Methods

To examine the longer term effect of antituberculous treatment given during the first two years, patients were followed for a total of five years from their date of entry to the study. They were seen routinely in the outpatient department at times determined by clinical need and also at yearly intervals. Patients continued on standard therapy throughout this period, the nature of which depended on the disease site, type, and activity. Treatment included various combinations of 5-acetylsalicylic acid preparations, steroids, immunosuppressants, and antibiotics. These were used in a routine way to treat both active episodes and maintain remission. At each annual review, events of the preceding 12 months were documented in a manner similar to that used in the initial 24 month study period. Specifically, the number of acute relapses of Crohn’s disease, defined as commencement or increase in the dose of prednisolone to control symptoms, was recorded; a note was also made of whether the patient had undergone surgery for the disease, and if so the type was documented. The number of hospital admissions for Crohn’s disease over the preceding 12 months, the disease activity index, full blood count, albumin, and current medication were also recorded.

The principal outcome measures of the follow up assessments were the number of clinical relapses, the number of patients who had surgery for Crohn’s disease over the preceding year, and the total disease activity score at each visit.

The analyses performed were simple comparisons of the active and placebo groups at 36, 48, and 60 months, and did not take into account any data collected at baseline or during the course of the first two years of the study. The most sensitive test of long term efficacy would be the cumulative number of relapses; however, due to missing data in the long term follow up this was not possible. Unpaired ttests, Mann-Whitney, or χ² tests were used to compare data between groups where appropriate.

Results

One hundred and thirty patients entered the initial study. Of these, 111 (81%) were followed regularly for five years from the time of entry to the study; follow up data for the remaining 19 patients were inadequate or incomplete and therefore were not included. Of the 111 patients, the number with records at 36, 48, and 60 months was 109, 100, and 88 respectively; other records were unavailable or insufficient. There were no deaths in the three years following completion of treatment; three patients died during the first 24 months—two in the active and one in the placebo group as reported previously.

Discussion

We report the five year follow up of a two year study of triple antituberculous chemotherapy. Although the number of relapses in those given active treatment was less than in the placebo group during the follow up period, the results only just achieved statistical significance at 60 months. There were other occasional significant differences between the groups for some of the other values, but none were consistent or sustained, and overall there was no other evidence of tangible benefit or disadvantage from antituberculous chemotherapy in any of the assessments made including the number of surgical episodes, hospital admissions, disease activity, and medication required for Crohn’s disease.

There are a number of obvious limitations with this study. We acknowledge that the most sensitive test of long term efficacy would be the cumulative relapse rate; however, due to missing data this cannot be calculated in any meaningful way—a problem observed with other long term follow up studies. We outlined in our original report the difficulties associated with clinical trials of this kind in patients with Crohn’s disease.6 It is quite possible that the response to therapy depends not only on the site of the disease, but also the type—for instance, whether the patient has predominantly stricturing, fistulating, or inflammatory disease. In future studies, disease should be stratified not only according to site but also type; this was not done in our study. With the benefit of hindsight the choice of antituberculous chemotherapy may be questioned. It is now eight years since the commencement of this study. During this time it has been recognised that some species of mycobacteria are resistant to conventional antituberculous treatment regimens; these include M paratuberculosis, the cause of Johne’s disease in cattle and a possible cause of Crohn’s disease. We chose rifampicin, ethambutol, and isoniazid after preliminary studies indicated that M paratuberculosis was sensitive to this combination of antituberculous drugs. The in vitro tests were carried out in the Mycobacterium Reference Unit of the Public Health Laboratory Service at the Cardiff Public Health Laboratory using a conventional model resistance method. The correlation between in vitro sensitivity of tubercle bacilli to antituberculous drugs and the in vivo response to treatment with these drugs has been well established by many controlled clinical trials. This is not the case with disease due to atypical mycobacteria such as M paratuberculosis. In addition, organisms of the Avium intracellulare complex which is closely related to M paratuberculosis are highly resistant in vitro to rifampicin and ethambutol but paradoxically a significant number of patients respond clinically. Even in situations where a mycobacterium has been identified as the cause of a disease, in vitro sensitivity tests are not necessarily a reliable guide to clinical response. The choice of antibiotics for atypical mycobacterial infections remains difficult and controversial.

Although conventional treatment was used in addition to trial medication throughout the five year period, this should not have influenced the results. Both groups of patients continued on standard medication which was used in a routine manner to treat active symptoms, maintain clinical remission, and was tailored to disease activity and site. There was no difference between groups in the type of medication taken at different time points or in the steroid dose used.

The primary objective of this trial was to examine whether there was net benefit for patients given triple antituberculous chemotherapy for two years. The absence of any benefit at the end of the two year trial period, and furthermore, after three years of subsequent follow up, fails to support the hypothesis that mycobacteria play an important part in the pathogenesis of this disease. It is not possible to say from this study whether there may be benefit in certain subgroups, for example those with stricturing disease, but with the available patient data it would be wrong and unhelpful to perform subgroup analyses in retrospect.

There have been anecdotal reports in the literature of clinical improvement in patients with Crohn’s disease given antituberculous treatment,^4 5 8 but randomised and open studies have yielded less convincing results. A total of 368 patients have taken part in nine published trials of antituberculous treatment in Crohn’s disease.6 ,9 ,10 Seven were randomised and placebo controlled,6 ,9 ,11 and two were open.10 ,12During these studies, treatment was given for three to 24 months, and various antituberculous regimens were used, from single to quadruple regimens. Five of the studies, including our own, showed no significant clinical benefit from active treatment.6 ,9-11 13-15Two studies showed possible benefit; of these one was a very small study in which clarithromycin alone was used12 while in the other, quadruple therapy was given. Although clinical relapses were fewer on active drugs, there were no differences in endoscopic or surgical recurrence.16 It is possible that the immediate improvement reported in some studies may be due to an effect of antibiotics on intestinal flora, rather than a direct effect on mycobacteria; in this case, one would not expect lasting benefit. Likewise, failure to improve with treatment initially does not exclude the possible role of mycobacteria initiating the disease. However, we have previously suggested that extension or recurrence of Crohn’s disease after a period of treatment may be a better indicator of a therapeutic effect on the underlying process—hence the need for an extended five year assessment. The complexity of Crohn’s disease as well as differences in the regimens used, make evaluation and comparison of trials difficult. However, a large number of patients have now been treated with antituberculous drugs for a significant period of time, and at present there are few data to suggest significant clinical benefit or support for the long term use of potentially toxic drugs.