Second-Line Rescue Therapy of Helicobacter Pylori Infection

Abstract

Helicobacter pylori infection is the main known cause of gastritis, gastroduodenal ulcer disease and gastric cancer. After more than 20 years of experience in H. pylori treatment, however, the ideal regimen to treat this infection has still to be found. Nowadays, apart from having to know well first-line eradication regimens, we must also be prepared to face treatment failures. Therefore, in designing a treatment strategy we should not focus on the results of primary therapy alone, but also on the final (overall) eradication rate. The choice of a ‘rescue’ treatment depends on which treatment is used initially. If a first-line clarithromycin-based regimen was used, a second-line metronidazole-based treatment (quadruple therapy) may be used afterwards, and then a levofloxacin-based combination would be a third-line ‘rescue’ option. Alternatively, it has recently been suggested that levofloxacin-based ‘rescue’ therapy constitutes an encouraging second-line strategy, representing an alternative to quadruple therapy in patients with previous PPI-clarithromycin-amoxicillin failure, with the advantage of efficacy, simplicity and safety. In this case, quadruple regimen may be reserved as a third-line ‘rescue’ option. Finally, rifabutin-based ‘rescue’ therapy constitutes an encouraging empirical fourth-line strategy after multiple previous eradication failures with key antibiotics such as amoxicillin, clarithromycin, metronidazole, tetracycline, and levofloxacin. Even after two consecutive failures, several studies have demonstrated that H. pylori eradication can finally be achieved in almost all patients if several ‘rescue’ therapies are consecutively given. Therefore, the attitude in H. pylori eradication therapy failure, even after two or more unsuccessful attempts, should be to fight and not to surrender.

Introduction

Helicobacter pylori infection is the main known cause of gastritis, gastroduodenal ulcer disease and gastric cancer. After more than 20 years of experience in H. pylori treatment, however, the ideal regimen to treat this infection has still to be found [Vakil, 2009]. Consensus conferences have recommended therapeutic regimens that achieve H. pylori cure rates higher than 80% on an intention-to-treat basis [Malfertheiner et al. 2007, 2002; Howden et al. 1998]. However, several large clinical trials and meta-analyses have shown that the most commonly used first-line therapies — including proton-pump inhibitors (PPIs) plus two antibiotics — may fail in up to 20% of patients [Gisbert et al. 2007d, 2000b], and in the clinical routine setting, the treatment failure rate might be even higher. Moreover, during the last few years, the efficacy of PPI-based regimens seems to be decreasing, and several studies have reported intention-to-treat eradication rates lower than 75% [Paoluzi et al. 2006; Calvet et al. 2005; Gisbert et al. 2005b; Vakil et al. 2004; Hawkey et al. 2003; Veldhuyzen Van Zanten et al. 2003; Laine et al. 2000, 1998] and even lower than 50% [Altintas et al. 2004; Gumurdulu et al. 2004; Della Monica et al. 2002]. Antibiotic resistance to clarithromy-cin has been identified as one of the major factors affecting our ability to cure H. pylori infection, and the rate of resistance to this antibiotic seems to be increasing in many geographical areas [Egan et al. 2008; Megraud, 2004; Vakil et al. 1998].

Papers dealing with retreatment of H. pylori after failure are difficult to analyze due to several reasons [Axon, 2000]. Firstly, patients who fail with their first treatment probably include a higher percentage of individuals who are unreliable tablet takers, others who have resistant organisms and also the ‘constitutional’ group, where failure will be inevitable. On the other hand, some patients submitted for ‘rescue’ therapy have already had more than one previous treatment for H. pylori, and this circumstance is not always clarified in the protocols. Furthermore, the original primary treatments vary among the different studies, not only with respect to the antibiotic type, but also in respect to the dose and duration of the regimen. Finally, only a few studies have directly compared, in the same protocol, two or more second-line therapies [Gisbert, 2008; Gisbert and Pajares, 2002, 2005].

Several ‘rescue’ therapies have been recommended, but they still fail to eradicate H. pylori in more than 20% of the cases [Gisbert and Pajares, 2002], and these patients constitute a therapeutic dilemma [Gisbert, 2008; Gisbert and Pajares, 2005]. Patients who are not cured with two consecutive treatments including clari-thromycin and metronidazole will have at least single, and usually double, resistance [Romano et al. 2008; Megraud, 2004]. Furthermore, bismuth salts are not available worldwide anymore and, therefore, management of first-line eradication failures is becoming challenging. Currently, a standard third-line therapy is lacking, and European guidelines recommend culture in these patients to select a third-line treatment according to microbial sensitivity to antibiotics [Malfertheiner et al. 2007, 2002]. However, cultures are often carried out only in research centers, and the use of this procedure as ‘routine practice’ in patients who failed several treatments seems not to be feasible [Gisbert, 2008; Gisbert and Pajares, 2005; Zullo et al. 2003b; Gisbert and Pajares, 2002]. Therefore, the evaluation of drugs without cross-resistance to nitroimidazole or macrolides as components of retreatment combination therapies would be worthwhile [Graham, 2009].

Probiotics have been proposed as a useful adjunct. Some studies prescribing probiotics with H. pylori eradication therapy had no effect on the side-effect profile but did increase the rates of eradication [Kim et al. 2008]. However, other studies on concurrent probiotic administration suggested the inverse with better side-effect profiles but no increase in eradication or rates of compliance with therapy [Cremonini et al. 2002].

All these issues are important at the present time, but they will be even more relevant in a near future, as therapy for H. pylori infection is becoming more and more frequently prescribed. Therefore, the evaluation of second or third ‘rescue’ regimens for these problematic cases seems to be worthwhile. In designing a treatment strategy we should not focus on the results of primary therapy alone; an adequate strategy for treating this infection should use several therapies which, if consecutively prescribed, come as close to the 100% cure rate as possible [Calvet et al. 2001; De Boer and Tytgat, 2000; Gisbert, 2008; Gisbert et al. 2005a; Gisbert and Pajares, 2002].

The aim of the present manuscript will be to review the experience dealing with ‘nonrespon-ders’ to H. pylori eradication therapy. As, at present, the current most prescribed first-line regimens include a combination of PPI plus two antibiotics, the present review will focus only in ‘rescue’ regimen when these triple combinations fail. Bibliographical searches were performed in the PubMed (Internet) database including studies available until July 2009, looking for the following words (all fields): pylori AND (retreatment OR re-treatment OR rescue OR failure OR salvage OR second-line OR third-line OR fourth-line). References of reviews on H. pylori eradication treatment, and from the articles selected for the study, were also examined in search of articles meeting inclusion criteria (i.e. dealing with H. pylori ‘rescue’ therapies).

Is it necessary to perform culture after failure of the first eradication treatment?

Pretreatment antibiotic resistance is the most important factor in nonresponse to initial treatment [Dore et al. 2000; Houben et al. 1999; Peitz et al. 1999; Van Der Wouden et al. 1999; Tompkins et al. 1997]. Thus, the choice of a second-line treatment depends on which treatment was used initially, as it would appear that retreatment with the same regimen cannot be recommended [Parente et al. 2003]. If a clari-thromycin-based regimen was used, a metronida-zole-based treatment (or at least a clarithromycin-free regimen) should be used afterwards, and vice versa [Battaglia et al. 1998]. This recommendation is based on the observation that acquired bacterial resistance to metro-nidazole or clarithromycin results primarily from the previous treatment failure [Kobayashi et al. 2006; Peitz et al. 1999; Deltenre et al. 1998], and therefore ‘rescue’ therapies should avoid these antibiotics and use different combinations.

An antimicrobial susceptibility test for H. pylori before second-line treatment is sometimes performed, although whether the test is truly necessary remains unknown. Some authors have evaluated the efficacy of susceptibility-guided versus empiric retreatment for H. pylori after a treatment failure. In the study by Yahav et al. [2006a], patients in whom at least one treatment regimen for H. pylori eradication had failed underwent gastric biopsy and culture, and were retreated according to the in vitro susceptibility results. Findings were compared with those for control patients (where culture was unavailable). Susceptibility-guided retreatment was associated with better eradication rates (86%) than empiric treatment (63%). However, several methodological drawbacks exist in this study. Firstly, more than 50% of the patients received first-line eradication treatment with both clarithromycin and metronidazole (instead of including clarithromy-cin and amoxicillin), which is not the generally recommended combination; consequently, no logical empirical treatment remained afterwards (levofloxacin-based regimens were not available at that time). In this respect, when only the eradication rates in control (culture unavailable) patients treated with a first regimen of PPI-amoxicillin-clarithromycin followed by a second empirical quadruple regimen were considered (the generally recommended first- and second-line strategies), the success figures were not significantly different from those reported in patients receiving susceptibility-guided retreat-ment. Secondly, because this study was nonran-domized, there might have been heterogeneity among the two groups with respect to the treatment regimens prescribed by the treating physicians. Finally, this study was limited by the lack of susceptibility data for the controls, which restricted the ability to analyze the reasons why empiric therapy did not work as well as the susceptibility-guided protocol.

