Rifaximin Reduces the Number and Severity of Intestinal Lesions Associated With Use of Nonsteroidal Anti-Inflammatory Drugs in Humans

Abstract

The intestinal microbiota might contribute to enteropathy associated with use of nonsteroidal anti-inflammatory drugs (NSAIDs), but there have been few human studies of this association. We performed a placebo-controlled study to determine whether a delayed-release antibiotic formulation (rifaximin-extended intestinal release [EIR]) prevents the development of intestinal lesions in subjects taking daily NSAIDs. Sixty healthy volunteers (median age, 26 y; 42% female) were given the NSAID diclofenac (75 mg twice daily) plus omeprazole (20 mg once daily), and either rifaximin-EIR (400 mg) or placebo, twice daily for 14 days. Subjects were assessed by videocapsule endoscopy at baseline and after 2 weeks of treatment. The primary end point was the proportion of subjects developing at least 1 small-bowel mucosal break at week 2. Secondary end points were the change in the mean number of mucosal lesions and the number of subjects with large erosions and/or ulcers after 14 days of exposure. We detected mucosal breaks in 20% of subjects given rifaximin and in 43% of subjects given placebo (P = .05 in the post hoc sensitivity analysis). None of the subjects in the rifaximin group developed large lesions, compared with 9 subjects in the placebo group (P < .001). Our findings indicate that intestinal bacteria contribute to the development of NSAID-associated enteropathy in human beings. Clinical trial no: EudraCT 2013-000730-36. The intestinal microbiota might contribute to enteropathy associated with use of nonsteroidal anti-inflammatory drugs (NSAIDs), but there have been few human studies of this association. We performed a placebo-controlled study to determine whether a delayed-release antibiotic formulation (rifaximin-extended intestinal release [EIR]) prevents the development of intestinal lesions in subjects taking daily NSAIDs. Sixty healthy volunteers (median age, 26 y; 42% female) were given the NSAID diclofenac (75 mg twice daily) plus omeprazole (20 mg once daily), and either rifaximin-EIR (400 mg) or placebo, twice daily for 14 days. Subjects were assessed by videocapsule endoscopy at baseline and after 2 weeks of treatment. The primary end point was the proportion of subjects developing at least 1 small-bowel mucosal break at week 2. Secondary end points were the change in the mean number of mucosal lesions and the number of subjects with large erosions and/or ulcers after 14 days of exposure. We detected mucosal breaks in 20% of subjects given rifaximin and in 43% of subjects given placebo (P = .05 in the post hoc sensitivity analysis). None of the subjects in the rifaximin group developed large lesions, compared with 9 subjects in the placebo group (P < .001). Our findings indicate that intestinal bacteria contribute to the development of NSAID-associated enteropathy in human beings. Clinical trial no: EudraCT 2013-000730-36. Nonsteroidal anti-inflammatory drugs (NSAIDs) are very effective medications, but their use is associated with a broad spectrum of adverse reactions involving the liver, kidney, cardiovascular system, skin, and gut. Gastrointestinal (GI) adverse effects are the most common and cover a wide clinical spectrum, ranging from dyspepsia, heartburn, and abdominal discomfort to more serious events, such as peptic ulcer with life-threatening ulcer complications of bleeding and perforation.1Scarpignato C. et al.BMC Med. 2015; 13: 55Crossref PubMed Scopus (173) Google Scholar Over the past decade, there has been a progressive change in the overall pattern of GI events leading to hospitalization, with a clear decreasing trend in upper GI events and a slight, but significant, increase in lower GI events.2Lanas A. et al.Am J Gastroenterol. 2009; 104: 1633-1641Crossref PubMed Scopus (429) Google Scholar Indeed, available studies3Graham D.Y. et al.Clin Gastroenterol Hepatol. 2005; 3: 55-59Abstract Full Text Full Text PDF PubMed Scopus (472) Google Scholar, 4Maiden L. et al.Gastroenterology. 