Abstract
Background: The presence of fatty liver (FL) has been suggested to influence the incidence of colorectal cancer (CRC). This study aimed to evaluate the predictive utility of six alternative indices of FL—namely, liver fat percentage (PLF), lipid accumulation product (LAP), hepatic steatosis index (HSI), United States fatty liver index (USFLI), fatty liver index (FLI), and Zhejiang University index (ZJU)—in assessing the risk of CRC. We aimed to determine their effectiveness in predicting CRC risk by comparing these surrogate indices. Methods: Data for this study were derived from the National Health and Nutrition Examination Survey (NHANES) conducted between 2003 and 2018, focusing on adults over 20 years old. The six FLIs were calculated using established methodologies outlined in prior research. To identify key variables, the Boruta algorithm was employed. The relationships between FLIs and CRC risk were assessed using multivariable logistic regression, generalized linear models (GLMs), and restricted cubic spline (RCS) models. Additionally, subgroup analyses were performed to investigate the effects of potential confounders. Results: Among the 16,250 individuals surveyed, 96 were diagnosed with CRC. Those with CRC exhibited significantly higher levels of PLF (4.65 vs. 3.31, P=0.004), LAP (55.63 vs. 42.34, P=0.04), USFLI (23.22 vs. 17.83, P vs. 50.86, P=0.048) compared to individuals without CRC. Multivariate logistic regression and RCS analyses indicated that, of the six indices, only USFLI was significantly associated with an increased risk of CRC. Notably, further stratification of USFLI revealed that this association was consistently stronger in individuals aged over 65 years [odds ratio (OR) =1.023; 95% confidence interval (CI): 1.005–1.041; P=0.01] and among non-smokers (OR =1.018; 95% CI: 1.003–1.033; P=0.02) after adjusting for multiple confounders. Conclusions: The USFLI index demonstrated a more significant association with the risk of CRC compared to the other five alternative FLIs, highlighting its potential utility in predicting CRC risk in clinical settings.

This publication has 51 references indexed in Scilit: