Article Text
PDF
Abstract
In most regions of the world, antimicrobial resistance has increased to the point where empirical standard triple therapy for Helicobacter pylori eradication is no longer recommended. The treatment outcome in a population is calculated as the sum of the treatment success in the subpopulation with susceptible infections plus treatment success in the subpopulation with resistant infections. The addition of bismuth (ie, 14-day triple therapy plus bismuth) can improve cure rates despite a high prevalence of antimicrobial resistance. The major bismuth effect is to add an additional 30%–40% to the success with resistant infections. The overall result is therefore dependent on the prevalence of resistance and the treatment success in the subpopulation with resistant infections (eg, with proton-pump inhibitor–amoxicillin dual therapy). Here, we explore the contribution of each component and the mechanisms of how bismuth might enhance the effectiveness of triple therapy. We also discuss the limitations of this approach and provide suggestions how triple therapy plus bismuth might be further improved.
- HELICOBACTER PYLORI
- HELICOBACTER PYLORI - TREATMENT
- PROTON PUMP INHIBITION
- ANTIBIOTICS
Statistics from Altmetric.com
Introduction
Helicobacter pylori eradication triple therapy consists of a proton-pump inhibitor (PPI), amoxicillin and a third drug (ie, either clarithromycin, metronidazole, a fluoroquinolone or possibly rifabutin). Because of increased antimicrobial resistance, empirical therapy with clarithromycin, levofloxacin and metronidazole triple therapies is no longer recommended.1–6 However, recent studies from China and Europe report that the addition of bismuth (ie, triple therapy plus bismuth) can result in excellent cure rates despite antimicrobial resistance.7–11 Here, we explore the mechanisms of enhanced effectiveness, its limitations and how triple therapy plus bismuth might be further improved.
The effect of resistance on traditional triple therapy
Treatment effectiveness is the sum of those with susceptible infection plus effectiveness among those with resistant infections.12–15 Resistance to clarithromycin, metronidazole, levofloxacin or rifabutin is essentially all-or-none, as functionally resistance reduces triple therapy to amoxicillin and PPI dual therapy.4 Those with susceptible strains receiving 14-day triple therapies typically achieve cure rates of >95%, whereas the cure rate in the resistant population typically ranges from 0% to 40%. For example, with the prevalence of clarithromycin resistance of 20%, the population cure rate would equal the cure rate with susceptible infections (eg, 97%) times the proportion with susceptible infections (eg, 80%) plus the cure rate with resistant strains (eg, 20%) times the success rate with resistant strains (eg, ∼10% for a US population) or ((80×0.97)+(20×0.10)=77.6+2) resulting in a cure rate of 79% (per protocol, PP).
Modern experience with triple therapy plus bismuth
In recent studies, the addition of bismuth to triple therapy with levofloxacin or clarithromycin produced high cure rates despite the presence of antimicrobial resistance.7–11 Four studies also provide susceptibility data.7–9 ,11 One study used omeprazole 20 mg, amoxicillin 1 g, clarithromycin 500 mg and bismuth potassium citrate 220 mg, all given two times per day for 7 or 14 days.8 Clarithromycin resistance was estimated to be 18%. The cure rate with 14-day therapy was 97.4% PP and 93.7% intention to treat (ITT; figure 1).8 With susceptible infections, the cure rates with 7 and 14 days were 92.5% and 100%, respectively. With clarithromycin-resistant strains, the cure rate with 14-day therapy was 84.6% rather than the expected rate of approximately 40%.
Cure rates of 7-day and 14-day bismuth plus triple therapy in a population where the prevalence of clarithromycin resistance was 18%.8 As expected the cure rates for 14-day therapy were higher than for 7-day therapy, however, despite the cure rates for 14-day therapy were greater than 90% despite resistance. The cure rate with clarithromycin (clari)-resistant infections was expected to be approximately 40% but was increased markedly by the addition of bismuth. *p<0.05 for comparisons. ITT, intention to treat; PP, per protocol.
