的辅助用药,对Hp根除率和相关性不良反应的评估.pdf
Meta-analysis:the effect of supplementation with probiotics oneradication rates and adverse events duringHelicobacter pylorieradication therapyJ.L.TONG*,Z.H.RAN*,J.SHEN*,C.X.ZHANG?&S.D.XIAO*Department of Gastroenterology,Shanghai Institute of Digestive Dis-ease,Renji Hospital,Shanghai JiaoTong University,School of Medicine,Shanghai,China;?Department ofPharmacy,Jiangni Hospital,Nanjing,ChinaCorrespondence to:Dr Z.H.Ran,Department ofGastroenterology,Shanghai Instituteof Digestive Disease,Renji Hospital,Shanghai Jiao Tong University,School of Medicine,Shanghai 200001,China.E-mail:z-Publication dataSubmitted 23 August 2006First decision 13 September 2006Resubmitted 26 September 2006Resubmitted 07 October 2006Accepted 13 October 2006SUMMARYBackgroundRecent evidence found probiotics could inhibit Helicobacter pylori col-onization from both in vitro and in vivo studies.AimTo systematically evaluate whether adding probiotics to anti-H.pyloriregimens could improve eradication rates and reduce side effects duringanti-H.pylori treatment.MethodsEligible articles were identified by searches of electronic databases.Weincluded all randomized trials comparing probiotics supplementation toplacebo or no treatment during anti-H.pylori regimens.Statistical ana-lysis was performed with Review Manager 4.2.8.Subanalysis/Sensitivityanalysis was also performed.ResultsWe identified 14 randomized trials(n 1671).Pooled H.pylori eradi-cation rates were 83.6%(95%CI 80.586.7%)and 74.8%(95%CI 71.178.5%)for patients with or without probiotics by intention-to-treatanalysis,respectively,the odds ratio(OR)was 1.84(95%CI 1.342.54);the occurrence of total side effects were 24.7%(95%CI 20.029.4%)and 38.5%(95%CI 33.044.1%)for groups with or withoutprobiotics,especially for diarrhoea,the summary OR was 0.44(95%CI 0.300.66).ConclusionsOur review suggests that supplementation with probiotics could be effect-ive in increasing eradication rates of anti-H.pylori therapy,and could beconsidered helpful for patients with eradication failure.Furthermore,pro-biotics show a positive impact on H.pylori therapy-related side effects.Aliment Pharmacol Ther 25,155168Alimentary Pharmacology&Therapeutics 2007 The Authors155Journal compilation 2007 Blackwell Publishing Ltddoi:10.1111/j.1365-2036.2006.03179.xINTRODUCTIONHelicobacter pylori(H.pylori)infection is a commonand significant public health problem;it can be foundin 7090%of the population in developing countriesand in 2550%in developed countries.1Eradication ofH.pylori is a primary goal in the management of gas-tric and duodenal ulcers,gastric cancer,atrophic gas-tritis,mucosa-associatedlymphomatissue(MALT)lymphoma and other H.pylori-related complication.24The standard treatment recommended for eradicationof H.pylori is a combination of proton-pump inhibitor(PPI)or ranitidine bismuth citrate,clarithromycin andeither amoxicillin or nitroimidazole.These regimenscould achieve eradication rates ranging from 65%to90%;however,approximately 1035%of patients willfail to eradicate the bacteria and remain H.pylori-pos-itive,and this value could be higher in clinical prac-tice.5The recommended second-line therapy is thequadruple regimen composed of tetracycline,metroni-dazole,bismuth salts and a PPI;however,the efficacyof this regimen is limited by poor patients compliancebecause of side effects,number of tablets per day andlong duration.6Antibiotic-associatedgastrointestinalsideeffectssuch as diarrhoea,nausea,vomitting,bloating andabdominal pain may represent a serious drawback ofanti-H.pylori therapies,although they are mild in mostcases,usually resulting in discontinue therapy.7Thesemanifestations have been related to quantitative andqualitativechangesintheintestinalmicroflorabecause of unabsorbed or secreted antibiotics in theintestinal content,with a resulting reduction in normalsaprophytic flora and overgrowth and the persistenceof potentially pathogenic antibiotic-resistant indigen-ous strains.8Several studies have previously reported that certainprobiotic bacteria,such as Lactobacillus spp.,exhibitinhibitory activity against H.pylori in vitro and invivo.9,10Therefore,the Maastricht 22000 ConsensusReport had speculated on the role of probiotics supple-mentation in the treatment of H.pylori infection.11Many clinical trials have obtained relatively goodresults by combining triple therapy with probioticswithout importantly adverse effects.7,1222However,the number of patients enrolled in these trials was toosmall to achieve statistically conclusive results.Theprimary aim of the meta-analysis was to study whe-ther probiotics could improve eradication rates andreduce side effects of anti-H.pylori treatment.METHODSBefore performing the meta-analysis,we developed asimplified protocol,including search strategies,cri-teria for study selection,how to exact related data,methods for assessing study quality and statisticalmethodology.Search strategyThe electronic databases,Pubmed(1966 to October2006),Embase(1980 