Randomized controlled trial of clarithromycin and ethambutol in the treatment of Crohn’s disease

Authors

  • R. W. Goodgame,

    1. Department of Medicine, Sections of Gastroenterology and Design and Analysis, and Department of Pathology, Texas Children’s Hospital, Baylor College of Medicine, Houston, TX, USA,
    Search for more papers by this author
  • K. Kimball,

    1. Department of Medicine, Sections of Gastroenterology and Design and Analysis, and Department of Pathology, Texas Children’s Hospital, Baylor College of Medicine, Houston, TX, USA,
    Search for more papers by this author
  • S. Akram,

    1. Department of Medicine, Sections of Gastroenterology and Design and Analysis, and Department of Pathology, Texas Children’s Hospital, Baylor College of Medicine, Houston, TX, USA,
    Search for more papers by this author
  • E. Ike,

    1. Pharmacy Administration, Harris County Hospital District
    Search for more papers by this author
  • C.-N. Ou,

    1. Department of Medicine, Sections of Gastroenterology and Design and Analysis, and Department of Pathology, Texas Children’s Hospital, Baylor College of Medicine, Houston, TX, USA,
    Search for more papers by this author
  • F. Sutton,

    1. Department of Medicine, Sections of Gastroenterology and Design and Analysis, and Department of Pathology, Texas Children’s Hospital, Baylor College of Medicine, Houston, TX, USA,
    Search for more papers by this author
  • D. Graham

    1. Department of Medicine, Sections of Gastroenterology and Design and Analysis, and Department of Pathology, Texas Children’s Hospital, Baylor College of Medicine, Houston, TX, USA,
    Search for more papers by this author

Dr R. W. Goodgame, Baylor College of Medicine, One Baylor Plaza, Room 525-D, Houston, TX 77030, USA. E-mail: goodgame@bcm.tmc.edu

Abstract

Background:

A mycobacterial infection may be the cause of Crohn’s disease in some patients. Measurement of intestinal permeability may identify Crohn’s disease patients with a high likelihood of relapse and may quantify the severity of intestinal injury.

Aim:

To assess the effect of 3 months of clarithromycin and ethambutol on the disease activity and intestinal permeability in patients with Crohn’s disease at high risk of relapse.

Methods:

Patients with Crohn’s disease, with a lactulose–mannitol permeability test above 0.03, were randomly assigned to receive either clarithromycin, 500 mg twice daily, and ethambutol, 15 mg/kg daily, or identically appearing placebo for 3 months in addition to their regular therapy. The Harvey–Bradshaw index and the lactulose–mannitol test were assessed in a blind fashion every 3 months for 12 months.

Results:

Thirty-one patients were randomized to receive either drugs (n=15) or placebo (n=16). The groups were similar in age, sex, duration of disease, location of disease, past complications and disease severity. Specifically, there was no difference between the drug or placebo groups in the mean Harvey–Bradshaw index (4.8 vs. 4.4), number with active disease (33% vs. 44%) and mean lactulose–mannitol test (0.06 vs. 0.10). During the 12-month follow-up period, there were no consistent, statistically significant differences in the mean Harvey–Bradshaw index or lactulose–mannitol test between treatment and placebo groups. Individual patients showed either improvement or worsening of these indices, but these were not related to study medication. Specifically, no ‘cures’ were noted with anti-mycobacterial treatment.

Conclusions:

Three months of treatment with clarithromycin and ethambutol does not benefit Crohn’s disease patients who are receiving standard medical therapy.

INTRODUCTION

The aetiology of Crohn’s disease is unknown. Several laboratories have isolated mycobacteria from the gut of Crohn’s disease patients.1–4 More recently, polymerase chain reaction was used to demonstrate an association between Mycobacterium paratuberculosis and Crohn’s disease.5–8 Humoral immune response to mycobacterial antigens was also used to support a relationship between Crohn’s disease and mycobacterial infection.9, 10 However, other studies relying on culture,11, 12 polymerase chain reaction11–16 and measurement of specific immune responses17–19 have not supported an aetiologic relationship between mycobacteria and Crohn’s disease.

Studies using anti-mycobacterial agents in the treatment of Crohn’s disease have also given conflicting and inconclusive results.20–24 One reason for the failure of past therapeutic trials may relate to the ineffectiveness of the antimicrobial agents studied. Clarithromycin, one of the newer macrolide antibiotics, has in vitro and in vivo bactericidal activity against many mycobacteria, including Mycobacterium paratuberculosis.25, 26 The extracellular and intracellular activity of clarithromycin against mycobacteria has been shown to be further potentiated by ethambutol.25 We hypothesized that 3 months of treatment with a combination of clarithromycin and ethambutol should reverse the pathologic process in patients whose disease was caused by Mycobacterium paratuberculosis-like organisms.

