Abstract
Objective
This study aimed to investigate the risk factors for colorectal sessile serrated lesions and to develop a nomogram prediction model that can assist in differentiating sessile serrated lesion from hyperplastic polyps during endoscopic and pathological diagnosis.
Methods
A total of 1628 eligible patients were included in this study. After collecting clinical information and data, they were randomly allocated into training and validation datasets at an 8:2 ratio. Univariate and multivariate logistic regression analyses were employed to identify risk factors and construct a nomogram prediction model. The model’s performance was comprehensively evaluated using receiver operating characteristic curves, calibration curves, and decision curve analysis.
Results
Multivariate logistic regression analysis indicated that age, hyperlipidemia, diarrhea, Helicobacter pylori infection, gastric neoplasm, polyp size, location, and laterally spreading morphology were all independent predictors of colorectal sessile serrated lesion. A nomogram prediction model was constructed and validated based on these variables. The predictive model demonstrated excellent discriminatory performance in both the training set and the independent validation set, with areas under the curve of 0.876 and 0.880, respectively. At the optimal diagnostic threshold of 0.687, the model achieved a specificity of 92.6% and a sensitivity of 61.7%, enabling effective identification with high specificity. The model showed good calibration, and decision curve analysis confirmed that it provides clear net clinical benefit across a wide range of threshold probabilities, indicating that this nomogram model has promising potential for clinical application.
Conclusion
This nomogram prediction model integrated eight independent predictors and demonstrated effective performance in differentiating colorectal sessile serrated lesions from hyperplastic polyps, showing favourable clinical utility and generalisation potential. It contributes to improving early detection rates, assists in optimising endoscopic resection strategies and provides a quantitative reference for pathological diagnosis, thereby facilitating more precise clinical intervention.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12876-026-04763-z.
Keywords: Colorectal cancer; Sessile serrated lesion; Hyperplastic polyp; Diagnosis, Differential; Logistic regression analysis; Nomogram; Predictive model
Introduction
Colorectal cancer (CRC) is the third most common malignancy and the second leading cause of cancer death worldwide [1]. In China alone, the 2022 age-standardized incidence and mortality rates were 30.6 and 13.9 per 100 000, respectively, underscoring its major public-health burden [2].Polyps serve as important precursors of CRC, with adenomatous polyps and serrated polyps being most closely associated with CRC development [3]. While most sporadic CRCs follow the “adenoma-carcinoma” sequence [4], serrated polyps, though not traditionally considered canonical precursor lesions, have increasingly been recognized for their malignant potential, accounting for approximately 10–15% of sporadic CRCs [5].
Serrated lesions are subdivided into hyperplastic polyp (HP), sessile serrated lesion (SSL), and traditional serrated adenoma (TSA) [6]. HPs are the most common type but rarely progress, whereas SSLs—although comprising only about 10% of serrated lesions—exhibit a significant propensity for progression to carcinoma [7]. TSAs are the rarest, but they display prominent morphological and histologic features that facilitate recognition [7, 8]. SSLs are endoscopically characterized by classic features like debris, or bubbles; however, due to their highly variable appearance, many lack specific endoscopic features [9], leading to underdiagnosis or misclassification as HPs. Furthermore, owing to technical variations in tissue sectioning and subjective interpretation, differentiating SSLs from HPs in pathological diagnosis remains challenging [10].
Therefore, this study aims to build a predictive model by comparing clinical profiles of SSL and HP patients, with the goal of providing endoscopists and pathologists with an objective and quantitative tool to improve preoperative endoscopic identification and postoperative pathological accuracy, ultimately enabling early detection and precise management of this high-risk precancerous lesion.
Methods
Patients
This study was conducted at Sir Run Run Shaw Hospital, Medicine School, Zhejiang University. We retrospectively collected data from 817 inpatients pathologically diagnosed with colorectal sessile serrated lesions between January 2019 and January 2025 as the case group, and selected 811 patients with hyperplastic polyps during the same period as the control group. The inclusion criteria were as follows: (1) age ≥ 18 years; (2) complete medical records and clinical data. The exclusion criteria were as follows: (1) incomplete colonoscopy or failure to reach the cecum; (2) coexistence of any other types of polyps (e.g., inflammatory or adenomatous polyps) in addition to the index SSL or HP; (3) patients with diagnosis of inflammatory bowel disease or intestinal infection; (4) patients with concurrent SSL and HP; (5) incomplete medical records or clinical data; (6) duplicate cases due to previous colonoscopies.
Data collection
This study identified the following candidate variables with potential discriminatory significance through a systematic review of epidemiological, endoscopic, and pathological literature on colorectal sessile serrated lesions, combined with clinical expert opinion.
The clinical data of the following participants were collected: (1) General information: age, gender, body mass index (BMI), education background, job nature, lifestyle (smoking history, drinking history), chronic disease history (hypertension, diabetes, coronary heart disease, hyperlipidemia), previous medical history (history of colorectal cancer, family history of colorectal cancer, history of other system cancers); (2) Main discomfort symptoms at the time of consultation: constipation, diarrhea, loose stools, blood in stool; (3) Laboratory tests: fecal occult blood test (FOBT), tumor markers, total cholesterol (TC) level, triglyceride (TG) level; (4) Concurrent gastroscopy conditions: Helicobacter pylori(Hp) infection, gastric polyps, gastroesophageal reflux, atrophic gastritis, neoplasm; (5) Colonoscopy conditions: the size, location, morphology, and total count of polyp, the quality of bowel preparation (assessed by the Boston Bowel Preparation Scale), and the performing endoscopist’s level of experience (categorized as junior or senior).