In a French multicenter study [Lamouliatte et al. 2003], patients in whom one previous H. pylori eradication therapy (mainly with PPI-amoxicillinclarithromycin) has failed were randomized to receive one of three empirical triple therapy regimens or a strategy based on antibiotic susceptibility. The empirical regimens were PPI-amoxicillin-clarithromycin (for 7 or 14 days) or PPI-amoxicillin-metronidazole (for 14 days).

In the susceptibility-based strategy, patients with clarithromycin-susceptible strains received PPI-amoxicillin-clarithromycin, whilst the others received PPI-amoxicillin-metronidazole. The eradication rates for empirical therapies were low, while the cure rate was higher (74%) for the susceptibility-based treatment. If the H. pylori strain was clarithromycin-susceptible (which occurred in approximately one-third of the cases), a high success rate was obtained with the PPI-clarithromycin-amoxicillin ‘rescue’ regimen. The study, however, was done in France, where bismuth is banned, so that the use of quadruple therapy with a PPI, bismuth, tetracycline and metronidazole, as recommended by the updated Maastricht Consensus Report [Malfertheiner et al. 2007], was not tested. In fact, as will be reviewed later, several studies have obtained relatively good results with this quadruple regimen empirically prescribed, with mean eradication rate of 77% (i.e. a similar figure than the 74% achieved for the susceptibility-based treatment in the present study). Thus, in this study, instead of not readministering any of the antibiotics against which H. pylori has probably become resistant, the authors insist on prescribing again clarithromycin (or metronida-zole) for the second-line treatment. Furthermore, statistically significant differences were not demonstrated when comparing the efficacy of the empirical PPI-amoxicillin-metronidazole and the susceptibility-based strategy, suggesting that the metronidazole-based combination may be an effective empirical alternative after failure of a clarithromycin-based combination.

In the updated Maastricht Consensus Report [Malfertheiner et al. 2007], it was recommended that culture and antimicrobial sensitivity testing should be routinely performed only after two treatment failures with different antibiotics. According to this statement, some studies have suggested that an antimicrobial susceptibility test for H. pylori before administering second-line treatment is not necessary. In this respect, in the study by Avidan et al. [2001], after failure of first-line eradication treatment, half of the patients were randomly assigned to treatment with a different PPI-based triple regimen regardless of the culture obtained, and the other half were assigned to treatment with PPI and two antibacterial agents chosen according to a susceptibility test; the authors found that the culture results did not influence the treatment protocol employed. Similarly, in the study by Miwa et al. [2003], patients with H. pylori infection for whom first-line treatment with a PPI-amoxicillin-clarithromycin regimen had failed were randomly assigned to two groups: those having or not having the susceptibility test before retreatment. For those patients in the susceptibility-test group, the authors used what they considered the best regimen based on susceptibility testing; while for those patients in the group with no susceptibility testing, PPI-amoxicillin-metronidazole was prescribed. The cure rates in the groups with and without susceptibility testing were not different.

Second-line H. pylori ‘rescue’ therapy after failure of one eradication treatment

‘Rescue’ regimen after PPI-clarithromycin-amoxicillin failure

PPI, amoxicillin and metronidazole. After failure of a combination of PPI, amoxicillin and clari-thromycin, a theoretically correct alternative would be the use, as second option, of other PPI-based triple therapy including amoxicillin (that does not induce resistance) and metronida-zole (an antibiotic not used in the first trial), and several authors have reported encouraging results with this strategy [Ueki et al. 2009; Murakami et al. 2008, 2006; Shirai et al. 2007; Matsuhisa et al. 2006; Nagahara et al. 2004; Shimoyama et al. 2004; Isomoto et al. 2003; Miwa et al. 2003]. However, in our experience, when this therapy has been administered twice-daily for 1 week, eradication rates lower than 50% have been obtained [Gisbert et al. 1999a]; the subsequently use of higher (three times per day) antibiotic doses, was followed only by a mild increase in eradication rate (58%), which was still unacceptable [Gisbert et al. 1999a]. However, if ranitidine bismuth citrate (RBC) is used instead of PPI, also plus amoxicillin and nitroimidazole, encouraging results have been reported (81% cure rate), although in this protocol antibiotics were administered for 14 instead of 7 days [Perri et al. 2001b]. In this same study, the readministration of clarithromycin, even when coprescribed with RBC, was associated with poor eradication rates. In the same way, Nagahara et al. [2001] studied a group of patients who, after failure of first-line PPI-clarithromycin-amoxicillin therapy, had received second-line therapy with the same regimen (for 14 days) or had received PPI-amoxicillin-metronidazole (for 10 days). The eradication rates for second-line therapy with the same regimen (thus readminis-tering clarithromycin) was of only 53%, while it was of 81% with PPI-amoxicillin-metronidazole. These observations underlie the idea that antibiotics, and specifically clarithromycin, should not be readministered in successive treatments.

Quadruple therapy. Another alternative, the use of a quadruple regimen (i.e. PPI, bismuth, tetra-cycline and metronidazole), has been generally used as the optimal second-line therapy after PPI-clarithromycin-amoxicillin failure, and has been the recommended ‘rescue’ regimen in several guidelines [Malfertheiner et al. 2007; Gisbert et al. 2000a; Lam and Talley, 1998]. Several studies have obtained relatively good results with this quadruple regimen, the results being summarized in Table 1 [Usta et al. 2008; Uygun et al. 2008; Chung et al. 2007; Navarro-Jarabo et al. 2007; Choung et al. 2006, Wu et al. 2006; Lee et al. 2005, 1999; Marko et al. 2005; Bilardi et al. 2004; Nista et al. 2003; Orsi et al. 2003; Perri et al. 2003, 2001a; Wong et al. 2003; Baena Diez et al. 2000; Michopoulos et al. 2000; Sicilia et al. 2000; Gasbarrini et al. 2000b; Gomollon et al. 1999; Gisbert et al. 1999a, 1999b; Elizalde et al. 1998]. Thus, the weighted mean eradication rate with this ‘rescue’ therapy, calculated from the studies included in the table, is of 77%. In this combination regimen, PPI should be prescribed in the usual dose and twice a day, colloidal bismuth subcitrate 120 mg four times per day, tetracycline 500 mg four times per day, and metronidazole is probably best prescribed at high doses (i.e. 500 mg three times per day). Precisely, the study with the lowest efficacy [Gisbert et al. 1999b] administered metronida-zole at low doses (250mg four times per day). Limited experience suggests that quadruple therapy may also be effective when the first (failed) regimen included RBC instead PPI. Thus, Beales [2001] reported that four of the five patients initially failing RBC-clarithromycin-amoxicillin therapy were successfully treated with quadruple therapy. A 7-day treatment duration seems to be sufficient when quadruple therapy is used after a failed first regimen, as quite similar eradication rates with 7, 10 and 14 days have been reported (mean figures, calculated from Table 1, of 74%, 72% and 81%, respectively). Furthermore, in a recent retrospective study, patients who failed to the standard triple therapy (PPI, amoxicillin, clarithromycin) received 1 or 2 weeks quadruple therapy, and the eradication rate was similar between the two regimens [Choung et al. 2006].

Table 1.

Eradication rates with quadruple therapy (proton-pump inhibitor, bismuth, tetra-cycline and a nitroimidazole) as ‘rescue’ therapy for proton-pump inhibitor-clarithromycin-amoxicillin failure.

Author	Number of patients	Duration (d)	Eradication rate (%)
Baena et al. [2000]	31	14	90
Bilardi et al. [2004]	46	7	37
Elizalde et al. [1998]	31	7	87
Choung et al. [2006]	56	7	77
Choung et al. [2006]	99	14	88
Chung et al. [2007]	87	7	84
Gasbarrini et al. [2000b]	9	7	88
Gisbert et al. [1999b]	30	7	57
Gisbert et al. [1999a]	9	7	78
Gomollon et al. [1999]	21	7	95
Lee et al. [1999]	20	7	68
Lee et al. [2005]	63	7	75
Marko et al. [2005]	27	7	63
Michopoulos et al. [2000]	38	14	76
Navarro-Jarabo et al. [2007]	54	7	70
Nista et al. [2003]	70	7	63
Nista et al. [2003]	70	14	68
Orsi et al. [2003]	50	12	88
Perri et al. [2001a]	45	10	67
Perri et al. [2003]	60	7	83
Sicilia et al. [2000]	21	10	83
Usta et al. [2008]	89	14	67
Uygun et al. [2008]	100	14	82
Wong et al. [2003]	53	7	91
Wu et al. [2006]	47	7	77

Eradication rates by intention-to-treat analysis when available. H. pylori eradication rate (weighted mean) with quadruple therapy: 77%.

