Colorectal cancer and noncancer patients have similar labeling indices by microscopy and computed image analysis

Abstract

The labeling index (LI), a microscopic measurement of proliferative activity in colonic crypts, is proposed as an indicator of colonic cancer risk. Computed image analysis of proliferative regions is less labor intensive and more objective than is direct microscopy but has not been validated for labeling indices by direct comparison. The authors compared colonic crypt proliferation in 26 cancer and 13 noncancer patients by using Ki-67 monoclonal antibody (McAb) labeling of flat mucosa obtained from surgically removed, frozen specimens. In cancer patients, the mucosa specimen was excised 10 cm away from the tumor, and the LI was determined microscopically for the whole crypt, the upper two thirds, and the upper one third of 15 crypts. Nuclear antigen levels of 15 whole crypts were determined by using the CAS-200 computed image analyzer (Cell Analysis Systems, Elmhurst, IL). Cancer and noncancer specimens were compared as were microscopically determined LI and stained nuclei specimens by using image analysis. No statistically significant difference in proliferative activity of whole crypts, or the upper two thirds of crypts, was observed between cancer specimens and noncancer specimens from using either technique. However, a significant correlation existed between microscopic analysis and computed image analysis of labeled nuclei. Computed image analysis using Ki-67 McAb labeling can be used instead of microscopy to determine crypt LI, but neither method can be used to distinguish cancer specimens from noncancer specimens.