Patients with Crohn’s disease have increased intestinal permeability. The severity of the permeability abnormality correlates with the activity of the disease.27 It has been suggested that the lactulose–mannitol permeability test may reliably identify Crohn’s disease patients at high risk of relapse.28 Wyatt et al. studied 72 patients with quiescent Crohn’s disease.28 Patients were followed for 1 year after the test. Twenty-six of 36 patients with raised permeability (greater than 0.030) relapsed, but only six of 35 with normal permeability relapsed. The sensitivity of the test as a predictor for relapse was 81%.29, 30 Subsequent studies have found that Crohn’s disease patients with abnormal permeability have about twice the likelihood of relapse within 1 year than patients with normal permeability.

We hypothesized that patients with a high risk of relapse (identified by a high lactulose–mannitol permeability test) would have their risk of relapse reduced, and subsequent lactulose–mannitol test improved, by adding clarithromycin and ethambutol to their standard medical therapy. We further hypothesized that the lactulose–mannitol test would be a sensitive method of quantifying intestinal injury in Crohn’s disease and would be useful in evaluating the effects of therapy. Therefore, we carried out a prospective, randomized trial to evaluate the efficacy of 3 months of clarithromycin and ethambutol in treating and preventing the relapse of Crohn’s disease.

MATERIALS AND METHODS

The study was a prospective, randomized, placebo-controlled trial. Patients over 18 years of age with documented Crohn’s disease (radiologically, endoscopically and/or pathologically proven), irrespective of the site of intestinal involvement, were identified in medical and gastroenterology clinics at the Ben Taub General Hospital, Houston, Texas. Patients were excluded if they had any of the following: unable or unwilling to fulfil the requirements of the study; known allergy to ethambutol, clarithromycin or any other macrolide antibiotic; active complications of Crohn’s disease such as intestinal obstruction, abscess or fistula; pregnant women or women of child-bearing age not on effective birth control; established retinopathy; a history of mycobacterial disease or recent or current treatment for mycobacterial disease. After informed consent for screening, the patients underwent a complete medical history, physical examination, complete blood count, renal and hepatic chemistries and calculation of the Harvey–Bradshaw index.31 Patients also underwent a detailed review of the past history of Crohn’s disease, including previous site of involvement, complications, operations and medications. A test of intestinal permeability using the differential absorption of lactulose and mannitol (lactulose–mannitol test) was also performed. After producing a baseline urine specimen, a fasting patient ingested a 150-mL aqueous solution containing 5 g of mannitol, 10 g of lactulose and 20 g of sucrose. Urine was collected for 6 h and the total volume was measured. A 40-mL aliquot was stored at – 20 °C for subsequent analysis. Lactulose and mannitol were measured simultaneously by the high pressure liquid chromatography method described by Catassi et al.32 The total amount of lactulose and mannitol in the 6-h collection was determined and expressed as the percentage recovery of the administered amount. The test results are expressed as a ratio calculated as the percentage recovery of lactulose/percentage recovery of mannitol.32

Patients with a lactulose–mannitol test greater than 0.03 were stratified according to whether they had active or inactive disease, using a Harvey–Bradshaw index greater than or equal to 5 to indicate active disease.31 These patients were then randomized, using random numbers, to receive ethambutol, 15 mg/kg daily, and clarithromycin, 500 mg twice daily, or identically appearing placebo for 3 months. The study drugs were given as an adjunct to therapy prescribed and supervised by the primary physicians (not the study physicians). The patients, study physicians and primary physicians were blind to whether patients were receiving anti-mycobacterial treatment or placebo.

Patients were followed for 1 year. Monthly telephone calls were made and, if this conversation suggested a relapse or change in Crohn’s disease activity, the patient was asked to come for a clinic visit. The Harvey–Bradshaw index and the lactulose–mannitol test were repeated on the third, sixth, ninth and twelfth months. Standard tests for visual acuity and colour vision were administered monthly during treatment to all patients to monitor for ethambutol-related retinal toxicity. The primary endpoints were the changes in the lactulose–mannitol test and the Harvey–Bradshaw index. Secondary endpoints were diarrhoea, abdominal pain, the need for surgery and the need for admission to hospital.

To maximize the available data, results were analysed from baseline to each follow-up. Changes from baseline to each follow-up in the drug and placebo groups were compared using the Kruskal–Wallis test.

RESULTS

The study was approved by the Baylor Institutional Review Board for Human Research and all patients gave written informed consent. A total of 50 patients had some form of screening, but only 42 agreed to long-term follow-up. Of these, 31 had a lactulose–mannitol test greater than 0.03 and were eligible for randomization. All wanted to be randomized, but one did not return for medication. Table 1 shows the important demographic and clinical features of the patients who were randomized to receive treatment. There were no significant differences between the study groups.

