Radiotherapy-related Gastrointestinal Adverse Events in Rectal Cancer: Risk Factor Analysis and Predictive Modeling Using Clinical and Small Bowel Dosimetric Features

Radiotherapy-related Gastrointestinal Adverse Events in Rectal Cancer: Risk Factor Analysis and Predictive Modeling Using Clinical and Small Bowel Dosimetric Features | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Radiotherapy-related Gastrointestinal Adverse Events in Rectal Cancer: Risk Factor Analysis and Predictive Modeling Using Clinical and Small Bowel Dosimetric Features Qiqi Huang, Jiali Meng, Chunyan Liang, Xinling Qin, Siyi He, Weimei Huang, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7461232/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 11 You are reading this latest preprint version Abstract Purpose To identify key risk factors for radiation-induced enteritis and construct a predictive model integrating dosimetric and clinical parameters. Methods We analyzed 206 colorectal cancer patients treated with pelvic radiotherapy (2015–2020). Clinical variables, dosimetric parameters and anatomical factors were evaluated. Toxicity endpoints included acute upper/lower gastrointestinal reactions (CTCAE v5.0) and late proctitis (LENT/SOMA). Statistical analyses included logistic regression and machine learning (Random Forest/XGBoost). Results The incidences of acute upper gastrointestinal reactions, acute lower gastrointestinal reactions, and late proctitis were 33.5% (69/206), 53.9% (109/206), and 38.3% (79/206), respectively. The incidence of acute upper gastrointestinal toxicity was higher in females compared with males (50% vs. 28%, P = 0.0088). A higher fraction number (OR = 0.74, P = 0.0485) and a lower radiation dose (OR = 0.999, P = 0.0293) were associated with a reduced incidence of acute lower gastrointestinal reactions.. For late proctitis, Vsmall intestine was a significant risk factor (AUC = 0.638, P = 0.006), while tumor-to-anal verge distance was protective (AUC = 0.583, P = 0.058). Machine learning models showed superior performance: random forest achieved an AUC of 0.72 for acute upper gastrointestinal reactions, XGBoost/AUC = 0.64 for acute lower gastrointestinal reactions, and LightGBM/AUC = 0.76 for late proctitis, outperforming conventional logistic regression (AUC range: 0.49–0.57 for acute endpoints; 0.57 for late proctitis). Conclusion Gender, fraction number, Vsmall intestine, and tumor-to-anal verge distance are associated with an increased risk of radiation-induced gastrointestinal toxicities. Machine learning models, particularly LightGBM for late proctitis (AUC = 0.76) and random forest for acute reactions (AUC = 0.72), provide robust tools for risk stratification. These findings may support gender-specific care, moderate hypofractionation, and stringent small bowel dose constraints to help optimize personalized radiotherapy. Rectal cancer radiotherapy gastrointestinal toxicity dosimetric predictors late proctitis small intestine volume Figures Figure 1 Figure 2 Figure 3 Introduction Colorectal cancer (CRC) ranks as the third most commonly diagnosed malignancy worldwide, with rectal tumors accounting for approximately 30% of cases [ 1 ] . Radiotherapy (RT) is increasingly employed in rectal cancer patients, either as neoadjuvant therapy for locally advanced disease or as palliative treatment for metastases [ 2 ] . While technological advancements such as intensity-modulated radiotherapy (IMRT) and volumetric modulated arc therapy (VMAT) have optimized tumor targeting and reduced radiation exposure to healthy tissues—combined with multimodal strategies that allow lower total doses—gastrointestinal (GI) toxicities remain a major challenge. Notably, a substantial proportion of patients receiving pelvic radiotherapy experience acute reactions (e.g., nausea, vomiting, or diarrhea during or within 3 months post-treatment) or late complications (e.g., proctitis or intestinal obstruction after 3 months). These adverse effects can significantly impair quality of life and often necessitate treatment interruption [ 3 , 4 ] . Although several predictive models for gastrointestinal toxicities have been proposed, such as studies focusing on dosimetric predictors for radiation proctitis or incorporating clinical factors to estimate enteritis risk [ 5 , 6 ] , these models still exhibit notable limitations: (1) a narrow focus on isolated toxicity endpoints rather than composite outcomes, (2) reliance on single-center datasets with poor generalizability, and (3) inadequate integration of dosimetric-anatomical correlations, despite evidence that toxicity risk arises from complex interplay between treatment parameters (e.g., dose-volume metrics) and patient-specific factors (e.g., bowel motility or comorbidities) [ 5 – 7 ] . Moreover, most existing models are based on single-center, small-sample data, limiting their external validation and clinical applicability. They also often fail to comprehensively consider the interaction between anatomical parameters and clinical factors, resulting in insufficient predictive accuracy [ 6 ] . Clinical observations have revealed significant differences in radiosensitivity among different parts of the gastrointestinal tract, with the small intestine, colon, and rectum exhibiting distinct radiation tolerance [ 8 ] . These anatomical differences suggest the need to establish separate predictive models for gastrointestinal reactions in different regions. Furthermore, the pathophysiological mechanism of radiation-induced gastrointestinal reactions is complex, involving multiple molecular pathways and cellular damage processes. Acute reactions are primarily associated with intestinal mucosal epithelial cell injury and inflammatory responses, while late reactions are closely linked to progressive vascular damage and fibrosis [ 9 – 11 ] . The complexity of this pathological process determines the diversity of its predictive factors. In clinical practice, accurate prediction of the risk of radiation-induced gastrointestinal reactions is of great value. On one hand, it helps identify high-risk patients and implement targeted preventive measures; on the other hand, it can provide a basis for formulating personalized radiotherapy regimens, minimizing the risk of adverse effects while ensuring tumor control [ 12 ] . However, existing predictive tools are still inadequate in terms of accuracy and practicality. They often fail to fully utilize detailed dose-volume parameters provided by modern radiotherapy planning systems and lack comprehensive assessment of multiple types of gastrointestinal reactions, making them difficult to meet actual clinical needs [ 13 ] . To address these limitations, this study simultaneously analyzes three outcome indicators—acute upper gastrointestinal reactions, acute lower gastrointestinal reactions, and late radiation proctitis—to provide more comprehensive predictive information. We focus particularly on small bowel dosimetry due to its high radiosensitivity and physiological mobility, which lead to considerable inter- and intra-fractional positional variations during radiotherapy [ 14 ] . In contrast, the colon is anatomically more fixed and less prone to positional changes, and its radiation tolerance dose is generally higher than that of the small bowel [ 15 ] . Therefore, small bowel dose parameters may serve as more sensitive and early indicators of GI toxicity. Our study comprehensively assesses multi-dimensional factors including treatment parameters, anatomical parameters, and patient characteristics to establish a more accurate predictive model, and employs multiple statistical analysis methods (including correlation analysis, logistic regression, and receiver operating characteristic (ROC) curve analysis) to systematically evaluate the predictive value of each factor. The present study aims to identify factors potentially associated with adverse reactions in rectal cancer patients undergoing radiotherapy and to construct a predictive model for estimating the risk of these toxicities. Our findings indicate that several dosimetric and clinical parameters show significant correlation with gastrointestinal reactions, and the established model demonstrates promising predictive performance. In the era of precision medicine, multifactor-based predictive models are expected to become valuable instruments for preventing and managing radiation-induced injuries [ 13 , 16 – 19 ] . This study contributes to this evolving field by providing new insights and methodologies for the precise prediction, prevention, and management of radiation-induced gastrointestinal reactions. Materials and Methods 1. Patient Selection and Data Collection This retrospective study included 206 colorectal cancer patients who underwent pelvic radiotherapy between January 2015 and December 2020. The inclusion criteria were: (1) histologically confirmed colorectal cancer, (2) receipt of curative-intent radiotherapy, and (3) availability of complete clinical and dosimetric records. Patients were excluded if they had pre-existing inflammatory bowel disease, or insufficient follow-up (< 6 months). This study was approved by the Ethics Committee of the First Affiliated Hospital, Guangxi Medical University (Approval No. ZWSCILWTGQSHB-2025-1722), and the requirement for informed consent was waived due to the retrospective nature of the study. Baseline demographics (age, gender, BMI), tumor characteristics (location, TNM stage), and treatment parameters (total dose, fractionation, technique) were collected. Dosimetric variables included small intestine V5–V45, maximum dose (Dmax), mean dose, and tumor-to-anal-verge distance. Toxicity was graded per Common Terminology Criteria for Adverse Events (CTCAE v5.0; acute, ≤ 3 months) and Late Effects Normal Tissues/Subjective, Objective, Management, Analytic (LENT/SOMA; late, > 3 months), Patients were stratified into three groups based on toxicity grading: acute upper gastrointestinal reaction group (Acute Upper GI toxicity), acute lower gastrointestinal reaction group (Acute Lower GI toxicity), and late radiation proctitis group (Late radiation proctitis). The data processing pipeline is illustrated in Fig. 1 . The workflow begins with patient selection and data collection of clinical and dosimetric variables. All variables are first analyzed by univariate logistic regression, followed by multivariate logistic regression. Variables with P < 0.05 in univariate analysis are further evaluated by ROC analysis. In parallel, SHAP analysis ranks all features, with the top 10 features used for machine learning model development (logistic regression, random forest, XGBoost, LightGBM). Model performance is primarily assessed by AUC, ensuring robust predictive evaluation. 2. Feature Correlation Analysis All continuous variables underwent normality assessment using the Shapiro-Wilk test. Normally distributed variables were expressed as mean ± standard deviation, while non-normally distributed variables were presented as median with interquartile range (IQR). The associations between continuous predictors and binary outcomes were examined using point-biserial correlation coefficients. For categorical variables, relationships with outcomes were evaluated through χ² tests or Fisher's exact tests, as appropriate based on expected cell frequencies. In all analyses, outcome presence was coded as 1 and absence as 0. 3. Univariate and Multivariate Logistic Regression 3.1 Univariate Analysis All independent variables were first subjected to univariate logistic regression to evaluate their individual associations with the clinical outcomes. This step provided an initial assessment of the potential predictive value of each variable. 3.2 Multivariable Model Development Subsequently, all variables were incorporated into a multivariable logistic regression model. A backward elimination procedure was applied, with a retention threshold of P 0.10, in order to achieve a parsimonious model while retaining relevant predictors. This approach enabled assessment of the joint effects of variables while accounting for potential confounding factors. 4. Receiver Operating Characteristic (ROC) Analysis Variables that demonstrated statistical significance (P < 0.05) in the univariate logistic regression were further evaluated using ROC curve analysis. The area under the ROC curve (AUC) was calculated as the primary metric of predictive performance, with values closer to 1.0 indicating stronger discriminative ability. This approach allowed for a more detailed assessment of the predictive value of individual significant variables. 5. Feature Selection Using SHapley Additive exPlanations (SHAP) All candidate variables were analyzed using SHAP, with SHAP values derived from eXtreme Gradient Boosting (XGBoost) models to quantify feature importance. Based on the mean absolute SHAP values, the top 10 ranked predictors were identified and subsequently incorporated into the development of predictive models. 6. Predictive Modeling and Evaluation Four machine learning algorithms—logistic regression, random forest, XGBoost, and LightGBM—were developed using the top 10 features identified through SHAP analysis. Each model was trained and tested, and their predictive performance was primarily assessed by the area under the ROC curve (AUC), providing a quantitative measure of discriminative ability. 