Another study used lansoprazole 30 mg, amoxicillin 1 g, two times per day and levofloxacin 500 mg once daily, with or without bismuth potassium citrate 220 mg two times per day for 14 days (figure 2).7 Levofloxacin resistance was estimated to be 30.3%. Treatment success with susceptible infections was approximately 97% with or without bismuth. The cure rate with resistant strains of 37.5% improved to 70.6% with bismuth.7 A recent trial added bismuth to levofloxacin-containing triple therapy and cured 85.4% PP (83% ITT) despite levofloxacin resistance of approximately 40%.11
Cure rates of two times per day 14-day levofloxacin (levo) containing triple therapy with or without the addition of bismuth.7 The cure rates both intention to treat (ITT) and per protocol (PP) were higher with the addition of bismuth. Susceptibility testing showed that with susceptible infections the results were high and equivalent but markedly different with resistant strains, where the addition of bismuth increases the cure rate approximately by 30%. *p<0.05 for comparisons.
Another study evaluated lansoprazole 30 mg, amoxicillin 1 g and bismuth potassium citrate 220 mg, two times per day plus metronidazole 400 mg four times daily (1600 mg) or clarithromycin 500 mg two times per day (figure 3).9 The dose of metronidazole was based on prior evidence that at least 1500 mg of metronidazole could potentially overcome resistance.16 Approximately, 95% cure rates PP were achieved with both clarithromycin and metronidazole (figure 3).9 The cure rate with metronidazole susceptible and resistant infections was similar (eg, approximately 90%; figure 3),9 suggesting that higher doses of metronidazole could overcome metronidazole resistance.
Cure rates of 14-day bismuth plus metronidazole (met) or clarithromycin (clari) triple therapy.9 The cure rates were similar with both despite relatively high prevalences of resistance (26.5% clarithromycin and 45.5% metronidazole). The results with clarithromycin paralleled the results shown in figure 1, in contrast, with metronidazole (400 mg four times per day) the results with resistant and susceptible strains were equivalent suggesting with being able to overcome metronidazole resistance. *p<0.05 for comparisons. ITT, intention to treat; PP, per protocol.
The high rate of success with 14-day triple therapy with susceptible strains (ie, >95%; figures 1⇑–3)7–9 ,11 prevented assessing whether bismuth further improved the outcome with susceptible infections. In China, the expected cure rates with resistant strains without bismuth are in the range of 35–40%.7 ,8 ,17 The addition of bismuth appeared to increase the cure rate by an additional 30–40%. As the effect of clarithromycin and levofloxacin resistance is all-or-none, one can conclude that the majority of the benefit was the result of improved effectiveness with resistant infections (ie, with the dual PPI–amoxicillin component of triple therapy). The issue is more complex with metronidazole as resistance can be partially overcome increasing the dose and duration of therapy.16 ,18 ,19 However, as discussed later, the concept of overcoming resistance simply describes that a high proportion of those with resistant infections were cured but cannot distinguish whether the organisms being killed by metronidazole versus some killed by metronidazole and some by bismuth.
Prior experience with triple therapy plus bismuth therapy
To our knowledge, the first study of triple therapy plus bismuth was performed in Italy as the addition of a PPI to bismuth–amoxicillin–tinidazole triple therapy.20 Therapy consisted of lansoprazole 30 mg once daily, amoxicillin 1 g thrice daily, tinidazole 500 mg two times per day and bismuth subcitrate 240 mg two times per day for 14 days and cured 90%.20 In 1997, a Spanish group ‘devised a new two times per day triple therapy by adding bismuth to a PPI-based triple therapy’.21 Their 7-day therapy consisted of two times per day omeprazole 20 mg, amoxicillin 1 g and tinidazole 500 mg along with bismuth subcitrate 240 mg. The PP cure rate was 86% (95% CI 71 to 94) and ITT was 84.1%. Susceptibility was not examined and the regimen was considered ‘probably not good enough for its use to be recommended or to justify further trials’.21
A recent European (Spanish and Italian) study evaluated esomeprazole 40 mg, amoxicillin 1 g and bismuth 240 mg all two times per day plus, levofloxacin 500 mg once daily for 14 days.10 PP and ITT eradication rates were 91.1% (95% CI 87% to 95%) and 90% (95% CI 86% to 94%). Although levofloxacin resistance had been increasing in the areas where the study was done, susceptibility was not examined preventing direct assessment of the benefits accrued by adding bismuth. We estimated the prevalence of levofloxacin resistance as approximately 15%, which is consistent with the author’s estimate.10 ,22
A study from Thailand used lansoprazole 30 mg, amoxicillin 1 g and bismuth subsalicylate 1048 mg all two times per day plus clarithromycin 1 g slow release once daily for 7 or 14 days.23 The cure rate for 7 days was 92% and for 14 days was 96%. Clarithromycin resistance was 0% for the 7-day group and 4% for the 14-day group making the additional effect of bismuth on outcome impossible to assess.
