Endoscopic characterization and development of a prediction model for colorectal adenomatous polyps

Abstract

BACKGROUND

Colorectal cancer carries a high mortality rate worldwide. Adenocarcinoma is its most common type, which is progressed following an adenoma-dysplasia-carcinoma sequence. Early accurate identification of individuals with a high risk for adenomas could facilitate prompt removal of adenomas and reduce colorectal cancer incidence.

AIM

To investigate the endoscopic features of colorectal adenomatous polyps and develop a prediction model for their occurrence.

METHODS

A total of 202 individuals undergoing their first colonoscope between January 2023 and January 2024 were selected. Endoscopic characteristics, demographics, clinical information, and laboratory results were compared between individuals with or without colorectal adenomas. Logistic regression was performed to identify risk factors for colorectal adenomatous polyps and construct a prediction model. A nomogram was developed using R (v3.5.2). Internal validation was conducted with 1000 bootstrap resamples to assess calibration and discrimination. External valuation was conducted in another 240 individuals.

RESULTS

In 202 participants, 75 (37.13%) and 127 (62.87%) had or had no colorectal adenomas, respectively. These adenomas varied in shape, size, and locations. Age ≥ 60 years, smoking, alcohol, fatty liver, gallbladder polyps, and abnormal glycosylated hemoglobin (HbA1c) were risk factors and were used to construct nomogram Logit(P) = -5.602 + 1.791 × age ≥ 60 years + 1.115 × smoking + 1.894 × alcohol + 1.727 × fatty liver + 1.749 × gallbladder polyps + 1.903 × abnormal HbA1c. Internal validation demonstrated excellent agreement between apparent and bias-corrected curves, strong discriminative ability (area under the curve of 0.889, 95% confidence interval: 0.841-0.938), and high net clinical benefit (threshold probability range 0.1-0.9). External validation also suggested excellent model performance.

CONCLUSION

Colorectal adenomas exhibit diverse characteristics. A nomogram incorporating age, smoking, alcohol, fatty liver, gallbladder polyps, and HbA1c provides a robust and accurate prediction for the risk of colorectal adenomas.

Key Words: Colorectal adenomatous polyps; Endoscopic characteristics; Risk factors; Prediction model; Colorectal cancer prevention

Core Tip: Colorectal adenomas varied significantly in shape, size, and locations. Age ≥ 60 years, smoking, alcohol, fatty liver, gallbladder polyps, and abnormal glycosylated hemoglobin were associated with the presence of colorectal adenomas. A nomogram Logit(P) = -5.602 + 1.791 × age ≥ 60 years + 1.115 × smoking + 1.894 × alcohol + 1.727 × fatty liver + 1.749 × gallbladder polyps + 1.903 × abnormal glycosylated hemoglobin showed excellent performance and could provide robust and accurate prediction for the risk of colorectal adenomatous polyps.

INTRODUCTION

Colorectal cancer (CRC) ranks third among all cancers in incidence and second in cancer-related mortality worldwide. The incidence of CRC is gradually shifting toward younger age groups[1]. Adenocarcinoma is among the most prevalent histological type of CRC. Adenomas are the most frequently detected neoplastic polyps of the colorectal mucosa. According to the proportion of villous architecture, colorectal adenomas are classified as tubular adenomas (< 25% villous features), villous adenomas (> 75% villous features), and tubulovillous adenomas (25%-75% villous features). The progression from adenoma to carcinoma is the principal pathway of CRC development. Most CRC lesions originate from adenomatous polyps[2]. Adenomatous polyps carry a 2.9-fold higher risk of malignant transformation compared with other types of polyps[3], and their cancer risk increases with lesion size[4]. High-risk or advanced adenomas show particularly strong associations with CRC development and progression[5,6]. The adenoma-dysplasia-carcinoma sequence typically evolves over 10-15 years, providing a critical window for prevention and early intervention. Timely detection and removal of colorectal adenomas are therefore essential to reducing CRC incidence and improving patient prognosis.