These results are in agreement with those reported previously with quadruple therapy as first-line regimen, where 1-week therapy appeared sufficient, and prolonging treatment did not increase efficacy [De Boer et al. 1994]. Finally, although PPIs are generally prescribed as the anti-secretors in quadruple therapy, some authors have shown, in a randomized study, that omeprazole 20 mg twice daily. and ranitidine 300 mg twice daily. were equally effective as antisecretory agents combined in a second-line quadruple eradication regimen after failure with previous regimens without metronidazole (although the power of the study to find statistically significant differences was limited) [Michopoulos et al. 2000]. Nevertheless, these regimens were administered during 14 days and, therefore, remains to be demonstrated if the equivalence between both antisecretors — PPIs and H₂-blockers — is also observable with 7-day regimens.

The question may be suggested whether treatment with PPI-clarithromycin-amoxicillin followed by ‘rescue’ with quadruple therapy if failure is preferable to the inverse strategy. To analyze this interesting aspect, Gomollon et al. [2000b] randomized consecutive patients to one of two strategies: (1) treatment during 7 days with quadruple therapy, and if failure second-line treatment with omeprazole-clarithromycin-amoxicillin during 7 days; and (2) initial treatment with omeprazole-clarithromycin-amoxicillin and if failure treatment with quadruple therapy. Direct and indirect costs were estimated, and a cost-effectiveness analysis using a decision-tree model was undertaken after real clinical data. Eradication was obtained (intention-to-treat) in 73% with the first strategy, versus 92% with the second one. Furthermore, cost per case eradicated was lower in the second group (320 versus 296 euros). However, in a similar but more recent study, Marko et al. [2005] assessed the usefulness and the cost-effectiveness of these two treatment strategies, performing a decision analysis. The effectiveness of ‘triple first’ and ‘quadruple first’ strategies was similar, although the latter seemed slightly more cost-effective.

RBC, tetracycline and metronidazole. More recently, it has been reported that replacing the PPI and the bismuth compound of the quadruple therapy by RBC also achieves good results as ‘rescue’ regimen [Gisbert et al. 2007a, 2005c, 1999; Koksal et al. 2005; Hojo et al. 2001, Zullo et al. 2001a; Rinaldi et al. 1999]. RBC is a compound that has, on one hand, the antise-cretory activity of ranitidine, and, on the other, the mucosal protective and anti-H. pylori effects of certain other bismuth salts [Gisbert et al. 2005d, 1999c, Van Oijen et al. 2000]. To date, several studies have evaluated 7—14 day RBC-based second-line regimens after PPI-based triple therapy failures, achieving encouraging results, with eradication rates of 67% [Gisbert et al. 2005c], 68% [Gisbert et al. 2007a], 76% [Thong-Ngam and Mahachai, 2006], 82% [Rinaldi et al. 1999], 83% [Gisbert et al. 1999b], 86% [Koksal et al. 2005] and 96% [Zullo et al. 2001a]. Furthermore, one randomized study has demonstrated that triple RBC-based therapy, when prescribed to patients with previous PPI-clarithromycin-amoxicillin failure, achieves even a higher efficacy than quadruple therapy, with additional advantages of a lower number of drugs and a simpler dose scheme [Gisbert et al. 1999b]. Nevertheless, the eradication rate with quadruple regimen in this last study was remarkably low, which was possibly explained by the low dose of metronidazole prescribed. The favorable results obtained with RBC in aforementioned studies are probably explained, at least in part, by the fact that RBC-based therapies may overcome the impact of metronidazole-resistant and clarithromycin-resistant strains on H. pylori eradication treatment [Gisbert et al. 2005d, 1999c; Megraud, 2004; Van Oijen et al. 2001]. In summary, due to the aforementioned encouraging results, quadruple therapy (as well as RBC-based regimens) may be considered as the preferred regimen after initial treatment failure with PPI-clarithromycin-amoxicillin [Malfertheiner et al. 2007; Hojo et al. 2001; Kearney, 2001; Gisbert and Pajares, 2002]. However, bismuth salts, including RBC, are no longer available worldwide, and some National Guidelines have been accordingly changed [Caselli et al. 2007].

PPI, amoxicillin and levofloxacin. As previously mentioned, after failure of a combination of a PPI-based triple regimen, the use of the quadruple therapy has been generally recommended as the optimal second-line therapy based on the relatively good results reported by several authors [Malfertheiner et al. 2007; Gisbert and Pajares, 2002; Hojo et al. 2001; Kearney, 2001]. However, this quadruple regimen requires the administration of four drugs with a complex scheme (bismuth and tetracycline usually prescribed every 6 hours, and metronidazole every 8 hours) and is associated with a relatively high incidence of adverse effects [Gisbert and Pajares, 2002]. Furthermore, this quadruple regimen still fails to eradicate H. pylori in approximately 20—30% of the patients, and these cases constitute a therapeutic dilemma, as patients who are not cured with two consecutive treatments including clarithromycin and metronidazole will usually have double resistance [Gisbert and Pajares, 2002].

Levofloxacin is a fluoroquinolone antibacterial agent with a broad spectrum of activity against Gram-positive and Gram-negative bacteria and atypical respiratory pathogens [Croom and Goa, 2003]. Recently, some studies have evaluated the efficacy of new fluoroquinolones, such as levofloxacin, that could prove to be a valid alternative to standard antibiotics not only as first-line therapies but, more interesting, as second-line regimens [Vaira et al. 2007; Gisbert and Morena, 2006c; Gisbert and Pajares, 2005; Saad et al. 2006]. In this respect, levofloxacin-based second-line therapies represent an encouraging strategy for eradication failures, as some studies have demonstrated that levofloxacin has, in vitro, remarkable activity against H. pylori [Sanchez et al. 2000], and that primary resistances to such antibiotic are (still) relatively infrequent (when compared with metronidazole or clarithromycin) [Chang et al. 2009, Chisholm and Owen, 2009; Antos et al. 2006; Kumala and Rani, 2006; Gatta et al. 2005; Zou et al. 2003; Xia et al. 2002]. A recent in vitro study also showed a synergistic effect of quinolone antimicrobial agents and PPIs on strains of H. pylori [Tanaka et al. 2002]. Furthermore, it has been shown in vitro that levofloxacin retains its activity when H. pylori strains are resistant to clarithromycin and metronidazole [Antos et al. 2006; Yahav et al. 2006b; Watanabe et al. 2003]. These favorable results have been confirmed in vivo, indicating that most of the patients with both metronidazole and clarithromycin resistance are cured with the levofloxacin-based regimen [Gatta et al. 2005; Matsumoto et al. 2005; Bilardi et al. 2004].

A combination of a PPI, amoxicillin and levofloxacin, as first-line regimen, has been associated with favorable results, with mean eradication rates of about 90% [Rispo et al. 2007; Gisbert et al. 2007b; Antos et al. 2006; Lee et al. 2006; Marzio et al. 2006; Di Caro et al. 2002; Cammarota et al. 2000]. Later, other authors studied this same regimen in patients with one previous eradication failure, also reporting exciting results, with H. pylori cures rates ranging from 60% to 94% [Di Caro et al. 2009; Eisig et al. 2009; Kuo et al. 2009; Schrauwen et al. 2009; Gisbert et al. 2007c; Page et al. 2007; Giannini et al. 2006; Lee et al. 2006; Wong et al. 2006; Matsumoto et al. 2005; Bilardi et al. 2004; Nista et al. 2003; Orsi et al. 2003; Perri et al. 2003; Watanabe et al. 2003; Wong et al. 2003]. A recent systematic review showed a mean eradication rate with levofloxacin-based ‘rescue’ regimens (combined with amoxicillin and a PPI in most studies) of 80%, which represents a relatively high figure when considering that this regimen was evaluated as a ‘rescue’ therapy [Gisbert et al. 2006c]. This systematic review found higher H. pylori cure rates with 10-day than with 7-day regimens, both in general (81% versus 73%) and also with the levofloxacin-amoxicillin-PPI combination in particular (80% versus 68%), suggesting that the longer (10-day) therapeutic scheme should be chosen.

