were randomly reviewed to verify IBD diagnoses. In the 1999 manuscript, this protocol yielded a sensitivity and specificity of approximately 90% for accurate IBD diag- nosis compared to the gold standard of chart review (Am J Epidemiol 1999;149:916 –924). These estimates were based on a patient sample that included all patients with 1 visit with an ICD9 code for IBD. Because the population for the current study includes a general, un- selected population in which the overall rate of IBD is expected to be low, the authors claim a specificity ap- proaching 100%. The second crucial factor is the degree to which isotretinoin prescription data accurately reflect actual isotretinoin use. Duration of isotretinoin therapy was based on the amount of time between first and last prescription of isotretinoin; these data are used as proxy for actual isotretinoin use. The authors do not report the regularity with which patients filled prescriptions, so it is possible that exposure to isotretinoin was overestimated. This should not affect results, because such overestima- tion should occur equally among cases and controls. Of concern is the date range used to evaluate isotretinoin prescriptions. The authors state that the drug network database has been in effect “since 1995,” but the specific date of database inception is not stated. Diagnosis dates were collected beginning April 1, 1995, so some patients could have remote isotretinoin use not captured in the study. Because this should not affect cases and controls differentially, it is unlikely to affect study results. Several biologic pathways have been proposed as po- tential mechanistic links between isotretinoin and IBD, including an exaggerated adaptive immune system re- sponse in the gut (owing to isotretinoin’ s impairment of the innate immune response to luminal bacteria), and alterations in T-cell gut-homing potential (Gut 2009;58: 737–741). These potential pathways remain theoretical and have not been directly studied in the case of IBD and isotretinoin. In the present study, a lag time of nearly 3 years between last isotretinoin prescription and diagnosis of IBD argues against a plausible biologic link between isotretinoin and IBD. Although often implicated, there is surprisingly little hard evidence supporting associations between use of specific medications and IBD incidence or disease flares. Oral contraceptive medications have been weakly associ- ated with increased risk of Crohn’s disease in women (Gut 2009;58:737–741). Nonsteroidal anti-inflammatory drugs are implicated in disease flares among patients with known IBD (Autoimmun Rev 2004;3:394 – 400). The use of antibiotics has been loosely associated with higher IBD risk at the population level (World J Gastroenterol 2008;14:165–173). The fact that decades of research has not found more convincing associations between medi- cation use and IBD underscores 2 important issues. First, IBD is a markedly heterogeneous disease, and it is likely that certain triggers are relevant only in certain patients. Second, designing studies to accurately identify associa- tions between medication use and uncommon outcomes, such as IBD, is extremely difficult. Despite minor concerns, the present study likely rep- resents the best study that can be feasibly performed to evaluate for associations between isotretinoin use and IBD. The authors’ conclusion that isotretinoin use is unlikely to be associated with true idiopathic IBD there- fore seems to be sound. Prior studies suggesting an as- sociation may suffer from confounding and/or simple coincidence. Although no study can completely exclude the possibility of isotretinoin acting as a trigger for IBD in rare patients, the magnitude of risk is insufficient to recommend withholding isotretinoin among patients in whom it is clearly indicated. Providers should discuss these issues as part of the informed consent process before initiating isotretinoin therapy. They should also be aware that several lawsuits by patients who used isotretinoin and developed colitis have been successful, despite the lack of evidence. JUSTIN L. SEWELL, MD, MPH UMA MAHADEVAN, MD UCSF Center for Colitis and Crohn’s Disease University of California, San Francisco San Francisco, California NOVEL INSIGHTS INTO THE ASSOCIATION BETWEEN HLA-DP VARIANTS AND PERSISTENT HEPATITIS B VIRUS INFECTION: A GENOME-WIDE ASSOCIATION STUDY Kamatani Y, Wattanapokayakit S, Ochi H, et al. (Labo- ratory of Molecular Medicine, Human Genome Center, Institute of Medical Science, the University of Tokyo, Tokyo, Japan). A genome-wide association study iden- tifies variants in the HLA-DP locus associated with chronic hepatitis B in Asians. Nature Genetics 2009;41: 591–595. With 400 million people worldwide chronically in- fected with hepatitis B virus (HBV), HBV infection is a globally challenging health problem. This issue is of particular significance in the Asia-Pacific region and Africa, where the incidence ranges from 5% to 15%. There are a few important observations epidemiologically, which strongly suggest underlying host genetic factors affecting the chronicity and clinical outcome of HBV infection. First, even in the most endemic area, the incidence of HBV infection seldom exceeds 20%. Apparently, a strong genetic pressure in nature has prevented it from exceed- ing this percentage. Second, although chronic HBV in- fection runs in families, usually it is the male subjects who suffer from the diseases; female siblings have a much lower chance of developing serious clinical sequela, such as cirrhosis or hepatocellular carcinoma. In addition, the 394 SELECTED SUMMARIES GASTROENTEROLOGY Vol. 138, No. 1