Colonoscopy is the most widely used method for both preliminary and follow-up screening of CRC and is an indispensable tool for the diagnosis and treatment of adenomas in clinical practice. Effective screening through colonoscopy in the early stages can reduce CRC mortality by approximately 18%[5]. However, variation in operator skill and technique contributes to a high rate of missed adenomas, which limits the overall effectiveness of colonoscopy. As endoscopic technology becomes more widely adopted, each 1% increase in the detection rate of adenomatous polyps in a clinical setting is associated with a 3% reduction in subsequent CRC risk[7]. Furthermore, early endoscopic resection of adenomas can lower CRC incidence by 21% and reduce mortality by 26%[8]. In addition to colonoscopy, other screening methods, such as fecal occult blood test and multi-target stool DNA test, were also applied in the clinic for CRC. However, fecal occult blood test carries low sensitivity and multi-target stool DNA test has low specificity[9]. Recently, peripheral blood-based screening tests, such as those detecting circulating tumor DNA and cell-free DNA and looking for fragments of tumor-specific DNA in the blood specimen, were also reported. However, these tests were less sensitive for adenoma and their wide clinical application are still under investigation[10]. Therefore, early detection and resection of precancerous lesions are key imperatives for CRC prevention, and a detailed characterization of colorectal adenomatous polyps during endoscopy is crucial to guide early diagnosis in clinical practice.

Many clinical studies have examined the risk factors associated with colorectal adenomatous polyps. Clinical risk prediction models were also reported to guide screening intensity. These models considered variable risk factors but all have their limitations. For example, Shaukat et al[11] incorporated age, sex, body mass index, family history, and smoking to develop and validate a risk-scoring system for colorectal adenoma, but its performance was unsatisfactory with the area under the receiver operating characteristic curve (AUC) of only 0.64. Li et al[12] applied regression analysis and constructed a nomogram to determine the risk of colorectal adenoma. This nomogram had better performance with a higher AUC of 0.78. However, it included ten variables (age, sex, hyperlipidemia, smoking, consumptions of red meat, salt, and dietary fiber, and Helicobacter pylori infection, non-alcoholic fatty liver disease, and chronic diarrhea) and certain variable, such as Helicobacter pylori infection requires specialized test[12]. A recent study reported another nomogram showing that elderly adults, male, smoking history, chronic drinking, frequent pickle intake, and irregular defecation could indicate a high risk for colorectal adenoma. However, information on pickle intake and irregular defecation may not be accurate due to recall bias from patients[13]. Another recent study reported a simplified nomogram with only four variables (age, daily number of defecation, thrombin time, and number of polyps). The model was easy to perform in the clinic, but information on day-to-day number of defecations could vary significantly, limiting its accurate risk predication in the clinical practice[14]. Therefore, more studies are required to determine an easy-to-use model with a high accuracy to predict the risk of colorectal adenoma.

To address this gap, the present study incorporates a broader range of variables informed by previous research to identify key risk factors and develop a clinical prediction model that is convenient, accurate, and practical. This model is intended to estimate the probability of colorectal adenomatous polyps and provide a reference to guide clinical treatment decisions and support the prevention of CRC.

MATERIALS AND METHODS

General data

A simple random sampling method was used to select 202 patients who underwent digestive endoscopy for the first time at the People’s Hospital of Guanghan, China, between January 2023 and January 2024. The study protocol was approved by the hospital ethics committee. There were 90 asymptomatic individuals (attending voluntarily for routine physical examinations or colorectal disease screening without any colorectal symptoms (such as abdominal pain, hematochezia, or altered bowel habits) and 11 were symptomatic referral patients (presenting with abdominal pain, hematochezia, mucoid stools, or abnormal bowel habits). The remaining 101 cases were individuals who presented for medical care due to various symptoms. In addition, between February and August 2024, additional 240 individuals meeting the same inclusion and exclusion criteria were selected from three local medical institutions for external validation.

