To explore the feasibility of predicting KRAS gene mutation in rectal cancer by preoperative multi-slice spiral CT (MSCT).

From October 2016 to December 2017, clinical data of fifty-two patients with rectal cancer in the First Affiliated Hospital of Sun Yat-sen University were analyzed, including the preoperative imaging parameters, pathological data and KRAS gene detection results. Binary Logistic regression models were established to speculate the possibility of predicting KRAS gene mutation by CT parameters, and the diagnostic effect and critical point were analyzed by receiver operating characteristic curve (ROC).

Among the 52 cases, 18 cases (34.6%) had KRAS gene mutation and were prone to lymph node metastasis (67.6% vs 32.4%, χ²=7.53, P=0.010). The diameter of superior rectal vein (SRV) was (5.07±1.00) mm in mutation patients, which was significantly larger than that in non-mutation patients (4.23±1.02) mm, the difference was statistically significant (t=2.84, P=0.010). Logistic regression model showed that the probability of gene mutation in patients with lymph node metastasis was 5.69 times higher than that in patients without lymph node metastasis (95% CI: 1.34-24.14). For every 1 mm increase in SRV, the probability of gene mutation increased by 2.56 times (95% CI: 1.14-5.73). The best threshold of SRV was 4.5 mm. Combined with imaging lymph node metastasis, the sensitivity, specificity and accuracy of MSCT in predicting KRAS mutation were 72.22% (13/18), 85.29% (29/34) and 80.77% (42/52), respectively.

KRAS mutations in rectal cancer can be predicted noninvasively using MSCT to measure SRV diameter in the axial venous phase. Combined with the presence or absence of lymph node metastasis, the diagnostic efficacy can be improved.