Association Between Natural Killer Cell Activity and Colorectal Cancer in High-Risk Subjects Undergoing Colonoscopy

Low activity of natural killer (NK) cells has been associated with increased risk of cancer and has been reported in patients with colorectal cancer (CRC). Activity of NK cells can be measured in a small volume of whole blood by a commercially available test. We investigated whether this test could be used to identify patients with CRC, using findings from colonoscopy as a reference standard.

We performed an open-label, prospective, cross-sectional study of 872 high-risk subjects (more than 40 years old) screened for CRC by colonoscopy at a university hospital in Montreal, Canada from October 2014 through January 2016. Blood samples were collected on the day of colonoscopy, prior to the procedure. The test involves stimulation of whole blood with cytokine that induces NK cells to secrete interferon gamma (IFNG), which is quantified by an ELISA. Tissue samples were taken from lesions during the colonoscopy and analyzed histologically; subjects were classified as having no evidence of disease, adenomatous polyps of less than 10 mm, of 10 mm or more, or CRC. We used the non-parametric Mann-Whitney test to compare NK cell activity between subjects with no evidence of CRC and subjects found to have CRC. Receiver operating characteristic curve analysis was used to assess the ability of the test to identify individuals with CRC. The primary objective was to determine the difference in NK cell activity between subjects with vs without CRC. The secondary objective was the test performance, based on receiver operating characteristic analysis, and cut-off value that most accurately identified individuals with CRC.

We found a significant difference in NK cell activity between the 23 subjects with CRC (based on pathology analysis) and the 849 subjects without CRC: subjects found to have CRC by colonoscopy had a median level of 86.0 pg IFNG/mL (inter-quartile range, 43.3-151.0 pg IFNG/mL), whereas subjects without CRC had a median level of 298.1 pg IFNG/mL (inter-quartile range, 100.4-920.2 pg IFNG/mL) (P = .0002). The cut-off value that most accurately identified subjects with CRC was 181 pg/mL. The NK cell activity test identified subjects with CRC with 87.0% sensitivity, 60.8% specificity, a positive predictive value of 5.7%, and a negative predictive value of 99.4%. The odds ratio for detection of CRC in subjects with low NK cell activity vs subjects with higher NK cell activity was 10.3 (95% CI, 3.03-34.9).

Using colonoscopy as the reference standard, a test for NK cell activity in whole blood samples identified patients with CRC with 87.0% sensitivity and a negative predictive value of 99.4%. Subjects with low NK cell activity had a 10-fold higher risk of CRC compared with subjects with high NK cell activity. This test might be used in clinical practice to assess patients for risk of CRC. Clinicaltrials.gov number: NCT02291198.