Keywords: Proton pump inhibitor; Peptic ulcer; Nonvariceal upper gastrointestinal bleeding; Rebleeding; Mortality; Management

Article Outline

Rationale for acid suppression pharmacotherapy

H₂-Receptor Antagonists in peptic ulcer bleeding

Proton Pump Inhibitors in peptic ulcer bleeding

Effects of PPIs on Clinical Outcomes: Evidence From Meta-analyses of Randomized Controlled Trials

Effects of Proton Pump Inhibitors on Clinical Outcomes: Evidence From Recent Randomized Controlled Trials

Effects of Proton Pump Inhibitors on Clinical Outcomes: Evidence From Observational Studies

Are Proton Pump Inhibitors More Efficacious in Asia?

What is the Optimal Regimen of Proton Pump Inhibitors Treatment for Peptic Ulcer Bleeding?

Proton Pump Inhibitors for Upper Gastrointestinal Bleeding Before Endoscopy

Summary

References

Acute upper gastrointestinal (GI) bleeding is a common emergency, associated with substantial morbidity, mortality, and health care cost. Patients with portal hypertension who bleed from esophageal or gastric varices comprise a small proportion of bleeders,¹ but have worse prognosis and require a different therapeutic approach than patients with nonvariceal bleeding.² Peptic ulcer bleeding is the principal cause of nonvariceal upper GI bleeding.^{[1] and [3]}

Prompt resuscitation and fluid replacement are of major importance⁴ especially in severe bleeding. Concurrent major diseases should be carefully monitored and treated; there may be further decompensation following the bleeding episode, and comorbidity is the major cause of mortality.³ Endoscopic hemostatic treatment is of established efficacy in patients with nonvariceal upper GI bleeding.^{[5] and [6]} Proton pump inhibitor (PPI) treatment is widely used as pharmacotherapy in nonvariceal upper GI bleeding, but its effect on mortality is not clear and is difficult to interpret.^{[7] and [8]}

This article will summarize and critically evaluate the evidence from randomized controlled trials (RCTs) and meta-analyses of RCTs on the role of PPIs in nonvariceal upper GI bleeding, with specific emphasis on peptic ulcer bleeding.

Rationale for acid suppression pharmacotherapy

Endoscopic hemostatic therapy reduces but does not eliminate the risk for adverse outcomes in patients with nonvariceal upper GI bleeding.^{[5] and [6]} Thus, there is room for further improvement in outcomes. Pharmacotherapy is a good candidate for adjuvant therapy, because it does not require any technical skills or specialty training to administer.

Furthermore, there is a biologically plausible underlying principle with regard to acid suppression therapy in this situation. Hemostasis in the stomach and duodenum is antagonized by gastric acid, which inhibits clot formation and promotes clot lysis. In vitro studies have shown that plasma coagulation and platelet aggregation are compromised by 50% in the presence of gastric juice at pH 6.4. At pH 6.0, previously formed platelet aggregates disaggregate; at pH 5.4, plasma coagulation and platelet aggregation are practically abolished; and at pH 4.0, previously formed fibrin clots are dissolved.⁹ Pepsin can further inhibit coagulation by promotion of clot lysis in an acidic environment, since its proteolytic activity is maximal at pH 2 and negligible at pH above 5.^{[9] and [10]} Such findings provided the rationale for rigorous acid suppression treatment attempting to maintain intragastric pH above 6 during the first 1–3 days following a bleeding episode. It is plausible that hydrochloric acid and pepsin can also provoke further bleeding from an ulcer by inducing ongoing tissue damage. However, this mechanism can be easily addressed even with standard doses of histamine H₂-receptor antagonists (H₂RAs) or PPIs; for healing of duodenal and gastric ulcers, maintaining the intragastric pH above 3 and 4 respectively, is adequate.^{[11], [12] and [13]}

Several RCTs have assessed the efficacy of acid suppression therapy at more than standard doses in nonvariceal upper GI bleeding. These studies had included patients with peptic ulcer bleeding either exclusively or predominantly.

