Establishment and validation of predictive models by mutational and transcriptional factors for the prognosis of stage IV colorectal cancer patients with liver metastasis who undergo palliative surgery on primary tumors.

Background: The prognostic factors at mutational and transcriptional levels are not clear for stage IV colorectal cancer (CRC) patients with liver metastasis who undergo primary cancer palliative surgery with post-surgical adjuvant therapy. We aimed to establish and validate models for predicting the prognosis of these patients by combining mutational, transcriptional and clinicopathological information.

Methods: Data of 45 stage IV CRC patients with liver metastasis were downloaded from the cBioportal database as the training cohort. Another 30 patients from our hospital were retrospectively recruited as the external validation cohort. Patients were followed up to 4,699 days (median: 823 days) and 2,380 days (median: 980 days) for the overall survival (OS) in the training and validation cohort, respectively. Patients were followed up to 4,699 days (median: 264 days) and 2,259 days (median: 272 days) for the progression-free survival (PFS) in the training and validation cohort, respectively. Tissue samples of the primary CRC were collected and sequenced. Data were analyzed and figures were plotted using the R software.

Results: The mutational and transcriptional alterations and their features were characterized. APC, TP53, TTN, KRAS and SYNE1 were genes with the highest mutational frequency. Significantly upregulated and downregulated genes can be found in transcription. Age, APC2 mutations, and ADRB1, ASTL, MRPL23-AS1 and PDZK1 transcription significantly stratified patient OS, while the KRAS, PTPRF, FREM2, and CLOCK mutations and LY6H, TMEM163, RFX8, ARHDGDIG, TECTA and MYEOV transcriptions significantly stratified patient PFS. Multivariate analyses identified age, APC2 mutations and ADRB1 transcription as independent risk factors for OS. KRAS and PTPRF mutations and RFX8 and MYEOV transcriptions were independent risk factors for PFS. The above independent risk factors were used to establish prediction models for OS and PFS. For the OS model, the 1-, 3- and 5-year area under the curve (AUC) reached 0.858, 0.774 and 0.907 in internal validation, and reached 0.810, 0.778 and 0.924 in external validation, respectively. For the PFS model, the 6 months, 1-year and 1.5-year AUC reached 0.950, 0.803 and 0.847 in internal validation, and reached 0.919, 0.949 and 0.944 in external validation, respectively.

Conclusions: The prognostic factors for stage IV CRC patients with liver metastasis were identified. Models for predicting the OS and PFS were successfully established and validated. The models may help to establish the personalized therapeutic strategies before treatment.