7. Statistical Analysis Analyses used SPSS 26.0 and Python 3.11. Two-tailed p-values < 0.05 were significant (* P < 0.05, ** P < 0.01, *** P < 0.001). Results 1. Patient Characteristics and Distribution of Gastrointestinal Reactions A total of 206 patients with colorectal cancer were enrolled in this study. As shown in Table 1 and Supplementary Tables 1–3, the incidences of acute upper gastrointestinal reactions, acute lower gastrointestinal reactions, and late radiation proctitis were 33.5% (69/206), 53.9% (109/206), and 38.3% (79/206), respectively. The corresponding incidences among female patients were 50%, 62%, and 41%, whereas those among male patients were 28%, 51%, and 38%, indicating higher rates in females.. Table 1 Incidence of Gastrointestinal Toxicities and Sex-Based Distribution Outcome N Incidence (%) Male (%) Female (%) Acute upper GI toxicity 69 33.50% 28% 50% Acute lower GI toxicity 109 53.90% 51% 62% Late radiation proctitis 79 38.30% 38% 41% 2. Correlation Between Clinical/Dosimetric Features and Gastrointestinal Toxicities Gender was significantly associated with the incidence of acute upper gastrointestinal reactions (P = 0.0088). Radiation dose showed a negative correlation with the incidence of acute lower gastrointestinal reactions (r = − 0.16, P = 0.0233). The incidence of late radiation proctitis was positively correlated with the irradiated volume of the small intestine (Vsmall intestine) (r = 0.2108, P = 0.0073) and negatively correlated with the distance from the tumor to the anal verge (r = − 0.1775, P = 0.0181). Detailed results are presented in Table 2 , and the P values for all patient characteristics are provided in Supplementary Tables 1–3. Table 2 Correlations between clinical variables/Dosimetric Features and the incidence of acute and late radiation proctitis Outcome Variable Correlation Incidence (%) Mean (Event = 1) Mean (Event = 0) P -value Acute upper GI toxicity Gender (Male vs. Female) - Male: 59.4% Female: 40.6% 0.0088 ** Acute lower GI toxicity Radiation dose (cGy) -0.16 5222.48 5347.83 0.0233 * Late radiation proctitis Vsmall intestine (cc) 0.2108 775.44 509.58 0.0073 ** Distance from the anal verge (cm) -0.1775 4.96 5.8 0.0181 * Note: 1 indicates event occurrence 3. Univariate Logistic Regression and Multivariable Logistic Regression Analysis In the univariate logistic regression (ULR) analysis, although the P value did not reach statistical significance (P = 0.0838), the incidence of acute upper gastrointestinal reactions showed a trend toward association with the maximum irradiation dose to the small intestine. A higher fraction number (OR = 0.74, 95% CI: 0.549–0.998, P = 0.0485) and a lower radiation dose (OR = 0.999, 95% CI: 0.998–1.000, P = 0.0293) were associated with a reduced risk of acute lower gastrointestinal reactions. For late radiation proctitis, a larger irradiated small intestine volume (Vsmall intestine) was identified as a risk factor (OR = 1.001, 95% CI: 1.000–1.002, P = 0.0113), whereas a greater distance from the anal verge appeared protective (OR = 0.842, 95% CI: 0.727–0.974, P = 0.0207) (Table 3 ). Table 3 Univariate analysis of gastrointestinal toxicity predictors Outcome Variable OR CI_lower CI_upper P -value Acute upper GI toxicity small intestine Dmax 1 0.999 1 0.0838 Acute lower GI toxicity Fraction number 0.74 0.549 0.998 0.0485 * Radiation dose 0.999 0.998 1 0.0293 * Late radiation proctitis Vsmall intestine 1.001 1 1.002 0.0113 * Distance from the anal verge 0.842 0.727 0.974 0.0207 * All variables were included in the multivariable logistic regression (MLR) model, regardless of their significance in univariate analysis, to ensure comprehensive adjustment for potential confounders. While no independent risk factors were identified in the final MLR model (all P ≥ 0.05), Table 4 presents factors demonstrating marginal significance ( P < 0.1). Specifically, acute upper gastrointestinal reactions showed borderline associations with irradiated small bowel volume (V10cc-V50cc) and tumor length. Both acute lower gastrointestinal reactions and late radiation proctitis appeared related to total small bowel volume and small intestine_V30cc. Table 4 Significant predictors of gastrointestinal toxicities in multivariable analysis Outcome Variable P- value Acute upper GI toxicity Smal lintestine_V50cc 0.069 Small intestine_V20cc 0.088 Small intestine_V10cc 0.093 tumor length 0.093 Acute lower GI toxicity Vsmall intestine 0.096 Small intestine_V30cc 0.097 Late radiation proctitis Vsmall intestine 0.096 Small intestine_V30cc 0.097 4. Discriminatory Ability of Significant Features (ROC Analysis) Since no independent risk factors were identified in the final multivariable logistic regression model (all P ≥ 0.05), we performed ROC analysis on variables with P < 0.05 from univariate analysis. The Vsmall intestine was a significant predictor for the development of late radiation proctitis following radiotherapy, with an AUC of 0.638 ( P = 0.006), the sensitivity and specificity of this predictor were 68.5% and 61.2%, respectively (Table 5 ). In contrast, other variables such as the maximum dose to the small intestine (small intestine Dmax) (AUC = 0.522, P = 0.602), the fraction number (AUC = 0.521, P = 0.308) and radiation dose (AUC = 0.562, P = 0.12) did not show significant predictive value. Additionally, the distance of tumer from the anal verge exhibited a marginal predictive trend for late radiation proctitis (AUC = 0.583, P = 0.058). The ROC curve for Vsmall intestine, small intestine Dmax, the fraction number, radiation dose, the distance of tumer from the anal verge is illustrated in Fig. 2 . These findings suggest that Vsmall intestine may serve as a useful dosimetric parameter for assessing the risk of late proctitis in patients undergoing radiotherapy. Table 5 ROC analysis of predictive factors for acute and late gastrointestinal toxicities Target Variable AUC P - value Acute upper GI toxicity Small intestine Dmax 0.522036337 0.602 Acute lower GI toxicity Fraction number 0.520550698 0.308 Radiation dose 0.561655322 0.12 Late radiation proctitis Vsmall intestine 0.638283143 0.006 ** Distance from the anal verge 0.582863661 0.058 5. SHAP-Based Feature Selection To develop clinically applicable prediction models, we analyzed three clinical outcomes along with all associated independent variables. The SHAP (SHapley Additive exPlanations) algorithm was applied to quantify the relative contribution of each feature to model prediction and to establish their importance ranking. The top 10 most influential features for each outcome were identified, and the corresponding SHAP-derived importance coefficients are summarized in Table 6 . Table 6 Feature importance scores Feature Importance Acute upper GI toxicity Acute lower GI toxicity Late radiation proctitis Age 0.076944536 0.057365568 0.07637748 Gender 0.017450645 0.017811126 0.007691761 diagnosis 0.003345289 0.031966019 0.002909093 tumor length 0.051676975 0.061859894 0.043795084 T 0.015732347 0.017469137 0.016789894 N 0.011508084 0.020889569 0.006966925 Distance from the anal verge 0.064223713 0.047162697 0.050465991 Radiation dose 0.021150435 0.030587085 0.017130751 Fraction number 0.027875196 0.004471481 0.004028049 Beams number 0.00519017 0.009205198 0.005689738 Vsmallintestine 0.0502723 0.064452136 0.074505525 smallintestineDmax 0.06140331 0.058077671 0.03901494 smallintestineDmean 0.056778825 0.066267478 0.039331547 smallintestine_V10% 0.064881243 0.055227158 0.05425022 smallintestine_V20% 0.059158727 0.05682254 0.035975262 smallintestine_V30% 0.061939084 0.057880534 0.041564078 smallintestine_V40% 0.064507682 0.052539495 0.045141976 smallintestine_V45% 0.057351498 0.04194902 0.038446466 smallintestine_V10cc 0.040853897 0.05772681 0.048539133 smallintestine_V20cc 0.039764882 0.051735429 0.039208383 smallintestine_V30cc 0.047808232 0.042364449 0.035047878 smallintestine_V40cc 0.046576064 0.049141464 0.04732744 smallintestine_V45cc 0.053606867 0.047028044 0.047214372 6. Predictive Model Performance Subsequently, we constructed predictive models using multiple machine learning algorithms, including logistic regression, random forest, XGBoost, and LightGBM, trained on the selected top 10 features. As illustrated in Fig. 3 , the ROC curves demonstrate that for the outcome of acute upper GI toxicity, the random forest algorithm exhibited superior predictive performance, achieving an AUC of 0.72. Similarly, for acute lower GI toxicity, random forest again showed the highest discriminative ability with an AUC of 0.64. In contrast, for late radiation proctitis, logistic regression achieved the best performance, with an AUC of 0.79. These findings suggest that different machine learning algorithms may be more suitable for specific clinical outcomes, and that the predictive value of the selected features can be effectively leveraged to improve risk stratification and facilitate personalized treatment planning. Discussion Our study provides a comprehensive analysis of clinical and dosimetric predictors for radiotherapy-induced GI toxicity in rectal cancer patients, addressing critical gaps identified in prior research [ 5 – 7 ] . By systematically evaluating three distinct toxicity endpoints through multivariable and machine learning approaches, we offer clinically actionable insights while highlighting persistent challenges in toxicity prediction. The significantly higher acute upper GI toxicity in female patients (50% vs 28%, P = 0.0088) aligns with emerging evidence of estrogen-mediated mucosal sensitivity [ 20 ] , suggesting the need for gender-specific antiemetic protocols as recommended by National Comprehensive Cancer Network (NCCN) guidelines [ 21 ] . Our finding that increased fraction number reduced acute lower GI risk (OR = 0.74, P = 0.0485) supports the ongoing shift toward moderate hypofractionation (e.g., 25–28 fractions) [ 22 , 23 ] , though the lack of association with late toxicity reinforces the distinct pathophysiology of chronic radiation injury [ 24 ] . Notably, the analysis revealed no statistically significant association between total radiotherapy dose (50–56 Gy) and acute upper gastrointestinal toxicity (Acute Upper GI toxicity: P = 0.1269) or late proctitis (Long: P = 0.4596). Although a marginal inverse correlation was observed for acute lower gastrointestinal toxicity (Acute Lower GI toxicity: P = 0.0233, with lower doses paradoxically linked to higher toxicity), this finding may reflect confounding factors such as treatment fractionation or patient heterogeneity rather than a true dose-effect relationship. Critically, the absence of dose-dependent toxicity for Acute Upper GI toxicity and Long endpoints supports the safety of dose escalation to 56 Gy for tumor targets. This approach enhances local tumor control rates without elevating the risk of clinically significant gastrointestinal toxicities, thereby providing survival benefits. The dosimetric parameter Vsmall intestine emerged as the most robust predictor of late proctitis (AUC = 0.638, P = 0.006), with our SHAP analysis confirming its dominance over clinical factors (importance score = 0.074). This refines existing Quantitative Analyses of Normal Tissue Effects in the Clinic (QUANTEC) guidelines [ 25 ] by suggesting a stricter < 30cc threshold for modern IMRT techniques, achievable through prone positioning [ 26 ] or adaptive planning [ 27 ] . The inverse correlation between anal verge distance and late toxicity (r = -0.1775, P = 0.0181) further emphasizes the importance of anatomical considerations, consistent with EMBRACE-II findings [ 28 ] . Our dual statistical/machine learning (ML) approach revealed critical insights: while logistic regression identified conventional dose-volume relationships (Table 3 ), tree-based models uncovered nonlinear interactions - particularly LightGBM's detection of prior acute toxicity as a late effect predictor (importance = 0.133), supporting the "consequential damage" hypothesis [ 7 ] . The AUC improvement from 0.63 (logistic regression) to 0.72 (Random Forest) aligns with trends reported in similar studies [ 29 ] . Three key opportunities emerge: (1) validation of our < 30cc small bowel constraint in prospective cohorts, (2) development of gender-adapted supportive care protocols, and (3) integration of biological markers with dosimetric data. The unexpected superiority of logistic regression for late toxicity prediction (AUC = 0.786) suggests that conventional models retain value when combined with judicious feature selection. Several limitations warrant consideration. First, the retrospective single-center design with a limited sample size may introduce selection bias and reduce statistical power, thereby limiting the generalizability of our findings [ 30 ] . Second, the absence of biological markers (e.g., TGF-β polymorphisms [ 31 ] or fecal microbiota [ 32 ] ) limits mechanistic insights. Third, dichotomizing CTCAE grades may obscure subtle dose-response relationships [ 33 ] . These gaps highlight the need for prospective trials integrating radiogenomics [ 34 ] and patient-reported outcomes [ 35 ] . Conclusion This study establishes a triad of actionable interventions: gender-specific supportive care, moderate hypofractionation, and stringent small bowel constraints (< 30cc). While dosimetry explains most toxicity variance, achieving precision radiotherapy requires combining these measures with emerging biomarkers [39,40]. Our findings contribute to the future development of a template for personalized risk mitigation and highlight the development of biologically-informed predictive models as a critical next step. Abbreviations CRC Colorectal cancer RT Radiotherapy I MRT Intensity-modulated radiotherapy VMAT volumetric modulated arc therapy GI Gastrointestinal ROC receiver operating characteristic CTCAE Common Terminology Criteria for Adverse Events LENT/SOMA Late Effects Normal Tissues/Subjective, Objective, Management, Analytic IQR interquartile range AUC The area under the ROC curve SHAP SHapley