H₂-Receptor Antagonists in peptic ulcer bleeding

H₂RAs were the first class of acid suppression medications to be tested in this patient population. In 1985 Collins and Langman published a seminal meta-analysis of 27 RCTs that had compared either cimetidine (Tagamet) or ranitidine (Zantac) with placebo in a total of 2670 patients with nonvariceal upper GI bleeding.¹⁴ There were marginally significant reductions in mortality and surgical intervention rates in the H₂RA group, but no evidence of an effect on rebleeding rates. When the analysis was restricted to studies that reported outcomes separately for patients who had bled from gastric and duodenal ulcers, the beneficial effect of H₂RAs was confined to patients with gastric ulcers; H₂RAs significantly reduced mortality, rebleeding and surgical intervention rates in patients with gastric ulcer bleeding, but there was no evidence of an effect on clinical outcomes in patients with duodenal ulcer bleeding.¹⁴ That meta-analysis was updated in 2002 by Levine and colleagues,¹⁵ who pooled 30 RCTs that had compared intravenous cimetidine, ranitidine, or famotidine (Pepcid) with placebo in patients with peptic ulcer bleeding. The meta-analysis found no evidence of an effect of intravenous H₂RAs on mortality (odds ratio [OR] 0.81; 95% confidence intervals [CI] 0.62–1.06), rebleeding (OR 0.86; 95% CI 0.74–1.00), or surgical intervention rates (OR 0.83; 95% CI 0.68–1.00), even though there were trends toward significance in rebleeding and surgical intervention rates. Subgroup analysis showed that H₂RAs significantly reduced rebleeding and surgical interventions in patients with gastric ulcer bleeding, but had no verifiable effect on mortality. As in the previous meta-analysis, H₂RAs did not significantly affect any clinical outcome in patients with duodenal ulcer bleeding.¹⁵

A planned subgroup analysis of 17 RCTs from a recent Cochrane meta-analysis showed that PPIs compared with H₂RAs significantly reduced rebleeding rates in patients with peptic ulcer bleeding (OR 0.61; 95% CI 0.48–0.78; number needed to treat [NNT] 20; 95% CI 13–34). There was no verifiable difference in mortality and surgical intervention rates.^{[7] and [8]}

Therefore, there is insufficient evidence to recommend H₂RA treatment for peptic ulcer bleeding. This was also the conclusion of a consensus meeting that recommended against the use of H₂RAs in the management of nonvariceal upper GI bleeding.⁴

Proton Pump Inhibitors in peptic ulcer bleeding

Effects of PPIs on Clinical Outcomes: Evidence From Meta-analyses of Randomized Controlled Trials

A recent Cochrane meta-analysis aimed to evaluate the efficacy of PPIs in treating peptic ulcer bleeding, using evidence from RCTs that had compared PPIs with either H₂RAs or placebo and had been published in November 2004.^{[7] and [8]} Twenty-four RCTs comprising 4373 patients were included.

The meta-analysis revealed a highly significant and robust reduction in rebleeding rates with PPI treatment. PPIs significantly reduced 30-day rebleeding rates compared with control treatment; unweighted pooled rates were 10.6% and 17.3%, respectively (OR 0.49, 95% CI 0.37–0.65; NNT 13, 95% CI 10–25) (Fig. 1). Three-day rebleeding rates were also significantly reduced by PPI treatment (8.3%) compared with control (14.2%); OR 0.39; 95% CI 0.19–0.80; NNT 13, 95% CI 8–34. The reduction of 30-day rebleeding rates remained statistically significant in all predetermined subgroup analyses. That is, PPIs significantly reduced rebleeding independent of methodological quality of the trials, severity of baseline endoscopic signs of recent hemorrhage, type of control treatment (placebo or H₂RA), geographic location of the trials (conducted in Asia or elsewhere), mode of PPI administration (oral or intravenous), dose of PPI (high-dose defined as at least 80 mg bolus followed by an intravenous infusion of 8 mg/h for 72 hours; low-dose defined as any lesser dose intravenous or oral), and whether or not endoscopic hemostatic treatment was given.^{[7] and [8]}