Furthermore, two recent meta-analyses have suggested that after H. pylori eradication failure, levofloxacin-based ‘rescue’ regimen is more effective than the generally recommended quadruple therapy [Saad et al. 2006; Gisbert et al. 2006c]. In one of these meta-analyses [Gisbert et al. 2006c], higher H. pylori cure rates with the levo-floxacin-based triple regimens than with the quadruple combinations were found (81% versus 70%), but with borderline statistical significance (Figure 1). Nevertheless, results were heterogeneous, mainly due to the discordant results of the study by Perri et al. [2003], who reported a cure rate of only 63% with the levofloxacin regimen, the lowest reported in the literature, a figure that contrasts with the mean eradication rate of 80% calculated in a systematic review [Gisbert et al. 2006c]. Nevertheless, when that single outlier study [Perri et al. 2003] was excluded from the meta-analysis, the difference between cure rates with both regimens reached statistical significance and heterogeneity markedly decreased. Furthermore, when only high-quality studies were considered, the advantage of the levofloxa-cin regimen over the quadruple regimen increased (88% versus 64%), also achieving statistical significance, and heterogeneity among studies almost disappeared [Gisbert et al. 2006c].

Figure 1.

Meta-analysis comparing H. pylori eradication efficacy with levofloxacin-based triple regimens versus quadruple therapy, as second-line ‘rescue’ regimen after failure of a proton-pump inhibitor (PPI)-amoxicillin-clarithromycin.

As previously mentioned, the quadruple regimen requires the administration of a complex scheme [Gisbert et al. 2002]. On the contrary, levoflox-acin-based regimens (with amoxicillin and PPIs administered twice daily, and levofloxacin every 12 or 24 hours) represents an encouraging alternative to quadruple therapy, with the advantage of simplicity. Furthermore, the quadruple regimen is associated with a relatively high incidence of adverse effects [Gisbert et al. 2002].

In contrast, levofloxacin is generally well tolerated, and most adverse events associated with its use are mild to moderate in severity and transient [Croom and Goa, 2003]. The most frequent adverse effects affect the gastrointestinal tract [Croom and Goa, 2003]. Occasional cases of tendinitis and tendon rupture have been reported in the literature with levofloxacin therapy [Bilardi et al. 2004; Croom and Goa, 2003]. However, data derived from more than 15 million prescriptions in the US indicated the rate is fewer than 4 per million prescriptions [Kahn, 2001]. Clostridium difficile infection may be associated with the use of this broad spectrum activity antibiotic [Croom and Goa, 2003]. In the aforementioned systematic review [Gisbert et al. 2006c], adverse effects were reported, overall, by 18% of the patients treated with levofloxacin-based therapies, and these adverse effects were severe (defined so by the authors or explaining treatment discontinuation) in only 3% of the cases. Furthermore, the incidence of adverse effects was not different when levofloxacin-amoxicillin-PPI was administered for 7 or 10 days, supporting the aforementioned recommendation of prescribing the more effective 10-day regimen. Moreover, two meta-analyses have demonstrated a lower incidence of adverse effects with levofloxacin-based treatments than with the quadruple combinations [Gisbert et al. 2006c; Saad et al. 2006].

Unfortunately, it has been shown that resistance to quinolones is easily acquired, and in countries with a high consumption of these drugs, the resistance rate is increasing and is already relatively high [Carothers et al. 2007; Cattoir et al. 2007; Glocker et al. 2007; Perna et al. 2007; Zullo et al. 2007; Bogaerts et al. 2006; Kim et al. 2006; Marzio et al. 2006; Miyachi et al. 2006; Wong et al. 2006; Coelho et al. 2005; Gatta et al. 2005; Cammarota et al. 2004; Cabrita et al. 2000]. More importantly, it has been demonstrated that the presence of levofloxacin resistance significantly reduce the eradication rate following a therapy with this antibiotic [Perna et al. 2007; Marzio et al. 2006; Gatta et al. 2005; Krakowka et al. 1996]. Therefore, it has been suggested to reserve levofloxacin for ‘rescue’ treatment to avoid the increase of the resistance phenomenon.

‘Rescue’ regimen after PPI-amoxicillin-nitroimidazole failure

After PPI-amoxicillin-nitroimidazole failure, retreatment with PPI-amoxicillin-clarithromycin has proved to be very effective, and it seems to be a logical strategy, as while amoxicillin is maintained (which does not induce resistance), clari-thromycin is substituted for metronidazole. Furthermore, the absence of cross-resistance among nitroimidazoles and clarithromycin favors this position. With this therapy, some authors [Gisbert et al. 1999a] have achieved H. pylori eradication in 85% of cases, while others have reported success rates of 86% [Reilly et al. 1995] or even 100% [Lerang et al. 1997]. In favor of this strategy is the study by Magaret et al. [2001], who studied a group of 48 patients after failure of previous H. pylori therapy with a metronidazole-containing regimen, and randomized them to either lansoprazole, amoxicillin and clarithromycin twice daily for 14 days (i.e. the logical approach with triple therapy not repeating metronidazole), or to lansopra-zole, bismuth, metronidazole and tetracycline for 14 days (i.e. the quadruple therapy repeating metronidazole). Intention-to-treat efficacies were 75% for triple regimen and 71% for quadruple. Although this difference did not reach statistical significance, the small sample size of this study does not preclude the possibility of a small but clinically significant difference in efficacy between the regimens. Finally, preliminary studies have suggested that RBC may be used instead of PPI in this triple second-line strategy (i.e. RBC-clarithromycin-amoxicillin), with similar or even better results [Beales, 2001]

‘Rescue’ regimen after PPI-clarithromycin-nitroimidazole failure

As previously mentioned, acquired bacterial resistance to metronidazole or clarithromycin results primarily from the previous treatment failure [Peitz et al. 1999; Deltenre et al. 1998], and therefore the first choice probably should not be a regimen that combines these two antibiotics in the same regimen [Calvet et al. 2001; Axon, 2000; De Boer et al. 2000b]. Although this regimen is very effective [Gisbert et al. 2000b], patients who are not cured will probably have double resistance [Peitz et al. 2002, 1999], and no logical empirical treatment remains afterwards (although, more recently, the levofloxacin-based regimens may represent an option). Thus, some authors have demonstrated that initial regimens containing both clarithromycin and nitroimida-zole are associated with significantly worse results overall, with lower eradication rates after logically chosen second-line therapy and sensitivity-directed third-line therapy; these poor results were due to the emergence of multiply resistant strains as evidenced by the results of culture testing after the second failed course [Beales, 2001]. In summary, due to problems with resistance it could be suggested that both key antibiotics — clarithromycin and metronidazole — should not be used together until a valid empirical back up regimen is available [De Boer et al. 2000b].

Nevertheless, if culture is not performed after failure of PPI-clarithromycin-metronidazole, and hence antibiotic susceptibility is unknown, several ‘rescue’ options may be suggested. Firstly, omeprazole plus amoxicillin, with a high dose of both the antibiotic and the antisecretor, could in theory be recommended [Miehlke et al. 2006; Axon, 2000]; however, we must remember that this ‘old-fashioned’ dual combination has achieved disappointing results in many countries. Therefore, a second antibiotic should be added, and at this point a difficult decision appears, as both antibiotics used in the first trial (clarithromycin and metronidazole) are capable of inducing secondary resistance to H. pylori, playing a negative role in future efficacy [Saad and Chey, 2008; Dore et al. 2000; Pilotto et al. 2000; Houben et al. 1999; Peitz et al. 1999; Van Der Wouden et al. 1999; Tompkins et al. 1997]. Nevertheless, the following possibilities exist.

Readministering metronidazole. Due to the fact that metronidazole resistance is frequent and clinically relevant [Dore et al. 2000; Houben et al. 1999; Peitz et al. 1999; Van Der Wouden et al. 1999], if this antibiotic is readministered, it should be used within bismuth-based quadruple regimen (thus PPI might reduce the negative effect of metronidazole resistance [Graham et al. 2000; Gomollon et al. 1999, Van Der Wouden et al. 1999; De Boer et al. 1995]). With this regimen, eradication rates up to 80% have been achieved [Gisbert et al. 1999a]. RBC, which may overcome the impact of resistance to metronidazole [Gisbert et al. 1999c], may also play a role in this regimen. Thus, some authors have reported an 88% cure rate with 2-week regimen or RBC-tetracycline-tinidazole in patients who had previously failed a clarithromycin-tinidazole based triple therapy [Zullo et al. 2001a].

Readministering clarithromycin. Several studies have underlined the relevance of clarithromycin resistance [Dore et al. 2000; Houben et al. 1999; Peitz et al. 1999; Tompkins et al. 1997], which advise against readministering this antibiotic. Therefore, a further option which has been proposed is to add (e.g. to PPI-amoxicillin-clarithromycin) a fourth medication (as bismuth) with bactericidal effect against H. pylori, with which a 70% eradication rate has been achieved [Gisbert et al. 1999a].