Additive exPlanations XGBoost the eXtreme Gradient Boosting NCCN National Comprehensive Cancer Network QUANTEC Quantitative Analyses of Normal Tissue Effects in the Clinic ML Machine Learning Declarations Ethics approval and consent to participate This study was approved by the Ethics Committee of the First Affiliated Hospital, Guangxi Medical University (Approval No. 2025-E0675), and the requirement for informed consent was waived due to the retrospective nature of the study. Consent for publication Not Applicable. Availability of data and materials The datasets generated and analyzed during the current study are not publicly available due to patient privacy and ethical restrictions imposed by the Ethics Committee of the First Affiliated Hospital of Guangxi Medical University. However, de-identified data may be made available from the corresponding author (Shanshan Ma) upon reasonable request and with permission of the institutional ethics committee. Competing Interests The author reports no conflicts of interest in this work. Funding This study was supported by grants from the National Natural Science Foundation of China (Grant No. 82102710), the Science and Technology Department of Guangxi Zhuang Autonomous Region (Grant No. 2023GXNSFBA026007), the Health Commission of Guangxi Zhuang Autonomous Region (Grant No. Z-A20230491), and the Education Department of Guangxi Zhuang Autonomous Region (Grant No. 2025KY0159). Authors' contributions Qiqi Huang, Jiali Meng, Chunyan Liang contribute equally to the article. They conducted the majority of the data collection, analyzed the data, and drafted the manuscript. Xinling Qin contributed to the data analysis and interpretation. Siyi He provided the analysis tools and were also involved in revising the manuscript critically for important intellectual content. Shanshan Ma, Weimei Huang, the co-corresponding authors, made substantial contributions to the conception and design of the work, and they also revised the manuscript and gave final approval of the version to be published. All authors read and approved the final manuscript. Acknowledgements The authors would like to express their sincere gratitude to the National Natural Science Foundation of China, the Science and Technology Department of Guangxi Zhuang Autonomous Region, the Health Commission of Guangxi Zhuang Autonomous Region, and the Education Department of Guangxi Zhuang Autonomous Region for their generous financial support through the respective grants, which laid a solid foundation for the smooth implementation of this study—including the collection of research samples, the purchase of experimental reagents, and the analysis of key data. References MORGAN E, ARNOLD M, GINI A, et al. Global burden of colorectal cancer in 2020 and 2040: incidence and mortality estimates from GLOBOCAN [J]. Gut, 2023, 72(2): 338-44. YANG Y, PANG K, LIN G, et al. Neoadjuvant chemoradiation with or without PD-1 blockade in locally advanced rectal cancer: a randomized phase 2 trial [J]. Nat Med, 2025, 31(2): 449-56. HOLM M O, BYE A, FALKMER U, et al. The effect of nutritional interventions in acute radiation-induced diarrhoea in patients with primary pelvic cancer: A systematic review [J]. Crit Rev Oncol Hematol, 2023, 188: 104038. SAPIENZA L G, GOMES M J L, CALSAVARA V F, et al. Does para-aortic irradiation reduce the risk of distant metastasis in advanced cervical cancer? A systematic review and meta-analysis of randomized clinical trials [J]. Gynecol Oncol, 2017, 144(2): 312-7. ZENG C, JI J, HUANG Y, et al. A Novel Scoring System to Predict Acute Radiation Enteritis Recovery in Cervical Cancer Patients Undergoing Concurrent Chemoradiotherapy: A Southwest China Cohort Study [J]. Int J Gen Med, 2024, 17: 5907-19. QIN Q, HUANG B, WU A, et al. Development and Validation of a Post-Radiotherapy Prediction Model for Bowel Dysfunction After Rectal Cancer Resection [J]. Gastroenterology, 2023, 165(6): 1430-42.e14. DE MOL VAN OTTERLOO S R, CHRISTODOULEAS J P, BLEZER E L A, et al. Patterns of Care, Tolerability, and Safety of the First Cohort of Patients Treated on a Novel High-Field MR-Linac Within the MOMENTUM Study: Initial Results From a Prospective Multi-Institutional Registry [J]. Int J Radiat Oncol Biol Phys, 2021, 111(4): 867-75. LI L, YANG Z, YI Y, et al. Gut microbiota and radiation-induced injury: mechanistic insights and microbial therapies [J]. Gut Microbes, 2025, 17(1): 2528429. O'SULLIVAN B, LEVIN W. Late radiation-related fibrosis: pathogenesis, manifestations, and current management [J]. Semin Radiat Oncol, 2003, 13(3): 274-89. TAN G, LIN C, HUANG C, et al. Radiosensitivity of colorectal cancer and radiation-induced gut damages are regulated by gasdermin E [J]. Cancer Lett, 2022, 529: 1-10. FAN J, LIN B, FAN M, et al. Research progress on the mechanism of radiation enteritis [J]. Front Oncol, 2022, 12: 888962. GE Z, CHEN C, CHEN J, et al. Gut Microbiota-Derived 3-Hydroxybutyrate Blocks GPR43-Mediated IL6 Signaling to Ameliorate Radiation Proctopathy [J]. Adv Sci (Weinh), 2024, 11(28): e2306217. ONG A L K, KNIGHT K, PANETTIERI V, et al. Dose-volume analysis of planned versus accumulated dose as a predictor for late gastrointestinal toxicity in men receiving radiotherapy for high-risk prostate cancer [J]. Physics and Imaging in Radiation Oncology, 2022, 23: 97-102. XIE L W, CAI S, ZHAO T S, et al. Green tea derivative (-)-epigallocatechin-3-gallate (EGCG) confers protection against ionizing radiation-induced intestinal epithelial cell death both in vitro and in vivo [J]. Free Radic Biol Med, 2020, 161: 175-86. BERTELLI G, PATAUNER S, GORGATTI T, et al. Right-side fixation of the sigmoid colon causing internal herniation with closed-loop obstruction of both small and large bowel: a case report and review of the literature [J]. J Med Case Rep, 2022, 16(1): 323. NIE T, CHEN Z, CAI J, et al. Integration of dosimetric parameters, clinical factors, and radiomics to predict symptomatic radiation pneumonitis in lung cancer patients undergoing combined immunotherapy and radiotherapy [J]. Radiother Oncol, 2024, 190: 110047. BAO D, ZHAO Y, LI L, et al. A MRI-based radiomics model predicting radiation-induced temporal lobe injury in nasopharyngeal carcinoma [J]. Eur Radiol, 2022, 32(10): 6910-21. ANDERSSON B, LANGEN B, LIU P, et al. Development of a machine learning framework for radiation biomarker discovery and absorbed dose prediction [J]. Front Oncol, 2023, 13: 1156009. BIJMAN R, ROSSI L, SHARFO A W, et al. Automated Radiotherapy Planning for Patient-Specific Exploration of the Trade-Off Between Tumor Dose Coverage and Predicted Radiation-Induced Toxicity-A Proof of Principle Study for Prostate Cancer [J]. Front Oncol, 2020, 10: 943. GUO Y, ZHU Y, ZHANG J, et al. Facilitation of mucosal healing by estrogen receptor β in ulcerative colitis through suppression of branched-chain amino acid transport and subsequent triggering of autophagy in colonic epithelial cells [J]. Acta Pharm Sin B, 2025, 15(1): 168-87. BENSON A B, VENOOK A P, ADAM M, et al. Colon Cancer, Version 3.2024, NCCN Clinical Practice Guidelines in Oncology [J]. J Natl Compr Canc Netw, 2024, 22(2 d). DEVLIN L, GROCUTT L, HUNTER B, et al. The in-silico feasibility of dose escalated, hypofractionated radiotherapy for rectal cancer [J]. Clin Transl Radiat Oncol, 2022, 36: 24-30. TURRI G, OSTUZZI G, VITA G, et al. Treatment of Locally Advanced Rectal Cancer in the Era of Total Neoadjuvant Therapy: A Systematic Review and Network Meta-Analysis [J]. JAMA Netw Open, 2024, 7(6): e2414702. SHURYAK I, BRENNER D J. Quantitative modeling of multigenerational effects of chronic ionizing radiation using targeted and nontargeted effects [J]. Sci Rep, 2021, 11(1): 4776. BENTZEN S M, CONSTINE L S, DEASY J O, et al. Quantitative Analyses of Normal Tissue Effects in the Clinic (QUANTEC): an introduction to the scientific issues [J]. Int J Radiat Oncol Biol Phys, 2010, 76(3 Suppl): S3-9. SCOBIOALA S, KITTEL C, NIERMANN P, et al. A treatment planning study of prone vs. supine positions for locally advanced rectal carcinoma : Comparison of 3‑dimensional conformal radiotherapy, tomotherapy, volumetric modulated arc therapy, and intensity-modulated radiotherapy [J]. Strahlenther Onkol, 2018, 194(11): 975-84. QIU Z, OLBERG S, DEN HERTOG D, et al. Online adaptive planning methods for intensity-modulated radiotherapy [J]. Phys Med Biol, 2023, 68(10). BERGER T, SEPPENWOOLDE Y, PöTTER R, et al. Importance of Technique, Target Selection, Contouring, Dose Prescription, and Dose-Planning in External Beam Radiation Therapy for Cervical Cancer: Evolution of Practice From EMBRACE-I to II [J]. Int J Radiat Oncol Biol Phys, 2019, 104(4): 885-94. BIBAULT J E, GIRAUD P, HOUSSET M, et al. Deep Learning and Radiomics predict complete response after neo-adjuvant chemoradiation for locally advanced rectal cancer [J]. Sci Rep, 2018, 8(1): 12611. HOWE C J, COLE S R, LAU B, et al. Selection Bias Due to Loss to Follow Up in Cohort Studies [J]. Epidemiology, 2016, 27(1): 91-7. WU F, WEIGEL K J, ZHOU H, et al. Paradoxical roles of TGF-β signaling in suppressing and promoting squamous cell carcinoma [J]. Acta Biochim Biophys Sin (Shanghai), 2018, 50(1): 98-105. CUI B, LUO H, HE B, et al. Gut dysbiosis conveys psychological stress to activate LRP5/β-catenin pathway promoting cancer stemness [J]. Signal Transduct Target Ther, 2025, 10(1): 79. EICHKORN T, BAUER J, BAHN E, et al. Radiation-induced contrast enhancement following proton radiotherapy for low-grade glioma depends on tumor characteristics and is rarer in children than adults [J]. Radiother Oncol, 2022, 172: 54-64. SU G H, XIAO Y, YOU C, et al. Radiogenomic-based multiomic analysis reveals imaging intratumor heterogeneity phenotypes and therapeutic targets [J]. Sci Adv, 2023, 9(40): eadf0837. MANZ C R, SCHRIVER E, FERRELL W J, et al. Association of Remote Patient-Reported Outcomes and Step Counts With Hospitalization or Death Among Patients With Advanced Cancer Undergoing Chemotherapy: Secondary Analysis of the PROStep Randomized Trial [J]. J Med Internet Res, 2024, 26: e51059. Additional Declarations No competing interests reported. Supplementary Files SupplementarymaterialV.2.0.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 17 Dec, 2025 Reviewers agreed at journal 12 Nov, 2025 Reviews received at journal 02 Nov, 2025 Reviewers agreed at journal 14 Oct, 2025 Reviewers agreed at journal 13 Oct, 2025 Reviewers agreed at journal 12 Oct, 2025 Reviewers invited by journal 12 Oct, 2025 Editor assigned by journal 10 Sep, 2025 Editor invited by journal 08 Sep, 2025 Submission checks completed at journal 06 Sep, 2025 First submitted to journal 06 Sep, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7461232","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":533957768,"identity":"ad23e7b6-d3ca-4779-badf-5d46a641a6d9","order_by":0,"name":"Qiqi Huang","email":"","orcid":"","institution":"The First Affiliated Hospital of Guangxi Medical University","correspondingAuthor":false,"prefix":"","firstName":"Qiqi","middleName":"","lastName":"Huang","suffix":""},{"id":533957770,"identity":"50da78a1-264b-44b8-8c96-fbae79010df0","order_by":1,"name":"Jiali Meng","email":"","orcid":"","institution":"The First Affiliated Hospital of Guangxi Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jiali","middleName":"","lastName":"Meng","suffix":""},{"id":533957774,"identity":"9e673b0b-abfd-4a21-a003-507bc2e8e8f1","order_by":2,"name":"Chunyan Liang","email":"","orcid":"","institution":"The First Affiliated Hospital of Guangxi Medical University","correspondingAuthor":false,"prefix":"","firstName":"Chunyan","middleName":"","lastName":"Liang","suffix":""},{"id":533957775,"identity":"815df606-fa6d-4cb2-9dbc-47f330fb36f9","order_by":3,"name":"Xinling Qin","email":"","orcid":"","institution":"The First Affiliated Hospital of Guangxi Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xinling","middleName":"","lastName":"Qin","suffix":""},{"id":533957776,"identity":"e8b68db3-c7a5-4305-a94d-25c4d515d7a9","order_by":4,"name":"Siyi He","email":"","orcid":"","institution":"The First Affiliated Hospital of Guangxi Medical University","correspondingAuthor":false,"prefix":"","firstName":"Siyi","middleName":"","lastName":"He","suffix":""},{"id":533957777,"identity":"82bf5404-cb8e-4bd5-a818-5dea0c724557","order_by":5,"name":"Weimei Huang","email":"","orcid":"","institution":"The First Affiliated Hospital of Guangxi Medical University","correspondingAuthor":false,"prefix":"","firstName":"Weimei","middleName":"","lastName":"Huang","suffix":""},{"id":533957778,"identity":"9e751ed8-3023-4715-af53-7bf34540ad48","order_by":6,"name":"Shanshan Ma","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA7UlEQVRIiWNgGAWjYHACNhCRwMDAfAAqkEC0FjaYUuK18BgQp0W3/fizBz931OXxz+75/Jo35zADP3uOAcPPHbi1mJ3JMTfsPXO4WOLO2W2WM7cdZpDseWPA2HsGj5YDOWwSvG0HEhtu5G4z+AjUYnAjx4CZsQ2PlvPPn0n+batLnH8j55lBIlCLPUEtNxLMpHnbmBM33MhhfgC2RYKgljdm0rJthxM33kgzY5y5LZ1H4syzgoO9eB2W/kzyLdBh824kP/7Mu81ajr89eeODn3i0IAM2CSDBA2IdIE4DMMV8IFblKBgFo2AUjCwAACKTWDpByHlfAAAAAElFTkSuQmCC","orcid":"","institution":"The First Affiliated Hospital of Guangxi Medical University","correspondingAuthor":true,"prefix":"","firstName":"Shanshan","middleName":"","lastName":"Ma","suffix":""}],"badges":[],"createdAt":"2025-08-26 09:23:32","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7461232/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7461232/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":94473743,"identity":"f5ee472b-2b4e-4ed0-8a13-4e1cee6383f2","added_by":"auto","created_at":"2025-10-27 