to October 2006),the CochraneControlled Trials Register(Issue 2,2006),the ScienceCitation Index(1945 to October 2006)and the ChineseBiomedical Database(1981 to October 2006),wereused for systematic literature searches.We employedbothMedicalsubjectheadingsandfree-languageterms for Helicobacter pylori(H.pylori)combiningwith each of the following:probiotic,probiotics,yeast,yogurt,Lactobacillus,Bifidobacterium,Saccharomyces,Lactococcus,Streptococcus,Enterococcus,as searchterms.A comprehensive search of reference lists of allreview articles and original studies retrieved by thismethod was performed to identify additional reports.Furthermore,we had searched abstracts of major gast-roenterological meetings,such as the Digestive DiseaseWeek of the American Gastroenterological Association,the World Congress of Gastroenterology,EuropeanHelicobacter Study Group.Authors of some identifiedtrials were asked whether they knew of additionalstudies,including unpublished randomized ones.Inclusion criteria and Exclusion criteriaThe selection criteria were as follows:(i)Only randomized controlled trials(RCTs)wereincluded.(ii)The study population consisted of subjects whohad never been treated for H.pylori infection before.(iii)Regardless of age and symptoms at the time ofenrollment.(iv)Confirmation of infection eradication by his-tology or13C-urea breath test at least 4 weeks aftertherapy.(v)Studies including at least two branches of treat-ment consisting of(a)Patients in the control groupreceived proton pump inhibitor plus two antibioticswith placebo or no additional intervention;(b)Patientsin experimental groups received probiotics plus thesame eradication regimen.156J.L.TONG et al.2007 The Authors,Aliment Pharmacol Ther 25,155168Journal compilation 2007 Blackwell Publishing Ltd(iv)Studies evaluating these therapies as a second-line treatment for previous eradication failures werealso included.(vii)The data of successful eradication or/and sideeffects in H.pylori eradication was available.Furthermore,articles published in English and Chi-nese were included.Studies in other languages wereexcluded unless a translation was available.Data extractionStandardized data abstraction sheets were prepared.Data were exacted for study quality,type and durationof probiotic treatment,anti-H.pylori regimens,andthe number,sex and age of enrolled subjects,diagnos-tic methods of testing H.pylori infection before enrol-ling and after completing study,score systems forassessing side effects;meanwhile,key outcome data,such as eradication rates,occurrence of diarrhoea,nausea,taste disturbance and epigastric pain wereabstracted from all included studies.All articles wereexamined independently for eligibility by two review-ers(Tong and Shen).Disagreements were resolved byconsulting a third reviewer(Ran).Quality was assessedusing the Jadad score system based on three items,randomization,double blinding and description ofwithdrawals and dropouts,to generate scores from 0to 5.The selected studies were scored independentlyby two investigators(Tong and Zhang),and if therewere disagreements,then consulted a third reviewer(Ran)to obtain the final scores.We considered thatthey were low quality when theirs scores were lesserthan 3.When the results of a particular study werereported in more than one publication,only the mostrecent and complete data were included in the meta-analysis.Statistical analysisData were entered into the Cochrane Collaborationreview manager programme RevMan 4.2.8(released on8 July 2005).The outcome measure examined was theodds ratios(ORs)of improving H.pylori eradicationrates and reducing side effects with probiotics vs.without probiotics combining with triple or quadrupleregimens.Eradication ratesandsideeffectswereanalysed based on a fixed-effects model using themethods of Mantel-Haenszel both by an intention-to-treat and by a per-protocol.Heterogeneity between thestudies was assessed by chi-square test.Statistical sig-nificance for the test of heterogeneity was set at 0.10.If significant heterogeneity exists,it would be inappro-priate to combine the data for further analysis using afixed-effects model,while the random model was usedfor calculations.Subanalyses for the meta-analysis were planneddepending on symptoms before enrollment,age of sub-jects and probiotic preparations.To evaluate the stabil-ity of the results of this meta-analysis,we performed asensitivity analysis.In this analysis,we evaluated theinfluence of low quality studies by estimating the ORsin the absence of one or more studies.Finally,we usedfunnel plot asymmetry to detect any publication biasin the meta-analysis,and Eggers regression test tomeasure funnel plot asymmetry.RESULTSDescription of the studiesThe bibliographical search yielded a total of 382 stud-ies.Of these studies,103 articles were excluded owingto publication types,i.e.84 reviews,12 case reports,two comments,two editorials,two congresses and onepiece of news,and 74 trials were excluded by usinganimals as