The participant inclusion criteria were as follows: (1) Colonoscopy specimens confirmed by histopathological biopsy performed by the hospital’s Department of Pathology; (2) Complete pathological report; and (3) Clear indications for colonoscopy. Exclusion criteria included patients with mental or intellectual disorders impairing effective communication or severe major organ (heart, lungs, kidneys, liver, or brain) dysfunction that might lead to difficulty in completing conscious sedation and colonoscope examination. In addition, we also excluded patients with inflammatory bowel disease, autoimmune disease, or intestinal tuberculosis, since these disorders were reported to impact the risk of CRC[15-17].

All participants provided written informed consent form for gastrointestinal endoscopy. Bowel preparation was performed one day before the examination using polyethylene glycol electrolyte powder dissolved in water, administered orally 12 hours prior to the procedure. Patients were encouraged to engage in light activity after administration. Adequate preparation was defined as the passage of clear, water-like stool with no visible fecal residue to ensure optimal visualization. Patients fasted from food and water for 3-4 hours before the procedure. Endoscopic examinations were performed at the hospital’s Gastrointestinal Endoscopy Center. The location, size, number, morphology, and base characteristics of all polyps in the colon and rectum were documented. Partial or complete polyp tissue samples were submitted to the Department of Pathology for histopathological examination. The examination was conducted by more than two senior physicians from this department, who were responsible for the diagnosis and issuance of reports. The patients were grouped based on the results of endoscopic and pathological examinations.

Data collection

Baseline patient information was extracted from the hospital’s electronic medical record system and included sex, age, smoking history, alcohol consumption, comorbidities, presence of fatty liver or gallbladder polyps, and family history of malignancy. Fasting venous blood samples (5 mL) were collected three days before endoscopy to measure glycosylated hemoglobin (HbA1c), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), triglyceride (TG), total cholesterol (TC), and other parameters. Reference ranges were: HbA1c 5.7%-6.4%, LDL-C 0-3.4 mmol/L, HDL-C 1.0-1.6 mmol/L (males) and 1.3-1.9 mmol/L (females), TG 0.45-1.70 mmol/L, and TC 2.86-5.98 mmol/L. In this study, HbA1c values were stratified by quartiles. HbA1c < 5.7% was defined as normal, whereas HbA1c ≥ 5.7% was considered abnormal as recommended by the American Diabetes Association[18].

Statistical analysis

All statistical analyses were performed using SPSS version 22.0. Normally distributed continuous variables were presented as mean ± SD and analyzed using the t-test. Categorical variables were expressed as percentages and analyzed using the χ² test. Logistic regression was employed to identify risk factors for colorectal adenomatous polyps, and receiver operator characteristic curves were plotted to evaluate the diagnostic performance of each factor. A nomogram prediction model based on the risk factors was developed and validated using appropriate R packages (version 3.5.2). Statistical significance was set at α = 0.05.

RESULTS

Detection of colorectal adenomatous polyps and their endoscopic characteristics

Among the 202 subjects who underwent endoscopy, 75 were diagnosed with colorectal adenomatous polyps (37.13%) and 127 had no colorectal adenomatous polyps (62.87%). Most polyps exhibited elevated or flat morphology, measured < 10 mm in diameter, and occurred as one to 2 Lesions per patient. Polyps were more frequently located in the left colon. Detailed characteristics are provided in Table 1.

Table 1 Endoscopic characteristics of colorectal adenomatous polyps among 75 patients.