Nevertheless, there was statistically significant heterogeneity among trials for rebleeding (P = .04) and this had to be explained. The abovementioned subgroup analyses provided clues about the effect of population and study characteristics. Two of the subgroup analyses, according to route of administration of PPI and geographic location of trials, resulted in groups of trials that were statistically homogeneous for rebleeding. Moreover, when the influence of predefined study characteristics on the effect of treatment on rebleeding was assessed by metaregression, only the geographic location of the studies had a significant influence: PPIs produced quantitatively greater reductions in rebleeding among RCTs that had been conducted in Asia compared with RCTs that had been conducted elsewhere. The findings of the subgroup analysis and the metaregression not only explained the heterogeneity for the outcome of rebleeding, but also provided indirect evidence of increased efficacy of PPIs for peptic ulcer bleeding in Asian trials.^{[7] and [8]}

The trials were nonheterogeneous for the other outcomes that were assessed. Surgical interventions were significantly less common with PPI treatment (6.1%) than with control treatment (9.3%); OR 0.61, 95% CI 0.48–0.78; NNT 34, 95% CI 20–50. Further endoscopic hemostatic treatment (after randomization) was also reduced with PPIs (5.6%) compared with control treatment (15.7%); OR 0.32, 95% CI 0.20–0.51; NNT 10, 95% CI 8–17.^{[7] and [8]}

Despite the beneficial effect of PPI treatment on the above outcomes, there was no evidence of an effect on all-cause mortality rates (OR 1.01, 95% CI 0.74–1.40; Fig. 2). Unweighted pooled rates were 3.9% for PPI treatment and 3.8% for control treatment. It is plausible that any beneficial or detrimental effect of PPIs on mortality could have been diluted by the inclusion of patients with low-risk in the studies. A planned subgroup analysis was restricted to patients with high-risk endoscopic findings of active bleeding or a nonbleeding visible vessel—the patient population that clinicians are mainly concerned about. In this population PPIs significantly reduced mortality; OR 0.53, 95% CI 0.31–0.91, NNT 50, 95% CI 34–100. Among such patients, the reduction of mortality by PPI treatment remained significant when the analysis was confined to the RCTs that consistently used initial endoscopic hemostatic treatment, which is the accepted standard of care for such patients (OR 0.54, 95% CI 0.30–0.96; NNT 50, 95% CI 34–100), but was not significant among trials that did not consistently do so.^{[7] and [8]}

Separating the trials according to geographic location was the only other subgroup analysis that showed a statistically significant effect of PPI treatment on mortality. PPIs significantly reduced mortality among trials that had been conducted in Asia (OR 0.35, 95% CI 0.16–0.74; NNT 34, 95% CI 20–100), but had no verifiable effect among trials that had been conducted elsewhere (OR 1.36, 95% CI 0.94–1.96). Similarly to the outcome of rebleeding, a higher treatment effect of PPIs in Asian trials was confirmed by metaregression.^{[7] and [8]}

Funnel plots for mortality, rebleeding and surgical interventions were visually asymmetric, suggesting the presence of publication bias because of some missing small trials with negative results. However, this was only confirmed statistically for the outcome of surgical interventions.^{[7] and [8]}

There have been three other published meta-analyses of RCTs that have assessed the role of PPIs in peptic ulcer bleeding.^{[16], [17] and [18]} The most consistent finding of these meta-analyses, which is also in agreement with the Cochrane meta-analysis, is that PPIs compared with H₂RAs or placebo significantly reduce rebleeding rates in patients with peptic ulcer bleeding.

Khuroo and colleagues pooled RCTs on acute nonvariceal upper GI bleeding that had been published till 2002.¹⁸ They found no evidence that PPI treatment compared with H₂RAs or placebo affected all-cause mortality. PPI treatment significantly reduced rebleeding and surgical intervention rates. However, a predetermined subgroup analysis showed that these beneficial effects of PPIs were restricted to trials on patients who had bled from peptic ulcers with active bleeding, nonbleeding visible vessels, or adherent clots. There was no verifiable effect of PPI treatment on rebleeding or surgical intervention rates among trials that had included patients with all causes of nonvariceal upper GI bleeding.¹⁸