15:45:27","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1166416,"visible":true,"origin":"","legend":"","description":"","filename":"articleV.2.0.docx","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/fbfa31e81d691615b2c2f554.docx"},{"id":94473664,"identity":"4ca1f4f2-1028-4350-bf9b-a5b1bbd503af","added_by":"auto","created_at":"2025-10-27 15:45:08","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9305,"visible":true,"origin":"","legend":"","description":"","filename":"4861073a5ac54619b37d9cdf38f9bb6c.json","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/8ce8b210bc8a5f258e5f0408.json"},{"id":94489753,"identity":"c502f32d-1ac2-4614-baef-0b8b6d672f66","added_by":"auto","created_at":"2025-10-27 17:05:53","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":30344,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementarymaterialV.2.0.docx","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/1f752c4f3e4c59397a776b28.docx"},{"id":94473380,"identity":"288c5c62-32ae-4a08-93d5-4e3cb075f9d6","added_by":"auto","created_at":"2025-10-27 15:44:12","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":115310,"visible":true,"origin":"","legend":"","description":"","filename":"4861073a5ac54619b37d9cdf38f9bb6c1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/c4ac712031dc0621e5f9d376.xml"},{"id":94473704,"identity":"4478706d-4ed3-4846-849e-6f262c8b9f64","added_by":"auto","created_at":"2025-10-27 15:45:18","extension":"jpeg","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1501672,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/a85efc64a9b6914dbf3e8645.jpeg"},{"id":94473739,"identity":"a64c3c66-1e4c-4929-be9e-81d577a9db44","added_by":"auto","created_at":"2025-10-27 15:45:26","extension":"jpeg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3600722,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/7420f1ca90239550a1aa3077.jpeg"},{"id":94473498,"identity":"5baa71a9-a16d-46d6-9edd-de3332e7a6cd","added_by":"auto","created_at":"2025-10-27 15:44:37","extension":"jpeg","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2478562,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/dd7846ab3962e8601415323e.jpeg"},{"id":94473438,"identity":"229a9090-f1d5-4939-8be0-2d9e6809313e","added_by":"auto","created_at":"2025-10-27 15:44:19","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":92935,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/761f72bb0951036096c63eaf.png"},{"id":94473505,"identity":"0ab058ef-3c1b-4d4c-9670-9cbdf4f08f99","added_by":"auto","created_at":"2025-10-27 15:44:39","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":49589,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/e4d232df50ff935c46fd9759.png"},{"id":94473615,"identity":"00b6e6f2-ad33-4665-a1ee-b3ed3a3946bf","added_by":"auto","created_at":"2025-10-27 15:45:03","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":52156,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/24cd10dadaa668545eb2335c.png"},{"id":94473710,"identity":"ee5ac044-19cc-4e8d-938f-986b67ec69fd","added_by":"auto","created_at":"2025-10-27 15:45:18","extension":"xml","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":114021,"visible":true,"origin":"","legend":"","description":"","filename":"4861073a5ac54619b37d9cdf38f9bb6c1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/dfe6a7e3322945eafeb067a2.xml"},{"id":94473712,"identity":"3f1f9455-2010-40a5-bbc6-ac529566309e","added_by":"auto","created_at":"2025-10-27 15:45:19","extension":"html","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":122585,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/19c78d4c63aea7f927f457cd.html"},{"id":94473775,"identity":"7340390f-66cb-4525-be7d-e22699a83966","added_by":"auto","created_at":"2025-10-27 15:45:40","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":495968,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eWorkflow of the analytical process.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe workflow begins with patient selection and data collection of clinical and dosimetric variables. All variables are first analyzed by univariate logistic regression, followed by multivariate logistic regression. Variables with P \u0026lt; 0.05 in univariate analysis are further evaluated by ROC analysis. In parallel, SHAP analysis ranks all features, with the top 10 features used for machine learning model development (logistic regression, random forest, XGBoost, LightGBM). Model performance is primarily assessed by AUC, ensuring robust predictive evaluation.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/1f6d4b8a7be4ffd9db081264.png"},{"id":94472974,"identity":"768a747d-a42f-4042-89bb-c49e17c715e7","added_by":"auto","created_at":"2025-10-27 15:42:20","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":200418,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePredictive performance of individual clinical and dosimetric factors for acute and late radiation proctitis. \u003c/strong\u003eROC curves are shown for key predictors: \u003cstrong\u003e(a)\u003c/strong\u003e Small intestine Dmax for Acute upper GI toxicity (AUC = 0.522, \u003cem\u003eP\u003c/em\u003e = 0.6020), \u003cstrong\u003e(b)\u003c/strong\u003e Radiation dose for Acute lower GI toxicity (AUC = 0.562, \u003cem\u003eP\u003c/em\u003e = 0.1200), \u003cstrong\u003e(c)\u003c/strong\u003e Number of radiotherapy sessions for Acute lower GI toxicity (AUC = 0.521, \u003cem\u003eP\u003c/em\u003e = 0.3080), \u003cstrong\u003e(d)\u003c/strong\u003e Vsmall intestine for \u003cstrong\u003elate radiation proctitis\u003c/strong\u003e (AUC = 0.638, \u003cem\u003eP\u003c/em\u003e = 0.0060), and \u003cstrong\u003e(e)\u003c/strong\u003e Distance from the anal verge for \u003cstrong\u003elate radiation proctitis\u003c/strong\u003e (AUC = 0.583, \u003cem\u003eP\u003c/em\u003e = 0.0580). The dashed line represents random guess performance. Vsmall intestine was the only statistically significant predictor (\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05) for late toxicity, while other factors showed limited discriminative ability (AUC range: 0.521–0.583).\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/c13a0f6e81f945be89d44073.png"},{"id":94473672,"identity":"4c1b0f3d-f635-4c88-8dab-8a38b97034f5","added_by":"auto","created_at":"2025-10-27 15:45:10","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":279030,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFeature importance and predictive performance for acute and late radiation proctitis.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(a) Feature importance for predicting Acute Upper GI toxicity. (b) Feature importance for predicting Acute Lower GI toxicity. (c) Feature importance for predicting late radiation proctitis. (d) ROC curves of the top 10-feature model for Acute Upper GI toxicity, with AUC values for Logistic Regression (0.53), Random Forest (0.72), XGBoost (0.62), and LightGBM (0.69). (e) ROC curves for Acute Lower GI toxicity prediction, with AUC values of 0.49 (Logistic Regression), 0.64 (Random Forest), 0.62 (XGBoost), and 0.59 (LightGBM). (f) ROC curves for late radiation proctitis prediction, demonstrating highest performance with LightGBM (AUC = 0.76), followed by Random Forest (0.69), XGBoost (0.68), and Logistic Regression (0.57). LightGBM showed superior predictive capability for late radiation proctitis, while Random Forest performed best for acute toxicity prediction.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/55505c4394eb0e85a8cebde8.png"},{"id":94491299,"identity":"d1dae8e0-1481-4a12-a085-f86440cf3226","added_by":"auto","created_at":"2025-10-27 17:24:22","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2445179,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/564dd0f2-e0b3-440f-91a6-7bade2714ed2.pdf"},{"id":94473434,"identity":"f56a3411-6481-47a4-a573-05dfe66fd731","added_by":"auto","created_at":"2025-10-27 15:44:18","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":30344,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementarymaterialV.2.0.docx","url":"https://assets-eu.researchsquare.com/files/rs-7461232/v1/8fb023eb2698c105a259a197.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Radiotherapy-related Gastrointestinal Adverse Events in Rectal Cancer: Risk Factor Analysis and Predictive Modeling Using Clinical and Small Bowel Dosimetric Features","fulltext":[{"header":"Introduction","content":"\u003cp\u003eColorectal cancer (CRC) ranks as the third most commonly diagnosed malignancy worldwide, with rectal tumors accounting for approximately 30% of cases\u003csup\u003e[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]\u003c/sup\u003e. Radiotherapy (RT) is increasingly employed in rectal cancer patients, either as neoadjuvant therapy for locally advanced disease or as palliative treatment for metastases\u003csup\u003e[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/sup\u003e. While technological advancements such as intensity-modulated radiotherapy (IMRT) and volumetric modulated arc therapy (VMAT) have optimized tumor targeting and reduced radiation exposure to healthy tissues\u0026mdash;combined with multimodal strategies that allow lower total doses\u0026mdash;gastrointestinal (GI) toxicities remain a major challenge. Notably, a substantial proportion of patients receiving pelvic radiotherapy experience acute reactions (e.g., nausea, vomiting, or diarrhea during or within 3 months post-treatment) or late complications (e.g., proctitis or intestinal obstruction after 3 months). These adverse effects can significantly impair quality of life and often necessitate treatment interruption\u003csup\u003e[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eAlthough several predictive models for gastrointestinal toxicities have been proposed, such as studies focusing on dosimetric predictors for radiation proctitis or incorporating clinical factors to estimate enteritis risk\u003csup\u003e[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/sup\u003e, these models still exhibit notable limitations: (1) a narrow focus on isolated toxicity endpoints rather than composite outcomes, (2) reliance on single-center datasets with poor generalizability, and (3) inadequate integration of dosimetric-anatomical correlations, despite evidence that toxicity risk arises from complex interplay between treatment parameters (e.g., dose-volume metrics) and patient-specific factors (e.g., bowel motility or comorbidities)\u003csup\u003e[\u003cspan additionalcitationids=\"CR6\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/sup\u003e. Moreover, most existing models are based on single-center, small-sample data, limiting their external validation and clinical applicability. They also often fail to comprehensively consider the interaction between anatomical parameters and clinical factors, resulting in insufficient predictive accuracy\u003csup\u003e[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eClinical observations have revealed significant differences in radiosensitivity among different parts of the gastrointestinal tract, with the small intestine, colon, and rectum exhibiting distinct radiation tolerance\u003csup\u003e[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]\u003c/sup\u003e. These anatomical differences suggest the need to establish separate predictive models for gastrointestinal reactions in different regions. Furthermore, the pathophysiological mechanism of radiation-induced gastrointestinal reactions is complex, involving multiple molecular pathways and cellular damage processes. Acute reactions are primarily associated with intestinal mucosal epithelial cell injury and inflammatory responses, while late reactions are closely linked to progressive vascular damage and fibrosis\u003csup\u003e[\u003cspan additionalcitationids=\"CR10\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]\u003c/sup\u003e. The complexity of this pathological process determines the diversity of its predictive factors.\u003c/p\u003e\u003cp\u003eIn clinical practice, accurate prediction of the risk of radiation-induced gastrointestinal reactions is of great value. On one hand, it helps identify high-risk patients and implement targeted preventive measures; on the other hand, it can provide a basis for formulating personalized radiotherapy regimens, minimizing the risk of adverse effects while ensuring tumor control\u003csup\u003e[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]\u003c/sup\u003e. However, existing predictive tools are still inadequate in terms of accuracy and practicality. They often fail to fully utilize detailed dose-volume parameters provided by modern radiotherapy planning systems and lack comprehensive assessment of multiple types of gastrointestinal reactions, making them difficult to meet actual clinical needs\u003csup\u003e[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eTo address these limitations, this study simultaneously analyzes three outcome indicators\u0026mdash;acute upper gastrointestinal reactions, acute lower gastrointestinal reactions, and late radiation proctitis\u0026mdash;to provide more comprehensive predictive information. We focus particularly on small bowel dosimetry due to its high radiosensitivity and physiological mobility, which lead to considerable inter- and intra-fractional positional variations during radiotherapy\u003csup\u003e[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]\u003c/sup\u003e. In contrast, the colon is anatomically more fixed and less prone to positional changes, and its radiation tolerance dose is generally higher than that of the small bowel\u003csup\u003e[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]\u003c/sup\u003e. Therefore, small bowel dose parameters may serve as more sensitive and early indicators of GI toxicity. Our study comprehensively assesses multi-dimensional factors including treatment parameters, anatomical parameters, and patient characteristics to establish a more accurate predictive model, and employs multiple statistical analysis methods (including correlation analysis, logistic regression, and receiver operating characteristic (ROC) curve analysis) to systematically evaluate the predictive value of each factor.