Main endoscopic characteristics	Number of cases	Percentage
Gross morphology
Elevated	35	46.67
Flat	36	48.00
Depressed	4	5.33
Pit pattern
Type I	41	54.67
Type II	15	20.00
Type III-L	7	9.33
Type III-S	7	9.33
Type IV	5	6.67
Type V	0	0
Surface vascular pattern
Type 1	34	45.33
Type 2A	28	37.33
Type 2B	13	17.34
Type 3	0	0
Size
< 10 mm	60	80.00
≥ 10 mm	15	12.00
Number of lesions
1-2	63	84.00
≥ 3	12	16.00
Location
Left colon	57	76.00
Right colon	18	24.00

Risk factors for colorectal adenomatous polyps

Univariate analysis: Significant differences were observed between the adenomatous and non-adenomatous groups in terms of age, smoking status, alcohol consumption, fatty liver, gallbladder polyps, and HbA1c level (P < 0.05), but not in sex, hypertension, diabetes, coronary heart disease, hyperlipidemia, family history of malignancy, LDL-C, HDL-C, TG, and TC (P > 0.05) (Table 2).

Table 2 Univariate analysis of risk factors for colorectal adenomatous polyps, n (%).

Clinical data	Colorectal adenomatous polyp group (n = 75)	Non-colorectal adenomatous polyp group (n = 127)	χ2 or t value	P value
Age			20.054	< 0.001
< 60 years	24 (32.00)	82 (64.57)
≥ 60 years	51 (68.00)	45 (35.43)
Sex			0.061	0.806
Male	40 (53.33)	70 (55.12)
Female	35 (46.67)	57 (44.88)
Smoking			10.129	0.001
No	24 (32.00)	70 (55.12)
Yes	51 (68.00)	57 (44.88)
Drinking			13.467	< 0.001
No	31 (41.33)	86 (67.72)
Yes	44 (58.67)	41 (32.28)
Underlying disease
Hypertension	10 (13.33)	21 (16.54)	0.372	0.542
Diabetes	18 (24.00)	23 (18.11)	2.143	0.143
Coronary heart disease	25 (33.33)	36 (28.35)	0.556	0.456
Hyperlipidemia	13 (17.33)	20 (15.75)	0.087	0.768
Fatty liver	46 (61.33)	37 (29.13)	20.197	< 0.001
Gallbladder polyp	38 (50.67)	22 (17.32)	25.108	< 0.001
Family history of malignancy			0.094	0.342
No	18 (24.00)	27 (21.26)
Yes	47 (76.00)	100 (78.74)
HbA1c			14.993	< 0.001
Normal	15 (20.00)	60 (47.24)
Abnormal	60 (80.00)	67 (52.76)
LDL-C (mmol/L), mean ± SD	2.71 ± 0.30	2.79 ± 0.32	1.757	0.081
HDL-C (mmol/L), mean ± SD	1.26 ± 0.21	1.20 ± 0.24	1.796	0.074
TG (mmol/L), mean ± SD	1.40 ± 0.26	1.35 ± 0.26	1.321	0.188
TC (mmol/L), mean ± SD	4.47 ± 0.58	4.50 ± 0.51	0.384	0.702

HbA1c: Glycosylated hemoglobin; LDL-C: Low-density lipoprotein cholesterol; HDL-C: High-density lipoprotein cholesterol; TG: Triglycerides; TC: Total cholesterol.

Multivariate analysis: Variables with P < 0.05 in the univariate analysis were entered into a logistic regression model, with the presence of colorectal adenomatous polyps (dependent variable) coded as 1 and the absence coded as 0 (value assignment shown in Table 3). Logistic regression identified age, smoking, alcohol consumption, fatty liver, gallbladder polyps, and HbA1c as independent risk factors for colorectal adenomatous polyps (P < 0.01; Table 4).

Table 3 Value assignment for variables.

Variable	Assignment
Age	< 60 years = 0; ≥ 60 years = 1
Smoking	Absence = 0; presence = 1
Drinking	Absence = 0; presence = 1
Fatty liver	Absence = 0; presence = 1
Gallbladder polyp	Absence = 0; presence = 1
HbA1c	Normal = 0; abnormal = 1

HbA1c: Glycosylated hemoglobin.