For all variables based on the presence or absence of a characteristic as the outcome, the coding scheme consistently assigned “No”, “Negative” or “Normal” as 0 and “Yes”, “Positive” or “Abnormal"as 1. The type of work is simply classified as physical (coded as 0) or mental (coded as 1). Polyp size was measured as the maximum diameter of the largest polyp identified during colonoscopy. This approach was chosen because the largest lesion is often the most clinically significant and is considered to carry the highest malignant potential. Polyp number was defined as the total number of polyps detected in each patient during the entire procedure. The location of polyps was defined based on the proximal (coded as 0) and distal (coded as 1) division method. The proximal colon includes the ileocecal region, cecum, ascending colon, and transverse colon; the distal colon includes the descending colon, sigmoid colon, and rectum. This study employed a hierarchical, integrated framework for endoscopic morphological assessment. First, the gross morphology of the lesions—including sessile, pedunculated, and semi-pedunculated—was recorded according to the Paris Classification. Second, the special category of laterally spreading morphology was identified. Building upon this, further observation and documentation of subtle surface features were conducted, including mucus cap, carpet-like elevation, and cloud-like appearances. The Boston Bowel Preparation Score (BBPS) is currently a common standard for evaluating the quality of intestinal preparation. If the total score of the Boston Bowel Preparation Score is ≥ 6 points, it is considered to have good intestinal preparation quality (coded as 1); if the total score is < 6 points, it is defined as insufficient intestinal preparation (coded as 0). In this study, endoscopists were stratified based on seniority level into senior and junior groups. The high-seniority group (coded as 1) comprised associate chief physicians and chief physicians, while the low-seniority group (coded as 0) consisted of residents and fellows.
Variable screening in this study adhered to the principle of “clinical guidance, statistical verification.” All candidate variables were initially proposed based on the biological mechanisms of colorectal carcinogenesis pathways, particularly the serrated pathway, and clinical diagnostic and treatment consensus. For instance, “Helicobacter pylori infection” was included based on the hypothesis that it may affect the colorectal mucosal environment through systemic chronic inflammation; “history of gastric neoplasm” was included as a surrogate indicator for susceptibility to gastrointestinal tumors. Subsequently, statistical verification was performed through univariate and multivariate analyses. This strategy ensured that the final model not only possesses predictive power but also that its variable composition has clear pathophysiological or clinical logic interpretability, avoiding a purely data-driven “black-box” model.
Pathological diagnosis
First, a pathology resident performed initial microscopic examination and diagnosis. Subsequently, the slide was submitted to a more experienced pathologist holding the title of attending physician or higher for independent systematic review. All diagnostic assessments were strictly based on the histological classification criteria and diagnostic guidelines for colorectal serrated lesions as outlined in the World Health Organization (WHO) Classification of Tumours of the Digestive System, 5th edition (2019).
All pathological slides were independently reviewed by two pathologists under double-blind conditions. If their diagnoses were consistent, the result was adopted. If there was a discrepancy between the initial and review diagnoses, the final diagnosis was recorded and reported according to the conclusion of the more senior and experienced reviewing pathologist.
Statistical analysis
All statistical analyses were performed using R software 4.4.2 and IBM SPSS Statistics 25.0. Normally distributed continuous variables were expressed as mean±standard deviation (x̄±s) and compared using independent samples t-tests. Non-normally distributed data were presented as median (interquartile range) and analyzed using Mann-Whitney U tests. Categorical variables were described as numbers (percentages) [n (%)], with between-group differences examined by χ² tests. Variables showing statistical significance (P value < 0.05) in univariate analysis were incorporated into multivariate logistic regression models to identify independent risk factors for SSLs. Based on the regression results, a nomogram prediction model was constructed. The model’s discriminative ability was evaluated using receiver operating characteristic (ROC) curve analysis, with calculation of the area under the curve (AUC) and its 95% confidence interval. The model’s calibration was assessed using bootstrap resampling (1000 repetitions) to generate calibration curves. Decision curve analysis (DCA) was additionally performed to systematically evaluate the clinical net benefit. A P value < 0.05 was considered statistically significant (Fig. 1).
Fig. 1.
The research flow chart
Results
Patient characteristics and group comparisons
The study enrolled a total of 1,628 participants, comprising 817 patients with colorectal sessile serrated lesions and 811 with hyperplastic polyps. These subjects were randomly allocated to either the training set (n = 1,303) or the validation set (n = 325) at an 8:2 ratio. As demonstrated in Table 1, comparative analysis of baseline characteristics revealed no statistically significant differences (P > 0.05) between the training and validation cohorts, confirming balanced randomization.
Table 1.
Clinical baseline characteristics in the training and validation sets and group comparisons
