No Excess Risk of Breast Cancer among Female Users of Systemic Glucocorticoids

Recently published evidence suggests that hyperinsulinemia, insulin-like growth factors, and abnormal glucose intolerance may promote mammary carcinogenesis (1-6) and that women with type 2 diabetes are at increased risk for postmenopausal breast cancer (1, 3). Insulin and insulin-like growth factors are major determinants of proliferation and apoptosis (5, 6). Steroid hormones are also important in the etiology of breast cancer: both estrogens and androgens are dominant hormones in breast cancer development (7). Glucocorticoids are immunosuppressive steroid hormones widely used in the treatment of allergic and autoimmune diseases. Metabolic changes, including decreased glucose tolerance, are well-known side effects of glucocorticoids (8). Because these drugs are commonly used and breast cancer is fairly common, any relation between the two would have major public health implications. Therefore, we examined the risk of breast cancer in a large cohort of female glucocorticoids users.

We used the population-based Prescription Database of North Jutland County, Denmark (9) (500,000 inhabitants) to identify the study cohort (Table 1). This database is generated by a computerized pharmacy accounting system that sends data to the Danish National Health Service. In addition to providing free access to medical care by general practitioners and at hospitals, the Health Service refunds part of the costs associated with outpatient prescribed drugs so prescription data for inhabitants is nearly complete.

The Prescription Database contains information on all prescriptions eligible for reimbursement filled in the county after January 1, 1989, including the patient's personal identification number, the drug prescribed, and the prescription date. During the prescription exposure period from January 1, 1989, to December 31, 1996, patients prescribed the following drugs (used either p.o. or by injection) were included in the study: betamethasone, dexamethasone, fludrocortisone, hydrocortisone, methylprednisolone, prednisolone, prednisone, and triamcinolone. We also obtained data on all prescriptions for other cytostatic and immunosuppressive drugs (azathioprine, methotrexate, cyclosporin, mycoplexnolate, and tacrolimus) to take into account use of these agents (e.g., for cancer treatment or organ transplant).

During the prescription exposure period, 32,673 females >15 years old received a glucocorticoid prescription. Of these, 337 were excluded because they received a prescription for cytostatic or immunosuppressive drugs before their first prescription for a glucocorticoid. We linked the remaining 32,336 patients to the Danish Cancer Registry, which collects information on all individuals diagnosed with cancer, including breast cancer. Use of the personal identification number ensured a complete prescription history and unambiguous record linkage.

We followed patients for breast cancer from the first glucocorticoid prescription until death, receipt of a prescription for a cytostatic or other immunosuppressive drug, or until December 31, 1998, whichever came first. We censored 787 persons because they received a prescription for cytostatic or other immunosuppressive drugs during follow-up. Person-years were stratified by the number of reimbursed prescriptions to create categories of cumulative exposure (1, 2-4, 5-9, 10-14, and 15 or more prescriptions). The expected number of breast cancer cases was calculated as the product of the county-specific cancer incidence rates among women obtained from the Cancer Registry in strata of age (fifteen 5-year groups) and calendar year (1988-1992, 1993-1997, and 1998) and the person-years accumulated in the corresponding stratum of steroid users. The 95% confidence intervals for the standardized incidence ratios (SIR; i.e., the ratio of observed to expected cancers) were computed by Byar's approximation. Exact limits were used when the observed number was <10. We stratified the analysis by age at the first prescription (<45, 45 + years of age) to explore whether the risk differed between premenopausal and postmenopausal women.

Of the 122,779 total corticosteroid prescriptions, 75,658 (62%) were for tablets, 47,080 (38%) were for injections, and 41 were for unknown formulations; 40% were for prednisone, 19% were for prednisolone, 14% were for triamcinolone, 17% were for betamethasone, 8% were for methylprednisone, and 2% were for other drugs. Over the study period, 367 cases of female breast cancer were observed in the cohort. The overall SIR was very near one (1.03; 95% confidence interval, 0.93-1.14). We found the lowest SIR among women with the highest number of prescriptions (SIR = 0.86; 95% confidence interval, 0.47-1.44). We did not find any substantial modification of the SIR in the analysis stratified by age at the first prescription. Our study had 80% power to detect a relative risk of 1.14 associating breast cancer with glucocorticoid prescription.

Prescription for systemic glucocorticoids was not associated with an increased incidence of breast cancer. Our study had the advantage of being a large population-based follow-up study with very little loss to follow-up. Our data lacked clinical details, in particular of the underlying diseases for glucocorticoid treatment and confounding factors aside from age. However, glucocorticoid treatment would not ordinarily be prescribed to treat breast cancer or its side effects, and confounding by indication would not ordinarily bias relative risks toward the null. Confounding factors would have to be related to glucocorticoid prescription and inversely related to breast cancer risk, conditional on adjustment for age, an unlikely combination.