\u003c/p\u003e\u003cp\u003eThe present study aims to identify factors potentially associated with adverse reactions in rectal cancer patients undergoing radiotherapy and to construct a predictive model for estimating the risk of these toxicities. Our findings indicate that several dosimetric and clinical parameters show significant correlation with gastrointestinal reactions, and the established model demonstrates promising predictive performance. In the era of precision medicine, multifactor-based predictive models are expected to become valuable instruments for preventing and managing radiation-induced injuries\u003csup\u003e[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan additionalcitationids=\"CR17 CR18\" citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]\u003c/sup\u003e. This study contributes to this evolving field by providing new insights and methodologies for the precise prediction, prevention, and management of radiation-induced gastrointestinal reactions.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\n\u003ch3\u003e1. Patient Selection and Data Collection\u003c/h3\u003e\n\u003cp\u003eThis retrospective study included 206 colorectal cancer patients who underwent pelvic radiotherapy between January 2015 and December 2020. The inclusion criteria were: (1) histologically confirmed colorectal cancer, (2) receipt of curative-intent radiotherapy, and (3) availability of complete clinical and dosimetric records. Patients were excluded if they had pre-existing inflammatory bowel disease, or insufficient follow-up (\u0026lt;\u0026thinsp;6 months). This study was approved by the Ethics Committee of the First Affiliated Hospital, Guangxi Medical University (Approval No. ZWSCILWTGQSHB-2025-1722), and the requirement for informed consent was waived due to the retrospective nature of the study.\u003c/p\u003e\u003cp\u003eBaseline demographics (age, gender, BMI), tumor characteristics (location, TNM stage), and treatment parameters (total dose, fractionation, technique) were collected. Dosimetric variables included small intestine V5\u0026ndash;V45, maximum dose (Dmax), mean dose, and tumor-to-anal-verge distance. Toxicity was graded per Common Terminology Criteria for Adverse Events (CTCAE v5.0; acute, \u0026le;\u0026thinsp;3 months) and Late Effects Normal Tissues/Subjective, Objective, Management, Analytic (LENT/SOMA; late, \u0026gt;\u0026thinsp;3 months), Patients were stratified into three groups based on toxicity grading: acute upper gastrointestinal reaction group (Acute Upper GI toxicity), acute lower gastrointestinal reaction group (Acute Lower GI toxicity), and late radiation proctitis group (Late radiation proctitis). The data processing pipeline is illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe workflow begins with patient selection and data collection of clinical and dosimetric variables. All variables are first analyzed by univariate logistic regression, followed by multivariate logistic regression. Variables with P\u0026thinsp;\u0026lt;\u0026thinsp;0.05 in univariate analysis are further evaluated by ROC analysis. In parallel, SHAP analysis ranks all features, with the top 10 features used for machine learning model development (logistic regression, random forest, XGBoost, LightGBM). Model performance is primarily assessed by AUC, ensuring robust predictive evaluation.\u003c/p\u003e\n\u003ch3\u003e2. Feature Correlation Analysis\u003c/h3\u003e\n\u003cp\u003eAll continuous variables underwent normality assessment using the Shapiro-Wilk test. Normally distributed variables were expressed as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation, while non-normally distributed variables were presented as median with interquartile range (IQR). The associations between continuous predictors and binary outcomes were examined using point-biserial correlation coefficients. For categorical variables, relationships with outcomes were evaluated through χ\u0026sup2; tests or Fisher's exact tests, as appropriate based on expected cell frequencies. In all analyses, outcome presence was coded as 1 and absence as 0.\u003c/p\u003e\n\u003ch3\u003e3. Univariate and Multivariate Logistic Regression\u003c/h3\u003e\n\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e3.1 Univariate Analysis\u003c/h2\u003e\u003cp\u003eAll independent variables were first subjected to univariate logistic regression to evaluate their individual associations with the clinical outcomes. This step provided an initial assessment of the potential predictive value of each variable.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e3.2 Multivariable Model Development\u003c/h2\u003e\u003cp\u003eSubsequently, all variables were incorporated into a multivariable logistic regression model. A backward elimination procedure was applied, with a retention threshold of P\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and an elimination criterion of P\u0026thinsp;\u0026gt;\u0026thinsp;0.10, in order to achieve a parsimonious model while retaining relevant predictors. This approach enabled assessment of the joint effects of variables while accounting for potential confounding factors.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003e4. Receiver Operating Characteristic (ROC) Analysis\u003c/h3\u003e\n\u003cp\u003eVariables that demonstrated statistical significance (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) in the univariate logistic regression were further evaluated using ROC curve analysis. The area under the ROC curve (AUC) was calculated as the primary metric of predictive performance, with values closer to 1.0 indicating stronger discriminative ability. This approach allowed for a more detailed assessment of the predictive value of individual significant variables.\u003c/p\u003e\n\u003ch3\u003e5. Feature Selection Using SHapley Additive exPlanations (SHAP)\u003c/h3\u003e\n\u003cp\u003eAll candidate variables were analyzed using SHAP, with SHAP values derived from eXtreme Gradient Boosting (XGBoost) models to quantify feature importance. Based on the mean absolute SHAP values, the top 10 ranked predictors were identified and subsequently incorporated into the development of predictive models.\u003c/p\u003e\n\u003ch3\u003e6. Predictive Modeling and Evaluation\u003c/h3\u003e\n\u003cp\u003eFour machine learning algorithms\u0026mdash;logistic regression, random forest, XGBoost, and LightGBM\u0026mdash;were developed using the top 10 features identified through SHAP analysis. Each model was trained and tested, and their predictive performance was primarily assessed by the area under the ROC curve (AUC), providing a quantitative measure of discriminative ability.\u003c/p\u003e\n\u003ch3\u003e7. Statistical Analysis\u003c/h3\u003e\n\u003cp\u003eAnalyses used SPSS 26.0 and Python 3.11. Two-tailed p-values\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were significant (*\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05, **\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01, ***\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e"},{"header":"Results","content":"\n\u003ch3\u003e1. Patient Characteristics and Distribution of Gastrointestinal Reactions\u003c/h3\u003e\n\u003cp\u003eA total of 206 patients with colorectal cancer were enrolled in this study. As shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and Supplementary Tables\u0026nbsp;1\u0026ndash;3, the incidences of acute upper gastrointestinal reactions, acute lower gastrointestinal reactions, and late radiation proctitis were 33.5% (69/206), 53.9% (109/206), and 38.3% (79/206), respectively. The corresponding incidences among female patients were 50%, 62%, and 41%, whereas those among male patients were 28%, 51%, and 38%, indicating higher rates in females..\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eIncidence of Gastrointestinal Toxicities and Sex-Based Distribution\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eOutcome\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eN\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eIncidence (%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eMale (%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eFemale (%)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eAcute upper GI toxicity\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e69\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e33.50%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e28%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e50%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eAcute lower GI toxicity\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e109\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e53.90%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e51%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e62%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eLate radiation proctitis\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e79\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e38.30%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e38%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e41%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\n\u003ch3\u003e2. Correlation Between Clinical/Dosimetric Features and Gastrointestinal Toxicities\u003c/h3\u003e\n\u003cp\u003eGender was significantly associated with the incidence of acute upper gastrointestinal reactions (P\u0026thinsp;=\u0026thinsp;0.0088). Radiation dose showed a negative correlation with the incidence of acute lower gastrointestinal reactions (r = \u0026minus;\u0026thinsp;0.16, P\u0026thinsp;=\u0026thinsp;0.0233). The incidence of late radiation proctitis was positively correlated with the irradiated volume of the small intestine (Vsmall intestine) (r\u0026thinsp;=\u0026thinsp;0.2108, P\u0026thinsp;=\u0026thinsp;0.0073) and negatively correlated with the distance from the tumor to the anal verge (r = \u0026minus;\u0026thinsp;0.1775, P\u0026thinsp;=\u0026thinsp;0.0181). Detailed results are presented in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, and the P values for all patient characteristics are provided in Supplementary Tables\u0026nbsp;1\u0026ndash;3.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eCorrelations between clinical variables/Dosimetric Features and the incidence of acute and late radiation proctitis\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"7\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eOutcome\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eVariable\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eCorrelation\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eIncidence (%)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eMean (Event\u0026thinsp;=\u0026thinsp;1)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eMean (Event\u0026thinsp;=\u0026thinsp;0)\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e-value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eAcute upper GI toxicity\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGender (Male vs. Female)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eMale: 59.4%\u003c/p\u003e\u003cp\u003eFemale: 40.6%\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.0088\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eAcute lower GI toxicity\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eRadiation dose (cGy)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-0.16\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e5222.48\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e5347.83\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.0233\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eLate radiation proctitis\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eVsmall intestine (cc)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.2108\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e775.44\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e509.58\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.0073\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eDistance from the anal verge (cm)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-0.1775\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e4.96\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e5.8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e\u003cp\u003e0.0181\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"7\"\u003eNote: 1 indicates event occurrence\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\n\u003ch3\u003e3. Univariate Logistic Regression and Multivariable Logistic Regression Analysis\u003c/h3\u003e\n\u003cp\u003eIn the univariate logistic regression (ULR) analysis, although the P value did not reach statistical significance (P\u0026thinsp;=\u0026thinsp;0.0838), the incidence of acute upper gastrointestinal reactions showed a trend toward association with the maximum irradiation dose to the small intestine. A higher fraction number (OR\u0026thinsp;=\u0026thinsp;0.74, 95% CI: 0.549\u0026ndash;0.998, P\u0026thinsp;=\u0026thinsp;0.0485) and a lower radiation dose (OR\u0026thinsp;=\u0026thinsp;0.999, 95% CI: 0.998\u0026ndash;1.000, P\u0026thinsp;=\u0026thinsp;0.0293) were associated with a reduced risk of acute lower gastrointestinal reactions. For late radiation proctitis, a larger irradiated small intestine volume (Vsmall intestine) was identified as a risk factor (OR\u0026thinsp;=\u0026thinsp;1.001, 95% CI: 1.000\u0026ndash;1.002, P\u0026thinsp;=\u0026thinsp;0.0113), whereas a greater distance from the anal verge appeared protective (OR\u0026thinsp;=\u0026thinsp;0.842, 95% CI: 0.727\u0026ndash;0.974, P\u0026thinsp;=\u0026thinsp;0.0207) (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eUnivariate analysis of gastrointestinal toxicity predictors\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eOutcome\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eVariable\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eOR\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eCI_lower\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eCI_upper\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e-value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eAcute upper GI toxicity\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003esmall intestine Dmax\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.999\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.0838\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eAcute lower GI toxicity\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFraction number\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.74\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.549\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.998\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.0485\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eRadiation dose\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.999\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.998\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.0293\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eLate radiation proctitis\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eVsmall intestine\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e1.001\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e1.002\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.0113\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eDistance from the anal verge\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.842\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.727\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.974\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.0207\u003csup\u003e*\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eAll variables were included in the multivariable logistic regression (MLR) model, regardless of their significance in univariate analysis, to ensure comprehensive adjustment for potential confounders. While no independent risk factors were identified in the final MLR model (all \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026ge;\u0026thinsp;0.05), Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e presents factors demonstrating marginal significance (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.1). Specifically, acute upper gastrointestinal reactions showed borderline associations with irradiated small bowel volume (V10cc-V50cc) and tumor length. Both acute lower gastrointestinal reactions and late radiation proctitis appeared related to total small bowel volume and small intestine_V30cc.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eSignificant predictors of gastrointestinal toxicities in multivariable analysis\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"3\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eOutcome\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eVariable\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cem\u003eP-\u003c/em\u003evalue\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e\u003cp\u003e\u003cb\u003eAcute upper GI toxicity\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSmal lintestine_V50cc\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.069\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSmall intestine_V20cc\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.088\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSmall intestine_V10cc\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.093\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003etumor length\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.093\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eAcute lower GI toxicity\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eVsmall intestine\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.096\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSmall intestine_V30cc\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.097\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eLate radiation proctitis\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eVsmall intestine\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.096\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSmall intestine_V30cc\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.097\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\n\u003ch3\u003e4. Discriminatory Ability of Significant Features (ROC Analysis)\u003c/h3\u003e\n\u003cp\u003eSince no independent risk factors were identified in the final multivariable logistic regression model (all \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026ge;\u0026thinsp;0.05), we performed ROC analysis on variables with P\u0026thinsp;\u0026lt;\u0026thinsp;0.05 from univariate analysis. The Vsmall intestine was a significant predictor for the development of late radiation proctitis following radiotherapy, with an AUC of 0.638 (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.006), the sensitivity and specificity of this predictor were 68.5% and 61.2%, respectively (Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). In contrast, other variables such as the maximum dose to the small intestine (small intestine Dmax) (AUC\u0026thinsp;=\u0026thinsp;0.522, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.602), the fraction number (AUC\u0026thinsp;=\u0026thinsp;0.521, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.308) and radiation dose (AUC\u0026thinsp;=\u0026thinsp;0.562, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.12) did not show significant predictive value. Additionally, the distance of tumer from the anal verge exhibited a marginal predictive trend for late radiation proctitis (AUC\u0026thinsp;=\u0026thinsp;0.583, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.058).\u003c/p\u003e\u003cp\u003eThe ROC curve for Vsmall intestine, small intestine Dmax, the fraction number, radiation dose, the distance of tumer from the anal verge is illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. These findings suggest that Vsmall intestine may serve as a useful dosimetric parameter for assessing the risk of late proctitis in patients undergoing radiotherapy.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eROC analysis of predictive factors for acute and late gastrointestinal toxicities\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTarget\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eVariable\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAUC\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cem\u003eP\u003c/em\u003e - value\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eAcute upper GI toxicity\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eSmall intestine Dmax\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.522036337\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.602\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eAcute lower GI toxicity\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFraction number\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.520550698\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.308\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eRadiation dose\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.561655322\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.12\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eLate radiation proctitis\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eVsmall intestine\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.638283143\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.006\u003csup\u003e**\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eDistance from the anal verge\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.582863661\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.058\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\n\u003ch3\u003e5. SHAP-Based Feature Selection\u003c/h3\u003e\n\u003cp\u003eTo develop clinically applicable prediction models, we analyzed three clinical outcomes along with all associated independent variables. The SHAP (SHapley Additive exPlanations) algorithm was applied to quantify the relative contribution of each feature to model prediction and to establish their importance ranking. The top 10 most influential features for each outcome were identified, and the corresponding SHAP-derived importance coefficients are summarized in Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eFeature importance scores\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003eFeature\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e\u003cp\u003eImportance\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eAcute upper GI toxicity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAcute lower GI toxicity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eLate radiation proctitis\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eAge\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.076944536\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.057365568\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.07637748\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eGender\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.017450645\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.017811126\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.007691761\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003ediagnosis\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.003345289\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.031966019\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.002909093\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003etumor length\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.051676975\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.061859894\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.043795084\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eT\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.015732347\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.017469137\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.016789894\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eN\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.011508084\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.020889569\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.006966925\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eDistance from the anal verge\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.064223713\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.047162697\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.050465991\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eRadiation dose\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.021150435\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.030587085\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.017130751\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eFraction number\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.027875196\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.004471481\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.004028049\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eBeams number\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.00519017\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.009205198\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.005689738\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eVsmallintestine\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.0502723\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.064452136\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.074505525\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003esmallintestineDmax\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.06140331\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.058077671\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.03901494\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003esmallintestineDmean\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.056778825\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.066267478\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.039331547\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003esmallintestine_V10%\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.064881243\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.055227158\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.05425022\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003esmallintestine_V20%\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.059158727\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.05682254\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.035975262\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003esmallintestine_V30%\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.061939084\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.057880534\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.041564078\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003esmallintestine_V40%\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.064507682\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.052539495\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.045141976\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003esmallintestine_V45%\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.057351498\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.04194902\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.038446466\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003esmallintestine_V10cc\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.040853897\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.05772681\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.048539133\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003esmallintestine_V20cc\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.039764882\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.051735429\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.039208383\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003esmallintestine_V30cc\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.047808232\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.042364449\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.035047878\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003esmallintestine_V40cc\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.046576064\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.049141464\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.04732744\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003esmallintestine_V45cc\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e0.053606867\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.047028044\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.047214372\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\n\u003ch3\u003e6. Predictive