Table 4 Multivariate analysis of risk factors for colorectal adenomatous polyps.

Factor	Β	SE	Wald χ2	P value	OR	95%CI
Age	1.791	0.430	17.378	< 0.001	5.993	2.582-13.908
Smoking	1.115	0.408	7.481	0.006	3.051	1.372-6.785
Drinking	1.894	0.442	18.405	< 0.001	6.647	2.798-15.792
Fatty liver	1.727	0.419	16.971	< 0.001	5.623	2.473-12.787
Gallbladder polyp	1.749	0.432	16.375	< 0.001	5.750	2.464-13.416
HbA1c	1.903	0.467	16.636	< 0.001	6.705	2.687-16.729
Constant	-5.602	0.785	50.926	< 0.001	0.004	-

CI: Confidence interval; OR: Odds ratio; HbA1c: Glycosylated hemoglobin.

Construction and validation of nomogram prediction model: A prediction model was developed using the above six independent risk factors. According to the G*Power simulation formula for logistic regression (α = 0.05; six variables, 75 cases and 127 controls), if the true odds ratio of the core variable (e.g., HbA1c) is 1.6, the test power is approximately 82%, satisfying the general requirement of ≥ 80%.

The resulting risk prediction model equation was as follows: Logit(P) = - 5.602 + 1.791 × (age ≥ 60 years) + 1.115 × smoking+ 1.894 × alcohol consumption + 1.727 × fatty liver + 1.749 × gallbladder polyps + 1.903 × abnormal HbA1c. This equation was visualized as a nomogram. Internal validation was performed using the enhanced bootstrap method. The apparent curve of the model closely matched the bias-corrected curve, and the slope of the calibration curve approached 1, indicating excellent agreement between predicted and observed risks (Table 5 and Figure 1). After 1000 bootstrap resamples, the AUC for the prediction model was 0.889 (95% confidence interval: 0.841-0.938), indicating strong discriminative ability (Figure 2).

Figure 1 Bias-corrected calibration curve of the nomogram prediction model. The curve slope approaches 1, consistent with excellent agreement between predicted and observed risks for colorectal adenoma.

Figure 2 Receiver operating characteristic curves for risk factors associated with colorectal adenomatous polyps. The area under the curve for the prediction model that includes all five individual variables is 0.889, indicating excellent performance to estimate the risk of colorectal adenoma.

Table 5 Diagnostic performance of risk factors for colorectal adenomatous polyps.

Factor	AUC	SE	P value	95%CI
Age	0.663	0.040	< 0.001	0.585-0.741
Smoking	0.616	0.041	0.006	0.536-0.695
Drinking	0.632	0.041	0.002	0.552-0.712
Fatty liver	0.661	0.040	< 0.001	0.582-0.740
Gallbladder polyp	0.667	0.041	< 0.001	0.586-0.747
HbA1c	0.636	0.040	0.001	0.559-0.714
Prediction probability	0.886	0.025	< 0.001	0.841-0.938

CI: Confidence interval; AUC: Area under the curve; HbA1c: Glycosylated hemoglobin.

Clinical utility of nomogram prediction model: Decision-curve analysis indicated that the nomogram prediction model yielded the greatest net benefit within a threshold probability range of 0.1-0.9.

External validation of nomogram prediction model: Among 240 individuals selected for external validation, 90 (37.50%) and 150 (62.50%) had or had no colorectal adenomatous polyps, respectively. The results of the external validation showed that the model still maintained satisfactory predictive ability in the independent cohort, with an AUC of 0.852 indicating good discriminatory ability, and the slope of the calibration curve close to 1suggesting good agreement between the predicted probability and the actual risk. In addition, the P value of the Hosmer-Leme show test was greater than 0.05, indicating that the model had a good fit.