2005; 128: 1172-1178Abstract Full Text Full Text PDF PubMed Scopus (411) Google Scholar have shown that approximately 75% of NSAID users display intestinal mucosal injury, ranging from denuded areas (seen mainly in the proximal small bowel) to the so-called mucosal breaks (erosions and ulcers), observed in its distal part. Although the incidence of upper GI injury can be reduced by proton pump inhibitors, this is not the case for NSAID-associated intestinal lesions, which actually may be aggravated by acid suppression.5Washio E. et al.Clin Gastroenterol Hepatol. 2016; 14: 809-815Abstract Full Text Full Text PDF PubMed Scopus (109) Google Scholar The pathogenesis of small intestinal damage is complex and still not completely understood. Several lines of experimental evidence have implicated commensal enterobacteria in the pathogenesis of NSAID enteropathy and suggest that luminal bacteria may represent a potential target for prevention and/or treatment.6Scarpignato C. Gut. 2008; 57: 145-148Crossref PubMed Scopus (50) Google Scholar In this context, the efficacy of a delayed-release formulation of rifaximin, a poorly absorbed antibiotic,7Scarpignato C. et al.Chemotherapy. 2005; 51: 36-66Crossref PubMed Scopus (202) Google Scholar in the prevention of NSAID-associated lesions was evaluated in human beings by means of videocapsule endoscopy (VCE). Sixty healthy volunteers (median age, 26 y; 42% female) were randomized to receive diclofenac slow release 75 mg twice daily plus omeprazole 20 mg once daily and either rifaximin-EIR 400 mg or rifaximin-EIR matching placebo twice daily for 14 days (Supplementary Figure 1). The primary end point was the proportion of subjects developing at least 1 small-bowel mucosal break at final VCE. Secondary end points were the change at VCE in the mean number of mucosal lesions and the number of subjects with large erosions and/or ulcers at the end of treatment (see the Supplementary Materials and Methods section and Supplementary Table 1). Six patients in the rifaximin group and 13 patients in the placebo group (12 of the per-protocol set) developed at least 1 mucosal lesion in the small bowel (primary end point). This provided an odds ratio of 0.33 (95% confidence interval [CI], 0.10–1.03) and 0.35 (95% CI, 0.11–1.12) in the modified full analysis (mFA) and per-protocol sets, respectively, for subjects in the rifaximin group to develop at least 1 mucosal lesion in the small bowel at logistic regression analysis. The efficacy on primary end point was not significant. However, when post hoc sensitivity analysis, using age as covariate, was performed, the difference between the rifaximin and placebo groups became significant (Table 1). The average mucosal score (mFA set) at the end of treatment was 0.87 ± 0.13 vs 1.83 ± 0.28 for rifaximin and placebo, respectively (P = .021).Table 1Primary and Secondary End Points of the StudyPrimary end pointRifaximinPlaceboP valueSubjects developing at least 1 mucosal lesion, n6/30 (20%)13/30 (43%).0566aLogistic regression analysis.Sensitivity analysis (treatment and sex as fixed effect and age as covariate).0490Secondary end pointsChange in the mean number of mucosal lesions, ±SEM0.3 ± 0.71.2 ± 2.3.0103bNegative binomial regression analysis.Subjects with large erosions and/or ulcers at the end of treatment, n09.001cChi-square test.a Logistic regression analysis.b Negative binomial regression analysis.c Chi-square test. Open table in a new tab Both secondary end points were reached successfully in the study (Table 1). Subjects in the placebo group developed a higher number of mucosal lesions compared with those in the rifaximin group (1.2 ± 2.3 vs 0.3 ± 0.7 in the mFA set). Negative binomial regression analysis proved a protective effect of rifaximin on mean changes from baseline in total number of lesions (treatment effect point estimate, -1.41; 95% CI, -2.49 to -0.34; P = .010). A similar result was detected for the change from baseline in the number of lesions without hemorrhage (treatment effect point estimate, -1.44; 95% CI, -2.53 to -0.35; P = .009). Both treatments were well tolerated and no serious adverse events were recorded. This was a randomized controlled trial investigating the effect of an antibiotic on NSAID-induced intestinal mucosal injury in human beings. This study showed that fewer rifaximin-treated volunteers developed small-bowel lesions compared with placebo-treated subjects. The