Model Performance\u003c/h3\u003e\n\u003cp\u003eSubsequently, we constructed predictive models using multiple machine learning algorithms, including logistic regression, random forest, XGBoost, and LightGBM, trained on the selected top 10 features. As illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, the ROC curves demonstrate that for the outcome of acute upper GI toxicity, the random forest algorithm exhibited superior predictive performance, achieving an AUC of 0.72. Similarly, for acute lower GI toxicity, random forest again showed the highest discriminative ability with an AUC of 0.64. In contrast, for late radiation proctitis, logistic regression achieved the best performance, with an AUC of 0.79. These findings suggest that different machine learning algorithms may be more suitable for specific clinical outcomes, and that the predictive value of the selected features can be effectively leveraged to improve risk stratification and facilitate personalized treatment planning.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eOur study provides a comprehensive analysis of clinical and dosimetric predictors for radiotherapy-induced GI toxicity in rectal cancer patients, addressing critical gaps identified in prior research\u003csup\u003e[\u003cspan additionalcitationids=\"CR6\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/sup\u003e. By systematically evaluating three distinct toxicity endpoints through multivariable and machine learning approaches, we offer clinically actionable insights while highlighting persistent challenges in toxicity prediction.\u003c/p\u003e\u003cp\u003eThe significantly higher acute upper GI toxicity in female patients (50% vs 28%, P\u0026thinsp;=\u0026thinsp;0.0088) aligns with emerging evidence of estrogen-mediated mucosal sensitivity \u003csup\u003e[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]\u003c/sup\u003e, suggesting the need for gender-specific antiemetic protocols as recommended by National Comprehensive Cancer Network (NCCN) guidelines\u003csup\u003e[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]\u003c/sup\u003e. Our finding that increased fraction number reduced acute lower GI risk (OR\u0026thinsp;=\u0026thinsp;0.74, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.0485) supports the ongoing shift toward moderate hypofractionation (e.g., 25\u0026ndash;28 fractions) \u003csup\u003e[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]\u003c/sup\u003e, though the lack of association with late toxicity reinforces the distinct pathophysiology of chronic radiation injury\u003csup\u003e[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]\u003c/sup\u003e. Notably, the analysis revealed no statistically significant association between total radiotherapy dose (50\u0026ndash;56 Gy) and acute upper gastrointestinal toxicity (Acute Upper GI toxicity: \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.1269) or late proctitis (Long: \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.4596). Although a marginal inverse correlation was observed for acute lower gastrointestinal toxicity (Acute Lower GI toxicity: \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.0233, with lower doses paradoxically linked to higher toxicity), this finding may reflect confounding factors such as treatment fractionation or patient heterogeneity rather than a true dose-effect relationship. Critically, the absence of dose-dependent toxicity for Acute Upper GI toxicity and Long endpoints supports the safety of dose escalation to 56 Gy for tumor targets. This approach enhances local tumor control rates without elevating the risk of clinically significant gastrointestinal toxicities, thereby providing survival benefits.\u003c/p\u003e\u003cp\u003eThe dosimetric parameter Vsmall intestine emerged as the most robust predictor of late proctitis (AUC\u0026thinsp;=\u0026thinsp;0.638, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.006), with our SHAP analysis confirming its dominance over clinical factors (importance score\u0026thinsp;=\u0026thinsp;0.074). This refines existing Quantitative Analyses of Normal Tissue Effects in the Clinic (QUANTEC) guidelines\u003csup\u003e[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]\u003c/sup\u003e by suggesting a stricter\u0026thinsp;\u0026lt;\u0026thinsp;30cc threshold for modern IMRT techniques, achievable through prone positioning\u003csup\u003e[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]\u003c/sup\u003e or adaptive planning\u003csup\u003e[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]\u003c/sup\u003e. The inverse correlation between anal verge distance and late toxicity (r = -0.1775, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.0181) further emphasizes the importance of anatomical considerations, consistent with EMBRACE-II findings\u003csup\u003e[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eOur dual statistical/machine learning (ML) approach revealed critical insights: while logistic regression identified conventional dose-volume relationships (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e), tree-based models uncovered nonlinear interactions - particularly LightGBM's detection of prior acute toxicity as a late effect predictor (importance\u0026thinsp;=\u0026thinsp;0.133), supporting the \"consequential damage\" hypothesis\u003csup\u003e[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]\u003c/sup\u003e. The AUC improvement from 0.63 (logistic regression) to 0.72 (Random Forest) aligns with trends reported in similar studies\u003csup\u003e[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eThree key opportunities emerge: (1) validation of our\u0026thinsp;\u0026lt;\u0026thinsp;30cc small bowel constraint in prospective cohorts, (2) development of gender-adapted supportive care protocols, and (3) integration of biological markers with dosimetric data. The unexpected superiority of logistic regression for late toxicity prediction (AUC\u0026thinsp;=\u0026thinsp;0.786) suggests that conventional models retain value when combined with judicious feature selection.\u003c/p\u003e\u003cp\u003eSeveral limitations warrant consideration. First, the retrospective single-center design with a limited sample size may introduce selection bias and reduce statistical power, thereby limiting the generalizability of our findings\u003csup\u003e[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]\u003c/sup\u003e. Second, the absence of biological markers (e.g., TGF-β polymorphisms\u003csup\u003e[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]\u003c/sup\u003e or fecal microbiota\u003csup\u003e[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]\u003c/sup\u003e) limits mechanistic insights. Third, dichotomizing CTCAE grades may obscure subtle dose-response relationships\u003csup\u003e[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]\u003c/sup\u003e. These gaps highlight the need for prospective trials integrating radiogenomics\u003csup\u003e[\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]\u003c/sup\u003e and patient-reported outcomes\u003csup\u003e[\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]\u003c/sup\u003e.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study establishes a triad of actionable interventions: gender-specific supportive care, moderate hypofractionation, and stringent small bowel constraints (\u0026lt;\u0026thinsp;30cc). While dosimetry explains most toxicity variance, achieving precision radiotherapy requires combining these measures with emerging biomarkers [39,40]. Our findings contribute to the future development of a template for personalized risk mitigation and highlight the development of biologically-informed predictive models as a critical next step.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"582\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCRC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003eColorectal cancer\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eRT\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003eRadiotherapy\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eI\u003c/strong\u003e\u003cstrong\u003eMRT\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003eIntensity-modulated radiotherapy\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVMAT\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003evolumetric modulated arc therapy\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eGI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003eGastrointestinal\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eROC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003ereceiver operating characteristic\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCTCAE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003eCommon Terminology Criteria for Adverse Events\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLENT/SOMA\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003eLate Effects Normal Tissues/Subjective, Objective, Management, Analytic\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eIQR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003einterquartile range\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAUC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003eThe area under the ROC curve\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSHAP\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003eSHapley Additive exPlanations\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eXGBoost\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003ethe eXtreme Gradient Boosting\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNCCN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003eNational Comprehensive Cancer Network\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eQUANTEC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003eQuantitative Analyses of Normal Tissue Effects in the Clinic\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 189px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eML\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 392px;\"\u003e\n \u003cp\u003eMachine Learning\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was approved by the Ethics Committee of the First Affiliated Hospital, Guangxi Medical University (Approval No. 2025-E0675), and the requirement for informed consent was waived due to the retrospective nature of the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot Applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated and analyzed during the current study are not publicly available due to patient privacy and ethical restrictions imposed by the Ethics Committee of the First Affiliated Hospital of Guangxi Medical University. However, de-identified data may be made available from the corresponding author (Shanshan Ma) upon reasonable request and with permission of the institutional ethics committee.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe author reports no conflicts of interest in this work.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was supported by grants from the National Natural Science Foundation of China (Grant No. 82102710), the Science and Technology Department of Guangxi Zhuang Autonomous Region (Grant No. 2023GXNSFBA026007), the Health Commission of Guangxi Zhuang Autonomous Region (Grant No. Z-A20230491), and the Education Department of Guangxi Zhuang Autonomous Region (Grant No. 2025KY0159).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eQiqi Huang, Jiali Meng, Chunyan Liang contribute equally to the article. They conducted the majority of the data collection, analyzed the data, and drafted the manuscript. Xinling Qin contributed to the data analysis and interpretation. Siyi He provided the analysis tools and were also involved in revising the manuscript critically for important intellectual content. Shanshan Ma, Weimei Huang, the co-corresponding authors, made substantial contributions to the conception and design of the work, and they also revised the manuscript and gave final approval of the version to be published.