Analysis of the 2007 Australian Selby Study

In Selby’s protocol, 213 patients were randomized to AMAT (n=102) or to placebo (n=111), and over the first 16 weeks of the study, all patients underwent a 16 week tapering course of prednisone therapy. Patients unable to achieve remission at week 16 (CDAI < 150), and those unable to tolerate full doses of study medication were considered treatment failures and were withdrawn. Although not the primary efficacy endpoint of the study, the 16 week data are worth examining, because all randomized patients were assessed through this timeframe. The findings were that 67/102 AMAT (65.7%) patients met the CDAI cutoff and had tolerated treatment, compared to 55/111 (49.5%) placebo subjects. This difference cannot claim statistical significance, as it was not prespecified as an endpoint, however it yielded a nominal p-value of 0.02. This promising data is suggestive of efficacy, but only in an exploratory way, and additional trials, well designed, would be needed to replicate the finding and assure efficacy in a prospective way.

Beyond week 16, the central flaw in the Selby protocol is that only those subjects attaining remission at week 16 were followed. Therefore, only the 67 AMAT and 55 placebo patients continued to be followed from that a point of the study forward. These are no longer randomized groups of patients, and any conclusions drawn between arms for these patients would therefore be flawed. Theoretically, for instance, the 55 placebo patients who were in remission could be very healthy placebo responders, while the 67 AMAT patients may be a different population. Conversely, any variety of hypotheses could support a variety of biases that would remain in the subset of overall randomized patients. The primary efficacy endpoint in Selby was the proportion of patients experiencing at least one relapse from week 16 to week 52, but again, only a nonrandomized subset of the overall randomized group was followed for this, with 39% of subjects on AMAT experiencing a relapse versus 56% of those on placebo (p-value 0.054).

One might conclude that this “just missed” statistical significance and is persuasive, however conclusions based on these non-randomized groups of patients are not valid. The lack of ongoing treatment and follow-up of the ITT randomized population from beginning to end of study is a central flaw in the Selby study report. Meaningful conclusions from week 16 forward, therefore, cannot be made. One can only note that in both arms, CDAI remission rates fell over time for the remaining subjects, and admittedly very few patients experienced endoscopic remissions over time, a disappointing result for all subjects in the trial. Unruptured clofazamine capsules, and subtherapeutic dosing of clarithromycin and rifabutin have been raised as possible contributors, of note.

Overall, the Selby study neither refutes nor supports a benefit of AMAT in Crohn’s patients. The initial 16 week data, however, do support the possibility of efficacy, and support the need for additional studies. Future trials of AMAT should be well designed, with complete follow-up of all randomized patients through to the end of the study. Perhaps most important, the link between AMAT and clinical response should be supported by clear microbiologic evidence of efficacy on the organism load. Without the clear pathophysiologic link between how AMAT might be rendering its benefits, the AMAT will face questions regarding exactly how it is working, and whether long term triple therapy is necessary. The circumstantial evidence supporting AMAT in Crohn’s is persuasive, yet much more work needs to be done to validate this link and the use of such aggressive antimicrobial therapy.