antibiotic also reduced the mean number of lesions and appeared to abolish the larger lesions, with a tolerability profile overlapping that of placebo. Although short-term studies, such as this one, and mucosal breaks may not have straightforward clinical implications,8Bar-Meir S. Aliment Pharmacol Ther. 2006; 24: 19-21Crossref PubMed Scopus (35) Google Scholar the results of this proof-of-concept study strongly suggest the role of enteric bacteria in the pathogenesis of NSAID enteropathy and call for a prospective trial in patients on long-term NSAID therapy. The entero-protective effect of rifaximin most likely depends on its broad spectrum of antibacterial activity.7Scarpignato C. et al.Chemotherapy. 2005; 51: 36-66Crossref PubMed Scopus (202) Google Scholar Experimental studies9Kim M.S. et al.Dig Dis Sci. 2013; 58: 1676-1682Crossref PubMed Scopus (37) Google Scholar, 10Xu D. et al.Gastroenterology. 2014; 146: 484-496.e4Abstract Full Text Full Text PDF PubMed Scopus (222) Google Scholar, 11Fornai M. et al.Pharmacol Res. 2016; 104: 186-196Crossref PubMed Scopus (26) Google Scholar show not only that this antibiotic reduces the total bacterial load, but also modulates bacterial community composition, an effect associated with a reduction of intestinal inflammation10Xu D. et al.Gastroenterology. 2014; 146: 484-496.e4Abstract Full Text Full Text PDF PubMed Scopus (222) Google Scholar, 11Fornai M. et al.Pharmacol Res. 2016; 104: 186-196Crossref PubMed Scopus (26) Google Scholar and improvement of gut barrier function.10Xu D. et al.Gastroenterology. 2014; 146: 484-496.e4Abstract Full Text Full Text PDF PubMed Scopus (222) Google Scholar Recent evidence points out that—besides nonantimicrobial activities (for instance, the anti-inflammatory property)—rifaximin also may show “eubiotic” properties. Indeed, in patients with inflammatory conditions (such as inflammatory bowel disease, colonic diverticular disease, or hepatic encephalopathy), the drug, although not altering the overall structure of human colonic microbiota, increased the relative abundance of Bifidobacteria and Lactobacilli.12Maccaferri S. et al.J Antimicrob Chemother. 2010; 65: 2556-2565Crossref PubMed Scopus (141) Google Scholar, 13Ponziani F.R. et al.Dig Dis. 2016; 34: 269-278Crossref PubMed Scopus (105) Google Scholar NSAID enteropathy is associated with significant GI complications,14Bjarnason I. et al.Lancet. 1987; 2: 711-714Abstract PubMed Scopus (211) Google Scholar, 15Chan F.K. et al.Lancet. 2010; 376: 173-179Abstract Full Text Full Text PDF PubMed Scopus (301) Google Scholar but there are no proven strategies for the prevention or treatment of this subtle clinical condition. However, there is now good evidence that intestinal bacteria play a pathogenic role in NSAID enteropathy in human beings. Treatments aimed at correcting the shift of intestinal microbiota toward proinflammatory gram-negative bacteria6Scarpignato C. Gut. 2008; 57: 145-148Crossref PubMed Scopus (50) Google Scholar are therefore potential avenues to explore in the prevention and treatment of NSAID enteropathy. After almost 40 years, with the advancement of knowledge on the pathogenic role of gut microbiota in NSAID enteropathy,6Scarpignato C. Gut. 2008; 57: 145-148Crossref PubMed Scopus (50) Google Scholar the time now seems ripe to study in well-designed, large, randomized, clinical trials, microbiota-directed interventions to protect the small bowel from NSAID injury and to allow safer anti-inflammatory therapy. Werner Dolak and Peter Matzneller had full access to all of the data in the study and take responsibility for the integrity of the data; Carmelo Scarpignato was responsible for the study concept and obtained funding; Carmelo Scarpignato, Ingvar Bjarnason, Angel Lanas, Maria Grimaldi, Werner Dolak, and Markus Zeitlinger were responsible for the study design; Werner Dolak, Ingvar Bjarnason, Angel Lanas, and Carmelo Scarpignato were responsible for the analysis and interpretation of data; Carmelo Scarpignato, Werner Dolak, and Markus Zeitlinger were responsible for drafting the manuscript; Ingvar Bjarnason and Angel Lanas were responsible for critical revision of the manuscript for important intellectual content; Current Research Organization for Scientific Services (CROss) Alliance and Cecilia Renzulli performed the statistical analysis; CROss Alliance provided administrative, technical, or material support; and Maria Grimaldi provided study supervision. This double-blind, randomized, placebo-controlled, phase IIa trial was conducted according to the CONSORT guidelines1Schulz K.F. et al.Ann Intern Med. 2010; 152: 726-732Crossref PubMed Scopus (2682) Google Scholar at the Department of Clinical Pharmacology of the General Hospital of Vienna in Vienna, Austria. The study protocol was approved by the internal review board of the Medical University of Vienna (EK 1244/2013) and the national competent authority (ref no. 718031). This study was registered at EudraCT (2013-000730-36). The trial was performed according to the International Conference on Harmonization of Technical Requirements for Registration of Pharmaceuticals for Human Use, guidelines for Good Clinical Practice,2ICH. https://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Efficacy/E6/E6_R1_Guideline.pdf, 1996. Accessed March 3, 2017.Google Scholar and the Declaration of Helsinki (1996 version, amended October 2000).3WMA. http://www.wma.net/en/30publications/10policies/b3/17c.pdf, 2008. Accessed March 3, 2017.Google Scholar Healthy male and female volunteers (age, 18–64 y) provided their availability and willingness to participate in the study. Sixty subjects met all the inclusion criteria (ie, unremarkable physical examination and medical history, normal vital signs, and all laboratory test values, including hematology, blood chemistry, virology, urinalysis, and screened for drug of abuse were within the reference ranges). Female subjects with child-bearing potential were required to have a negative urine pregnancy test. Exclusion criteria included any serious neurologic, psychiatric, cardiovascular, respiratory, gastrointestinal, renal, or musculoskeletal disease, drug misuse including alcohol, pregnancy, and age outside the range of 18–64 years. Subjects who used steroids, nonsteroidal anti-inflammatory drugs, bisphosphonates, sulfasalazine, biological drugs, treatments targeting gut microbiota (antibiotics, prebiotics, or probiotics), as well as gastrointestinal prokinetic drugs within the previous 30 days also were excluded. The use of these drugs, even occasionally, was prohibited during the study. All participants were admitted to the Investigational Unit. After an overnight stay at the Department of Clinical Pharmacology, Medical University of Vienna, with appropriate preparation (see later), eligible subjects underwent VCE of the small bowel (baseline VCE). To be included in the study, VCE should have shown no abnormality, but a maximum of 1 mucosal break (erosion or ulcer) in the small bowel was allowed. After re-assessment of inclusion/exclusion criteria, subjects were randomized according to a computer-generated list on a 1:1 basis to receive diclofenac slow release 75 mg twice daily (Voltaren; Novartis Farma SpA, Origgio, VA, Italy) plus omeprazole 20 mg once daily (Losec; AstraZeneca SpA, Basiglio, MI, Italy), plus either rifaximin-EIR 400 mg (Alfa Wasserman SpA, Bologna, Italy) or rifaximin-EIR matching placebo twice daily (Alfa Wasserman SpA) for a total of 14 days. Among different NSAIDs, diclofenac was selected because its administration is associated with a definite risk of small intestinal injury4Ishihara M. et al.Aliment Pharmacol Ther. 2014; 40: 538-547PubMed Google Scholar and because it has been used in almost all of the videocapsule studies that have been performed on VCE (for review see Fujimori et al5Fujimori S. et al.Pharmaceuticals. 2010; 3: 1187-1201Crossref Scopus (15) Google Scholar). In a previous study,6Maiden L. et al.Gastroenterology. 