Development And Validation Of A Risk Prediction Model For Colorectal Polyps

Background and Purpose:Colorectal cancer(CRC)is one of the most common malignancies of the digestive system.2020 global cancer statistics show that CRC has the third highest incidence of malignancies with more than 1.9million cases and the second highest mortality rate with more than 900,000 cases.CRC is one of the major cancers affecting people’s lives and poses a serious health burden.And early diagnosis and treatment are generally considered to be the most effective way to reduce the mortality of CRC.Colorectal polyps are often considered to be the most common precancerous lesion of CRC in clinical practice,so early detection and removal of colorectal polyps is expected to further reduce the incidence of CRC,and therefore it is important to study the risk factors for the development of colorectal polyps.Risk prediction models are a hot topic in current research,and there are many models for CRC,but relatively few risk prediction models for colorectal polyps have been developed.Therefore,this study analyzes the risk factors of colorectal polyps,constructs and validates risk prediction models for colorectal polyps,so as to improve the screening rate of colorectal polyps and play a guiding role in preventing the occurrence of CRC.Methods:A retrospective analysis was performed among a total of 1952 subjects who were hospitalized between January 2021 and October 2022 in the Department of Gastroenterology of Huaihe Hospital,Henan University and underwent colonoscopy.978 subjects were diagnosed with colorectal polyps and 974 subjects with no significant findings;the basic characteristics of the included 1952 patients were The basic characteristics of the included 1952 patients were described,and single-and multifactorial analyses of the occurrence of colorectal polyps were performed.The colorectal polyp group was divided into adenomatous polyps(744 cases in the CAP group)and non-adenomatous polyps(234 cases in the NCAP group)according to the results of pathological types,and then the basic characteristics of the polyp group were described,and the single-and multifactorial analyses of adenomatous polyp rows were performed.For the risk modeling of colorectal polyps,the data of 1952 cases were completely randomly split into a modeling cohort(1464 cases)and a validation cohort(488 cases)using R language at 3:1,and the risk factors for colorectal polyps were screened by univariate and multifactor regression analysis,and the risk prediction model for colorectal polyps was established based on this,and the clinical prediction model was presented in the form of a column line graph with The clinical prediction model was evaluated by differentiation,calibration and clinical validity.Results:(1)A total of 1952 patients were included in this study,and baseline characteristics of cases in the polyp group were compared with those in the no-polyp group.There were statistically significant differences between the two groups in age,sex,TC,TG,LDL-C,TG/HDL,CEA,BMI,waist circumference,waist height ratio,history of smoking,history of alcohol consumption,and history of hypertension and diabetes mellitus(P < 0.05),while HDL-C,TC/HDL,LDL/ HDL,BA,CRP,ALB,TP,uric acid,and education were not statistically significant(P > 0.05).All factors were included for univariate analysis,and the results showed that age,gender,TC,TG,LDL,TG/HDL,BMI,waist circumference,waist height ratio,history of hypertension,history of diabetes,history of smoking,and history of alcohol consumption were associated with the development of colorectal polyps(P < 0.05),while HDL-C,TC/HDL,LDL/HDL,BA,CRP,ALB,TP,and CEA,uric acid,and education were not statistically significant(P > 0.05);inclusion of the above single-factor significant(P < 0.05)indicators in the multifactor showed that age,TG,waist circumference,history of alcohol consumption,and history of hypertension were independent risk factors for colorectal polyps.(2)The 978 patients with colorectal polyps were divided into the CAP group NCAP group according to the type of pathology,and the comparison results between the two groups showed statistical differences in age,TC,LDL,gender,smoking,polyp size,number of polyps,and polyp site(P < 0.05),while there were no statistical differences in TG,HDL-C,TC/HDL,TG/HDL,LDL/HDL,BA,CRP,ALB,TP,CEA,BMI,waist circumference,waist height ratio,history of alcohol consumption,history of hypertension,history of diabetes mellitus,hyperuric acid,and education were not statistically different(P >0.05).The results of univariate analysis showed that advanced age,male,TC,LDL-C,smoking,number of polyps,size of polyps,and site of polyps were associated with the development of adenomatous polyps(P< 0.05),while TG,HDL,TC/HDL,TG/LDL,LDL/HDL,BA,CRP,Al B,TP,CEA,BMI,waist circumference,waist height ratio,history of alcohol consumption,history of history of hypertension,history of diabetes mellitus,high uric acid,and education did not correlate with adenomatous polyps(P >0.05).The above significant indicators(P < 0.05)were included in the multifactorial analysis,which showed that TC,gender,number of polyps,and polyp size were independent risk factors for adenomatous polyps.(3)Data from 1952 cases were split into a modeling cohort(1464 cases)and a validation cohort(488 cases)at a complete randomization of 3:1 using R language,and the results showed no statistical differences between the modeling and validation cohorts in terms of general clinical information and laboratory tests(P > 0.05),indicating consistent baseline characteristics and good balance between the two cohorts;age,gender,TC,TG,LDL-C,in the polyp and non-polyp groups in the modeling cohort and TG/HDL,BA,BMI,waist circumference,waist height ratio,history of smoking,history of alcohol consumption,history of hypertension,and history of diabetes mellitus were statistically different between the polyp group and the non-polyp group(P < 0.05),whereas there were no statistical differences in HDL-C,TC/HDL,LDL/HDL,CRP,ALB,TP,CEA,uric acid,and education(P > 0.05).Univariate analysis in the modeling cohort showed that age,male,TC,TG,LDL-C,BA,TG/HDL,BMI,waist circumference,waist height ratio,history of smoking,history of alcohol consumption,history of hypertension,and history of diabetes mellitus were associated with the development of colorectal polyps(P< 0.05).Multifactorial regression analysis showed that age,TG,waist circumference,and history of alcohol consumption were all independent risk factors for colorectal polyps.The risk factors in the multifactorial were plotted by R language to predict the occurrence of colorectal polyps with column line graphs,calibration degree evaluation and decision curves,and the results showed that the area under the ROC curve(AUC)for the modeling cohort was 0.706 and the area under the ROC curve(AUC)for the validation cohort was 0.711,indicating that the model has some predictive effect;moreover,the calibration curves of both data sets were close to the ideal curve,indicating that the model has good predictive power.shows that the model has good calibration ability;the decision curve shows that the column line graph prediction model has some clinical application value.Conclusions:(1)Age,TG,waist circumference,and alcohol consumption were all independent risk factors for colorectal polyps;(2)Female,TC,number of polyps,and size of polyps were independent risk factors for adenomatous polyps;(3)Establishing a clinical prediction model for colorectal polyps and presenting it in the form of a columnar line graph can help improve the detection rate of colorectal polyps.