KRAS testing of patients with metastatic colorectal cancer in a community-based oncology setting: a retrospective database analysis

BACKGROUND: In 2009, treatment guidelines were updated to recommend KRAS testing at diagnosis for patients with metastatic colorectal cancer (mCRC). We investigated KRAS testing rates over time and compared characteristics of KRAS-tested and not-tested patients in a community-based oncology setting. METHODS: Adult patients with a diagnosis of mCRC from 2008–2011 were selected from the ACORN Data Warehouse (ACORN Research LLC, Memphis, TN). Text mining of physician progress notes and full chart reviews identified KRAS-tested patients, test dates, and test results (KRAS status). The overall proportion of eligible patients KRAS-tested in each calendar year was calculated. Among KRAS-tested patients, the proportion tested at diagnosis (within 60 days) was calculated by year. Univariate and multivariate analyses were used to compare patient characteristics at diagnosis between tested and not-tested cohorts, and to identify factors associated with KRAS testing. RESULTS: Among 1,363 mCRC patients seen from 2008–2011, 648 (47.5%) were KRAS-tested. Among newly diagnosed mCRC patients, the rate of KRAS testing increased from 5.9% prior to 2008, to 13.9% in 2008, and then jumped dramatically to 32.3% in 2009, after which a modest yearly increase continued. The proportions of KRAS-tested patients who had been diagnosed in previous years but not tested previously increased from 17.7% in 2008 to 27.0% in 2009, then decreased to 19.0% in 2010 and 17.6% in 2011. Among patients who were KRAS-tested, the proportions tested at the time of diagnosis increased annually (to 78.4% in 2011). Patients more likely to have been tested included those with lung metastases, poor performance status, more comorbidities, and mCRC diagnosis in 2009 or later. CONCLUSIONS: The frequency of KRAS testing increased over time, corresponding to changes in treatment guidelines and epidermal growth factor receptor inhibitor product labels; however, approximately 50% of eligible patients were untested during the study period.