Khuroo and colleagues also reported that all-cause mortality in nonvariceal upper GI bleeding was significantly increased with intravenous PPI treatment, although it was significantly reduced with oral PPI treatment. They also reported that PPI treatment significantly reduced deaths directly caused by bleeding, while it significantly increased deaths caused by associated diseases or of unknown etiology.¹⁸ However, the validity of these analyses on mortality was undermined by several factors: some eligible RCTs were missed, noneligible trials were inadvertently included, and there were mistakes in the data extraction process.¹⁹ When the authors re-extracted raw data and repooled the trials, it was found that PPI treatment marginally reduced “bleeding related” deaths and had no effect on “non–bleeding related” deaths.¹⁹ The fact that the vast majority of published trials did not provide sufficient data to allow for a reliable differentiation of causes of deaths renders even this “corrected” analysis unreliable. It would be very useful if future studies were prospectively designed to categorize causes of death in patients with nonvariceal upper GI bleeding in a systematic way (as only one of the published studies did²⁰), but until then it would be premature to draw conclusions regarding any differential effect of PPIs on different causes of death.

Andriulli and colleagues¹⁶ conducted a series of meta-analyses of RCTs published till August 2003 on PPI treatment for peptic ulcer bleeding. It was concluded that PPI treatment, compared with H₂RA or placebo, significantly reduced rebleeding but not surgical intervention or all-cause mortality rates.

The meta-analysis by Bardou and colleagues¹⁷ included RCTs published up to April 2003 that compared PPIs with H₂RAs or placebo in patients with peptic ulcers with active bleeding, non-bleeding visible vessels, or adherent clots. The trials were divided into three groups according to the dose of PPI treatment: high-dose intravenous PPI (40–80 mg bolus followed by intravenous infusion of at least 6 mg/h), high-dose oral PPI (at least twice the standard dosage), and non-high-dose PPI (any other dose of PPI). In each group, trials were further subgrouped according to the control treatment used. PPI treatment significantly reduced rebleeding rates in all comparisons apart from that of high-dose intravenous PPI versus H₂RA. The effect of PPIs on surgical intervention rates was less consistent. Mortality was significantly reduced in the analyses of high-dose intravenous versus placebo and non-high-dose PPI versus placebo. The authors have expressed concerns²¹ about the exclusion of two studies^{[22] and [23]} from the former analysis and the misclassification of another study²⁴ in the latter analysis. Nevertheless, this was the first meta-analysis that showed that PPIs significantly reduced mortality in a subgroup of patients with high-risk endoscopic stigmata.

Effects of Proton Pump Inhibitors on Clinical Outcomes: Evidence From Recent Randomized Controlled Trials

Since the completion of the most recent of the above meta-analyses,^{[7] and [8]} two RCTs that had been included in the meta-analysis as abstracts were published in full.^{[25] and [26]} Furthermore, eight new RCTs that assessed the efficacy of PPI treatment in peptic ulcer bleeding have been published.^{[27], [28], [29], [30], [31], [32], [33] and [34]} The characteristics and main results of these trials are shown in Table 1. PPI treatment produced statistically significant reductions in rebleeding rates in all but 1 trial,³³ did not have a verifiable effect on mortality in any trial, and significantly reduced surgical interventions in one trial.²⁸ Since these results are consistent with the results of the Cochrane meta-analysis,^{[7] and [8]} it is unlikely that the inclusion of the above trials will significantly affect the update of the meta-analysis scheduled for 2009. The trial by Sung and colleagues³² is of particular importance. It had great methodological quality (in fact the protocol had been registered in ClinicalTrials.gov and prospectively published³⁵), was adequately powered, and was the first trial to use high-dose intravenous esomeprazole (Nexium). Among the 18 RCTs that had compared PPI treatment with H₂RAs or placebo in predominantly Caucasian populations, it is the only trial that has shown a highly significant reduction of rebleeding rates with PPI treatment; 3-day rebleeding rates were 5.9% and 10.3% (P = .03) and 30-day rebleeding rates were 7.7% and 13.6% (P = .01), respectively.³² The results of a predetermined analysis confined to Caucasian patients (87% of total population) were very similar to the results from the total study population, which also included Asians, blacks, and others.³⁶ The only other trial on Caucasian patients that reported that PPIs significantly reduced rebleeding rates is a small trial by Naumovski-Mihalic and colleagues,³¹ but the statistical significance of this finding is dependent on the statistical test used, and is marginal at best.

