Cigarette Smoking as a Risk Factor for Pancreatic Cancer in Patients With Hereditary Pancreatitis
- Markus M. Lerch, MD
To the Editor: Smoking approximately doubles the risk of pancreatic cancer,1 the fourth most common cause of death from cancer in the United States. In addition, about 5% to 10% of patients have an inherited genetic predisposition to this tumor.2 A large variety of germ-line defects have been detected in families with pancreatic cancer, including the familial pancreatic cancer syndrome, mutations in the BRCA2 gene, the familial atypical multiple mole melanoma syndrome, hereditary nonpolyposis colorectal carcinoma syndrome, the Peutz-Jeghers syndrome, and hereditary pancreatitis.
We previously reported an exceptionally high risk of pancreatic cancer in patients with hereditary pancreatitis, a rare autosomal dominant disorder.3 We now report information on smoking as an additional risk factor in these indviduals.
Methods
By contacting gastroenterologists, pancreatologists, and surgeons, we have established a registry that now includes 497 patients with hereditary pancreatitis. Each respondent has submitted data about demographics, family history, diagnosis, disease course, life-style factors, including smoking and drinking habits, and vital status. The expected number of cancers was determined from available sex, age, and country-specific incidence data.
Results
Of the 497 patient with herditary pancreatitis, 19 biopsy-proven pancreatic cancers were reported, compared with an expected number of 0.33, yielding a risk ratio of 57 (95% confidence interval [CI], 35-90). In this cohort, smoking status was available for 88% of the patients: 40% were either current or former smokers. Of the 19 patients with pancreatic cancer, 11 were current or former smokers, 6 were nonsmokers, and smoking information was missing for 2 patients. As in the general US population, smoking resulted in an approximately 2-fold (age- and sex-adjusted odds ratio [OR], 2.1; 95% CI, 0.7-6.1) increased risk of pancreatic cancer compared with nonsmoking members of the cohort. An unexpected finding was that pancreatic cancer developed 20 years earlier in smokers than in nonsmokers (P = .02). (Figure 1)
Figure. Age Distribution of Pancreatic Cancer Cases According to Smoking Status
Median age (horizontal lines) at onset of cancer: ever smokers, 50 years and never smokers, 70 years (P = .02, Mann-Whitney test).
Forty-three percent of adults in the cohort had consumed alcohol, and drinking status was known for 17 of the 19 patients with pancreatic cancer. The mean (SD) age of onset of pancreatic cancer in 11 alcohol consumers was 55 (14.7) years, which was nearly identical to the mean age of onset of pancreatic cancer in 6 nonconsumers (56 [17.8] years). In multivariate analysis, a nonsignificant increased risk of pancreatic cancer was observed in consumers vs nonconsumers of alcohol (OR, 2.1; 95% CI, 0.7-6.3).
Comment
Hereditary pancreatitis usually begins in childhood and is now known to be caused by defects in the trypsinogen gene located at 7q35.4 Although the trypsinogen gene is not considered to be a cancer gene, the high risk of pancreatic cancer in these patients may be related to progressive glandular destruction over a long time period.
Our data suggest that in this sample of patients with hereditary pancreatitis, as in the general population, smoking doubles the risk of pancreatic cancer and accounts for approximately 25% to 30% of all pancreatic tumors. In addition, smoking appears to result in a dramatically earlier age of diagnosis of pancreatic cancer, perhaps because of a gene-environment interaction. We believe it is unlikely that either selective overreporting of smoking in young patients or underreporting of smoking in older patients with pancreatic cancer occurred. Evidence has suggested that there is a smoking-genetic interaction in patients with familial pancreatic cancer,5 but additional studies will be required to learn whether smoking has a similar effect in other patients who have an inherited susceptibility to pancreatic cancer. Meanwhile, our data indicate that in this germ-line disorder with a high risk of pancreatic cancer, smoking acts as an independent risk factor and seems to lower the age of onset of this aggressive cancer by approximately 2 decades. In addition, since alcohol can be injurious to the pancreas, we suggest that persons with a predisposition to pancreatic cancer should avoid drinking alcohol.
Acknowledgments
Funding/Support: This study was supported by grants from the C.D. Smithers Foundation; Solvay Pharmaceuticals Inc; the Italian Association for Cancer Research; the National Institute of Health (grant DK54709); the German Research Council, and the Interdisciplinary Center for Clinical Research (IZKF), Münster.