| Training set (n = 1303) | Validation set (n = 325) | Statistics | P value | |
|---|---|---|---|---|
| Sex (%) | ||||
| Female | 542(41.596) | 121(37.231) | χ2 = 2.054 | 0.152 |
| Male | 761(58.403) | 204(62.769) | ||
| Age, years | 59(49,67) | 59(50,66) | Z =−0.102 | 0.919 |
| BMI, kg/m2 | 23.9(21.7,26.1) | 23.9(21.8,26.4) | Z =−0.039 | 0.969 |
| Degree of education (%) | ||||
| College degree or above | 398(30.545) | 102(31.385) | ||
| High school | 571(43.822) | 155(47.692) | ||
| Primary school or illiteracy | 334(25.633) | 68(20.923) | ||
| Type of work (%) | ||||
| Intellectual work | 964(73.983) | 255(78.462) | χ2 = 2.773 | 0.096 |
| Physical work | 339(26.017) | 70(21.538) | ||
| Smoking history (%) | ||||
| Yes | 461(35.380) | 131(40.308) | χ2 = 2.730 | 0.099 |
| No | 842(64.620) | 194(59.692) | ||
| Drinking history (%) | ||||
| Yes | 333(25.556) | 89(27.385) | χ2 = 0.453 | 0.501 |
| No | 970(74.444) | 236(72.615) | ||
| Previous disease history(%) | ||||
| Hypertension | 461(35.380) | 122(37.538) | χ2 = 0.527 | 0.468 |
| Diabetes | 134(10.284) | 38(11.692) | χ2 = 0.546 | 0.460 |
| hyperlipidemia | 350(26.861) | 79(24.308) | χ2 = 1.164 | 0.281 |
| Coronary heart disease | 100(7.675) | 24(7.385) | χ2 = 0.031 | 0.860 |
| History of CRC | 104(7.982) | 21(6.462) | χ2 = 0.848 | 0.357 |
| Family history of CRC | 12(0.921) | 3(0.923) | χ2 <0.01 | 0.997 |
| Other systemic cancer | 95(7.291) | 25(7.692) | χ2 = 0.061 | 0.804 |
| Main discomfort symptoms (%) | ||||
| Constipation | 54(4.144) | 12(3.692) | χ2 = 0.137 | 0.712 |
| Diarrhea | 85(6.523) | 24(7.385) | χ2 = 0.309 | 0.578 |
| Loose stools | 59(4.528) | 15(4.615) | χ2 = 0.005 | 0.946 |
| Hemafecia | 42(3.223) | 11(3.385) | χ2 = 0.021 | 0.883 |
| FOBT (%) | ||||
| Positive | 192(14.735) | 49(15.077) | χ2 = 0.024 | 0.877 |
| Negative | 1111(85.265) | 276(84.923) | ||
| Tumor marker (%) | ||||
| Increased | 273(20.952) | 84(25.846) | χ2 = 3.640 | 0.056 |
| Normal | 1030(79.048) | 241(74.154) | ||
| Gastroscopy report (%) | ||||
| Hp infection | 167(12.817) | 42(12.923) | χ2 = 0.003 | 0.959 |
| Gastric polyps | 252(19.340) | 65(20.000) | χ2 = 0.072 | 0.788 |
| Gastroesophageal reflux | 181(13.891) | 42(12.923) | χ2 = 0.206 | 0.650 |
| Atrophic gastritis | 469(35.994) | 112(34.462) | χ2 = 0.266 | 0.606 |
| Gastric neoplasm | 56(4.298) | 18(5.538) | χ2 = 0.923 | 0.337 |
| Polyp number, count | 3(2,5) | 3(2,5) | Z =−1.034 | 0.301 |
| Polyp size, cm | 0.5(0.4,1.0) | 0.5(0.4,1.0) | Z =−0.181 | 0.857 |
| Polyp location (%) | ||||
| Distal colon | 652(50.038) | 165(50.769) | χ2 = 0.056 | 0.814 |
| Proximal colon | 651(49.962) | 160(49.231) | ||
| Endoscopic morphology of polyps (%) | ||||
| Sessile | 120(9.210) | 37(11.385) | χ2 = 1.412 | 0.235 |
| Pedunculated | 23(1.765) | 3(0.923) | χ2 = 1.174 | 0.279 |
| Subpedunculated | 21(1.612) | 6(1.846) | χ2 = 0.088 | 0.767 |
| Mucus cap | 2(0.153) | 1(0.308) | χ2 = 0.336 | 0.562 |
| Cloudy appearance | 6(0.460) | 0(0) | χ2 = 1.502 | 0.220 |
| Laterally spreading | 27(2.072) | 8(2.462) | χ2 = 0.187 | 0.665 |
| Carpet-like elevation | 29(2.226) | 5(1.538) | χ2 = 0.601 | 0.438 |
| Bowel preparation (%) | ||||
| Well-prepared | 1164(89.332) | 289(88.923) | χ2 = 0.045 | 0.831 |
| Poor-prepared | 139(10.668) | 36(11.077) | ||
| Seniority of endoscopist (%) | ||||
| High-seniority group | 689(52.878) | 171(52.615) | χ2 = 0.007 | 0.932 |
| Low-seniority group | 614(47.122) | 154(47.385) | ||
Abbreviations: BMI Body mass index, CRC Colorectal cancer, FOBT Fecal occult blood test, Hp Helicobacter pylori
Prior to inclusion in the multivariate model, we assessed the linear relationship between the continuous predictor variables and the outcome. The results indicated that all continuous predictor satisfied the linearity assumption (nonlinear P-value > 0.05); therefore, they were treated as continuous linear variables in the final model. Multicollinearity among all continuous predictors was assessed using Pearson’s correlation analysis. All pairwise correlations were weak (|r| < 0.20), far below the threshold of 0.7 for strong correlation. It is particularly noteworthy that the two key predictors, polyp size and polyp number, also demonstrated only a negligible correlation (r = 0.051). These results confirm the absence of substantial multicollinearity, and therefore all variables were included in the multivariate logistic regression model.
Certain original endoscopic morphological variables exhibited quasi-complete separation, meaning some features were overwhelmingly present in the SSL group while being nearly absent in the HP group. This phenomenon prevented the acquisition of stable and interpretable parameter estimates in the regression models. To address this issue, we synthesized these variables into novel composite variables based on clinical relevance. Conventional morphology: referring to sessile, pedunculated, or subpedunculated lesions. Predictive surface features: referring to the presence of mucous cap, and cloudy appearance, these features are highly specific for SSLs. Laterally spreading morphology, carpet-like elevation were retained as independent categories.
Risk factors and logistic regression analyses
Univariate logistic regression analysis of relevant influencing factors revealed that within the training datasets, statistically significant differences were observed between the SSLs and HPs groups in terms of age, hyperlipidemia, diarrhea symptoms, Helicobacter pylori infection, gastric neoplasm, polyp size, polyp location, and endoscopic laterally spreading morphology. In contrast, no significant differences were found in other clinical features including diabetes, hypertension, history of colorectal cancer, or other specific endoscopic features (Table 2).
Table 2.
