Abstract
BACKGROUND: The optimal laboratory assay for detecting KRAS mutations in different biospecimens from patients with metastatic colorectal cancer (mCRC), and the clinical relevance of these gene alterations is still in question. We analyzed the prognostic-predictive relevance of KRAS status, determined in tumor and plasma DNA by two different assays, in a large mono-institutional series of mCRC patients. METHODS: DNA sequencing and peptide-nucleic-acid-mediated-polymerase chain reaction clamping (PNA-PCR) were used to determine KRAS status in 416 tumor and 242 matched plasma DNA samples from mCRC patients who received chemotherapy only. Relationships with outcomes were analyzed with respect to the different assays and tissue types. RESULTS: PNA-PCR was significantly more sensitive in detecting KRAS mutations than sequencing (41% vs. 30%, p < 0.001). KRAS mutations were more frequent in tumor tissue than in plasma (sequencing, 38% vs. 17%, p < 0.001; PNA-PCR, 47% vs. 31%, p < 0.001). Median OS was consistently shorter in KRAS-mutated patients than KRAS wild-type patients, independent from the assay and tissue tested; the largest difference was in plasma samples analyzed by PNA-PCR (KRAS mutated vs. wild-type: 15.7 vs. 19.1 months, p = 0.009). No association was observed between KRAS status and other outcomes. When tumor and plasma results were considered together, median OS in patients categorized as tissue/plasma KRAS negative/negative, tissue/plasma KRAS discordant, and tissue/plasma KRAS positive/positive were 21.0, 16.9 and 15.4 months, respectively (p = 0.008). CONCLUSIONS: KRAS mutation status is of prognostic relevance in patients with mCRC. KRAS mutations in both tumor tissue and plasma are a strong prognostic marker for poor outcomes.