Effects of Proton Pump Inhibitors on Clinical Outcomes: Evidence From Observational Studies

The effectiveness of PPI treatment in a “real-life” setting was shown in an analysis of the Canadian Registry of patients with Upper Gastrointestinal Bleeding and Endoscopy by Barkun and colleagues.³ They analyzed data from 1869 randomly selected patients who underwent endoscopy for nonvariceal upper GI bleeding at 18 community and tertiary care institutions from across Canada between 1999 and 2002. Peptic ulcer was the commonest cause of bleeding (responsible for 55% of episodes). Logistic regression models identified that PPI use was significantly and independently associated with decreased mortality in patients with high-risk endoscopic findings defined as active bleeding, visible vessels, or visible vessels with clots (OR 0.53, 95% CI 0.37–0.77), and decreased rebleeding in all patients regardless of the severity of endoscopic findings (OR 0.18, 95% CI 0.04–0.80). Despite the large sample size and the superior methodological quality of this database analysis, the authors noted that “because of the nature of the study design, these findings should be considered exploratory and require prospective confirmation.”³ These results were confirmed by subsequent meta-analyses of RCTs, as described above.^{[7], [8] and [17]}

Are Proton Pump Inhibitors More Efficacious in Asia?

The suggestion that PPI therapy for peptic ulcer bleeding is more efficacious in trials that had been conducted in Asia than elsewhere originated from a post hoc analysis³⁷ of the initial Cochrane meta-analysis.³⁸ The difference in efficacy persisted in the updated Cochrane meta-analysis, as described above.^{[7] and [8]} PPI treatment compared with H₂RA treatment or placebo significantly reduced mortality among the 8 RCTs that had been conducted in Asia, but had no effect on mortality among the 16 RCTs that had been conducted elsewhere. (One trial had been performed in the US,³⁹ 14 in Europe, and one predominantly in Europe but also in Canada and South Africa⁴⁰). Although rebleeding rates were significantly reduced by PPI treatment in both subgroups, the reduction was quantitatively greater in the Asian trials. Separating Asian from non-Asian trials eliminated the statistical heterogeneity among trials for rebleeding. Furthermore, the increased efficacy of PPIs in Asian trials was confirmed by metaregression.^{[7] and [8]} Although the results of subgroup analyses and metaregression should be interpreted cautiously, the above findings seem to be robust.

There are a number of possible explanations for this apparent difference in efficacy. Asian patients seem to have a lesser parietal cell mass and an increased higher prevalence of Helicobacter pylori infection and genetically determined slow metabolism of PPIs. All these factors would tend to produce a greater antisecretory effect of PPI therapy in Asian than in non-Asian patients.³⁷

The recent trial by Sung and colleagues^{[32] and [36]} may shed more light on this issue. The outcomes provided by the abstract publications of the trial have shown that PPI therapy is beneficial in Caucasian patients with peptic ulcer bleeding. The full publication of the trial may allow for the comparison of outcomes between Asian and non-Asian patients, which, although less important from a practical point of view, is scientifically interesting.

What is the Optimal Regimen of Proton Pump Inhibitors Treatment for Peptic Ulcer Bleeding?

Omeprazole (Prilosec, Losec) had been the PPI most commonly used in RCTs in peptic ulcer bleeding, followed by pantoprazole (Protonix), esomeprazole and lansoprazole (Prevacid). There is no obvious indication that any one PPI was more efficacious than the others. Metaregression analysis found no evidence that using omeprazole as opposed to another PPI affected the treatment effect regarding mortality, rebleeding, or surgical interventions.^{[7] and [8]} Three small RCTs have reported clinical outcomes for similar doses of head-to-head comparisons of different PPIs in peptic ulcer bleeding.^{[41], [42] and [43]} None of these trials has provided conclusive evidence of clinical superiority of any one PPI over another. Therefore, it is reasonable to assume that the effect of PPIs on clinical outcomes of peptic ulcer bleeding is a class effect.