There have been other examples of triple therapy plus bismuth from Iran, Turkey, India and China24–39 (table 1), all regions in which high cure rates with triple therapy are rare and metronidazole resistance is generally high (eg, >40%).1 ,40–44 None assessed susceptibility and the results overall have been variable. Most have used metronidazole 500 mg two times per day. In Iran and Turkey, antimicrobial resistance in general is common and both countries typically report low cure rates irrespective of the therapy offered (eg, various triple therapies, bismuth quadruple and the non-bismuth quadruple therapies).1 ,3 ,44 High rates of antimicrobial resistance and even multidrug resistance have also been reported from both areas.13 We recently reviewed the published studies using bismuth quadruple therapy from Turkey and Iran and confirmed that the results were variable.13 It is unclear why, however, in developing countries it is more common to find drugs with reduced activity, outdated and even counterfeit,45 thus adding an additional unquantifiable variable. The lack of susceptibility data also makes it impossible to pinpoint why triple therapy plus bismuth produced such poor results in these areas. Below we offer our hypotheses to explain these different results based on the prevalence of resistance, the doses of metronidazole commonly used, issues with compliance and duration of therapy. One small study of levofloxacin triple therapy plus bismuth from Taiwan was complicated by a high prevalence of amoxicillin resistance46 and another from Hong Kong by multiple prior therapies.47
Summary of triple therapy plus bismuth from developing countries without susceptibility data
Understanding the variation in outcome with triple therapy plus bismuth
Rational use of bismuth plus triple therapy to improve outcome with resistant infections requires understanding the reasons for effectiveness and failure. We examine the contributions of each component of therapy separately and in combination using data from early attempts to identify effective anti-H. pylori therapies, which involved many different antimicrobials and antiulcer agents evaluated as single drugs and as combinations of 2, 3 and ultimately 4 drugs given simultaneously or sequentially. For details, the reader is referred to overlapping reviews by Axon in 198948 and 1991,49 and by Borsch and Graham,50 Chiba et al,51 van der Hulst et al52 and Wermeille et al.53
Anti-H. pylori therapy began with bismuth monotherapies and progressed to include two-drug (dual) and three-drug (triple) therapies.49–51 Treatment success with bismuth subcitrate alone (data from 338 subjects) averaged 20% (range 7–43%), with bismuth subsalicylate (87 subjects) averaged 16% (range 0–23%), with amoxicillin alone (106 subjects, average 17%, range 0–30%) and with metronidazole alone (16 subjects, average 6%, range 0–10%).49 Dual-therapy results were only slightly better: with bismuth subcitrate plus amoxicillin (267 subjects), the cure rate averaged 44% (range 0–67%) and with bismuth subsalicylate plus amoxicillin (57 subjects, average 33%, range 25–58%). Results with bismuth and an imidazole were better than with amoxicillin: bismuth subcitrate plus metronidazole (141 subjects, average 73%, range 57–85%), plus tinidazole (106 subjects, average 62%, range 53–80%) and bismuth subsalicylate plus metronidazole 19 subjects (average 79%).49
Bismuth improved the treatment success with both amoxicillin and metronidazole. With amoxicillin the effect appears additive, whereas with metronidazole the effect was greater than with either alone and thus appeared synergistic. Importantly, in vivo and in vitro results frequently do not correlate (eg, bismuth plus amoxicillin appeared additive in vivo and synergistic in some in vitro experiments54 and additive in others55).
Combining three drugs (bismuth, amoxicillin and metronidazole triple therapy) produced cure rate similar to bismuth plus an imidazole. For example, six studies with 1-week or 2-week triple therapy and a total of 130 subjects achieved an overall average cure rate of 73% (range 55–90%).51 Wermeille et al53 subsequently compiled data from eight studies of bismuth, amoxicillin and nitroimidazole therapy, including four reported previously by Chiba et al, for a total of 162 subjects and an average cure rate of 74.7% (range 50–90%). A meta-analysis confirmed that the effect of metronidazole resistance was strongly influenced by duration of therapy and recommended that when resistance was present therapy should be given for longer than 7 days.19 Fischbach and Evans’16 meta-analysis of bismuth, amoxicillin and nitroimidazole triple therapy included 319 subjects in 19 arms including treatments ranging from 3 to 16 days. She reported treatment success of 83% (95% CI 67.7% to 97.3%) in the absence of metronidazole resistance. In the presence of metronidazole resistance, the weighted mean average fell to 44% (95% CI to 7.8% to 80.4%). Most importantly, there was a dose–response effect on efficacy with a weighted mean of 64% (95% CI 56% to 71%) when 1500 mg of metronidazole or more was given versus 17% (95% CI 0% to 59%) with doses below 1500 mg.