subjects.We excluded another 174 articles(154 non-clinical trials and 20 unrelated articles)afterexamining the title and abstract,remaining 31 poten-tially relevant articles retrieved for more detailedassessment.Of these potential eligible articles,weexcluded another 19 articles,because of no combiningwith triple or quadruple eradication regimens(n 10),2332combing with antibiotics but not mentioneradication rates and/or side effects(n 1),33pub-lished in Russian(n 3),3436combining with doubleregimen(n 1)37and investigating relation to forma-tion of cholesterol gallstones(n 1)38and the influ-ence on ornithine decarboxylase activity(n 1)39andeffects on the intestinal microbiota(n 2).40,41Fur-thermore,we identified two additional articles fromthe Chinese Biomedical Database(1981 to October2006).42,43Finally,14 RCTs met the inclusion criteria.Two institutions published two articles each Armuzziet al.in Italy and Sheu BS et al.in Taiwan.This raiseda concern about dual publication.Close study of thearticles suggested that there were separate studiesreported in all these four articles.For example,sub-jects of the trial conducted by Armuzzi published inAlimentaryPharmacology&Therapeuticsenrolledfrom 1 September 1999 to 31 January 2000,but fromMETA-ANALYSIS:PROBIOTICS AND HELICOBACTER PYLORI ERADICATION157 2007 The Authors,Aliment Pharmacol Ther 25,155168Journal compilation 2007 Blackwell Publishing Ltd1 May to 31 July 1999 in the trial published in Diges-tion.We therefore included all four articles in themeta-analysis.The flowchart of reviews showed thedetailed process of selection(Figure 1).The character-istics of 14 trials included in meta-analysis are sum-marized in Tables 1 and 2,including quality score andstudies excluded from meta-analysis were listed inTable 3.Eradication ratesEleven studies described H.pylori eradication rates(excluded two studies with rescue therapy and onecontaining no data for eradication rates)were selec-ted for the meta-analysis.Of these selected studies,threereportedsignificantlyimprovederadicationrates,the remaining eight had similar efficacy forH.pylori eradication.Pooled eradication rates wereachievedin463of554patientswithprobioticssupplementation(83.6%;95%CI 80.586.7%)andin 389 of 520 patients without probiotics(74.8%;95%CI 71.178.5%)by intention-to-treat analysis,the OR was 1.84(95%CI 1.342.54)(Figure 2).Overall,per-protocol eradication rates were 85.4%(95%CI 82.588.4%)and 77.6%(95%CI 75.879.5%)for probiotics supplementation and withoutprobiotics,respectively(OR1.82;95%CI 1.302.56).For H.pylori eradication failures,the effect of pro-biotics supplementation on eradication rates was alsoevaluated.We identified two RCTs related to probioticssupplementationduringH.pylorieradicationforpatients with eradication failure.Pooled eradicationrates were achieved in 92 of 104 patients with probiot-ics supplementation(88.5%;95%CI 82.394.6%)and in 79 of 104 patients without probiotics(76.0%;95%CI 84.267.7%).The OR was 2.47(95%CI 1.165.29)(Figure 3).Side effectsTotal side effects were initially performed for meta-analysis.Data for occurrence of total side effects canobtain from seven RCTs.Of these studies,four trialsreported significant minimized the occurrence of gas-trointestinal side effects.Side effects in probiotic sup-plementation were lesser than to without probiotic:24.7%;95%CI 20.029.4%and 38.5%;95%CI 33.0%44.1%,respectively;the summary OR was 0.44(95%CI 0.300.66)(Figure4).Thenindividualsymptoms during eradication therapy,such as diar-rhoea(Figure 5),epigastric pain(Figure 6),nausea(Figure 7),taste disturbance(Figure 8)were also ana-lysed,respectively.Most of the trials reported lowerincidences of diarrhoea(6.0%vs.16.1%),OR were0.34(95%CI 0.220.52).Incidences of taste distur-bance,nausea and epigastric pain were similar in bothgroups,although marginally significant(OR 0.38 vs.0.58 vs.0.62;95%CI 0.170.85 vs.0.380.88 vs.0.390.97).Subanalysis/sensitivity analysisSubanalysesforthemeta-analysiswereplanneddepending on symptoms before enrollment,age ofsubjects and probiotic preparations.We divided all eli-gible trials into symptomatic subgroup and asympto-matic subgroup or child subgroup and adult subgroup.For H.pylori eradication rates,all trials performed on382 potentially relevant literatures identified from search strategy 103 papers excluded on the basis of publication type 84 reviews 12 caser reports Two comments Two editorials Two congresses One news 205 papers from Initial Screening 154 non-clinical trials 20 unrelated papers 31 potentially relevant papers retrieved for more detailed assessment 19 papers were excluded 10 no combining with triple or quadruple eradication regimens One combing with antibiotics but not mention eradication rates and side effects One investigating relation to formation of cholesterol gallstones Three published in Russian One combining with double regimen One the influence on ornithine decarboxylase activity Two effects on the intestinal microbiota 14 satisfied inclus
