) The role of immunosuppression in the development and progression of cancer has long been known. Malignant conditions are seen more frequently in the setting of primary and iatrogenic immunodeficiency, and in autoimmune disease. In 1981, the AIDS epidemic marked the date after which certain neoplasms such as Kaposi's sarcoma (KS) and non-Hodgkin's lymphoma (NHL) were noted to occur at higher rates among seropositive men and women. Studies reporting the excess risk for KS and NHL due to human immunodeficiency virus (HIV) infection have produced widely variable results, have typically not included women, and have failed to exclude HIV seropositive cases from incidence calculation in the comparison populations. Although invasive cervical carcinoma (ICC) is an acquired immunodeficiency syndrome-(AIDS) defining condition, little is known about its incidence among HIV-infected females because few studies have had sufficient sample size to assess it or cervical intraepithelial neoplasia (CIN). Similarly, little is known about how the incidence of KS, NHL, and other malignancies might vary between genders. The study aims to assess the incidence of malignancy in the context of immunosuppression due to HIV in a large managed care-based cohort of 5,574 HIV seropositive individuals, 582 (10.4 percent) of whom are female. The primary aims of this study are to: (1) Determine more precise estimates for the excess risk due to HIV infection for KS and NHL. (2) Assess the risk for: (a) ICC/CIN among HIV-infected women; and (b) Non-AIDS-related malignancies in HIV-infected males and females. We will use cancer cases diagnosed in the HIV cohort between September 1, 1991 and December 31, 1997, and expected cases will be calculated using two methods - an internal, population-based standard and published SEER rates. Observed case counts in the HIV cohort will be compared to expected case counts in each of the comparison groups to compute standardized incidence rates (SIR's) for each type and/or group of neoplasm, and separately for males and females when possible. We will then calculate exact confidence limits to assess the statistical significance of each SIR.
