Fig 1.—-Tomogram of case 1, with arrow indicating erosion of sella turcica and mass extending into sphenoid sinus. Fable 1.—Serial Thyroid-Stimulating Hormone’ (TSH) Levels in Two Patients With Pituitary Tumors Year Case 1° 1965 2.2 1967 1.0 1969 1.7 1972 110 1973 Case 2 Lae <2.5 Fig 2.—With immunoftuorescent staining, dark cytoplasmic reaction product, indicated by arrows, can be seen localizing prolactin in cells of pituitary adenoma of.case 2. Prolactin-secreting cells have round to oval nuclei and distinct nucleoli (X800). Prolactin assays performed on stored frozen sera obtained from her annual examinations have revealed prolactin elevations ranging from 209 to 609 ng/mL as far back as 1975, at which time she was 21 years of age. wan Results 1974 1.0 1975 1976 5.9 115 Lae 1978 1.8 1.7 1979 wae 0.3 1981 <2.5 <2.5¢ “Normal values are less than 5 pU/m_L. tCase 1 had a total thyroidectomy in 1969 for papillary carcinoma, and the elevated TSH levels in 1972 and 1976 were obtained when thyroxine therapy was discontinued prior to iodine 131 scanning. +Year pituitary tumor was diagnosed. turcica suggested the presence of a 1.2-cm pituitary adenoma with no suprasellar extension. Preoperative endocrine evaluation showed the following to be normal: one-hour adrenocorticotropic hormone (ACTH) stimulation test results, growth hormone and cortisol responses to an insulin tolerance test, free and total serum thyroxine levels, and thyroid-stimulating hormone values. Subsequently, transsphenoidal removal of an intrasellar adenoma was performed. A soft-tissue tumor that contained many small cystic foci was found. Immunofluorescent staining of the tumor was positive for prolactin (Fig 2). Two weeks postoperatively, thyroid functions and results of a one-hour ACTH stimulation test were normal, and serum prolactin values had decreased to 68 ng/ mL. She remains amenorrheic one year postoperatively. JAMA, Aug 3, 1984—Vol 252, No. 5 a Pal 201 K u 14 Analysis of Epidemiologic Data.— Incidencefigures for pituitary tumors in the United States are not available from the National Tumor Registry or the National Cancer Institute’s Biometric Division because they are not considered malignant. Several studies suggest an incidence ap- proaching one in 100,000 persons per year among those younger than 45 years, distributed equally between the sexes.” However, these retrospective studies occurred before modern methodsof pituitary tumor diagnosis were available. Data from the Olmsted County, Minnesota, epidemiologic study of pituitary tumors’ and the 1970 and 1980 census information for that county provide information that may be used (as a referent) in comparison with the Marshallese data. In Olmsted County, the incidence rate of pituitary tumors among women aged 15 to 44 years during the years 1971 through 1977 was 11 cases per 163,096 person years. For radiation-exposed Marshallese women in the same age 2 11 2,176 163,096 2 12 4,252 454,472 risk, 13.6; 95% confidence interval, 4, .01. risk, 17.8; 95% confidenceinterval, 6, .001. group, the rate is two cases per 2,176 person years. The incidence rate for the Marshallese womenis 13.6 times higher than that for the Olmsted County women; that is, the relative risk, as estimated by the incidence density ratio, is 13.6.’ If the propensity for pituitary tumors developing were the same for each population studied, then the relative risk should be near unity, allowing for random variation. An exact (binomial) test, which is appropriate where numerators are small, indicates that the relative risk is significantly greater than unity at the a=0.01 level. Table 2 provides relative risks, ' exact P values, and approximate confidence intervals for women aged between 15 and 44 years and for men and women combined who are younger than 45 years of age, using Olmsted County as the referent population. No allowance is made in the incidence data for any latent period in tumor induction because there are no published data on this point. The two pituitary tumors, however, were clearly present 21 and 22 years after radiation exposure. The statistical analyses also ignore the fact that the Marshallese people exposed to radioactive fallout have been followed up more closely than any large control population. Finally, the baseline frequency of pituitary tumors in the Marshall Islands is unknown. While no cases have been diagnosed in a comparison population of unexposed Marshallese, the number of personyears of observation is small (698 person-years for women aged 15 to 44 years; 1,527 person-yearsfor the total population younger than 45 years). This does not permit a meaningful’ statistical analysis of pituitary tumor incidence in the Marshal! Islands. Pituitary Tumors—Adams et al 665 7 "Relative 42; P value, +Relative 53; P value, Person Years RS Women, 15-44 yr* Marshail Islands Olmsted County Total population, 45 yrt Marshall Islands Olmsted County incidence . Population 2 aegS RPI "ARRMo RE TSA, BaD eT MY Pt Table 2.—Statistical Analysis of Relative Risk of Pituitary Tumor in Radiation-Exposed Marshallese