DISCUSSION

With ongoing changes in lifestyle and dietary habits, the global incidence of CRC continues to rise annually[19,20]. Colorectal adenomas are benign neoplasms originating from the glandular epithelium of the colorectal mucosa[21]. According to the widely accepted adenoma-carcinoma sequence theory, colorectal adenomatous polyps represent precancerous lesions of CRC, with 70%-90% of CRC cases arising from these polyps[22]. Early detection and removal of adenomatous polyps are therefore critical for the primary prevention of CRC[23]. Although the fecal immunochemical test can help reduce mortality rate of CRC by 14%, its utility for adenoma detection remains limited in clinical practice[24]. For patients who have undergone polypectomy, surveillance colonoscopy can help prevent CRC occurrence, and endoscopic screening lowers both incidence and mortality[25]. In the present study, we analyzed the endoscopic characteristics of colorectal adenomatous polyps. Most lesions were elevated or flat, measured less than 10 mm in diameter, occurred as one or two polyps per patient, and were primarily located in the left colon. These characteristics largely align with the typical benign nature of colorectal adenomatous polyps.

We also identified age, smoking, alcohol consumption, fatty liver, gallbladder polyps, and HbA1c as independent risk factors for colorectal adenomatous polyps. Several mechanisms may underlie these associations: (1) Age: International studies have consistently demonstrated an age-related increase in the prevalence of colorectal adenomas[26,27]. With advancing age, intestinal function declines and mucosa barrier integrity weakens, while chronic low-grade inflammation increases, all of which may predispose to adenoma formation[28]; (2) Smoking: Long-term smoking has been linked to a higher risk of high-risk adenomas larger than 10 mm[29]. Tobacco contains group 1 carcinogens such as benzene, nicotine, and tar, which can irritate and damage the colonic mucosa[30]; (3) Alcohol consumption: Intake exceeding 30 g of alcohol per day increases the risk of CRC[31]. Alcohol impairs folate absorption and inhibits key enzymes, leading to folate deficiency in the colon and rectum[32]. It may also suppress tumor immune surveillance, interfere with DNA repair, and activate other oncogenic pathways[33]; (4) Fatty liver: Adipose tissue secretes numerous cytokines and hormones, collectively referred to as adipokines, which have tumorigenic properties. Adipokines can modulate angiogenesis and apoptosis through various pathways, promoting colorectal tumorigenesis[34,35]. Patients with fatty liver often have metabolic issues, such as obesity, insulin resistance, chronic inflammation, and dysregulated cell death such as ferroptosis and copper-dependent cell death, which could stimulate the abnormal proliferation of colorectal mucosal cells and inhibit cell apoptosis, providing favorable environmental conditions for the formation of polyps also[36]. Fatty liver itself represents a state of chronic liver inflammation, increasing in the levels of systemic inflammatory factors and damaging the colorectal mucosal barrier to promote polyp development. In addition, fatty liver can alter the structure of the gut microbiota, leading to an increase in harmful bacteria, exacerbating the accumulation of intestinal toxins, further aggravating intestinal inflammation and metabolic disorders, and indirectly increasing the risk of polyp development[37]; (5) Gallbladder polyps: Previous studies have reported a significant association between gallbladder polyps and colorectal adenomatous polyps[38,39]. This relationship appears strongest for gallbladder polyps > 5 mm. Because the gallbladder and colonic mucosa share similar epithelial structures and are exposed to consistent risk factors, concomitant gallbladder polyps may signal an elevated risk of colorectal adenomatous polyps. In addition, these two diseases share common risk factors, such as obesity, hyperlipidemia, diabetes, and metabolic syndrome. These factors can interrupt tissue metabolism of the gallbladder and the colorectum, stimulate the abnormal proliferation of mucosal cells, cause hormonal and oxidative stress, and disrupt immune responses, predisposing affected individuals to both gallbladder polyps and colorectal tumor[40,41]. Moreover, the gallbladder is the site for bile storage and concentration. The presence of gallbladder polyps often indicates an overall abnormality in bile metabolism. Dysfunction of the liver and gallbladder can lead to changes in the synthesis, composition, and excretion of bile acids. The abnormal bile enters the intestine and can be converted into a higher proportion of secondary bile acids to have cytotoxic and genotoxic effects on the colorectal mucosal epithelium, promoting cell proliferation and creating a local micro-environment conducive to the formation of adenomas[42]. Therefore, concomitant gallbladder polyps may signal an elevated risk of colorectal adenomatous polyps; and (6) HbA1c: A cross-sectional study reported that among patients older than 40 years or 50 years, the probability of concomitant colorectal adenomatous polyps increases significantly when HbA1c levels exceed 5.44% and 4.81%, respectively[43]. Elevated HbA1c levels reflect underlying insulin resistance and hyperinsulinemia. Hyperinsulinemia enhances angiogenesis, promotes colorectal tumor growth, increases secretion of insulin-like growth factor-I and insulin-like growth factor-II), and stimulates colorectal epithelial cell proliferation while inhibiting apoptosis[44]. Based on these risk factors, we constructed a prediction model for colorectal adenomatous polyps. The model achieved an AUC of 0.889, markedly exceeding the discriminative ability of any single factor. This finding suggests that the model can be applied clinically to estimate individual risk for colorectal adenomatous polyps and guide subsequent preventive and treatment strategies.