2005; 128: 1172-1178Abstract Full Text Full Text PDF PubMed Scopus (450) Google Scholar we found that its slow-release formulation (combined with omeprazole) was associated—after 14 days—with intestinal mucosal lesions in 68% of healthy subjects. Drug intake was performed and witnessed at the study site at 8 AM and 8 PM (±1 h). Within 36 hours after the last study drug intake, the subjects underwent a second VCE assessment (final VCE), vital signs check, physical examination, and safety laboratory tests (hematology, blood chemistry, and urinalysis). In case of study withdrawal, an early termination visit including a final VCE was performed. All patients were followed up using a telephone interview 1 week after the final VCE visit to ensure capsule excretion. Baseline and post-treatment VCE were performed on an inpatient basis at the study site. On the day before VCE subjects were allowed to eat solid food until 1:00 PM and to drink clear liquids until 10:00 PM. In the evening they received 2 L of a polyethylene glycol cleansing solution (Klean-Prep; Norgine, Marburg, Germany), according to the current guidelines7Song H.J. et al.Gastroenterol Res Pract. 2016; 2016: 6802810PubMed Google Scholar and on-site protocol.8Dolak W. et al.Endoscopy. 2012; 44: 1012-1020Crossref PubMed Scopus (37) Google Scholar The next morning VCE was performed after attachment of the sensor belt and data recorder. Before swallowing the capsule (PillCam SB2; Given Imaging, Yokneam, Israel), all subjects received 70 mg of simethicone (Sab-Simplex; Pfizer m.b.H., Vienna, Austria) to avoid visual interference by intestinal air bubbles.9Ge Z.Z. et al.Endoscopy. 2006; 38: 836-840Crossref PubMed Scopus (70) Google Scholar, 10Spada C. et al.Dig Liver Dis. 2010; 42: 365-370Abstract Full Text Full Text PDF PubMed Scopus (45) Google Scholar Subjects were allowed to drink clear liquids 2 hours after capsule intake and to eat a light meal 2 hours later. The sensor belt and data recorder were detached after a total capsule operation time of 8 hours. All capsule videos were pre-analyzed on site by a gastrointestinal endoscopist (W.D.) with experience in VCE, who had no knowledge of the treatment regimen. VCE results were re-evaluated blindly by a panel of 3 independent VCE experts (I.B., A.L., and C.S.). In case of disagreement between initial evaluation and re-assessment, consensus was achieved on the final diagnosis between W.D. and the 3 VCE experts. Patient data remained blinded until final consensus was achieved. All findings were classified following the “five-point scoring system for endoscopic lesions with capsule endoscopy”11Graham D.Y. et al.Clin Gastroenterol Hepatol. 2005; 3: 55-59Abstract Full Text Full Text PDF PubMed Scopus (499) Google Scholar (Supplementary Table 1). The primary end point of the study was the proportion of subjects who developed at least 1 mucosal break, defined as an erosion or ulcer, in the small bowel at final VCE. Secondary end points were the change from baseline to final VCE in the number of mucosal breaks, those with and without hemorrhage, as well as factors reflecting safety and tolerability of the study treatment such as the distribution of treatment-emergent adverse events (defined as all adverse events occurring or worsening after the first dose of the study overall adverse events, or a change in clinical laboratory and vital on an of mucosal lesions of in the a of subjects subjects was to a reduction of approximately in the primary end point for subjects in the a of a of and a were used to baseline and final were performed for the following data mFA of all randomized subjects who received at least 1 dose of study and who successfully the final per-protocol of all randomized subjects included in the mFA without any protocol and safety of all randomized subjects who received at least 1 dose of study Logistic regression analysis and the test were used for of the primary end point between A negative binomial with the change from baseline as the and treatment as a fixed effect was used to the secondary efficacy end points between were only for was used for all statistical in the of this study. were only after the statistical analysis of primary and secondary end All had access to the study the statistical and and approved the final Table for of with of number of erosions number of erosion breaks erosion and/or Open table in a new tab in of Nonsteroidal Good with the by Scarpignato et the efficacy of rifaximin in intestinal lesions associated with nonsteroidal anti-inflammatory drugs The that rifaximin is effective in the of NSAID-induced intestinal lesions and that intestinal bacteria contribute to the development of NSAID-associated mucosal injury is very common in NSAID and effective treatment for NSAID-associated enteropathy is to PDF