Patient characteristics and registry variation associated with receipt of KRAS testing in colorectal cancer

Charlton ME, Karlitz J, Lynch C, Chen V. Patient characteristics and registry variation associated with receipt of KRAS testing in colorectal cancer. Poster session presented at: North American Association of Central Cancer Registries Annual Conference; 2014 Jun 25; Ottawa, Canada.

Background: In 2008, the National Comprehensive Cancer Network recommended KRAS testing for all patients with Stage IV colorectal cancer (CRC) at time of diagnosis, and only those with wild-type KRAS be treated with epidermal growth factor receptor inhibitors due to toxicity and cost. KRAS was collected by SEER registries as a CS site-specific factor beginning with 2010 cases. Purpose: To examine registry and patient-level variation in KRAS testing. Methods: The study population included all malignant cases with positive histology for adenocarcinoma of the colon/rectum diagnosed in a SEER region in 2010 (N = 30,416). Chi-square tests and logistic regression analyses were conducted to determine patient characteristics associated with KRAS testing, stratified by stage I-III vs. Stage IV. Results: Substantial variation between registries was detected; 39% of New Mexico Stage IV cases had documented KRAS testing vs. 15% of Louisiana cases. In the logistic model, younger age, being married, living in large metropolitan areas, tumor in sigmoid or descending colon and sub-stage IVA or IVB (vs. IVNOS) were associated with receipt of KRAS testing. Among those with stage I-III CRC for whom KRAS testing is not recommended, 5% had testing; registry-specific rates ranged from 11% in Seattle to < 2% in Iowa and San Francisco. Conclusion: There appears to be wide variation in documented KRAS testing in Stage IV CRC patients by SEER registry; it is unknown if variation exists primarily due to inconsistencies in data capture or due to differences in provider practices. The fact that age remained highly significant after controlling for registry suggests providers are selecting patients for testing based on this characteristic or that younger patients tend to select providers more likely to order testing. Further research is needed to determine drivers of registry and age variations. Reasons for the relatively high rate of testing in stage I-III also warrants further investigation.