Why some studies report variable results with triple therapy plus bismuth
As shown above, in China with 14-day triple therapy the addition of bismuth appeared to add 30%–40% to the cure rate with resistant infections to produce a combined cure rate of 70%–80%. In western populations, PPIs are generally metabolised rapidly blunting their ability to achieve prolonged periods of high pH such that, in the face of clarithromycin resistance, the cure rates with clarithromycin triple therapy generally range between 0% and 20%. The additive 30%–40% improvement with bismuth then results in a combined cure rate below 60% and thus producing a relatively low result (see below). The differences in outcome with 14-day triple therapy plus bismuth with clarithromycin or levofloxacin therapy in recent studies can probably be explained by differences in the prevalence of resistance, the cure rate with resistant infections or both. However, these explanations are insufficient to explain the poor results reported from some series from Iran and Turkey. In those countries, the most commonly used regimen contained metronidazole 500 mg two times per day (table 1) and, as noted above, success with metronidazole resistance is both dose and duration dependent. With low-dose metronidazole, 500 mg two times per day, resistance is effectively all-or-none thus producing results similar to that obtained with clarithromycin resistance. However, the cure rate in the presence of resistance without bismuth can increase to >60% when the dose is 1500 mg or greater.13 ,16 ,18 The good success in China probably reflects the higher dose of metronidazole used.9
The overall effect is dependent on both the cure rate with resistant strains and the proportion with resistance
We used the H. pylori therapy nomogram to model the effect of antimicrobial resistance on outcome (figure 4A,B).12 In China, the cure rate for susceptible strains was approximately 97% and with resistant infections the cure rate was approximately 40%, which increased to approximately 70% with bismuth. As shown in figure 4A, the cure rate for the entire population is expected to remain 90% or greater until the prevalence of resistance exceeds 25%.12 Figure 4B shows data for a simulated western population where the cure rate with clarithromycin resistance is lower. The addition of bismuth was modelled as an increase in cure rate of 30% (ie, from 10% to 40%). In that population, the overall cure rate would fall below 90% when the prevalence of resistance exceeded approximately 12%.
Hp nomogram.12 (A) The nomogram plots the cure rate with susceptible infections on the left vertical axis and those with resistant infections on the right vertical axis. The proportion with resistance is shown on the horizontal axis. The cure rates are connected with a line allowing one to visualise the population cure rates for any prevalence of resistance. This plot is of a typical Asian population. The solid line shows the effect without bismuth with a 97% cure rate for susceptible infections and 40% for resistant strains.7 ,8 ,17 The dotted line shows the increase in the cure rate associated with the addition of bismuth to 70% with resistant infections. The cure rate of 90% is shown in more detail to facilitate identification of the prevalence of resistance that results in a fall in the population cure rate to below 90%. (B) This plot is of a typical western population with a high proportion with rapid proton-pump inhibitor metabolisers and a corresponding cure rate with dual therapy of only 20% (solid line). The dotted line shows the effect of the addition of bismuth. The cure rate of 90% is shown in more detail to facilitate identification of the prevalence of resistance that results in a fall in the population cure rate to below 90%. As shown, the addition of bismuth allows the prevalence of resistance that results in a cure rate of <90% to increase from approximately 10% to approximately 15%.
Figure 4A also illustrates how a high prevalence of metronidazole resistance (>40%–50%)1 ,44 ,56 coupled with the low cure rates obtained with metronidazole 500 mg two times per day could be responsible for the poor results reported from Turkey and Iran. In figure 4A, the expected outcome is the point where the treatment success line intersects the cure rate desired. For example (figure 4B), the overall cure rate (PP) falls below 80% when metronidazole resistance increases >35% (arrow). The very low cure rates reported in some trials suggest an even more complex situation such as the presence of amoxicillin resistance,56 poor cure rates with resistant infections, poor patient adherence and possibly one or more drugs with poor bioavailability. For example, in Turkey, the addition of bismuth failed to improve the results with PPI, metronidazole and tetracycline therapy,13 ,57 suggesting poor bioavailability of at least one component. Study interpretation requires susceptibility data.
