Synopsis

KRAS mutations are well known as predictive markers of resistance to epidermal growth factor receptor-targeted antibodies in rectal cancers. Approximately 30%-40% colorectal cancers are observed with KRAS mutation, and rectal cancer accounts for 30%-35% among CRC ^1-3. The pre-operative neoadjuvant therapy including anti-EGFR chemotherapy demonstrated robust value is the current trend in the management of rectal cancer. Therefore, it is important to select suitable patients who would benefit from the aggressive multimodality approaches and tailor individual treatment to better combat disease. Currently, few data are available regarding the potential relationship between MRI characteristics and genetic biomarkers.

Purpose

To evaluate the MRI features and texture analysis of different KRAS mutation status in rectal cancer.

Material and Methods

This is an institutional review board approved retrospective case-control study. Totally, 158 patients(mean age,[60.66±13.38] years; range 26-87 years)including106 men and 52 women with pathologically proved rectal cancers and preoperative pelvic MRI examinations were enrolled. Patients were stratified into two groups: wild-type and mutant according to genomic DNA extraction and analysis. MRI features of rectal cancers (including histogram-based texture features onT2 weighted images) and relevant clinical characteristics were statistically evaluated to identify the differences between wild-type and mutant groups. The independent samples t test or Mann-Whitney U test were used for continuous variables. The differences of the remaining categorical other polytomous variables were analyzed using the chi-square test or Fisher exact test. A receiver operating characteristic (ROC) curve analysis was performed to evaluate the discriminatory power of MRI features in predicting tumor KRAS status. The area under the ROC curve (AUC) and the optimal cut-off values were calculated, meanwhile sensitivity, specificity, and accuracy rate were determined. The interobserver agreement for continuous variables was evaluated using the interclass correlation coefficient (ICC), and for categorical varivables using Kappa of agreement. P<0.05 was considered to indicate a statistically significant difference.

Results

Relatively good to excellent interobserver agreement was obtained for continuous and categorical variables. Statistically difference of tumor shape distribution was observed between wild-type and mutant group (=7.591, p=0.022). Higher T stage was observed more frequently in the mutant group than that of wild-type group (=7.086, p=0.029). Higher ADC values were observed in KRAS mutant group than those in wild-type group (t=-2.708, p=0.008). Textural parameters (Mean, Variance, Kurtosis, Entropy) value derived from T2-weighted images showed significant differences in rectal cancer with different KRAS status (p=0.008, p=0.046, p=0.031, p=0.001, respectively). The AUC values for MRI features including textural parameters ranged from 0.595 to 0.682.

Conclusion

The MRI features of rectal cancer including histogram-based texture features showed potential value in identifying KRAS status, however, more effort need to verify our results.

Acknowledgements

No acknowledgement found.