MK At least 10 published case reports or case series have described patients developing inflammatory bowel disease (IBD) following treatment with isotretinoin. More recently, Reddy and colleagues published a review of 85 voluntary reports submitted by manufacturers, health-care professionals, and consumers to the US Food and Drug Administration (FDA)'s Medwatch program and incorporated into the FDA's Adverse Events Reporting System. In this review, the authors applied the Naranjo ADR Probability Scale, a tool used to assess the potential connection between a medication and a particular adverse event. The review found that of the 85 cases included, 4 (5%) scored in the highly probable range for isotretinoin being the cause of IBD. As a caveat to this study, it is important to note that the Naranjo scale was not developed specifically for chronic illnesses such as IBD.

In addition, this concern about isotretinoin has been heightened by the inclusion of a warning on the package insert, which was added by the manufacturer following case reports and related lawsuits filed against the manufacturer.

MK Yes, it is always important to pay attention to case reports such as these. At the same time, case reports are among the weakest of epidemiologic study designs and should not be used to make assumptions about causality. Case reports are limited by reporting and publication bias, recall bias, incomplete and subjective presentation of case details, and lack of long-term follow-up. By definition, case reports include only individuals who have had both exposure (to isotretinoin, in this instance) and the outcome (IBD). To properly assess causality, additional information would be needed, such as the incidence of IBD among individuals not treated with isotretinoin and the number of people taking isotretinoin who have not experienced IBD.

Case reports are critically important in identifying what might be a possible signal for a causal association, and generally initiate hypothesis generation. When a patient is exposed to a drug and develops an illness, it is logical to investigate whether there is a connection between the two. Appropriately, the FDA has systems in place to track potential adverse events, where an individual develops a certain condition after receiving a certain medication.

However, in interpreting case reports, it is essential to keep in mind that some individuals included in the case reports may have developed IBD independent of their isotretinoin use, as IBD is not that uncommon of a disease. Furthermore, focusing solely on the case reports ignores the fact that millions of individuals have been safely treated with isotretinoin with no adverse sequelae.

MK Yes, and this is how epidemiology happens. The observations were made, and it is critically important to publish case reports and submit them to the FDA when these events occur. However, the report itself is not proof of a causative association—it is merely the suggestion of a possible signal.

MK From an epidemiologic standpoint, this question is crucial. Biologic plausibility is one of the criteria developed by Sir Austin Bradford Hill in 1965 to evaluate whether the relationship between two factors might be causal. However, the etiology of IBD, thought to involve an interplay between host genetics, immune response, and environmental triggers, has not been clearly elucidated. Therefore, it is impossible to answer the question of how this drug could lead to IBD.

Shale and colleagues have hypothesized that isotretinoin exposure might lead to IBD through the pleiotropic effects of retinoic acid on intestinal epithelial cells and immune functioning. These effects include impaired neutrophil function, favoring the balance of regulatory T cells rather than effector T cells, increasing intestinal homing markers, and inhibition of autophagy mediated through retinoic acid-inducible gene 1 (RIG-1).

However, our group and others have pointed out that some of these immunologic effects, particularly the balance between regulator and effector T cells, might actually prevent, not trigger, the development of IBD. Similarly, defects in intestinal barrier function have recently been postulated as a key factor in the development of IBD, and retinoic acid has been shown to enhance, not disrupt, barrier function. Considering these competing hypotheses, clearly further mechanistic research is needed.

MK We performed a systematic Medline search using these keywords: Accutane, isotretinoin, inflammatory bowel disease, ulcerative colitis, Crohn's disease, and colitis. We also hand-searched reference lists of identified articles for additional relevant publications and searched BIOSIS Previews for relevant scientific meeting abstracts using these same keywords.

Twelve published case reports and 1 case series (15 total cases) along with the 85 cases included in the study by Reddy and colleagues were summarized. Cases of IBD developed at a range of isotretinoin doses. In several instances, a definitive diagnosis of IBD was not reported.

Then, we assessed whether each of Hill's 9 criteria of causation—strength, consistency, specificity, temporality, biologic gradient, plausibility, coherence, experimental evidence, and analogy—had been fulfilled by the cases. At the time of this review, only the criterion of temporality had been met by the reports. Biologic plausibility was debatable, as noted above, and none of the other criteria had been met.

MK Yes. Bernstein and colleagues performed a case-control study by linking Manitoba Health administrative databases containing diagnosis codes for physician outpatient services and hospitalization abstracts with the Manitoba prescription drug registry. Cases of Crohn's disease and ulcerative colitis were identified using a previously validated case-finding algorithm and matched with non-IBD controls on the basis of age, gender, and postal area. Exposure to isotretinoin was compared between cases and controls. The study found weakly positive associations between isotretinoin and Crohn's disease (odds ratio [OR]: 1.15), ulcerative colitis (OR: 1.16), and IBD overall (OR: 1.16) that were not statistically significant.

The strengths of this study include the population-based design and the use of previously validated case definitions to measure the outcomes of interest. The limitations of the study were as follows: the sample size of approximately 2,000 IBD patients might not have been large enough to provide adequate power to detect small differences; dose and duration of isotretinoin use were not directly analyzed, and this would have been an indicator of biologic gradient, as proposed by the Hill criteria; and in many instances, isotretinoin exposure was assessed many years before the onset of IBD. This limitation might have underestimated the true effects of isotretinoin exposure if the expected onset of diseases is more proximal to the drug exposure.

A second case-control study was presented by our group at the American College of Gastroenterology and Advances in IBD meetings in late 2009. We performed a case-control study using a large, US-based health insurance claims database. The study found a positive association between isotretinoin and ulcerative colitis (OR: 4.36) but not Crohn's disease. The association was even stronger for those exposed to higher doses of isotretinoin over longer periods of time, suggesting a biologic gradient. In addition, we found that a diagnosis of acne was not associated with either Crohn's disease or ulcerative colitis.

MK The conflicting results of the previously described studies indicate that causality cannot be firmly established at the present time. Further work is needed, such as a prospective cohort study with careful measurement of exposure (including dose, duration, and adherence), outcomes (using clinical, endoscopic, and histopathologic criteria), and potential confounders.

MK It is important to note that regardless of the relative risks, as estimated by ORs in the above-mentioned case-control studies, the absolute risk of developing IBD after treatment with isotretinoin is likely to be very small. Given the unquestioned effectiveness of isotretinoin in the treatment of severe acne, the risk:benefit ratio does not appear to be unfavorable. Therefore, as of this article, I do not recommend withholding isotretinoin for those individuals with severe, refractory acne.

Whether there is effect modification between isotretinoin and any high-risk groups, such as those with a family history of IBD, has yet to be studied. However, it might be important to consider the possibility of a higher risk of IBD following isotretinoin among individuals with a first-degree family member with IBD.

Of course, there are segments of the population for whom this drug is absolutely contraindicated, such as expecting mothers and sexually active women of child-bearing age not using at least two forms of birth control. However, these risks are not related to gastrointestinal conditions and, thus, are beyond the scope of this interview.