Lifestyle factors and colorectal cancer prediction: A nomogram-based model

Lifestyle factors are important contributors to the risk of colorectal cancer (CRC). This study developed and validated an age-based CRC risk-prediction model incorporating lifestyle factors using the National Health Insurance Service (NHIS)-National Sample Cohort database.

Individuals who underwent the National Health Examination between 2009 and 2012 were eligible as study participants. Among them, 119,700 (30.38%) were aged 20-39, 190,645 (48.39%) were aged 40-59, and 83,611 (21.22%) were aged ≥ 60. Using the LASSO regression algorithm, we selected risk factors and fitted a Cox proportional hazards model to predict the 10-year CRC incidence. Nomogram-based risk scores were calculated for each age group. Candidate risk factors included sex, age, abdominal obesity, BMI, smoking, alcohol consumption, physical activity, presence of abnormal liver function, hypertension, hypercholesterolemia, and type 2 diabetes mellitus.

In each age group, higher risk scores had a higher incidence probability of CRC. In the discriminatory analysis, the concordance indices of the three models ranged from 0.60 to 0.70, indicating moderate discrimination power. The calibration plot from 10-fold cross-validation showed that the observed proportions of events and predicted probabilities overlapped the entire range of probabilities fairly well. Kaplan-Meier plots demonstrated that individuals in the high-risk group were more likely to develop CRC within 10 years than those in the low-risk group.

Lifestyle factors were identified as significant predictors of CRC incidence, with slight differences across age groups. Nomogram-based risk scores could be used as a tailored intervention to motivate individuals to modify their lifestyles.