Readministering no antibiotic. A final alternative, obviously, consists of no readministering either metronidazole or clarithromycin. Although only published in abstract form, one study has prescribed RBC, tetracycline and amoxicillin for 2 weeks and has reported the eradication in 89% of the cases who had previously failed PPI, clarithromycin and tinidazole [Cudia et al. 1997]. These encouraging results may be due, at least in part, to the use of RBC instead of bismuth in this regimen, as ‘classic’ triple therapy with bismuth, tetracycline and amoxicillin has been previously considered relatively ineffective. Finally, although not specifically evaluated in PPI-clarithromycin-metronidazole failures, rifabutin or levofloxacin-based regimens (e.g. PPI, amoxicillin and either levofloxacin or rifabutin) could play a role in this difficult situation.

Patients allergic to penicillin. When penicillin allergy is documented, replacing amoxicillin with metronidazole has been recommended in PPI-based triple combinations. H. pylori eradication is a challenge in patients allergic to penicillin, especially in those who have failed a first-eradication trial with these two key antibiotics, clarithro-mycin and nitroimidazoles. To date, only few studies have evaluated the efficacy of H. pylori eradication treatment specifically in those patients allergic to penicillin, which constitutes a relatively common subgroup [Gisbert et al. 2009, 2005e; Rodriguez-Torres et al. 2005]. In our previous study, H. pylori infected patients allergic to penicillin failing first-line treatment with PPI-clarithromycin-metronidazole therapy received a second-line treatment with RBC, tetracycline and metronidazole, but this rescue regimen cured the infection in only 47% of the patients [Gisbert et al. 2005e]. In this same study, it was shown that rifabutin-clarithromycin-PPI regimen was ineffective and poorly tolerated. However, a final rescue regimen including levofloxacin cured the infection in all treated patients [Gisbert et al. 2005e]. In a more recent study, it has been confirmed that a levofloxacin-containing regimen (together with omeprazole and clarithromycin) represents an encouraging second-line alternative in the presence of penicillin allergy [Gisbert et al. 2009].

Is it necessary to perform culture after failure of the second eradication treatment?

As previously mentioned, it has been generally recommended that performing culture after a first eradication failure is not necessary and therefore assessing H. pylori sensitivity to antibiotics only after failure of the second treatment may be suggested in clinical practice [Malfertheiner et al. 2007; De Boer et al. 2000b; European Helicobacter pylori Study Group 1997]. However, the utility of the culture (with consequent antibiotic susceptibility testing) and the moment when it must be performed after eradication failure are both controversial [Gisbert, 2008a, 2005b]. It is evident that, as pretreatment antibiotic resistance is the most important factor in nonresponse to initial treatment [Dore et al. 2000; Houben et al. 1999; Peitz et al. 1999; Van Der Wouden et al. 1999; Tompkins et al. 1997], knowledge of the organism's antibiotic susceptibility may represents an aid in selecting the therapy regimen. However, performing culture systematically after the second eradication failure also has some limitations, which are summarized as follows:

• (1)Culture implies, obviously, the performance of endoscopic exploration, which has several disadvantages: it is not free of risk, and, since endoscopy centers have been meeting increasing demand, usually involves prolonged waiting times.
• (2)H. pylori culture is expensive, due to the cost of the procedure itself, but mainly the costs of the associated endoscopy, which is necessary to obtain biopsy samples.
• (3)Culture is time-consuming, as H. pylori is a rather ‘fastidious’ bacterium at culture, especially when a low bacterial load is present, as generally occurs after eradication failure [Zullo et al. 2003b].
• (4)Culture is not always available on a routine basis.
• (5)The sensitivity of bacterial culture is not 100%, and therefore the antimicrobial susceptibility cannot be obtained in all cases. Indeed, even in the optimal conditions usually encountered in therapeutic trials — when both gastroenterologist and microbiologist are thoroughly motivated — a culture sensitivity of ‘only’ approximately 90% has been achieved in patients not previously treated [Zullo et al. 2003b]. Furthermore, in several studies enrolling patients who had failed one or more eradication treatments, the bacterium was isolated in less than 80% of cases [Zullo et al. 2003b]. Therefore, an even lower probability of isolating the bacterium is to be expected in routine clinical practice. This indicates that, even in the hands of experts, antimicrobial sensitivity would not be obtained in several eradication failure patients, who had undergone an upper endos-copy solely for bacterial culture [Zullo et al. 2003b].
• (6)Antibiotic susceptibility testing in clinical practice yields useful information only regarding a few antibiotics. Antibiotics effective and generally used against H. pylori are mainly amoxicillin, clarithro-mycin, metronidazole, and tetracycline. Resistance to amoxicillin has been estimated to be less than 1% in most studies [Megraud, 2004; Zullo et al. 2003b]. Hence, its role in clinical practice may even be marginalized. Similarly, resistance to tetracycline is also very low, or even absent, in most countries [Megraud, 2004; Zullo et al. 2003b]. Therefore, it may even be assumed that antibiotic susceptibility testing in clinical practice yields useful information only regarding the latter two antibiotics, namely clarithromycin and metronidazole [Zullo et al. 2003b].
• (7)In vitro antibiotic susceptibility does not necessarily lead to eradication in vivo. Even knowing the susceptibility of H. pylori, eradication rates do not achieve 100%, as the results observed in vivo by following in vitro susceptibility to anti-H. pylori antibiotics are often disappointing [Guslandi, 2001]. Some discrepancies between antibiotic susceptibility and H. pylori eradication may occur, due, for example, to the possibility of coinfection with different H. pylori strains [Kim et al. 2003]. Thus, a variable proportion of noneradicated patients is made of subjects who harbor strains sensitive to the administered drugs, and in these patients the reasons for treatment failure are unclear [Ali et al. 1999]. For example, Gomollon et al. [2000a] reported how third-line treatment often (in 50% of the cases) failed to eradicate H. pylori infection, in spite of giving 14-day, full-dose, quadruple culture-guided combination, showing that in vitro susceptibility did not predict eradication success. In the same way, Vicente et al. [2002] determined the effectiveness of a third, culture-guided, treatment of H. pylori infection after two unsuccessful attempts. Patients received a 2-week quadruple culture-guided therapy, and overall eradication was achieved in only 60% of the patients. In fact, paradoxically, the lowest eradication rate was obtained in patients with H. pylori strains sensitive to all antibiotics. In summary, it seems that despite the use of culture-guided combinations of drugs, a third treatment is frequently unsuccessful, indicating that other factors different from in vitro antibiotic susceptibility influence eradication rates. On the other hand, the reverse situation is also possible, as H. pylori eradication may nonetheless be achieved in the presence of metronidazole- or clarithromy-cin-resistant strains even with a drug combination including these antibiotics. Therefore, in vitro resistance to either clarithromycin or metronidazole could be overcome in vivo in a significant proportion of patients by prescribing the same antibiotics [Zullo et al. 2003b].
• (8)When a repeat (rescue) therapy must be selected, we have several data that will aid us in suspecting resistance to a particular antibiotic, without the necessity of a culture, based on the observation that acquired bacterial resistance to metronidazole or clari-thromycin results primarily from the previous treatment failure [Peitz et al. 1999; Deltenre et al. 1998]. Thus, when a therapy with clari-thromycin fails, resistance to this antibiotic appears in most cases, and the same is true when a nitroimidazole is the antibiotic first used [Megraud, 2004; Pilotto et al. 2000; Adamek et al. 1998; Tompkins et al. 1997]. Even if resistance to these antibiotics does not appear, it remains uncertain whether their readministration is adequate, as they were not efficacious (for unknown reasons) for the first time. Although some studies suggest that retreatment of H.pylori infection with the same combination is still a choice when the status of bacterial resistance to antibiotics is unknown, however, full doses and a longer treatment duration must be used and a poor eradication rate has usually been reported [Huang and Hunt, 1999]. Therefore, the position in the case of therapy failure would be clear: do not readminister any of the antibiotics against which H. pylori has probably become resistant [Howden and Hunt, 1998; Lam and Talley, 1998].
• (9)Finally, relatively high eradication rates have been obtained with empirical third-line treatment after two consecutive failures in several studies [Gisbert et al. 2008b, 2006a, 2006b, 2004, 2003; Dore et al. 2003; Zullo et al. 2003a; Treiber et al. 2002a; Canducci et al. 2001; Zullo et al. 2001a; Bock et al. 2000, Chan et al. 2000; Perri et al. 2000; Seppala et al. 2000].