Understanding how bismuth improves cure rates despite antimicrobial resistance
A variety of bismuth salts successfully used for H. pylori eradication appear to provide similar results despite differences in solubility and crystalline structure.58–60 The bismuth species most responsible for antimicrobial effect may also vary depending upon the intragastric pH. With the exception of bismuth citrate, commonly used bismuth preparations (subsalicylate and nitrate) are almost completely insoluble in water. In the stomach, bismuth compounds interact with gastric acid and form an amorphous precipitate consisting largely of bismuth oxychloride, the most active but very poorly soluble antimicrobial species.61–63
The antimicrobial effect of bismuth is rapid in an acidic stomach. Marshall et al compared gastric histology and ultrastructural changes of H. pylori following oral doses of liquid and tablet formulations of bismuth subcitrate or subsalicylate. Within 30 min, bismuth particles were noted within and on the surface of the organism as well as in the plane between the cell wall and the cytoplasmic membrane.64 By 1 h, most H. pylori had detached from the gastric epithelial cells and had undergone structural degradation. After 24 h of bismuth administration, only subcellular bacterial fragments remained visible. Bacterial damage was most evident near the surface of the stomach and less within the gastric pits possibly providing sanctuaries from which H. pylori might recolonise the stomach.64 ,65
The mechanisms responsible for rapid destruction of H. pylori remain unclear. A large number of in vitro studies have evaluated the effect of bismuth on various critical H. pylori functions.59 However, in vivo and in vitro observations correlate poorly possibly because in vivo studies are done in acidic conditions and most in vitro studies require neutral pH.59 ,63 ,66–72 Armstrong et al73 evaluated bismuth subcitrate and antibiotics (separately) on viability and ultrastructure of H. pylori in vitro. Amoxicillin produces a rapid onset of morphological changes and viability typically associated with β-lactams confirming the validity of approach. The bismuth concentrations used (1000 and 2400 mg/L of bismuth subcitrate as DeNol powder) included the intragastric concentration following administration of one DeNol tablet. The minimal inhibitory concentration (MIC) for these bismuth compounds assessed in vitro was estimated to be <25 mg/L.64 At neutral pH in vitro, Armstrong et al were unable to identify significant ultrastructural differences between bismuth treated and untreated H. pylori over a 24 h observation period. However, they confirmed the presence of electron-dense particulate bismuth deposits in the space between the cell wall and cytoplasmic membranes and directly adherent to the cytoplasmic membranes. The proportion of organisms with visible bismuth increased over time and reached 100% by 24 h. While there was no change in viability during the first 8 h of incubation, a 3 log drop in viability was evident at 24 h confirming bacteriocidal activity of bismuth. The most recent in vitro study also used bismuth subcitrate and reported that bismuth impeded proton entry into the organisms potentially impairing their ability to respond to acid and enhancing the efficacy of growth-dependent antibiotics.72 One suspects that this finding might be related to the deposition of bismuth into the periplasmic space described previously.64 ,73
An in vitro–in vivo correlation attempt assessed bismuth at different pH’s in vitro.74 As noted above, in vitro bismuth subcitrate was not associated with ultrastructural changes during incubation for 24 h. At pH 5, bismuth caused very minor cell-wall changes. At pH 4, exposure for only 10 min resulted in focal ballooning and major structural damage to the bacteria. At pH 3, the damage was marked and associated with cytoplasmic coagulation and flagellar fragmentation. At pH 2, the finding closely approximated those seen in vivo. The authors used DeNol bismuth subcitrate powder obtained from the manufacturer and noted that it was soluble in water.74 ,75 We previously showed that the antibacterial effect of bismuth was most likely related to bismuth oxychloride.61 ,62 Importantly, no in vitro study has attempted to identify the bismuth species present or the one(s) primarily responsible for the effects demonstrated.
In vitro experiments require a bismuth compound that is at least partially soluble in that environment. Phillips et al55 studied the solubility of colloidal bismuth subcitrate at pH 1.4 and after the pH was adjusted to 7. Even at pH 1.4, approximately 10% of bismuth was present in the supernatant and all was filterable (<229 nm). The amount in the supernatant increased to 45% at pH 7 with approximately 10% remaining filterable. In contrast, bismuth subnitrate was essentially not soluble or filterable at either pH. Overall, it seems likely that the bismuth species responsible for the in vitro effects at neutral pH likely differs from that responsible for effects in vivo. Elemental bismuth mirrors have been described forming on the agar plates used to assess the MIC of bismuth compounds in vitro, suggesting that a variety of bacteria or bacterial products can interact and release bismuth metal.64 It is unclear whether this is important in vivo.
