Norepinephrine antagonists and cancer risk

Fitzgerald1, 2 has hypothesized that norepinephrine is an etiological factor in some types of cancer and that drugs that lower norepinephrine level or block its receptors may prevent various cancers. We have followed his suggestion to test possible preventive effects of these drugs “through epidemiological investigation of medical records of large numbers of people who have taken these drugs.”1 This was done as part of our screening of commonly used prescription drugs for positive or negative associations with cancer,3 using the pharmacy records and cancer registry of the Kaiser Permanente Medical Care Program in Northern California, during the period from August 1994 to February 2008. To assess commonly prescribed drugs that were hypothesized as preventive, we studied clonidine, which reduces norepinephrine secretion, prazosin and terazosin, which are alpha-1 adrenergic receptor blockers, and atenolol, metoprolol and propranolol, which are beta-adrenergic receptor blockers. In addition, we analyzed the antihypertensive drugs, hydrochlorothiazide, lisinopril and nifedipine, which operate by different mechanisms. Case–control analysis used conditional logistic regression for matched sets within the cohort of program subscribers with at least partial coverage of payment for drug prescriptions. The cases were patients first diagnosed with any cancer (except nonmelanoma skin cancer) during the study period, and 50 control subjects were matched to each case for age, sex and year of joining the cohort, with their index date set to provide follow-back time equal to that from the date of the case's cancer diagnosis. The controls could be matched to more than one case and could develop cancer later than their index date. The use of a drug was defined as receipt of three or more prescriptions before the date of cancer diagnosis or the control's index date, with the first prescription dispensed at least 2 years before these dates. We also looked at the six hypothesized drugs in relation to the most common of the hypothesized cancer sites, colon, lung, breast and prostate. For all sites combined, both the hypothesized and nonhypothesized drugs showed a slightly elevated relative risk, as represented by odds ratios ranging from 1.05 to 1.11, and all but nifedipine were nominally statistically significant, as indicated by a lower 95% confidence limit greater than 1.0. The number of cases exposed to the hypothesized drugs ranged from 2,079 for clonidine to 15,553 for atenolol. The median months of use by controls were substantial, more than 2 years for all of the hypothesized drugs, suggesting that there was adequate use for the hypothesized preventive effect to occur. The median duration of use for atenolol, by far the most commonly prescribed of the hypothesized drugs analyzed, was 33.5 months. Altogether, 10.0% of cases and 9.6% of controls had received three or more prescriptions of atenolol; the prevalence of similar exposure to the other hypothesized drugs was much smaller, approximating 2%. Therefore, any apparent attenuation in a negative association for a hypothesized drug due to the use of other hypothesized drugs by those not exposed to the drug of interest would be small. In the site-specific analyses (Table 1), three of the odds ratios for colon cancer were above and three were below 1.0, with nominal statistical significance below for one drug, terazosin. For lung cancer, there were two odds ratios above 1.0—clonidine significantly above—and four odds ratios below 1.0. For breast cancer, there were five odds ratios above 1.0—atenolol significantly above—and one odds ratio below 1.0. For prostate cancer, there were three odds ratios above 1.0 and three odds ratios below 1.0. The odds ratios for prazosin and terazosin were significantly above 1.0 and the odds ratio for metoprolol was significantly below 1.0. All differences from odds ratio 1.0 were small, none exceeding 20%, i.e., greater that 1.20 or less than its inverse, 0.83. As lung cancer is strongly associated with cigarette smoking, and if this habit is also associated with the use of any of these drugs, the odds ratios for this cancer site may be lower if smoking habit could be accounted for. Although our findings for lung cancer are less certain, we could not confirm a preventive effect for drugs that lower norepinephrine level or block its receptors. Contributing to the slightly elevated risks for all cancers combined could be the known relationship of hypertension or its treatment with renal cancer4 and its associations with obesity and alcohol use, which are associated with cancer of several sites.5, 6 The associations of hypertension with various cancers have been inconsistent, and this cardiovascular condition has not been established as an important causal factor for cancer.7 Thus, it is unlikely that hypertension is masking a substantial preventive effect of the hypothesized drugs. Yours sincerely, Gary D. Friedman, Natalia Udaltsova, Laurel A. Habel.