However, limited experience suggests that endo-scopy with culture and susceptibility testing may be appropriate after failure of two eradication therapies; in this situation, a nonrandomized retrospective study suggests that third-line therapy directed by the results of sensitivity testing improve eradication compared to further empirical antibiotics, demonstrating that the success rate of sensitivity-directed therapy is superior to PPI-amoxicillin-rifabutin triple therapy, and therefore suggesting that endoscopy and sensitivity testing at this point may be worthwhile rather than more widespread use of rifabutin-based regimens [Beales, 2001]. Cammarota et al. [2004] assessed the efficacy of a third-line, culture-guided treatment approach for the eradication of H. pylori. After the first two eradication attempts, all patients were resistant to metroni-dazole, and 95% were resistant to clarithromycin. Consequently, most patients (89 out of 94) received quadruple regimen including PPI, bismuth, tetracycline and amoxicillin, and H. pylori eradication was obtained in 90% of the cases. Although the authors conclude that a culture-guided, third-line therapeutic approach is effective for the eradication of H. pylori, it would seem more appropriate to conclude, in fact, that the tetracycline- and amoxicillin-based quadruple regimen may be a good empirical third-line ‘rescue’ treatment option (as to choose such a regimen, which implies not readministering metronidazole or clarithromycin, it would not be necessary to know antibiotic susceptibilities).

In summary, when critically reviewing the role of culture in the management of H. pylori infection in clinical practice it may be concluded, in coincidence with other authors, that H. pylori culture is an invasive, time-consuming method, offering quite low sensitivity, requiring significant cost and which, in practice, tests very few antibiotics, with a questionable contribution to the management of nonresponder patients [Zullo et al. 2003b, Cianci et al. 2006]. Obviously, the importance of H. pylori culture remains unaltered both in epidemiological and pharmacological research fields. However, whether patients should undergo an upper endoscopy for bacterial culture after second-line therapy failure remains a debatable matter, and the role of culture in clinical practice requires a critical reappraisal [Cianci et al. 2006; Zullo et al. 2003b]. As it has been brilliantly expressed by Zullo et al. [2003b], regrettably, gastroenterologists need to accept that gastric biopsy culture is not as simple as filling a sample bottle!

Nevertheless, it is recommended that those prescribing H. pylori eradication therapies continually assess their success rate and adjust the relevant local practices and policies in line with the results and local bacterial resistance patterns. Thus, it would be recommendable that culture should be routinely performed after eradication failure in some specialized centers with special interest in H. pylori research and treatment, with the intention to study the incidence of resistances after failures and also to evaluate the influence of such resistances on the efficacy of ‘rescue’ regimens [Megraud and Lamouliatte, 2003]. Data coming from this experience on H. pylori resistance will be used as reference for the corresponding population. This preventive approach has been recommended to avoid an increase in refractory H. pylori infection in the future [Megraud and Lamouliatte, 2003].

Empirical third-line H. pylori‘rescue’ therapy after failure of two eradication treatments

If we have decided, finally, not to perform culture before the administration of the third-line ‘rescue’ treatment after failure of the first two trials (generally including clarithromycin and metronidazole), different possibilities of empirical treatment may be suggested. As eradication regimens may be less efficacious for retreatment as compared to their efficacy when used as primary treatment, it may be suggested that the course of the ‘rescue’ therapy should be extended to 10—14 days, at least when ‘rescue’ therapy fails and third-line regimens are therefore prescribed [De Boer and Brody, 2000]. As several studies have underlined the relevance of metronidazole [Dore et al. 2000; Houben et al. 1999; Peitz et al. 1999; Van Der Wouden et al. 1999] and clarithromycin [Dore et al. 2000; Houben et al. 1999; Peitz et al. 1999; Tompkins et al. 1997] resistance, these two antibiotics should not be readministered, and several regimens have been evaluated in this scenario, outlined as follows.

Amoxicillin ± tetracycline-based regimens

In a recent study, patients with at least one treatment failure and infected with H. pylori resistant to both metronidazole and clarithromycin, were treated with high doses of omeprazole (4 × 40 mg) and amoxicillin (4 × 750 mg) for 14 days, and the infection was cured in 76% of the cases [Miehlke et al. 2003]. This study suggests that, although the ‘old-fashioned’ dual combination of omeprazole plus amoxicillin is generally considered quite ineffective as first-line regimen, it may be associated with relatively good results if prescribed at high doses, even for H. pylori infection resistant to both metronidazole and clari-thromycin in patients who experienced previous treatment failures. Another possibility to avoid retreatment with clarithromycin or metronida-zole is to prescribe a quadruple combination of PPI, bismuth, tetracycline and amoxicillin (instead of metronidazole), which has been used by some authors with favorable results [Chi et al. 2003]. Nevertheless, this regimen has been tested only as second-line (and not third-line) therapy, and only after failure of PPI-clarithromycin-amoxicillin (and not after metronidazole-based therapy), emphasizing that the experience should be extended to patients with two previous eradication failures containing both clarithromy-cin and metronidazole. Finally, as previously mentioned, one study prescribed RBC, tetracy-cline and amoxicillin for 2 weeks and achieved eradication in 89% of the cases who had previously failed PPI, clarithromycin and tinidazole [Cudia et al. 1997].

Levofloxacin-based ‘rescue’ regimens

It has been suggested that levofloxacin-based therapies may also represent an alternative when two (or more) consecutive eradication treatments have failed to eradicate the infection [Nishizawa et al. 2009; Hsu et al. 2008; Rokkas et al. 2006; Gisbert et al. 2006a, 2006b, 2005a; Coelho et al. 2005; Gatta et al. 2005; Bilardi et al. 2004; Zullo et al. 2003a, 2001a]. As an example, a recent study by Zullo et al. [2003a] aimed to evaluate the efficacy of a levofloxacin-amoxicillin-PPI combination in patients who previously had failed two or more therapeutic attempts, and eradication rate was 83% (intention-to-treat analysis). More recently, Gisbert et al. [2006b] evaluated, in a multicenter study including 100 patients, the efficacy of a third-line levofloxacin-based regimen in patients with two consecutive H. pylori eradication failures. An intention-to-treat eradication rate was 66%, which represents a relatively high figure when considering that this regimen was evaluated as a third-line therapy. Other alternatives of ‘rescue’ therapies, different from levofloxacin-based regimens, have been suggested. Rifabutin-based ‘rescue’ therapy, as will be reviewed in the following section, also constitutes a possible strategy after previous eradication failures, although it has been recently shown that 10-day triple levofloxacin-based regimen is more effective than the same combination with rifabutin as a ‘rescue’ regimen [Gisbert et al. 2006b]. In summary, levofloxacin-based ‘rescue’ therapy constitutes an encouraging empirical third-line strategy after multiple previous H. pylori eradication failures with key antibiotics (such as amoxicillin, clarithromycin, metronida-zole and tetracycline).

Rifabutin-based ‘rescue’ regimens

As previously mentioned, the evaluation of drugs without cross-resistance to nitroimidazole or macrolides as components of retreatment combination therapies seem to be worthwhile. H. pylori has been proved to be highly susceptible in vitro to rifabutin, a rifamycin derivate of the established tuberculostatic drug [Akada et al. 1999; Heep et al. 1999; Brogden and Fitton, 1994]. Moreover, rifabutin is chemically stable at a wide pH range and its antibacterial activity is likely not to be hampered by the acidic environment of the stomach [Rossi, 1999]. Furthermore, selection of resistant H. pylori strains has been low in experimental conditions. Thus, until now, no rifabutin-resistant strain has been isolated from patients who were either treated or untreated for H. pylori infection [Heep et al. 1999].

As summarized in Table 2, rifabutin-based ‘rescue’ therapy constitutes an encouraging strategy after multiple previous eradication failures [Gonzalez Carro et al. 2007; Navarro-Jarabo et al. 2007; Van Der Poorten and Katelaris, 2007; Borody et al. 2006; Miehlke et al. 2006; Gisbert et al. 2006b, 2003; Toracchio et al. 2005; Beales, 2001; Canducci et al. 2001; Bock et al. 2000; Perri et al. 2000]. As an example, Perri et al. [2000, 1998] used a 1-week regimen of PPI, amoxicillin and rifabutin in patients who were still H. pylori infected after two or more courses of PPI-based triple therapies, and achieved a eradication rate of 71% by intention-to-treat analysis. Gisbert et al. [2003], in a prospective multicenter study, included patients in whom a first eradication trial with PPI, clarithro-mycin and amoxicillin and a second trial with PPI, bismuth, tetracycline and metronidazole had failed. A third 14-day eradication regimen with rifabutin, amoxicillin and a PPI was effective in 79% of the patients (intention-to-treat analysis). However, these encouraging results were not confirmed in a more recent study by these same authors [Gisbert et al. 2006b]. In the largest study up to now with rifabutin [Gonzalez Carro et al. 2007], 92 consecutive patients diagnosed with H. pylori infection resistant to two previous treatment regimens were treated with a PPI, rifabutin and amoxicillin for 10 days, and the intention-to-treat eradication rate was 61%. In summary, the weighted mean eradication rate with rifabutin-based ‘rescue’ therapy, calculated from the studies included in the Table 2, is of 69%.

