RISK OF LEUKEMIA AFTER TREATMENT WITH PITUITARY GROWTH-HORMONE

Objective.-To determine whether pituitary-derived human growth hormone treatment increases the subsequent risk of developing leukemia and lymphoma. Design.-Cohort study. Setting.-United States. Participants.-A total of 6284 recipients of pituitary-derived human growth hormone distributed by the National Hormone and Pituitary Program between 1963 and 1985. Main Outcome Measures.-Leukemia and lymphoma. Results.-Three cases of leukemia occurred in 59736 patient-years of follow-up from the start of growth hormone therapy to case ascertainment at interview; this number was not significantly higher (P=.23) than the 1.66 cases expected in the US age-, race-, and gender-matched general population. Three additional cases, found in an extended follow-up that provided 83 917 person-years of risk, yielded a minimum rate of leukemia that was significantly increased (six cases found, 2.26 expected; P=.028). The relative risk of leukemia in pituitary growth hormone recipients compared with the general population was 1.8 (90% confidence interval [CI], 0.82 to 7.5) for the defined follow-up and 2.6 (90% CI, 1.2 to 5.2) for the extended follow-up. Five of the six subjects who developed leukemia had antecedent cranial tumors (four craniopharyngioma, one astrocytoma) as the cause of growth hormone deficiency, and four had received radiotherapy. There was no increase in leukemia in patients with idiopathic growth hormone deficiency. The association of leukemia and craniopharyngioma was significant (P<.001). There was no excess of lymphoma in the cohort. Conclusions.-This cohort of growth hormone recipients had a significantly increased rate of leukemia compared with the age-, race-, and gender-matched general population. However, the upper bound CI of the relative risk in our population (5.2) is well below the other estimates (7.6). Compared with the general population, our study population had more possible risk factors for leukemia (radiation, tumor) that may have contributed to the excess observed. The clustering of cases of leukemia in craniopharyngioma patients should be further evaluated.