One of the most controversial issues in the management of peptic ulcer bleeding is on the optimal dose and route of administration (intravenous or oral) of PPI treatment. Predetermined subgroup analyses and metaregression from the Cochrane meta-analysis of RCTs showed no evidence that the effects of PPI treatment were dependent on the route of administration or the dose of the PPI.^{[7] and [8]} A post hoc subgroup analysis found that the reduction in mortality among patients with high-risk endoscopic signs remained significant when the analysis was confined to the four trials^{[39], [40], [44] and [45]} that had consistently applied initial endoscopic hemostatic treatment and also used high-dose intravenous PPI treatment (80 mg bolus followed by 8 mg/h for 72 hours); OR 0.46, 95% CI 0.24–0.90; NNT 50, 95% CI 34–100.⁸ There was no evidence of an effect on mortality among the trials on patients with high-risk endoscopic signs that had consistently applied initial endoscopic hemostatic treatment but used a lesser dose of intravenous PPI or an oral PPI (OR 1.01, 95% CI 0.26–3.83).⁸ Although this post hoc subanalysis suggests that high-dose intravenous PPIs are more efficacious than lower-dose PPIs, the suggestion should be regarded with caution since the comparison is not a direct one. Furthermore, the latter subgroup analysis may have been underpowered.

The authors are aware of at least 16 RCTs that had compared head-to-head different doses and/or routes of administration of PPIs in peptic ulcer bleeding and had reported clinical outcomes. Most of these have undoubtedly been underpowered. An ongoing Cochrane meta-analysis aims to assess the potential differences among various regimens of PPIs using evidence from such RCTs.

Defining the minimum effective dose of PPI will have important cost-saving implications to health care systems. Cost-effectiveness analyses have shown that either oral or intravenous PPI treatment is more effective and less costly than treatment with H₂RAs or placebo following endoscopic hemostatic treatment for peptic ulcer bleeding in the US and Canadian settings.^{[46] and [47]} However, cost-effectiveness analyses of oral versus intravenous administration of PPIs in the US, Canada, and the UK have yielded conflicting results.^{[46], [47] and [48]}

Proton Pump Inhibitors for Upper Gastrointestinal Bleeding Before Endoscopy

Since PPIs reduce rebleeding in patients with peptic ulcer bleeding, it might be argued that they should be administered to patients with upper GI bleeding at the time of presentation and while awaiting endoscopy. Since most patients with nonvariceal upper GI bleeding will bleed from a peptic ulcer, this seems a logical and justifiable approach. This issue was addressed by another Cochrane meta-analysis published in 2006⁴⁹ that had included four RCTs published in full^{[20], [50], [51] and [52]} and the preliminary results of a fifth RCT by Lau and colleagues⁵³ which, at that time, was only available as an abstract. These RCTs had studied the efficacy of PPI treatment initiated on admission in patients with upper GI bleeding, after having excluded patients suspected of having variceal bleeding. The patients in the comparator arms had received either an H₂RA^{[51] and [52]} or placebo.^{[20], [50] and [53]} Endoscopy was performed within 24^{[20], [50], [51] and [53]} or 48 hours⁵² of admission. Three trials had been conducted in Europe,^{[20], [50] and [52]} one in Turkey⁵¹ and one in China (Hong Kong).⁵³ Pooling of the trials that reported outcomes for all patients who were randomized (four trials^{[20], [50], [51] and [52]} with 1512 patients in total) showed no evidence of an effect of pre-endoscopic PPI treatment on mortality (OR 1.12; 95% CI 0.72–1.73), rebleeding (OR 0.81; 95% CI 0.61–1.09), requirement for endoscopic hemostatic treatment at subsequent index endoscopy (OR 0.93, 95% CI 0.53–1.64) or surgical intervention rates (OR 0.96, 95% CI 0.68–1.35). There was no detectable heterogeneity among the trials. The effect of pre-endoscopic PPI treatment on the abovementioned outcomes remained nonsignificant when the trials were pooled separately according to the type of control treatment (H₂RA or placebo), the route of PPI administration (intravenous or oral), and the methodological quality of the trials. The only significant effect of pre-endoscopic PPI treatment was a reduction in the proportion of patients with active bleeding, a nonbleeding visible vessel, or an adherent clot found at the index endoscopy; pooled (unweighted) rates were 37.2% on PPI treatment and 46.5% on control treatment; OR 0.67; 95% CI 0.54–0.84.⁴⁹

