Geospatial analyses identify regional hot spots of diffuse gastric cancer in rural Central America.

BACKGROUND

Geospatial technology has facilitated the discovery of disease distributions and etiology and helped target prevention programs. Globally, gastric cancer is the leading infection-associated cancer, and third leading cause of cancer mortality worldwide, with marked geographic variation. Central and South America have a significant burden, particularly in the mountainous regions. In the context of an ongoing population-based case-control study in Central America, our aim was to examine the spatial epidemiology of gastric cancer subtypes and H. pylori virulence factors.

METHODS

Patients diagnosed with gastric cancer from 2002 to 2013 in western Honduras were identified in the prospective gastric cancer registry at the principal district hospital. Diagnosis was based on endoscopy and confirmatory histopathology. Geospatial methods were applied using the ArcGIS v10.3.1 and SaTScan v9.4.2 platforms to examine regional distributions of the gastric cancer histologic subtypes (Lauren classification), and the H. pylori CagA virulence factor. Getis-Ord-Gi hot spot and Discrete Poisson SaTScan statistics, respectively, were used to explore spatial clustering at the village level (30-50 rural households), with standardization by each village's population. H. pylori and CagA serologic status was determined using the novel H. pylori multiplex assay (DKFZ, Germany).

RESULTS

Three hundred seventy-eight incident cases met the inclusion criteria (mean age 63.7, male 66.3%). Areas of higher gastric cancer incidence were identified. Significant spatial clustering of diffuse histology adenocarcinoma was revealed both by the Getis-Ord-GI* hot spot analysis (P-value < 0.0015; range 0.00003-0.0014; 99%CI), and by the SaTScan statistic (P-value < 0.006; range 0.0026-0.0054). The intestinal subtype was randomly distributed. H. pylori CagA had significant spatial clustering only in association with the diffuse histology cancer hot spot (Getis-Ord-Gi* P value ≤0.001; range 0.0001-0.0010; SaTScan statistic P value 0.0085). In the diffuse gastric cancer hot spot, the lowest age quartile range was 21-46 years, significantly lower than the intestinal cancers (P = 0.024).

CONCLUSIONS

Geospatial methods have identified a significant cluster of incident diffuse type adenocarcinoma cases in rural Central America, suggest of a germline genetic association. Further genomic and geospatial analyses to identify potential spatial patterns of genetic, bacterial, and environmental risk factors may be informative.