Comparison of baseline characteristics between patients with SSLs and HPs in the training set
| Regression Coefficient (β) | Standard error (SE) | Wald | P value | Odds Ratio (OR) | 95% CI | ||
|---|---|---|---|---|---|---|---|
| Lower bound | Upper bound | ||||||
| Sex | 0.277 | 0.200 | 1.911 | 0.167 | 1.319 | 0.891 | 1.952 |
| Age | -0.016 | 0.008 | 3.970 | 0.046 | 0.984 | 0.969 | 1.000 |
| BMI | -0.033 | 0.023 | 2.069 | 0.150 | 0.967 | 0.924 | 1.012 |
| Degree of education | 0.046 | 0.213 | 0.047 | 0.829 | 1.047 | 0.689 | 1.591 |
| Type of work | -0.353 | 0.198 | 3.173 | 0.075 | 0.703 | 0.476 | 1.036 |
| Smoking history | 0.145 | 0.206 | 0.493 | 0.483 | 1.156 | 0.772 | 1.731 |
| Drinking history | 0.061 | 0.209 | 0.085 | 0.771 | 1.063 | 0.705 | 1.601 |
| Hypertension | -0.144 | 0.177 | 0.664 | 0.415 | 0.866 | 0.612 | 1.224 |
| Diabetes | -0.003 | 0.255 | 0.000 | 0.991 | 0.997 | 0.605 | 1.644 |
| Hyperlipemia | 0.591 | 0.172 | 11.829 | 0.001 | 1.807 | 1.290 | 2.531 |
| Coronary heart disease | 0.513 | 0.280 | 3.362 | 0.067 | 1.670 | 0.965 | 2.889 |
| History of CRC | 0.174 | 0.296 | 0.346 | 0.556 | 1.190 | 0.666 | 2.128 |
| Family history of CRC | 1.731 | 0.926 | 3.494 | 0.062 | 5.648 | 0.919 | 34.696 |
| Other systemic cancer | 0.420 | 0.276 | 2.304 | 0.129 | 1.521 | 0.885 | 2.616 |
| Constipation | 0.263 | 0.375 | 0.493 | 0.483 | 1.301 | 0.624 | 2.716 |
| Diarrhea | 1.313 | 0.330 | 15.848 | 0.000 | 3.717 | 1.947 | 7.094 |
| Loose stools | -0.599 | 0.378 | 2.509 | 0.113 | 0.549 | 0.262 | 1.153 |
| Hemafecia | 0.072 | 0.455 | 0.025 | 0.875 | 1.074 | 0.440 | 2.623 |
| FOBT | -0.245 | 0.230 | 1.132 | 0.287 | 0.783 | 0.498 | 1.229 |
| Tumor marker | 0.331 | 0.190 | 3.022 | 0.082 | 1.392 | 0.959 | 2.021 |
| Hp infection | 0.633 | 0.228 | 7.695 | 0.006 | 1.883 | 1.204 | 2.945 |
| Gastric polyps | 0.337 | 0.191 | 3.088 | 0.079 | 1.400 | 0.962 | 2.038 |
| Gastroesophageal reflux | 0.122 | 0.215 | 0.323 | 0.570 | 1.130 | 0.741 | 1.724 |
| Atrophic gastritis | -0.178 | 0.167 | 1.130 | 0.288 | 0.837 | 0.603 | 1.162 |
| Gastric neoplasm | 1.319 | 0.393 | 11.276 | 0.001 | 3.741 | 1.732 | 8.081 |
| Polyp number | 0.051 | 0.026 | 4.067 | 0.044 | 1.053 | 1.001 | 1.107 |
| Polyp size | 3.792 | 0.293 | 167.270 | 0.000 | 44.351 | 24.965 | 78.792 |
| Polyp location | -1.332 | 0.154 | 75.053 | 0.000 | 0.264 | 0.195 | 0.357 |
| Conventional Morphology | 0.103 | 0.270 | 0.144 | 0.704 | 1.108 | 0.652 | 1.883 |
| Predictive Surface Features | -0.047 | 1.078 | 0.002 | 0.965 | 0.954 | 0.115 | 7.893 |
| Lateral spreading | -1.704 | 0.677 | 6.332 | 0.012 | 5.477 | 1.453 | 20.639 |
| Carpet-like elevation | 1.358 | 1.212 | 1.255 | 0.263 | 3.890 | 0.361 | 41.876 |
| Bowel preparation | -0.039 | 0.246 | 0.026 | 0.873 | 0.961 | 0.594 | 1.556 |
| Seniority of endoscopist | 0.163 | 0.154 | 1.121 | 0.290 | 1.177 | 0.870 | 1.592 |
Abbreviations: BMI Body mass index, CRC Colorectal cancer, FOBT Fecal occult blood test, Hp Helicobacter pylori
The variables showing statistical significance in the univariate analysis were subsequently included in a multivariate logistic regression model. The Hosmer-Lemeshow test indicated good model fit (χ² = 11.840, P = 0.158). Multivariate analysis demonstrated that age, hyperlipidemia, diarrhea, Helicobacter pylori infection, gastric neoplasm, polyp size, polyp location, and endoscopic laterally spreading morphology were all independent influencing factors for the occurrence of SSLs (P < 0.05, Table 3). Specifically, the risk of SSLs decreased with increasing age (OR = 0.987, 95% CI: 0.976–0.999, P = 0.029), while it significantly increased with larger polyp size (OR = 46.113, 95% CI: 27.569–77.130, P < 0.01). Compared with the control group, patients with hyperlipidemia (OR = 1.681, 95% CI: 1.217–2.322, P < 0.01), diarrhea symptoms (OR = 2.715, 95% CI: 1.504–4.899, P < 0.01), Helicobacter pylori infection (OR = 1.579, 95% CI: 1.037–2.405, P = 0.033) or complicated by gastric neoplasm (OR = 3.688, 95% CI: 1.764–7.711, P < 0.01) were more likely to develop SSLs. Furthermore, polyps located in the proximal colon (OR = 0.266, 95% CI: 0.199–0.355, P < 0.01) or exhibiting a laterally spreading morphology under endoscopy (OR = 0.146, 95% CI: 0.041–0.523, P < 0.01) were also more inclined to be SSLs.
Table 3.
