# Clinical calculator based on clinicopathological characteristics predicts local recurrence and overall survival following radical resection of stage II-III colorectal cancer

## TL;DR

This study developed a clinical calculator to predict local recurrence and survival in stage II-III colorectal cancer patients after surgery, using clinicopathological factors.

## Contribution

A novel clinical risk calculator and nomograms were developed to predict outcomes in stage II-III colorectal cancer patients.

## Key findings

- The calculator incorporated six variables and showed good discrimination with AUC values of 0.764 for local recurrence-free survival and 0.815 for overall survival in the training cohort.
- Calibration plots showed excellent agreement between predictions and actual observations for 3- and 5-year outcomes.
- Low-risk patients were more likely to undergo salvage surgery when recurrence occurred.

## Abstract

This study aimed to analyze the risk factors and survival prognosis of local recurrence in stage II-III colorectal cancer (CRC) and develop a clinical risk calculator and nomograms to predict local recurrence and survival in treated patients.

Patients who underwent radical surgery between January 2009 and December 2019 at the China National Cancer Center were included. Multivariate nomograms and a clinical risk calculator based on Cox regression were developed. Discrimination was measured with an area under curve (AUC) and variability in individual predictions was assessed with calibration curves. We stratified patients into different risk groups according to the established model to predict their prognosis and guide clinical practice.

The clinical risk calculator incorporated six variables: tumor thrombus, perineural invasion, tumor grade, pathology T-stage, pathology N-stage, and whether more than 12 lymph nodes were harvested. Our clinical risk calculator provided good discrimination, with AUC values of local recurrence-free survival (LRFS) (0.764) and overall survival (OS) (0.815) in the training cohort and LRFS (0.740) and OS (0.730) in the test cohort. Calibration plots illustrated excellent agreement between the clinical risk calculator predictions and actual observations for 3- and 5-year LRFS and OS. Recurrence risk-stratified analysis showed that low-risk patients were more likely to undergo salvage radical surgery when recurrent disease existed.

The clinical calculator can better account for tumor and patient heterogeneity, providing a more individualized outcome prognostication. The model is expected to aid in treatment planning, such as resectability evaluation, and it can be used in postoperative surveillance (https://oldcoloncancer.shinyapps.io/dynnomapp/).