In vivo studies have been limited and have not included patients in whom gastric acid secretion has been controlled. While it has been suggested that acid inhibition might reduce the effectiveness of bismuth therapy,74 ,76 clinically the combination of bismuth with an antisecretory agent such as ranitidine bismuth citrate proved to be effective as an antiulcer and anti-H. pylori drug.77 ,78 In addition, Pepto Bismol (bismuth subsalicylate) tablets also contain the antacid, calcium carbonate, and are effective in clearing H. pylori. Finally, PPIs clearly improve the outcome with bismuth quadruple therapy. However, PPIs rarely produce sustained high elevations in the intragastric pH. Recently, the tendency has been to increasingly inhibit acid secretion with the goal of enhancing dual-acid inhibitor—amoxicillin by maintaining the intragastric pH at 6 or above.79–81 Studies are needed that control pH in patients who receive bismuth after administration of a PPI or a potassium-competitive acid blocker to better understand the effects of bismuth relation to pH in vivo. If an acidic stomach is required for full effectiveness (eg, to generate bismuth oxychloride), options might include administering the bismuth salt with citric acid, malic acid or even Coca Cola to temporarily acidify the stomach,82 or administering bismuth oxychloride tablets.
The ideal compound, formulation, dose, dosing interval and relation to meals for administering bismuth remain unclear. It has also been suggested that bismuth be administered with meals to enhance it distribution throughout the stomach.59 ,83 ,84 The recent clinical trials in China have administered bismuth 30 min before meals. Early clinical studies suggested that treatment success with administration of bismuth four times daily was superior to two times per day.85 However, two times per day administration, 5–8 bismuth subsalicylate tablets, and low-dose bismuth citrate (108 mg) given four or five times daily have all proved to be effective when given as part of multidrug therapies.86 ,87 Although head-to-head comparisons are lacking, current four-drug regimens appear effective when given as two times per day therapy.13
Improving the results with bismuth by increasing the effect of dual PPI–amoxicillin therapy with resistant infections
As noted above, the population cure rate is dependent on the cure rate with susceptible strains, the cure rate with resistant strains and the proportion with resistant infections. The ability to achieve a population cure rate of >90% increases along with the increasing cure rate with resistant strains (ie, with PPI–amoxicillin dual therapy) (figure 4). Increasing the effectiveness with PPI–amoxicillin dual therapy should increase the effectiveness of bismuth cotherapy.
PPI–amoxicillin dual therapy was introduced by Unge et al88 in the late 1980s and followed by studies searching for the optimum doses and formulations of omeprazole and amoxicillin.48–50 Treatment success with dual PPI–amoxicillin therapy has varied widely and especially geographically. In 1996, van der Hulst et al52 analysed dual therapy data from 1260 patients (34 treatment arms); treatment success averaged 54.8% (range 52–57%), omeprazole two times per day plus amoxicillin in 2275 subjects (53 treatment arms) averaged 61.9% (range 60–64%). Omeprazole doses of 20 or 40 mg once or two times per day provided similar results and increasing dose from 20 to 60 mg two times per day did not reliably improve outcome.89 For example, Fleischmann et al90 gave 40 mg of omeprazole two times per day or thrice daily along with 2 g of amoxicillin for 14 days to patients with duodenal ulcer; approximately 80% were cured PP with both regimens. We gave 120 mg of a long-acting lansoprazole (dexlansoprazole) and 1 g amoxicillin two times per day for 14 days and achieved a cure rate of only 53.8%.91 Extending the duration of therapy does not reliably improve outcome; therapy with omeprazole 20 mg and 1 g of amoxicillin two times per day for 6 weeks achieved a cure rate of 62.5%.92
In the 1990s, during the height of interest in PPI–amoxicillin dual therapy, Labenz et al93 reported their experience with 405 subjects and 11 different protocols using 40–80 mg of omeprazole and amoxicillin (2–3 g daily) given for 1 or 2 weeks. The independent factors predicting treatment failure were poor adherence, short duration of therapy, smoking and omeprazole pretreatment. The adverse role of pretreatment with omeprazole was confirmed by Bayerdörffer et al89 but not by Miehlke et al94 and that issue remains unresolved. However, repeating dual therapy after failure with PPI–amoxicillin therapy proved unsuccessfully in two studies.95 ,96 In our studies, we also confirmed that development of resistance to amoxicillin or a change in the MIC for omeprazole was not responsible (unpublished data). Repeated treatment failure is consistent with the notion that treatment failure enriches the population in patients who originally had an inadequate pH response (eg, CYP2C19 rapid metabolisers, natural hyper secretors, etc). Pretreatment with omeprazole can also stimulate acid rebound and thus potentially bias the outcome towards increased treatment failures.