Multivariate logistic regression analysis in the training set
| Regression Coefficient (β) | Standard error (SE) | Wald | P value | Odds Ratio (OR) | 95% CI | ||
|---|---|---|---|---|---|---|---|
| Lower bound | Upper bound | ||||||
| Age | -0.013 | 0.006 | 4.796 | 0.029 | 0.987 | 0.976 | 0.999 |
| Hyperlipemia | 0.519 | 0.165 | 9.923 | 0.002 | 1.681 | 1.217 | 2.322 |
| Diarrhea | 0.999 | 0.301 | 10.995 | 0.001 | 2.715 | 1.504 | 4.899 |
| Hp infection | 0.457 | 0.215 | 4.525 | 0.033 | 1.579 | 1.037 | 2.405 |
| Gastric neoplasm | 1.305 | 0.376 | 12.029 | 0.001 | 3.688 | 1.764 | 7.711 |
| Polyp size | 3.831 | 0.262 | 213.077 | 0.000 | 46.113 | 27.569 | 77.130 |
| Polyp location | -1.325 | 0.148 | 80.603 | 0.000 | 0.266 | 0.199 | 0.355 |
| Lateral spreading | -1.927 | 0.652 | 8.735 | 0.003 | 6.867 | 1.914 | 24.644 |
Coding: For Hyperlipemia, Diarrhea, Hp infection Gastric neoplasm, and Lateral spreading, the coding scheme assigned “No” as 0 and “Yes” as 1; for Polyp location, proximal colon = 0, distal colon = 1
Abbreviations: Hp Helicobacter pylori
Nomogram construction and validation
Based on the aforementioned variables, a nomogram model was constructed to predict the risk of SSL occurrence, providing a visual representation of the prediction method. To use the nomogram, first locate each variable value on its corresponding axis and draw a vertical line upward to the “Points” axis to obtain the individual score for each variable. The scores for all variables are summed and projected onto the “Total Points” axis. A vertical line is then drawn downward to the “Prob of SSL” axis to determine the corresponding probability, enabling individualized risk assessment (Fig. 2).
Fig. 2.
The nomogram for predicting SSLs and differentiating them from HPs. For the variables of hyperlipidemia, gastric neoplasia, and laterally spreading morphology, the coding rule is as follows: 0 indicates "absent" or "none", and 1 indicates "present" or "detected". Polyp location is coded as 0 for proximal colon and 1 for distal colon
The discriminative ability of the model was evaluated using ROC curves. The AUC was 0.876 (95% CI: 0.857–0.895) in the training set and 0.880 (95% CI: 0.844–0.916) in the validation set, indicating good performance. The optimal diagnostic threshold for clinical application was determined as 0.687 based on Youden’s index from the validation ROC curve, which yielded a sensitivity of 61.7% and a specificity of 92.6% (Fig. 3). In calibration curve, the ideal line, depicted as a dashed diagonal, indicates perfect prediction. The solid line corresponds to the observed performance of the nomogram. Predictive accuracy improves as the solid line approaches the dashed ideal line. The x-axis represents the model-predicted event rate, while the y-axis shows the actual observed event rate. Our calibration curve showed close agreement with the diagonal line, and the Hosmer-Lemeshow test yielded a P value of 0.424, demonstrating excellent consistency between predicted probabilities and observed outcomes (Fig. 4). As shown in Fig. 5, DCA indicated that the model’s curve lay above both the “none” and “all” reference lines across a range of threshold probabilities, suggesting stable net clinical benefit. These results support the model’s strong potential for clinical application in SSL risk stratification.
Fig. 3.
ROC curves for training set (a) and validation set (b). Training set AUC: 0.876 (95% CI: 0.857 - 0.895); Validation set AUC: 0.880 (95% CI: 0.844 - 0.916)
Fig. 4.
Calibration curves for training set (a) and validation set (b)
Fig. 5.
DCA for training set (a) and validation set (b)
Discussion
In early clinical practice, serrated polyps were once simply categorized as lesions without malignant potential [11]. In 1996, researchers introduced the concept of “serrated adenomatous polyposis” based on pathological and immunohistochemical characteristics, emphasizing its distinction from conventional hyperplastic polyps [12]. With further research, molecular biological and clinical evidence increasingly indicated that serrated polyps represent not a single entity but a spectrum of lesions with varying malignant potential. Among these, sessile serrated lesions came to be regarded as critical precursor lesions in the serrated pathway to carcinogenesis, with a risk comparable to or even exceeding that of adenomas [13, 14]. It was not until 2019 that the World Health Organization (WHO) formally recommended the term “sessile serrated lesion” and established diagnostic criteria. These criteria require at least one architecturally distorted crypt, defined by the following features: horizontal growth above the muscularis mucosae, dilation of the crypt base, serrations extending into the crypt base, and asymmetrical proliferation [7, 10, 15].
According to molecular biology research, SSLs progress to CRC via the “serrated pathway,” which is initiated by the BRAF V600E mutation [16]. This mutation leads to sustained activation of the MAPK signaling pathway, predominantly manifested as the CpG island methylator phenotype-high (CIMP-H) [17–19]. Consequently, key tumor suppressor genes such as MLH1 are silenced, resulting in defective DNA repair function and high microsatellite instability (MSII-H), ultimately promoting malignant transformation [20, 21]. This pathway is distinct from the classical “adenoma-carcinoma” sequence and may proceed more rapidly and insidiously [22, 23].
Improving CRC survival rates and patient outcomes hinges on effective colonoscopic screening [2]. SSLs often present as inconspicuous, flat elevated lesions covered with mucus and exhibiting indistinct borders, which contribute to their frequent oversight during routine colonoscopy [7]. Although features such as a mucous cap or cloud-like surface have been proposed to aid in discriminating SSLs from HPs [24, 25], in practice, most SSLs closely resemble HPs under endoscopy (especially white-light endoscopy) [26]. Moreover, the subtle mucosal features of SSLs are often obscured by mucus and fecal residue [27], making differential diagnosis difficult even for experienced endoscopists [4]. While techniques such as narrow-band imaging (NBI) and computer-aided detection of polyps (CADe) have shown potential for improving the detection of colorectal SSLs, data analyses have not demonstrated statistically significant difference in recognition rates [28, 29]. To date, high-quality prospective studies on endoscopic technology for substantially enhancing SSL detection sensitivity remain lacking [26, 30].
At the histopathological level, the diagnosis of SSLs also faces challenges. Firstly, improper sectioning can obscure the crypt base, hindering accurate differentiation [10]. Secondly, the interpretation of key histological features involves considerable subjectivity, which is a major cause of low diagnostic consistency. Multiple previous studies have indicated considerable diagnostic disagreement among pathologists in classifying serrated polyps [31]. Over 25% of cases initially diagnosed as HPs were reclassified as SSLs upon review, and approximately one-quarter of cases could not be definitively distinguished between SSLs and HPs [32–34]. As a result, pathologists sometimes must rely on other information about the polyp, such as size and location, to achieve a more accurate differential diagnosis [7].