Table 2.

Rifabutin-based ‘rescue’ therapies (rifabutin-amoxicillin PPI) in patients with previously failed eradication treatments and/or resistance to clarithromycin and nitroimidazoles.

Author	Number of patients	Drugs and doses	Duration of treatment (d)	Eradication rate (%)
Beales [2001]	10	Rifabutin 300mg o.d. Amoxicillin 1 g b.i.d. Omeprazole 20mgb.i.d.	14	60
Bock et al. [2000]	25	Rifabutin 150mgb.i.d. Amoxicillin 1 g b.i.d. Lansoprazole 30mgb.i.d.	7	72
Borody et al. [2006]	67	Rifabutin 150 mg b.i.d. Amoxicillin 1—1.5 g t.i.d.
Pantoprazole 60 mg t.i.d	12	90
Canducci et al. [2001]	10	Rifabutin 300mg o.d. Amoxicillin 1 g b.i.d. Omeprazole 20mgb.i.d.	10	70
Gisbert et al. [2003]	14	Rifabutin 150 mg b.i.d. Amoxicillin 1 g b.i.d. Omeprazole 20 mg b.i.d.	14	79
Gisbert et al. [2006b]	20	Rifabutin 150mgb.i.d. Amoxicillin 1 g b.i.d. Omeprazole 20mgb.i.d.	10	45
Gonzalez Carro [2007]	92	Rifabutin 150mgb.i.d. Amoxicillin 1 g b.i.d. Pantoprazole 40mgb.i.d	10	61
Miehlke et al. [2006]	73	Rifabutin 150 mg b.i.d. Amoxicillin 1 g b.i.d. Esomeprazole 20 mg b.i.d.	7	74
Navarro-Jarabo et al. [2007]	45	Rifabutin 150mgb.i.d. Amoxicillin 1 g b.i.d. Omeprazole 20mgb.i.d.	7	44
Perri et al. [2000]	41	Rifabutin 300mg o.d. Amoxicillin 1 g b.i.d. Pantoprazole 40mgb.i.d.	7	71
Toracchio et al. [2005]	65	Rifabutin 150 mg b.i.d. Amoxicillin 1 g b.i.d. Pantoprazole 40 mg b.i.d.	10	78
Van der Poorten et al. [2007]	44	Rifabutin 150 mg b.i.d. Amoxicillin 1 g b.i.d. PPI b.i.d.	10	68

PPI, proton-pump inhibitor (omeprazole, pantoprazole, rabeprazole or esomeprazole) at the usual dose. Eradication rates by intention-to-treat analysis when available. H. pylori eradication rate (weighted mean) with rifabutin-based ‘rescue’ therapy: 69%. o.d., once daily; b.i.d., twice daily; t.i.d., three times per day.

These findings suggest that new rifabutin-based combinations are effective for H. pylori strains resistant to antibiotics, and specifically to clari-thromycin or metronidazole [Moayyedi et al. 2006; Miehlke et al. 2008]. Furthermore, rifabu-tin-based therapies have been compared with the widely used ‘classic’ quadruple therapy. Perri et al. [2001a] performed a randomized study where three groups of patients were treated for 10 days with pantoprazole, amoxicillin and rifa-butin 150mg once daily, or 300mg once daily, and quadruple therapy. On intention-to-treat analysis, eradication rates were 67% in the rifa-butin 150mg and quadruple groups, and higher (87%) in the rifabutin 300 mg group. Finally, in this comparative study, side effects were less frequent in rifabutin-treated patients than in those on quadruple therapy [Perri et al. 2001a].

Several concerns still remain, however, regarding rifabutin treatment. Firstly, this drug is very expensive. Secondly, severe leucopoenia and trombocytopenia have been reported in one patient treated with rifabutin, with myelotoxicity demonstrated by bone marrow aspirate [Canducci et al. 2001]. Although blood cell count returned to normal at day 15 after discontinuation of therapy, physicians should be aware of the risk of major side effects arising during rifabutin-based regimen [Gisbert et al. 2006b, 2003]. Finally, there is some concern about a widespread use of rifabutin, a member of a class of established antimycobacterial drugs, in patients with H. pylori infection. Because multi-resistant strains of Mycobacterium tuberculosis increase in numbers, indications for these drugs should be chosen very carefully to avoid further acceleration of development of resistance [Bock et al. 2000]. At present, therefore, rifabutin should be considered only as the last option (e.g. restricted to infected patients even after several eradication regimens including, among them, levofloxacin).

Furazolidone-based ‘rescue’ regimens

Furazolidone is an antimicrobial drug that belongs to synthetic nitrofurans and is active against a broad spectrum of gram-negative and gram-positive bacteria and protozoa. This antibiotic has demonstrated a high antimicrobial activity against H. pylori if given as a single drug [Xiao et al. 1990], and the majority of first-line furazolidone-based combination therapies revealed eradication rates above 80% [Xia et al. 2002]. Primary resistance to furazolidone is virtually absent [Megraud et al. 2003; Kwon et al. 2001; Haas et al. 1990], and its potential to develop resistanceisas lowasfor bismuth compoundsoramoxicillin [Treiber et al. 2002b]. Moreover, this drug has no cross-resistance potential to metronidazole [Haas et al. 1990]. Triple therapy in which furazolidone is used instead of metronidazole achieves high eradication rates, even in populations with a high prevalence of nitroimidazole resistance [Eisig et al. 2005; Nijevitch et al. 2005; Xiao et al. 2001; Ebrahimi-Dariani et al. 2003]. In this respect, a recent study has evaluated furazoli-done-based triple therapy (combined with bismuth and tetracycline) in the eradication of H. pylori resistant to metronidazole, with favorable results (86% eradication rate) [Isakov et al. 2002]. A few years ago, some authors tested a quadruple combination of furazolidone, bismuth, tet-racycline and PPI as second-line eradication therapy, and reported encouraging results [Sotoudehmanesh et al. 2001]. More recently, Treiber et al. [2002a] and Sanches et al. [2008] investigated whether this quadruple regimen containing furazolidone could be effective as third-line therapy in patients with H. pylori treatment failure after first-line (clarithromycin-metro-nidazole±amoxicillin) and/or second-line (PPI-bismuth-tetracycline-metronidazole) regimens, and H. pylori infection was cured in >90% of the cases. Furthermore, 7-day triple regimen comprising furazolidone, amoxicillin and a PPI achieved an eradication rate of60%in10 patients who failed first-line, second-line and even rifabutin-based triple therapy [Qasim et al. 2005].

A recent systematic review and meta-analysis of the effect of furazolidone- and nitrofurantoin-based regimens in the eradication of H. pylori infection has been performed [Buzas and Jozan, 2007]. The pooled eradication rate of primary PPI-based regimens containing furazolidone was 76%. Second-line schedules containing furazolidone obtained eradication rates of 76%. Finally, third-line ‘rescue’ therapies were effective in 65% of the cases. In summary, a quadruple regimen including furazolidone, bismuth, tetra-cycline and PPI seems to represent a promising alternative after two consecutive failures with regimens including both metronidazole and clari-thromycin. However, some aspects related with the use of furazolidone as a rescue therapy for H. pylori infection should be remarked, especially regarding its potential oncologic risk [De Francesco et al. 2009]. Furthermore, furazoli-done is not available worldwide.

Cumulative eradication rates with three (or more) consecutive eradication treatments

In patients with conditions where the indication for H. pylori eradication is definitively accepted, as is the case of peptic ulcer disease (or gastric MALT lymphoma), ‘rescue’ treatment after firstline failure is clearly advisable. Furthermore, if the second therapy fails, a third or even a fourth regimen should be prescribed, as infected patients continue to have high risk of ulcer recurrence and ulcer complications and are in an obviously disadvantageous situation in view of the enormous benefits that follow H. pylori eradication in peptic ulcer disease: increased ulcer healing, less ulcer recurrence and less ulcer complication. However, multiple repeated antibiotic treatment of patients where benefits of H. pylori eradication has not been so clearly established, such as those with functional dyspepsia [Moayyedi et al. 2006; Laine et al. 2001], may not be completely justified.