Different from previous studies[45,46], we did not find any statistically significant difference in the family history of malignant tumors between patients with colorectal adenomatous polyps and those without colorectal adenomatous polyps, suggesting that the family history of malignant tumors does not appear to be associated with the polyp occurrence. Several factors may explain this finding: (1) The sample size of this study was relatively limited and the number of cases with a family history of malignant tumors was small (18 cases in the colorectal adenomatous polyp group and 27 cases in the non-colorectal adenomatous polyp group). This might lead to insufficient statistical power to detect the actual differences between the two groups; (2) Colorectal adenomas predispose cancer and both illnesses have family aggregation. However, the association between other types of malignant tumors and colorectal adenomas is relatively weak. In this study, the family history of malignant tumors included all types of cancers, which might dilute the influence of the specific family history on colorectal adenoma and cancer; and (3) The acquisition of family history information mainly relied on self - reported history from patients, which may have been subject to recall bias and inaccurate information, thus affecting the accuracy of the prediction model.

Compared with previous models, our prediction model has a higher AUC of 0.889, suggesting its robust and accuracy performance to assess the risk of colorectal adenomatous polyps and guide subsequent clinical treatment. The nomogram prediction model constructed in this study has the following clinical application advantages: Firstly, the six predictive variables included in the model (age, smoking, alcohol consumption, fatty liver, gallbladder polyps, and HbA1c) are all measurements routinely collected during the clinical practice. Age and history of smoking and drinking can be available during patient interviews and medical chart review. Fatty liver and gallbladder polyp can be assessed by noninvasive ultrasound examination. HbA1c can be tested in many laboratories. All these examinations are easy to obtain and low-cost, facilitating wide application in various clinical settings. Secondly, through the visual form of the nomogram, clinicians can directly calculate the total score from the point of each variable, and then estimate the risk of colorectal adenomatous polyps. The process is simple and fast. All of these ensure that our model can make a great contribution during screening individuals for colorectal adenomas.

CONCLUSION

Colorectal adenomatous polyps display diverse endoscopic characteristics, and their occurrence is influenced by age, smoking, alcohol consumption, fatty liver, gallbladder polyps, and HbA1c. A prediction model created based on these variables has the potential to reliably identify high-risk individuals and to enhance clinical outcomes through early detection, diagnosis, and intervention. Nevertheless, this study was limited by relatively low power to detect weak associations. Future research with larger sample sizes is needed to validate these findings and refine the predictive model.