Following the publication of the above meta-analysis, Lau and colleagues⁵⁴ published in full the final results of their RCT, which was the only trial that had used high-dose pre-endoscopic PPI treatment (80 mg intravenous bolus of omeprazole followed by continuous intravenous infusion of 8 mg/h). Their results are in general agreement with those of the meta-analysis of the four previously published trials. PPI treatment before endoscopy compared with placebo did not significantly affect mortality, rebleeding, or surgical intervention rates.⁵⁴ Among the 377 patients with peptic ulcer bleeding (60% of study population), pre-endoscopic PPIs accelerated the resolution of stigmata of recent bleeding as assessed at index endoscopy performed at a mean of 15 hours after treatment initiation.⁵⁴ The requirement for endoscopic hemostatic treatment at index endoscopy among all patients who were randomized with nonvariceal upper GI bleeding was also significantly reduced by pre-endoscopic PPI treatment compared with placebo (19% versus 28%; P = .007). Subgroup analysis showed that this beneficial effect was driven by, and confined to, patients with bleeding from peptic ulcers; there was no difference in the need for endoscopic therapy among patients with other sources of bleeding.⁵⁴

Hence, is the glass half empty or half full? Does the downstaging of the stigmata of recent bleeding and possibly the reduction of the requirement for endoscopic hemostatic treatment at index endoscopy justify the use of pre-endoscopic PPIs, despite the lack of evidence of an effect on clinical outcomes? Given the efficacy of endoscopic hemostatic treatment and postendoscopic PPI treatment, it is unlikely that an RCT adequately powered to detect a further reduction in rebleeding or mortality will ever be conducted. Analyses of large multicenter databases using propensity score methods to adjust for overt biases may be able to detect such effects,⁵⁵ but such studies have yet to be conducted. The two small retrospective observational studies, both from Canada, have produced different results among them. Andrews and colleagues⁵⁶ reported that intravenous PPI treatment before endoscopy did not affect clinical outcomes in patients with peptic ulcers with high-risk endoscopic stigmata of recent hemorrhage. Keyvani and colleagues⁵⁷ found that pre-endoscopic PPI (which had been administered orally in 91% of the cases) significantly reduced mortality, rebleeding, and surgical intervention rates in patients with nonvariceal upper GI bleeding.

Cost-effectiveness analyses have not been in full agreement among them. Initiation of high-dose intravenous PPI treatment before endoscopy was both marginally more costly and effective than after endoscopy in the US and Canadian settings⁵⁸ but was less costly and more effective in Hong Kong.⁵⁹

A consensus meeting stated that “in patients awaiting endoscopy empiric therapy with a high-dose PPI should be considered,” although only 40% of the participants accepted this recommendation without reservations.⁴

Summary

Pre-endoscopic administration of PPIs in patients with nonvariceal upper GI bleeding is still of controversial efficacy. It downstages the severity of the endoscopic signs of recent bleeding and may reduce the requirement for endoscopic hemostatic therapy at index endoscopy. However, there is no evidence of an effect on mortality, rebleeding, or surgical intervention rates.

In contrast, the efficacy of PPIs in endoscopically diagnosed peptic ulcer bleeding is supported by high-quality evidence from numerous RCTs and meta-analyses of RCTs. PPIs compared with H₂RAs or placebo consistently reduce rebleeding rates regardless of dose, route of administration, application or not of endoscopic hemostatic treatment, and geographic location. Surgical intervention rates and the need for further endoscopic hemostatic treatment are also reduced by PPI treatment, although the results are not as robust as those for rebleeding. There is no evidence of an overall effect of PPI treatment on all-cause mortality. However, all-cause mortality is reduced among patients with high-risk endoscopic signs and among trials that had been conducted in Asia. The optimal dose and route of PPI administration has yet to be determined.

There was no financial support for this work.

Financial disclosure obligations: Dr Leontiadis has received speaker honoraria and reimbursement for expenses to attend scientific meetings from AstraZeneca, Sanofi-Aventis, Glaxo-SmithKline, and Janssen-Cilag. Dr Howden has served as a consultant to TAP Pharmaceuticals Inc, Takeda Global Research and Development, Santarus, Novartis, Otsuka, Biovail, Extera Partners, and KV Pharmaceuticals. He has received speaking honoraria from Santarus, AstraZeneca, and Otsuka. He has received research grant support from AstraZeneca.