The predictive model developed in this study offers dual auxiliary value for endoscopists and pathologists. For endoscopists, the model serves as a real-time decision support tool. During or after colonoscopy, rapid risk assessment can be performed by combining endoscopic findings with readily accessible common information from the electronic medical record. This supports the implementation of more thorough and standardized resection strategies for high-risk lesions—such as using endoscopic mucosal resection instead of cold snare polypectomy—thereby effectively reducing the risk of missed diagnosis and incomplete resection. Suspicious lesions can also be flagged to enhance pathological attention. For pathologists, when facing diagnostically uncertain cases, the model’s predictive score can be used to assist in judgment by integrating medical record information and endoscopic reports. This helps compensate for the subjectivity of relying solely on morphological analysis, improves diagnostic consistency and reduce misclassification.
Our model successfully identified eight predictors observably associated with SSLs, but their effects varied in direction and strength. Specifically, polyp size emerged as the strongest predictor, with larger polyps showing a great positive correlation with SSL risk [35]. The presence of hyperlipidemia [36], laterally spreading morphology [37], and proximal colon location [23, 35]also increased the predicted probability of SSLs. These factors are highly consistent with established biological behavior and endoscopic characteristics of SSLs. Furthermore, our analysis revealed previously unreported risk factors, including the presentation of diarrhea, Helicobacter pylori infection and the presence of gastric neoplasm.
Notably, multivariate logistic regression analysis in our model showed a negative correlation between age and SSL risk, which appears to contradict previous researches. However, this discrepancy can be explained by both study design and disease characteristics. Firstly, this study aimed to differentiate SSLs from HPs rather than predict SSL occurrence in the general population. Our control group consisted of HP patients, so conclusions from earlier studies using healthy controls [38] or pure detection rate statistics [39] are not directly applicable. More importantly, multiple case-control studies have indicated that the incidence of SSLs remains relatively stable across age groups, showing no statistically significant difference in detection rates between individuals above and below 50 years of age [40–42]. So aging has not been established as a strong independent risk factor for SSL. In this context, inverse association identified in our model carries dual implications. On the one hand, it suggests that the pathogenesis of SSLs may differ from that of conventional adenomas. On the other hand, it highlights the unique diagnostic value of youth: given the presence of a serrated lesion, the younger the patient, the higher the relative probability that the lesion will ultimately be diagnosed as an SSL. That is to say, serrated polyps in younger patients warrant greater suspicion for SSL. Meanwhile, the relatively modest overall weight assigned to age, hyperlipidemia, diarrhea, and Hp infection implies their limited predictive power when considered alone. But they can primarily serve as auxiliary “modulating factors” in the comprehensive assessment, particularly in borderline cases, where they can play a dynamic and alerting role.
The innovation of this research stems from its case-control design with HPs as the control, which directly targets the essential clinical challenge of differentiating SSLs from HPs. This approach facilitated the identification of discriminative predictors and led to a diagnostic model possessing clear clinical utility. By integrating multidimensional variables, the model establishes an objective and robust comprehensive evaluation system. Its presentation as a nomogram promotes user adoption and operational practicality.
However, this study has several limitations. First, to precisely analyze the independent features of SSL, the study excluded patients with concurrent adenomas during the design phase to avoid confounding interference. Although this strategy enhanced internal validity, it also limited the direct applicability of the model in real-world clinical scenarios where polyps of mixed types coexist. Therefore, the model should be regarded as an auxiliary diagnostic tool rather than a stand-alone decision instrument, particularly in mixed-polyp scenarios. Second, the variable selection strategy employed, involving univariate screening followed by multivariate regression, has demonstrated robustness but still carries a theoretical risk of data-driven selection. Future studies with larger sample sizes could employ penalized regression methods such as LASSO for comparison and optimization. Other limitations include: the findings from this single-center retrospective analysis may be influenced by selection and information bias, the inclusion of certain pathological parameters was restricted, and some predictors are not immediately accessible. Furthermore, as a retrospective study, not all enrolled patients underwent magnifying endoscopy, which prevented the acquisition of complete detailed endoscopic feature data. So a simplified version using only endoscopic variables could be feasible in future studies, subsequent efforts could consider integrating high-definition endoscopic features through prospective design to develop and validating a “pure endoscopic model” that complements the current “integrated model.” Future research should also conduct external validation in cohorts containing mixed polyps and explore the integration of more detailed indicators and artificial intelligence (AI) analysis to facilitate the clinical translation and iterative improvement of the model.
Conclusion
The nomogram model developed in this study provides an objective and practical quantitative tool for the clinical and pathological differentiation between SSLs and HPs. By integrating multidimensional features, the model effectively enhances the detection capability of SSLs, particularly raising vigilance for high-risk lesions in younger patients. It holds promise in assisting endoscopists and pathologists in achieving more precise diagnosis and stratified management in the future.
Supplementary Information
Acknowledgements
Not applicable.
Abbreviations
- CRC
Colorectal cancer
- HP
Hyperplastic polyp
- SSL
Sessile serrated lesion
- TSA
Traditional serrated adenoma
- BMI
Body mass index
- Hp
Helicobacter pylori
- BBPS
Boston Bowel Preparation Score
- ROC
Receiver operating characteristic
- AUC
Area under the curve
- DCA
Decision curve analysis
- WHO
World Health Organization
- CIMP-H
The CpG island methylator phenotype-high
- MSII-H
High microsatellite instability
- NBI
Narrow-band imaging
- CADe
Computer-aided detection of polyps
- EMR
Endoscopic mucosal resection
- AI
Artificial intelligence
Authors’ contributions
DZ: made substantial contributions to the conception, design of the work, acquisition, analysis, interpretation of data, drafted the work. XT: acquisition. WH: substantively revised the work.
Funding
National Key Research and Development Program of China (No. 2023YFF0720400).
National Natural Science Foundation of China (No. T2293751).
Data availability
The original contributions presented in the study are included in the article. Further inquiries can be directed to the corresponding author.
