Genotypic classification of colorectal adenocarcinoma. Biologic behavior correlates with K-ras-2 mutation type.

New measures enabling better prediction of biologic behavior of large bowel cancer are highly desirable. One hundred ninety-four consecutive primary, recurrent, and metastatic colorectal adenocarcinomas, accessioned during 1991 at Rhode Island Hospital, were classified according to the presence and specific type of K-ras-2 point mutation. An integrated histopathologic-genetic approach was used to detect mutations starting with minute, topographically selected, tissue samples from formaldehyde-fixed, paraffin-embedded specimens. Each colorectal adenocarcinoma exhibited either no or only one of seven specific types of K-ras-2 mutation. The mutation type of each primary tumor was present consistently in its metastatic deposits. Thirty-five percent of primary colorectal adenocarcinomas were found to be mutated (42 of 119). A significantly higher mutation rate (65%) was seen in lymphogenous-hematogenous metastases as a group (35 of 54; P < 0.005). By contrast, 22% of anastomotic recurrences and transcoelomic metastasis were mutated (4 of 18). Twenty-eight percent of adenocarcinomas with invasion limited to muscularis propria (Tis, T1, T2) were mutated (16 of 57), compared to 41% for more deeply invasive tumors (T3, T4; 26 of 63). When colorectal adenocarcinomas were analyzed by specific K-ras-2 mutation type, it was found that codon 13 mutated tumors did not progress to local or distant metastasis (P < 0.01). Tumors having a codon 12 valine substitution did not metastasize beyond pericolonic-perirectal lymph nodes. In contrast, colorectal cancers with codon 12 aspartic acid substitutions accounted for most of the distant hematogenous deposits (P < 0.01). Tumors with normal K-ras-2 accounted for most intraperitoneal deposits. Genotyping of colorectal adenocarcinoma by K-ras-2 status can identify subsets of patients likely to pursue indolent and aggressive forms of disease. The integrated histopathologic-genetic approach outlined is feasible for use in diagnostic pathology, providing information that together with clinicopathologic staging may individualize and optimize treatment.