CHICAGO – A genomewide association study has identified specific alleles that may pinpoint those at risk for thiopurine-induced pancreatitis in inflammatory bowel disease.
Patients heterozygous at rs2647087 are 2.5 times more likely to get pancreatitis if treated with thiopurine (Imuran), while those with homozygous risk alleles were 5 times more likely to do so, Dr. Graham Heap reported at the annual Digestive Disease Week.
Pancreatitis is a well-recognized, idiosyncratic complication of thiopurine that occurs in 4%-7% of patients, independent of dose, and usually stops treatment in its tracks. Thiopurines are a mainstay treatment in inflammatory bowel disease (IBD), particularly as an adjunct to anti–tumor necrosis factor–alpha drugs.
Dr. Heap and his associates at 168 research sites recruited 314 IBD patients with pancreatitis that developed within 3 months of starting a thiopurine for the genomewide linkage study and another 119 IBD patients for a case-control cohort. Thiopurine-induced pancreatitis was adjudicated by four independent experts, and determined to be definite in 61 patients, probable in 274, possible in 28, and unlikely in 70.
Among the definite and probable cases, pancreatitis developed within an average of 24 days after starting a thiopurine. Most cases were mild, but 70% of patients were hospitalized for an average of 5.7 days, noted Dr. Heap of Royal Devon and Exeter Hospital, Devon, England.
Analyses performed on three different genotyping platforms for 177 samples identified a significant single-nucleotide polymorphism (SNP) association in chromosome 6 within the class II HLA region. All four SNPs – rs7745656, rs2647087, rs6935723, and rs2647089 – had odds ratios of 2.6, with P values of 2.2 x 10–16.
The investigators then looked in more detail at the HLA region, previously linked to celiac disease, and identified a strong association between thiopurine-induced pancreatitis and HLA-DRB1:07:01 (OR, 2.55; P = 1.17 x 10–14) and HLA-DQA1:02:01 (OR, 2.54; P = 1.68 x 10–14), he said.
The results were replicated in an independent cohort of 78 thiopurine-induced pancreatitis cases and 472 IBD controls who did not develop pancreatitis after at least a year of thiopurine therapy, and again there was a significant association with rs2647078 (OR, 2.21; P = 4 x 10–6), Dr. Heap said.
Interestingly, there was no significant association between rs2647078 and known risk factors such as smoking status at the time of pancreatitis, thiopurine methyltransferase (TPMT) status, or inosine triphosphate pyrophosphatase genotype, he said.
What all this means for clinicians is that for every 1,000 patients tested, 77 risk allele homozygotes will be identified, and these individuals will have a 17% risk of pancreatitis, Dr. Heap said. If azathioprine and 6-mercaptopurine are avoided in these individuals, the number needed to treat is 76 patients to prevent one case of pancreatitis.
Session cochair and gastroenterologist Barrett Levesque of the University of California, San Diego,said in an interview that further testing is needed, but that the study approach was very robust, especially to show a number needed to treat of 76.
"Compared to what we use, TPMT activity levels, where there’s a few patients in a thousand who might not have the activity at all, this is very promising," he said. "I thought it was a great study."
In a separate interview, Dr. Heap said, "We do not currently feel this test has clinical utility on its own, but we will investigate its utility in combination with other pharmacogenetic tests in the future."
The study was funded by the International Serious Adverse Events Consortium. The authors reported no financial disclosures.