Declarations
Ethics approval and consent to participate
The studies involving humans were approved by Medical Ethics Committee, Sir Run Run Shaw Hospital, Medical School, Zhejiang University. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by the participant or their legal guardians/next of kin. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable or data included in this article.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Baidoun F, Elshiwy K, Elkeraie Y, Merjaneh Z, Khoudari G, Sarmini MT, et al. Colorectal Cancer Epidemiology: Recent Trends and Impact on Outcomes. Curr Drug Targets. 2021;22:998–1009. 10.2174/1389450121999201117115717. [DOI] [PubMed] [Google Scholar]
- 2.Matsuda T, Fujimoto A, Igarashi Y. Colorectal cancer: epidemiology, risk factors, and public health strategies. Digestion. 2025;106:91–9. 10.1159/000543921. [DOI] [PubMed] [Google Scholar]
- 3.Li Q, Geng S, Luo H, Wang W, Mo Y-Q, Luo Q, et al. Signaling pathways involved in colorectal cancer: pathogenesis and targeted therapy. Signal Transduct Target Ther. 2024;9:266. 10.1038/s41392-024-01953-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Mezzapesa M, Losurdo G, Celiberto F, Rizzi S, d’Amati A, Piscitelli D, et al. Serrated colorectal lesions: an up-to-date review from histological pattern to molecular pathogenesis. Int J Mol Sci. 2022;23:4461. 10.3390/ijms23084461. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Conteduca V, Sansonno D, Russi S, Dammacco F. Precancerous colorectal lesions (review). Int J Oncol. 2013;43:973–84. 10.3892/ijo.2013.2041. [DOI] [PubMed] [Google Scholar]
- 6.Dornblaser D, Young S, Shaukat A. Colon polyps: updates in classification and management. Curr Opin Gastroenterol. 2024;40:14–20. 10.1097/MOG.0000000000000988. [DOI] [PubMed] [Google Scholar]
- 7.Hagen R, Srivastava A, Anderson JC. The serrated pathway and colorectal cancer: what the gastroenterologist should know. Expert Rev Gastroenterol Hepatol. 2025;19:593–606. 10.1080/17474124.2025.2509797. [DOI] [PubMed] [Google Scholar]
- 8.Gui H, Husson MA, Mannan R. Correlations of morphology and molecular alterations in traditional serrated adenoma. World J Gastrointest Pathophysiol. 2020;11:78–83. 10.4291/wjgp.v11.i4.78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Tadepalli US, Feihel D, Miller KM, Itzkowitz SH, Freedman JS, Kornacki S, et al. A morphologic analysis of sessile serrated polyps observed during routine colonoscopy (with video). Gastrointest Endosc. 2011;74:1360–8. 10.1016/j.gie.2011.08.008. [DOI] [PubMed] [Google Scholar]
- 10.Tran TH, Nguyen VH, Vo DT. How to pick up colorectal serrated lesions and polyps in daily histopathology practice: from terminologies to diagnostic pitfalls. World J Clin Oncol. 2024;15:1157–67. 10.5306/wjco.v15.i9.1157. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Carballal S, Balaguer F, IJspeert JEG. Serrated polyposis syndrome; epidemiology and management. Best Pract Res Clin Gastroenterol. 2022. 10.1016/j.bpg.2022.101791. 58–59:101791. [DOI] [PubMed] [Google Scholar]
- 12.Torlakovic E, Snover DC. Serrated adenomatous polyposis in humans. Gastroenterology. 1996;110:748–55. 10.1053/gast.1996.v110.pm8608884. [DOI] [PubMed] [Google Scholar]
- 13.Torlakovic E, Skovlund E, Snover DC, Torlakovic G, Nesland JM. Morphologic reappraisal of serrated colorectal polyps. Am J Surg Pathol. 2003;27:65–81. 10.1097/00000478-200301000-00008. [DOI] [PubMed] [Google Scholar]
- 14.Farris AB, Misdraji J, Srivastava A, Muzikansky A, Deshpande V, Lauwers GY, et al. Sessile serrated adenoma: challenging discrimination from other serrated colonic polyps. Am J Surg Pathol. 2008;32:30–5. 10.1097/PAS.0b013e318093e40a. [DOI] [PubMed] [Google Scholar]
- 15.Booth AL, Taggart MW, Ono Y, Gonzalez RS. From mixed hyperplastic/adenomatous polyp to sessile serrated lesion: a long and winding road for long and winding crypts. Arch Pathol Lab Med. 2021;145:1289–96. 10.5858/arpa.2020-0591-RA. [DOI] [PubMed] [Google Scholar]
- 16.Vu NTH, Le HM, Vo DT-N, Vu HA, Le NQ, Ho DDQ, et al. Prevalence, risk factors, and BRAF mutation of colorectal sessile serrated lesions among vietnamese patients. World J Clin Oncol. 2024;15:290–301. 10.5306/wjco.v15.i2.290. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Lee JA, Park HE, Yoo S-Y, Jeong S, Cho N-Y, Kang GH, et al. CpG island methylation in sessile serrated adenoma/polyp of the colorectum: implications for differential diagnosis of molecularly high-risk lesions among non-dysplastic sessile serrated adenomas/polyps. J Pathol Transl Med. 2019;53:225–35. 10.4132/jptm.2019.03.12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bettington M, Walker N, Rosty C, Brown I, Clouston A, McKeone D, et al. Clinicopathological and molecular features of sessile serrated adenomas with dysplasia or carcinoma. Gut. 2017;66:97–106. 10.1136/gutjnl-2015-310456. [DOI] [PubMed] [Google Scholar]
- 19.De Palma FDE, D’Argenio V, Pol J, Kroemer G, Maiuri MC, Salvatore F. The Molecular Hallmarks of the Serrated Pathway in Colorectal Cancer. Cancers. 2019;11:1017. 10.3390/cancers11071017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.O’Brien MJ, Zhao Q, Yang S. Colorectal serrated pathway cancers and precursors. Histopathology. 2015;66:49–65. 10.1111/his.12564. [DOI] [PubMed] [Google Scholar]
- 21.Bettington M, Walker N, Clouston A, Brown I, Leggett B, Whitehall V. The serrated pathway to colorectal carcinoma: current concepts and challenges. Histopathology. 2013;62:367–86. 10.1111/his.12055. [DOI] [PubMed] [Google Scholar]