Some authors have evaluated, in the same study, different regimens after failure of two eradication treatments, which provide interesting information about cumulative, and not only absolute, eradication rates [Gisbert et al. 2008a, 2005b]. For example, in the study by Gasbarrini et al. [2000a], a total of 2606 patients were administered a PPI, tinidazole and clarithromycin for 1 week. Patients with continuing infection were then given a second 1-week course of amoxicillin, clarithromycin and RBC. Finally, patients still infected after the second course underwent upper gastrointestinal endoscopy with H. pylori culture, and then received a 1-week quadruple scheme established on antibiotic sensitivity. Eradication rates after the first, second and third treatment, were 79%, 77% and 52%, respectively. This algorithm led to overall per-protocol eradication rates of 99%. Chan et al. [2000] prescribed quadruple therapy to a group of patients who had failed to respond to RBC-based regimens (as first regimen) and PPI-clarithromycin-amoxicillin combination (as second regimen), and achieved successful eradication in 83% of the cases receiving quadruple regimen, finally achieving a 99% cumulative eradication rate. Beales [2001] evaluated 469 patients receiving eradication therapy in routine clinical practice. Second-line therapy was chosen empirically, using whichever of clarithromycin or metronidazole was not used initially. All patients requiring third-line therapy underwent endo-scopy, choice of therapy being guided by sensitivities. Overall success after one, two and three courses of therapy were 73%, 94% and 98% respectively. Zullo et al. [2001a] reported an 83% cure rate in patients who had previously failed two courses of clarithromycin-amoxicillin and clarithromycin-tinidazole-based triple therapies. Gomollon et al. [2000a] studied the effectiveness of third-line treatment of H. pylori infection with 2-week quadruple, culture-guided regimens. The combination of omeprazole, tetra-cycline, bismuth and clarithromycin showed an eradication rate of only 36%, but if amoxicillin was used the rate was 67%. In the study by Vicente et al. [2002], after two unsuccessful attempts at eradication, all patients underwent endoscopy and culture, and patients received a quadruple culture-guided therapy. Cumulative H. pylori eradication rate with this strategy was as high as 99.6%. Treiber et al. [2002a] investigated whether a quadruple regimen containing furazolidone could be effective as a third-line therapy in patients with two previous H. pylori treatment failures. Cure of H. pylori was achieved in 90% of the patients nonrespon-sive to second eradication trial, which gave a final eradication rate of 99%. In the study by Qasim et al. [2005], 3280 patients received standard first-line eradication therapy, which was successful in 77% of the cases. Second-line therapy (bismuth-based quadruple) or triple therapy (altering constituent antibiotics) was successful in 56% of treated patients. Subsequent eradication attempts using rifabutin-based regimen was successful in 38% of patients, giving a cumulative eradication rate of 94%. Gisbert et al. [2004] included consecutive patients in whom two eradication regimens had failed to eradicate H. pylori, prescribed empirical third-line ‘rescue’ regimens, and achieved H. pylori eradication in 71% of the cases (intention-to-treat analysis). Based on these results, with estimated efficacy of 85%, 75% and 71%, respectively with first, second and third regimens, H. pylori eradication could finally be achieved in 99% of the patients. Gisbert et al. [2008b] evaluated the efficacy of different ‘rescue’ therapies empirically prescribed during 10 years to 500 patients in whom at least one eradication regimen had failed to cure H. pylori infection. Antibiotic susceptibility was unknown (therefore ‘rescue’ regimens were chosen empirically). Overall, H. pylori cure rates with the second and third-line ‘rescue’ regimens were 70% and 74%, giving a cumulative eradication rate as high as 98%. Finally, Rokkas et al. [2009] treated initially their patients with a triple regimen (omeprazole, amoxicillin and clari-thromycin), and subsequently with a second-line quadruple. After two previous H. pylori eradication failures, patients received omeprazole, amoxicillin and levofloxacin, as a third-line empirical strategy. Out of 540 patients initially included in the study, H. pylori was finally eradicated in 90% (intention-to-treat) and in 98% (per-protocol) of them.

Therefore, a wider perspective of the benefits of retreating H. pylori infection can be obtained if cumulative eradication rates with successive treatments are taken into account. Thus, as represented in Figure 2, it can be concluded that H. pylori eradication can finally be achieved in almost 100% of the patients if three ‘rescue’ therapies are consecutively given [Rokkas et al. 2009; Gisbert et al. 2008b, 2004; Qasim et al. 2005; Vicente et al. 2002; Treiber et al. 2002a; Beales, 2001; Zullo et al. 2001a; Chan et al. 2000; Seppala et al. 2000; Gasbarrini et al. 2000a; Gomollon et al. 2000a].

Figure 2.

Cumulative H. pylori eradication rates with three consecutive eradication treatments.

Furthermore, these encouraging (cumulative) results have been obtained when more than three consecutive treatments have been prescribed [Gisbert et al. 2008a, 2005b]. As an example, Seppala et al. [2000] reported a cumulative eradication rate of 93% (intention-to-treat analysis) and even 100% (per-protocol analysis) after four empirical retreatments. We have recently confirmed that levofloxacin-based regimen can also be administered with good results after three previous eradication failures with antibiotics such as amoxicillin, clarithromycin, metronidazole, tetracycline and even rifabutin [Gisbert, 2005a]. Thus, we prospectively evaluated 10 patients with three consecutive H. pylori eradication failures (first treatment with PPI-clarithromycin-amoxicillin, second treatment with RBC-tetracycline-metronidazole, and third treatment with PPI-amoxicillin-rifabutin). A fourth eradication regimen with 10-day levofloxacin, amoxicillin and PPI was prescribed, and intention-to-treat eradication rates were 70%. When we reviewed our experience with different ‘rescue’ therapies empirically prescribed during 10 years to 500 patients, the cumulative H. pylori eradication rate with four successive treatments was 99.5% [Gisbert et al. 2008b].

Finally, reports of ‘ineradicable’ H. pylori infection after more than four eradicating treatments failed have been recently published. Dore et al. [2003] prescribed a quadruple combination of PPI, bismuth, tetracycline, and metronidazole to patients who had failed two or more treatment courses of H. pylori eradication therapy (33 patients had failed prior treatment twice, 19 had failed three times, and 16 had failed four or more times); despite this a priori difficult task, H. pylori eradication was finally achieved in 93% of the patients. Tucci et al. [1999] reported their experience of 13 patients with at least five eradication failures and H. pylori strains resistant to both clarithromycin and nitroimidazoles. The treatment was organized into three sequential schedules employing partially different drug combinations (to face the various resistant strains), suspension formulations were preferred to tablets (to improve the dispersal of the drugs into the stomach), antibiotics were administered after meals and a variation on a standard diet exceeding the normal fat composition was given (to increase the time of contact of the antimicrobials with gastric mucosa), and patients were invited to lie down after the meals, changing their position every 5 min (to facilitate the penetration of drugs amid the anfractuosities of fundic mucosa). With this particular therapy, eradication was successful in 70% of the patients. In another example of ‘ineradicable’ H. pylori infection, levofloxacin-amoxicillin combination was successfully employed in a patient with a clari-thromycin- and metronidazole-resistant strain, who previously failed eight consecutive therapeutic attempts [Zullo et al. 2001b].

Conclusion

Even with the current most effective treatment regimens, >20% of patients will fail to eradicate H. pylori infection. This issue seems important at the present time, as therapy for H. pylori infection is becoming more and more frequently prescribed. Nowadays, apart from having to know well first-line eradication regimens, we must also be prepared to face treatment failures. Therefore, in designing a treatment strategy we should not focus on the results of primary therapy alone, but also on the final (overall) eradication rate.

The choice of a ‘rescue’ treatment depends on which treatment is used initially. If a first-line clari-thromycin-based regimen was used, a second-line metronidazole-based treatment (such as the quadruple therapy) may be used afterwards, and then a levofloxacin-based combination would be a third-line ‘rescue’ option. Alternatively, it has recently been suggested that levofloxacin-based ‘rescue’ therapy constitutes an encouraging second-line strategy, representing an alternative to quadruple therapy in patients with previous PPI-clarithromycin-amoxicillin failure, with the advantage of efficacy, simplicity and safety. In this case, quadruple regimen may be reserved as a third-line ‘rescue’ option. Finally, rifabutin-based ‘rescue’ therapy constitutes an encouraging empirical fourth-line strategy after multiple previous eradication failures with key antibiotics such as amoxicillin, clarithromycin, metronidazole, tetracycline and levofloxacin (Figure 3).

Figure 3.

Choice of an empirical retreatment regimen, without culture and antimicrobial sensitivity testing, after failure of proton-pump inhibitor (PPI), amoxicillin and clarithromycin combination.

Note: quadruple therapy: combination of proton-pump inhibitor (PPI), bismuth, tetracycline and nitroimidazole (metronidazole or tinidazole).

Even after two consecutive failures, several studies have demonstrated that H. pylori eradication can finally be achieved in almost all patients if several ‘rescue’ therapies are consecutively given. As a final conclusion, therefore, the attitude in H. pylori eradication therapy failure, even after two or more unsuccessful attempts, should be to fight and not to surrender [Gisbert, 1998].

Acknowledgements

CIBEREHD is funded by the Instituto de Salud Carlos III.

Conflict of interest statement

None declared.