In the late 1990s, PPI–amoxicillin dual therapy was generally abandoned being unable to reliably achieve 90% or greater cure rates. However, work has continued on identifying factors influencing the effectiveness of PPI therapy in healing ulcers and erosive oesophagitis, especially the relationship with role of CYP2C19 polymorphisms and the metabolism of PPIs. Slow PPI metabolism results in prolonged duration of PPI effect resulting in enhanced ulcer cure rates and prolonged increases in gastric pH, which increases the effectiveness of dual therapy.80 ,81 ,97–103 Current data suggest the effectiveness of dual therapy most depends on the ability to consistently obtain and maintain an intragastric pH of 6 or greater.80 ,100 ,102 ,104–106
It has long been known that the effectiveness of β-lactam antibiotics is pH dependent and that β-lactams are most effective in a neutral pH environment.100 ,107 Elevated pH is also associated with improved stability of amoxicillin.108 Savarino et al tested different doses of omeprazole for maintaining the intragastric pH at 6 or above with single doses of 20, 40 and 60 mg. The 40 and 60 mg doses were equivalent and only maintained a pH of 6 or greater for 13 h and at pH 7 for 5 h.102 Yang et al81 evaluated the roles of once or two times per day omeprazole dosing and 250 and 500 mg of amoxicillin four times daily, CYP2C19 genotype, intragastric pH, and presence of gastric corpus inflammation on outcome of 14-day dual omeprazole–amoxicillin therapy. Based on the concept that the bactericidal activity of β-lactam antibiotics is time rather than concentration dependent,109 they used every 6 h amoxicillin dosing to attempt to maintain the MIC within the effective range.81 Furuta et al have long emphasised that in order to reliable achieve cure rates over 90%, it is important to administer amoxicillin approximately every 6 h and maintain the pH at 6 or above.80 ,110 Although several studies have suggested that dosing of amoxicillin such as every 8 h may also be effective,81 ,111 these studies have been done in Asia where atrophic gastritis and slow CYP2C19 metaboliser’s are common.
The need to maintain bactericidal activity throughout treatment is based on an old theory that has repeatedly been challenged (reviewed in ref. 112). However, the requirement may be organism-specific or disease-specific. At low pH, H. pylori enter a non-replicative state that can theoretically be reversed by increasing the pH to 6 or above.105 ,106 In the Yang et al81 study mentioned above, the factors that enhanced prolonged elevated intragastric pH were associated with improved cure rates including slow CYP2C19 metaboliser genotype, corpus gastritis and two times per day rather than once a day amoxicillin administration. Of note, only the slow CYP2C19 metabolisers reliably achieved cure rates >90%. The most comprehensive recent study (also from Asia) used four times daily rabeprazole and 750 mg of amoxicillin along with prohibiting acid foods and provided high cure rates irrespective of CYP2C19 genotype.113 However, in Europe, Miehlke et al114 gave omeprazole 40 mg four times daily before meals and amoxicillin 750 mg every 6 h for 2 weeks to 41 prior treatment failures who had dual clarithromycin and metronidazole resistance. Only 31 were cured with a PP cure rate of 83.8% (95% CI 68% to 94%) and ITT 75.6 (95% CI 60 to 88). All were susceptible to amoxicillin pretreatment and all treatment failures remained susceptible. The fact that 97.6% had failed at least two previous attempts at therapy suggests that rapid PPI metabolisers were likely over represented. In Western societies where rapid and ultrarapid CYP2C19 metabolisers are more frequent, it is difficult or even impossible to reliably achieve the goal of a pH of 6 or greater with oral PPI therapy.115 It is important to note that no US trial of dual therapy has been successful.91 ,95 ,96 ,116–120 However, none have used every 6 h administration of the amoxicillin and PPI.