Table 1.   Characteristics of patients randomized Thumbnail image of

The treatment period was 3 months, during which time the patients had to come monthly for eye examinations and evaluations. Twenty-two of 30 patients came for an evaluation after stopping treatment (evaluation at 3, 6, 9 or 12 months). The numbers of patients seen at the follow-up intervals were as follows: 3 months, 16 of 22; 6 months, 13 of 22; 9 months, 17 of 22; 12 months, 14 of 22. Patients who did not return for follow-up were similar in terms of demographic and clinical characteristics to patients who did return.

The primary outcomes were the Harvey–Bradshaw index and the lactulose–mannitol test. Table 2 shows the change from baseline of these two variables at each of the follow-up intervals. The only statistically significant differences were that the drug group had higher lactulose–mannitol tests at 3 and 6 months than the placebo groups. This did not persist at 6 and 12 months. There was a trend towards improved Harvey–Bradshaw index scores in the placebo group at 3 and 6 months, but this was not statistically significant (P=0.0756 and P=0.08, respectively). The data in Table 2 do not give a picture of the course of individual patients on drug or placebo followed over time. Figures 1 and 2 show the results for the individual patients with the most consistent follow-up. It is apparent that, in these few patients, drug treatment is not associated with obvious clinical or laboratory evidence of beneficial effects.

Table 2.   Change in the Harvey–Bradshaw index and lactulose–mannitol test in drug and placebo groups Thumbnail image of
Figure 1.

 Graphic representation of the data from Table 2 showing the change in the mean Harvey–Bradshaw index in placebo and drug groups over the 12-month follow-up period. The only statistically significant difference was found at 3 months, when the placebo group had a slightly lower mean Harvey–Bradshaw index. No differences at 6, 9 and 12 months in the clinical activity of disease based on the mean Harvey–Bradshaw index.

Figure 2.

 Graphic representation of the data from Table 2 showing the change in the mean permeability (lactulose–mannitol test) in placebo and drug groups over the 12-month follow-up period. There was no difference in the change at 3, 6 or 9 months. Also, there was no difference at 12 months, but a single value in the placebo group with a very high permeability score makes the mean very high. This was not a statistically significant difference.

Minor outcomes assessed included diarrhoea, abdominal pain, the need for surgery and the need for admission to hospital. There was no difference in the occurrence rate of these outcomes between the placebo- and drug-treated patients.

Six patients who had a lactulose–mannitol test below 0.03 were followed for 6–12 months without study drug or placebo. The mean Harvey–Bradshaw index increased from 3.3 to 5.2 (n=5) at 6 months and to 5.2 (n=5) at 12 months. Figure 3 shows the Harvey–Bradshaw scores of patients with a normal lactulose–mannitol test at baseline. Thus, the baseline lactulose–mannitol test does not reliably identify patients who are at risk of worsening clinical status.

Figure 3.

 Effect of drug vs. placebo on the Harvey–Bradshaw index at 3, 6, 9 and 12 months after starting treatment compared to baseline at 0 months.

DISCUSSION

Case series1–4, 13, 14 and case-control5–12, 15–19 studies have reported conflicting results regarding the possibility that Crohn’s disease is caused by a mycobacterial infection. If an infection is the cause of disease, effective antimicrobial treatment may be associated with clinical improvement and a change in the natural history of the disease. For example, eradication of Helicobacter pylori infection is associated with the cure of peptic ulcer disease and prevention of relapse. We, like others,20–24 tried to apply this principle to Crohn’s disease. We were unable to identify a difference in the activity of disease between those who had been treated for a mycobacterial infection and those who had not. However, these results can be contrasted with studies of H. pylori treatment for peptic ulcer disease in several ways. We did not make any attempt to identify active mycobacterial infection in patients or to assess the effective eradication of any organism by the treatment. In spite of the fact that the combination of clarithromycin and ethambutol is a potent and broad-spectrum anti-mycobacterial agent, it could be that mycobacterial infection plays a role in only a subset of patients. It could be that treatment has no effect on the causative organism.

We were hoping that our study would be enhanced by the use of the lactulose–mannitol test. Previous studies have suggested that a high permeability test would identify patients with a high likelihood of developing severe disease.28–30 This would allow us to demonstrate efficacy of therapy with fewer patients. Secondly, if a high lactulose–mannitol test is the result of mycobacterial infection and not just a predisposing factor for the development of infection, then reduced permeability could be measured as an effect of successful mycobacterial treatment. Neither of these benefits of using the lactulose–mannitol test were apparent. Mycobacterial treatment was not associated with decreased permeability. In addition, the six patients with low permeability that were followed for 1 year had clinical activity and permeability test results similar to those of the study patients.

While the evidence for an infectious aetiology of Crohn’s disease is substantial, proof of the causative role of a particular pathogen will require improved diagnostic tools to enable specific chemotherapy to be directed towards a subset of patients with proven infection. We were unable to show any benefit of adding general anti-mycobacterial treatment to standard Crohn’s disease therapy.

Ancillary