Characterizing Spatial and Environmental Determinants of Prostate Cancer Using a Geographic Information System

Abstract

ISEE-652 Objective: Prostate cancer (PrCA) incidence in South Carolina (SC) exceeds the national average, particularly among African Americans (AAs). A Geographic Information System (GIS) was used to evaluate spatial and environmental determinants of PrCA in SC, a cancer for which there are few established risk factors. Materials and Methods: Age-adjusted PrCA incidence, standardized incidence ratios (SIRs), and 95% confidence intervals were calculated for census tracts using SC Central Cancer Registry data (1996–2002). Demographic data (groundwater consumption, age, race, median income, land use) were obtained from census data. Metal concentrations in groundwater and soil were obtained from existing federal and state databases to test hypotheses that zinc and selenium are protective, and that cadmium and uranium increase PrCA incidence. Census tract-specific exposure estimates included: geometric mean groundwater and soil metal concentrations, proportion of samples exceeding background, and census tract-centroid concentration obtained via spatial modeling (kriging). A GIS was used to map incidence rates, SIRs, and exposures. Statistical procedures (linear and Poisson regression) were applied with and without censoring of tracts with <50% groundwater consumption per capita (N = 173 and 854, respectively). An adjacency matrix was generated and spatial statistical models were performed using WinBUGS software, with adjustment for correlated and uncorrelated heterogeneity. Results: There were 18,091 PrCA cases; 30% AA and 69% European American. Among uncensored tracts with elevated groundwater zinc, 7 had PrCA rates below expected incidence, and this inverse relationship also was observed using linear regression (P = 0.02). There were 96 census tracts with both elevated soil selenium concentrations and reduced SIRs. Three tracts had both elevated soil cadmium and higher SIRs. Conclusions: A decreasing trend in PrCA rates was observed among SC census tracts with elevated zinc in soil or groundwater. Relationships between environmental factors and cancer incidence can be efficiently explored using GIS, spatial statistics, and existing population databases.