While work continues on providing an effective dual antisecretory drug–amoxicillin dual therapy, it appears likely that a reliable regimen will require use of high dose of PPIs that are relatively unaffected by CYP2C19 metabolism (eg, esomeprazole or rabeprazole),121–123 long-acting PPIs, the combination of a PPI and an antacid,115 or a potassium-competitive acid blocker.124 ,125 The initial study of the potassium-competitive acid blocker vonoprazan in Japan used 20 mg of vonoprazan two times per day, clarithromycin 200 or 400 mg two times per day and amoxicillin 750 mg two times per day for 7 days.126 Both vonoprazan and PPI-containing triple therapies cured 97% with susceptible infection. However, among those with clarithromycin resistance, vonoprazan and amoxicillin cured only 82% (compared with 40% with 30 mg of lansoprazole two times per day). It is unclear whether the failure to achieve 90% or greater cure rates was related to inability of 20 mg of vonoprazan to maintain the pH at 6 or above, whether more frequent amoxicillin dosing is required or both.124 ,125
Overall, it remains unclear whether it is critical to maintain the pH at 6 or above and a continuous blood level of amoxicillin above the MIC or whether it is critical to have the blood level of amoxicillin above the MIC only during the time the pH is above 6. To address this question will likely require the use of potassium-competitive acid blockers. However, the addition of bismuth to current two times per day vonoprazan, amoxicillin and clarithromycin or vonoprazan and amoxicillin should further increase the cure unless, as mentioned earlier, the high pH reduces the effectiveness of bismuth. High-dose PPI (eg, esomeprazole or rabeprazole or dexlansoprazole)-containing triple therapy plus bismuth should also theoretically be able to meet or achieve the results currently obtained in China and provide reliably successful two times per day dosing. Further trials are needed from western countries.
The outcome of the addition of bismuth to triple therapy
The cure rate with triple therapy is calculated as the sum of the cure rate with susceptible infections plus the cure rate with resistant strains.
The cure rate with resistant infection is calculated as the sum of the cure rate with resistant infections plus the increase with bismuth.
As the cure rate with susceptible strains is very high, any significant improvement must come from improved cure rates with resistant infections.
Cure rates with clarithromycin, levofloxacin and rifabutin in resistant infections are all-or-none and cannot be improved by increasing the dose or duration of these drugs.
The cure rate with a proton-pump inhibitor (PPI) plus amoxicillin, the regimen seen by resistant infections, can be improved by increasing the time the gastric pH is above 6.
The effect of bismuth is primarily to improve the results of the PPI amoxicillin component, appears additive and is in the range of 30%–40% increase with 100% resistant infections. The actual effect depends on the prevalence of resistance.
Achieving high cure rates despite resistance
Bismuth plus triple therapy is not a panacea as its effectiveness is sensitive to cure rate with resistant infection and to the prevalence of resistance.
Further improvements depend upon being able to reliably increase the cure rate with resistant infections, by increasing the cure rate with bismuth, or both.
In order to achieve an overall cure rate of >90% when the prevalence of resistance is below 15%–18%, the combined effects of proton-pump inhibitor plus amoxicillin dual therapy and bismuth therapy must achieve a cure rate of at least 50%–60%. A combined cure rate of 80% will tolerate 40% resistant infections.
This regimen is excellent for increasing the cure rates for populations; however, those with resistant strains may still experience unacceptable low cure rates (eg, 60%–80%).
While, in the presence of resistance, there is no advantage to increasing the doses of clarithromycin, levofloxacin or rifabutin, the cure rate with metronidazole is increased when at least 1500 mg/day is given.
Acknowledgments
The authors thank Dr Lori Fischbach for her input regarding metronidazole resistance and for providing references regarding triple therapy plus bismuth from developing countries what we had been unable to find.
References
Footnotes
MPD and HL contributed equally and are co-first authors.
Funding DYG is supported in part by the Office of Research and Development Medical Research Service, Department of Veterans Affairs, Public Health Service grants DK067366 and DK56338, which funds the Texas Medical Center Digestive Diseases Center. HL is supported by the grant of National Natural Science Foundation of China (81170355 and 81370592).
Competing interests DYG is a paid consultant for RedHill Biopharma regarding novel H. pylori therapies, for Otsuka Pharmaceuticals regarding diagnostic testing and for BioGaia regarding use of probiotics for H. pylori infections. DYG has received royalties from Baylor College of Medicine patents covering materials related to 13C-urea breath test. MPD receives research support from BioGaia.
Provenance and peer review Commissioned; externally peer reviewed.