a novel potassium-competitive acid blocker, vs. lansoprazole for the treatment of erosive oesophagitis. Aliment Pharmacol Ther 2015;42(6):685–95. 14. Xiao Y, Zhang S, Dai N, et al. Phase III, randomised, double-blind, multicentre study to evaluate the efficacy and safety of vonoprazan compared with lansoprazole in Asian patients with erosive oesophagitis. Gut 2020;69(2):224–30. 15. Ashida K, Iwakiri K, Hiramatsu N, et al. Maintenance for healed erosive esophagitis: Phase III comparison of vonoprazan with lansoprazole. World J Gastroenterol 2018;24(14):1550–61. 16. Higgins JPT, Green S (eds). Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Cochrane Collaboration, 2011. Available from www.handbook.cochrane.org. 17. Deeks JJ.Issues in the selection of a summary statistic for meta-analysis of clinical trials with binary outcomes. Stat Med 2002;21(11):1575–600. 18. Cho YK, Kim JH, Kim HS, et al. Randomised clinical trial: Comparison of tegoprazan and lansoprazole as maintenance therapy for healed mild erosive oesophagitis. Aliment Pharmacol Ther 2023;57(1):72–80. 19. Chen S, Liu D, Chen H, et al. The efficacy and safety of keverprazan, a novel potassium-competitive acid blocker, in treating erosive oesophagitis: A phase III, randomised, double-blind multicentre study. Aliment Pharmacol Ther 2022;55(12):1524–33. 20. Okanobu H, Kohno T, Mouri R, et al. Efficacy of vonoprazan 10 mg compared with 20 mg for the initial treatment in patients with erosive esophagitis: A randomized pilot study. Esophagus 2021;18(3):669–75. 21. Castell DO, Kahrilas PJ, Richter JE, et al. Esomeprazole (40 mg) compared with lansoprazole (30 mg) in the treatment of erosive esophagitis. Am J Gastroenterol 2002;97(3):575–83. 22. Cho YK, Choi MG, Bak YT, et al. Efficacy of S-pantoprazole 20 mg compared with pantoprazole 40 mg in the treatment of reflux esophagitis: A randomized, double-blind comparative trial. Dig Dis Sci 2012;57(12):3189–94. 23. Fennerty MB, Johanson JF, Hwang C, et al. Efficacy of esomeprazole 40 mg vs. lansoprazole 30 mg for healing moderate to severe erosive oesophagitis. Aliment Pharmacol Ther 2005;21(4):455–63. 24. Kahrilas PJ, Falk GW, Johnson DA, et al. Esomeprazole improves healing and symptom resolution as compared with omeprazole in reflux oesophagitis patients: A randomized controlled trial. The esomeprazole study investigators. Aliment Pharmacol Ther 2000;14(10):1249–58. 25. Kinoshita Y, Hongo M, Japan TWICE Study Group.Efficacy of twice-daily rabeprazole for reflux esophagitis patients refractory to standard once-daily administration of PPI: The Japan-based TWICE study. Am J Gastroenterol 2012;107(4):522–30. 26. Kinoshita Y, Kato M, Fujishiro M, et al. Efficacy and safety of twice-daily rabeprazole maintenance therapy for patients with reflux esophagitis refractory to standard once-daily proton pump inhibitor: The Japan-based EXTEND study. J Gastroenterol 2018;53(7):834–44. 27. Labenz J, Armstrong D, Lauritsen K, et al. Esomeprazole 20 mg vs. pantoprazole 20 mg for maintenance therapy of healed erosive oesophagitis: Results from the EXPO study. Aliment Pharmacol Ther 2005;22(9):803–11. 28. Labenz J, Armstrong D, Lauritsen K, et al. A randomized comparative study of esomeprazole 40 mg versus pantoprazole 40 mg for healing erosive oesophagitis: The EXPO study. Aliment Pharmacol Ther 2005;21(6):739–46. 29. Laine L, Katz PO, Johnson DA, et al. Randomised clinical trial: A novel rabeprazole extended release 50 mg formulation vs. esomeprazole 40 mg in healing of moderate-to-severe erosive oesophagitis–the results of two double-blind studies. Aliment Pharmacol Ther 2011;33(2):203–12. 30. Metz DC, Howden CW, Perez MC, et al. Clinical trial: Dexlansoprazole MR, a proton pump inhibitor with dual delayed-release technology, effectively controls symptoms and prevents relapse in patients with healed erosive oesophagitis. Aliment Pharmacol Ther 2009;29(7):742–54. 31. Schmitt C, Lightdale CJ, Hwang C, et al. A multicenter, randomized, double-blind, 8-week comparative trial of standard doses of esomeprazole (40 mg) and omeprazole (20 mg) for the treatment of erosive esophagitis. Dig Dis Sci 2006;51(5):844–50. 32. Vcev A, Begić I, Ostojić R, et al. Esomeprazole