- 22.Kim SY, Kim TI. Serrated neoplasia pathway as an alternative route of colorectal cancer carcinogenesis. Intest Res. 2018;16:358–65. 10.5217/ir.2018.16.3.358. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Tan YY, Tay GSK, Wong YJ, Li JW, Kwek ABE, Ang TL, et al. Clinical features and predictors of dysplasia in proximal sessile serrated lesions. Clin Endosc. 2021;54:578. 10.5946/ce.2020.198. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Monreal-Robles R, Jáquez-Quintana JO, Benavides-Salgado DE, González-González JA. Serrated polyps of the colon and rectum: a concise review. Rev Gastroenterol México Engl Ed. 2021;86:276–86. 10.1016/j.rgmxen.2021.06.001. [DOI] [PubMed] [Google Scholar]
- 25.Kashida H. Endoscopic diagnosis of sessile serrated polyp: a systematic review. Dig Endosc. 2019;31:16–23. 10.1111/den.13263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Murakami T, Sakamoto N, Nagahara A. Endoscopic diagnosis of sessile serrated adenoma/polyp with and without dysplasia/carcinoma. World J Gastroenterol. 2018;24:3250–9. 10.3748/wjg.v24.i29.3250. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Wu Y, Wang Y, Zhao B, Li Y, Wang W, Sun W, et al. Clinical and endoscopic characteristics of colorectal sessile serrated lesion with dysplasia: a single-center cross-sectional comparative study. J Gastrointest Oncol. 2025;16:802–10. 10.21037/jgo-2024-901. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Rex DK, Clodfelter R, Rahmani F, Fatima H, James-Stevenson TN, Tang JC, et al. Narrow-band imaging versus white light for the detection of proximal colon serrated lesions: a randomized, controlled trial. Gastrointest Endosc. 2016;83:166–71. 10.1016/j.gie.2015.03.1915. [DOI] [PubMed] [Google Scholar]
- 29.Thoufeeq M, Mohanan V, Nishad N, Toh E, Shiha MG. Variation in proximal sessile serrated lesion detection rates during non-screening colonoscopies. Cureus 17:e82317. 10.7759/cureus.82317. [DOI] [PMC free article] [PubMed]
- 30.Parikh ND, Chaptini L, Njei B, Laine L. Diagnosis of sessile serrated adenomas/polyps with image-enhanced endoscopy: a systematic review and meta-analysis. Endoscopy. 2016;48:731–9. 10.1055/s-0042-107592. [DOI] [PubMed] [Google Scholar]
- 31.Payne SR, Church TR, Wandell M, Rösch T, Osborn N, Snover D, et al. Endoscopic detection of proximal serrated lesions and pathologic identification of sessile serrated adenomas/polyps vary on the basis of center. Clin Gastroenterol Hepatol Off Clin Pract J Am Gastroenterol Assoc. 2014;12:1119–26. 10.1016/j.cgh.2013.11.034. [DOI] [PubMed] [Google Scholar]
- 32.Erichsen R, Baron JA, Hamilton-Dutoit SJ, Snover DC, Torlakovic EE, Pedersen L, et al. Increased risk of colorectal cancer development among patients with serrated polyps. Gastroenterology. 2016;150:895–e9025. 10.1053/j.gastro.2015.11.046. [DOI] [PubMed] [Google Scholar]
- 33.Li D, Doherty AR, Raju M, Liu L, Lei NY, Amsden LB, et al. Risk stratification for colorectal cancer in individuals with subtypes of serrated polyps. Gut. 2021. 10.1136/gutjnl-2021-324301. :gutjnl-2021-324301. [DOI] [PubMed] [Google Scholar]
- 34.Anderson JC, Lisovsky M, Greene MA, Hagen C, Srivastava A. Factors associated with classification of hyperplastic polyps as sessile serrated adenomas/polyps on morphologic review. J Clin Gastroenterol. 2018;52:524–9. 10.1097/MCG.0000000000000840. [DOI] [PubMed] [Google Scholar]
- 35.Chavali LB, Hu K, Sheth A, Gao N, Xiong W, Zhang L. Colorectal sessile serrated lesion with large size or synchronous neoplasm: a prospective study. Eur J Gastroenterol Hepatol. 2020;32:199–204. 10.1097/MEG.0000000000001642. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Macarie M, Bataga S, Mocan S, Pantea M, Opaschi R, Voidazan S, et al. Correlation of metabolic risk factors with sessile serrated lesions. J Gastrointest Liver Dis JGLD. 2020;29:175–9. 10.15403/jgld-507. [DOI] [PubMed] [Google Scholar]
- 37.Tan C, Qin G, Wang Q-Q, Zhou Y-C, Yao S-K. Clinicopathologic and endoscopic features of sessile serrated lesions and conventional adenomas: a large inpatient population-based study in China. Front Oncol. 2024;14:1337035. 10.3389/fonc.2024.1337035. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Huang S, Liu S, Tan F, Chen H, Chen G. Construction and validation of a risk nomogram model for colorectal sessile serrated lesions. J Int Med Res. 2025;53:3000605251337577. 10.1177/03000605251337577. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Yao B-Y, Zhang L, Wu C-X, Zheng L, Duan B-S, Xu Q-W, et al. Sessile serrated lesions: searching for the true prevalence and risk factors in China. Turk J Gastroenterol. 2024;36:15–23. 10.5152/tjg.2024.24188. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Lall V, Ismail AGM, Ayonrinde OT. Disparate age and sex distribution of sessile serrated lesions and conventional adenomas in an outpatient colonoscopy population–implications for colorectal cancer screening? Int J Colorectal Dis. 2022;37:1569–79. 10.1007/s00384-022-04191-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Yeh J-H, Lin C-W, Hsiao P-J, Perng D-S, Chen J-C, Hung K-T, et al. Prevalence and predictive factors of colorectal sessile serrated lesions in younger individuals. Endoscopy. 2024;56:494–502. 10.1055/a-2272-1911. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Kim HY, Kim SM, Seo J-H, Park E-H, Kim N, Lee DH. Age-specific prevalence of serrated lesions and their subtypes by screening colonoscopy: a retrospective study. BMC Gastroenterol. 2014;14:82. 10.1186/1471-230X-14-82. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The original contributions presented in the study are included in the article. Further inquiries can be directed to the corresponding author.