\u0026nbsp;All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors would like to express their sincere gratitude to the National Natural Science Foundation of China, the Science and Technology Department of Guangxi Zhuang Autonomous Region, the Health Commission of Guangxi Zhuang Autonomous Region, and the Education Department of Guangxi Zhuang Autonomous Region for their generous financial support through the respective grants, which laid a solid foundation for the smooth implementation of this study\u0026mdash;including the collection of research samples, the purchase of experimental reagents, and the analysis of key data.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eMORGAN E, ARNOLD M, GINI A, et al. Global burden of colorectal cancer in 2020 and 2040: incidence and mortality estimates from GLOBOCAN [J]. Gut, 2023, 72(2): 338-44.\u003c/li\u003e\n\u003cli\u003eYANG Y, PANG K, LIN G, et al. Neoadjuvant chemoradiation with or without PD-1 blockade in locally advanced rectal cancer: a randomized phase 2 trial [J]. Nat Med, 2025, 31(2): 449-56.\u003c/li\u003e\n\u003cli\u003eHOLM M O, BYE A, FALKMER U, et al. The effect of nutritional interventions in acute radiation-induced diarrhoea in patients with primary pelvic cancer: A systematic review [J]. Crit Rev Oncol Hematol, 2023, 188: 104038.\u003c/li\u003e\n\u003cli\u003eSAPIENZA L G, GOMES M J L, CALSAVARA V F, et al. Does para-aortic irradiation reduce the risk of distant metastasis in advanced cervical cancer? A systematic review and meta-analysis of randomized clinical trials [J]. Gynecol Oncol, 2017, 144(2): 312-7.\u003c/li\u003e\n\u003cli\u003eZENG C, JI J, HUANG Y, et al. A Novel Scoring System to Predict Acute Radiation Enteritis Recovery in Cervical Cancer Patients Undergoing Concurrent Chemoradiotherapy: A Southwest China Cohort Study [J]. Int J Gen Med, 2024, 17: 5907-19.\u003c/li\u003e\n\u003cli\u003eQIN Q, HUANG B, WU A, et al. Development and Validation of a Post-Radiotherapy Prediction Model for Bowel Dysfunction After Rectal Cancer Resection [J]. Gastroenterology, 2023, 165(6): 1430-42.e14.\u003c/li\u003e\n\u003cli\u003eDE MOL VAN OTTERLOO S R, CHRISTODOULEAS J P, BLEZER E L A, et al. Patterns of Care, Tolerability, and Safety of the First Cohort of Patients Treated on a Novel High-Field MR-Linac Within the MOMENTUM Study: Initial Results From a Prospective Multi-Institutional Registry [J]. Int J Radiat Oncol Biol Phys, 2021, 111(4): 867-75.\u003c/li\u003e\n\u003cli\u003eLI L, YANG Z, YI Y, et al. Gut microbiota and radiation-induced injury: mechanistic insights and microbial therapies [J]. Gut Microbes, 2025, 17(1): 2528429.\u003c/li\u003e\n\u003cli\u003eO\u0026apos;SULLIVAN B, LEVIN W. Late radiation-related fibrosis: pathogenesis, manifestations, and current management [J]. Semin Radiat Oncol, 2003, 13(3): 274-89.\u003c/li\u003e\n\u003cli\u003eTAN G, LIN C, HUANG C, et al. Radiosensitivity of colorectal cancer and radiation-induced gut damages are regulated by gasdermin E [J]. Cancer Lett, 2022, 529: 1-10.\u003c/li\u003e\n\u003cli\u003eFAN J, LIN B, FAN M, et al. Research progress on the mechanism of radiation enteritis [J]. Front Oncol, 2022, 12: 888962.\u003c/li\u003e\n\u003cli\u003eGE Z, CHEN C, CHEN J, et al. Gut Microbiota-Derived 3-Hydroxybutyrate Blocks GPR43-Mediated IL6 Signaling to Ameliorate Radiation Proctopathy [J]. Adv Sci (Weinh), 2024, 11(28): e2306217.\u003c/li\u003e\n\u003cli\u003eONG A L K, KNIGHT K, PANETTIERI V, et al. Dose-volume analysis of planned versus accumulated dose as a predictor for late gastrointestinal toxicity in men receiving radiotherapy for high-risk prostate cancer [J]. Physics and Imaging in Radiation Oncology, 2022, 23: 97-102.\u003c/li\u003e\n\u003cli\u003eXIE L W, CAI S, ZHAO T S, et al. Green tea derivative (-)-epigallocatechin-3-gallate (EGCG) confers protection against ionizing radiation-induced intestinal epithelial cell death both in vitro and in vivo [J]. Free Radic Biol Med, 2020, 161: 175-86.\u003c/li\u003e\n\u003cli\u003eBERTELLI G, PATAUNER S, GORGATTI T, et al. Right-side fixation of the sigmoid colon causing internal herniation with closed-loop obstruction of both small and large bowel: a case report and review of the literature [J]. J Med Case Rep, 2022, 16(1): 323.\u003c/li\u003e\n\u003cli\u003eNIE T, CHEN Z, CAI J, et al. Integration of dosimetric parameters, clinical factors, and radiomics to predict symptomatic radiation pneumonitis in lung cancer patients undergoing combined immunotherapy and radiotherapy [J]. Radiother Oncol, 2024, 190: 110047.\u003c/li\u003e\n\u003cli\u003eBAO D, ZHAO Y, LI L, et al. A MRI-based radiomics model predicting radiation-induced temporal lobe injury in nasopharyngeal carcinoma [J]. Eur Radiol, 2022, 32(10): 6910-21.\u003c/li\u003e\n\u003cli\u003eANDERSSON B, LANGEN B, LIU P, et al. Development of a machine learning framework for radiation biomarker discovery and absorbed dose prediction [J]. Front Oncol, 2023, 13: 1156009.\u003c/li\u003e\n\u003cli\u003eBIJMAN R, ROSSI L, SHARFO A W, et al. Automated Radiotherapy Planning for Patient-Specific Exploration of the Trade-Off Between Tumor Dose Coverage and Predicted Radiation-Induced Toxicity-A Proof of Principle Study for Prostate Cancer [J]. Front Oncol, 2020, 10: 943.\u003c/li\u003e\n\u003cli\u003eGUO Y, ZHU Y, ZHANG J, et al. Facilitation of mucosal healing by estrogen receptor \u0026beta; in ulcerative colitis through suppression of branched-chain amino acid transport and subsequent triggering of autophagy in colonic epithelial cells [J]. Acta Pharm Sin B, 2025, 15(1): 168-87.\u003c/li\u003e\n\u003cli\u003eBENSON A B, VENOOK A P, ADAM M, et al. Colon Cancer, Version 3.2024, NCCN Clinical Practice Guidelines in Oncology [J]. J Natl Compr Canc Netw, 2024, 22(2 d).\u003c/li\u003e\n\u003cli\u003eDEVLIN L, GROCUTT L, HUNTER B, et al. The in-silico feasibility of dose escalated, hypofractionated radiotherapy for rectal cancer [J]. Clin Transl Radiat Oncol, 2022, 36: 24-30.\u003c/li\u003e\n\u003cli\u003eTURRI G, OSTUZZI G, VITA G, et al. Treatment of Locally Advanced Rectal Cancer in the Era of Total Neoadjuvant Therapy: A Systematic Review and Network Meta-Analysis [J]. JAMA Netw Open, 2024, 7(6): e2414702.\u003c/li\u003e\n\u003cli\u003eSHURYAK I, BRENNER D J. Quantitative modeling of multigenerational effects of chronic ionizing radiation using targeted and nontargeted effects [J]. Sci Rep, 2021, 11(1): 4776.\u003c/li\u003e\n\u003cli\u003eBENTZEN S M, CONSTINE L S, DEASY J O, et al. Quantitative Analyses of Normal Tissue Effects in the Clinic (QUANTEC): an introduction to the scientific issues [J]. Int J Radiat Oncol Biol Phys, 2010, 76(3 Suppl): S3-9.\u003c/li\u003e\n\u003cli\u003eSCOBIOALA S, KITTEL C, NIERMANN P, et al. A treatment planning study of prone vs. supine positions for locally advanced rectal carcinoma : Comparison of 3‑dimensional conformal radiotherapy, tomotherapy, volumetric modulated arc therapy, and intensity-modulated radiotherapy [J]. Strahlenther Onkol, 2018, 194(11): 975-84.\u003c/li\u003e\n\u003cli\u003eQIU Z, OLBERG S, DEN HERTOG D, et al. Online adaptive planning methods for intensity-modulated radiotherapy [J]. Phys Med Biol, 2023, 68(10).\u003c/li\u003e\n\u003cli\u003eBERGER T, SEPPENWOOLDE Y, P\u0026ouml;TTER R, et al. Importance of Technique, Target Selection, Contouring, Dose Prescription, and Dose-Planning in External Beam Radiation Therapy for Cervical Cancer: Evolution of Practice From EMBRACE-I to II [J]. Int J Radiat Oncol Biol Phys, 2019, 104(4): 885-94.\u003c/li\u003e\n\u003cli\u003eBIBAULT J E, GIRAUD P, HOUSSET M, et al. Deep Learning and Radiomics predict complete response after neo-adjuvant chemoradiation for locally advanced rectal cancer [J]. Sci Rep, 2018, 8(1): 12611.\u003c/li\u003e\n\u003cli\u003eHOWE C J, COLE S R, LAU B, et al. Selection Bias Due to Loss to Follow Up in Cohort Studies [J]. Epidemiology, 2016, 27(1): 91-7.\u003c/li\u003e\n\u003cli\u003eWU F, WEIGEL K J, ZHOU H, et al. Paradoxical roles of TGF-\u0026beta; signaling in suppressing and promoting squamous cell carcinoma [J]. Acta Biochim Biophys Sin (Shanghai), 2018, 50(1): 98-105.\u003c/li\u003e\n\u003cli\u003eCUI B, LUO H, HE B, et al. Gut dysbiosis conveys psychological stress to activate LRP5/\u0026beta;-catenin pathway promoting cancer stemness [J]. Signal Transduct Target Ther, 2025, 10(1): 79.\u003c/li\u003e\n\u003cli\u003eEICHKORN T, BAUER J, BAHN E, et al. Radiation-induced contrast enhancement following proton radiotherapy for low-grade glioma depends on tumor characteristics and is rarer in children than adults [J]. Radiother Oncol, 2022, 172: 54-64.\u003c/li\u003e\n\u003cli\u003eSU G H, XIAO Y, YOU C, et al. Radiogenomic-based multiomic analysis reveals imaging intratumor heterogeneity phenotypes and therapeutic targets [J]. Sci Adv, 2023, 9(40): eadf0837.\u003c/li\u003e\n\u003cli\u003eMANZ C R, SCHRIVER E, FERRELL W J, et al. Association of Remote Patient-Reported Outcomes and Step Counts With Hospitalization or Death Among Patients With Advanced Cancer Undergoing Chemotherapy: Secondary Analysis of the PROStep Randomized Trial [J]. J Med Internet Res, 2024, 26: e51059.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcan","sideBox":"Learn more about [BMC Cancer](http://bmccancer.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcan/default.aspx","title":"BMC Cancer","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Rectal cancer radiotherapy, gastrointestinal toxicity, dosimetric predictors, late proctitis, small intestine volume","lastPublishedDoi":"10.21203/rs.3.rs-7461232/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7461232/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003ePurpose\u003c/h2\u003e\u003cp\u003eTo identify key risk factors for radiation-induced enteritis and construct a predictive model integrating dosimetric and clinical parameters.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eWe analyzed 206 colorectal cancer patients treated with pelvic radiotherapy (2015\u0026ndash;2020). Clinical variables, dosimetric parameters and anatomical factors were evaluated. Toxicity endpoints included acute upper/lower gastrointestinal reactions (CTCAE v5.0) and late proctitis (LENT/SOMA). Statistical analyses included logistic regression and machine learning (Random Forest/XGBoost).\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eThe incidences of acute upper gastrointestinal reactions, acute lower gastrointestinal reactions, and late proctitis were 33.5% (69/206), 53.9% (109/206), and 38.3% (79/206), respectively. The incidence of acute upper gastrointestinal toxicity was higher in females compared with males (50% vs. 28%, P\u0026thinsp;=\u0026thinsp;0.0088). A higher fraction number (OR\u0026thinsp;=\u0026thinsp;0.74, P\u0026thinsp;=\u0026thinsp;0.0485) and a lower radiation dose (OR\u0026thinsp;=\u0026thinsp;0.999, P\u0026thinsp;=\u0026thinsp;0.0293) were associated with a reduced incidence of acute lower gastrointestinal reactions.. For late proctitis, Vsmall intestine was a significant risk factor (AUC\u0026thinsp;=\u0026thinsp;0.638, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.006), while tumor-to-anal verge distance was protective (AUC\u0026thinsp;=\u0026thinsp;0.583, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.058). Machine learning models showed superior performance: random forest achieved an AUC of 0.72 for acute upper gastrointestinal reactions, XGBoost/AUC\u0026thinsp;=\u0026thinsp;0.64 for acute lower gastrointestinal reactions, and LightGBM/AUC\u0026thinsp;=\u0026thinsp;0.76 for late proctitis, outperforming conventional logistic regression (AUC range: 0.49\u0026ndash;0.57 for acute endpoints; 0.57 for late proctitis).\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e\u003cp\u003eGender, fraction number, Vsmall intestine, and tumor-to-anal verge distance are associated with an increased risk of radiation-induced gastrointestinal toxicities. Machine learning models, particularly LightGBM for late proctitis (AUC\u0026thinsp;=\u0026thinsp;0.76) and random forest for acute reactions (AUC\u0026thinsp;=\u0026thinsp;0.72), provide robust tools for risk stratification. These findings may support gender-specific care, moderate hypofractionation, and stringent small bowel dose constraints to help optimize personalized radiotherapy.\u003c/p\u003e","manuscriptTitle":"Radiotherapy-related Gastrointestinal Adverse Events in Rectal Cancer: Risk Factor Analysis and Predictive Modeling Using Clinical and Small Bowel Dosimetric Features","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-27 14:26:47","doi":"10.21203/rs.3.rs-7461232/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"177592219495589454066423493362591926926","date":"2025-12-17T08:44:35+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"66086509342262764380683755301730588051","date":"2025-11-12T17:25:37+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-11-02T07:12:13+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"238266816045150836412091811998142466706","date":"2025-10-14T18:32:44+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"114415160046017670648572718041338600947","date":"2025-10-13T08:58:19+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"254000552701780477233742678026686978839","date":"2025-10-12T20:25:15+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-10-12T14:16:14+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-09-10T10:11:15+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-09-08T09:54:51+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-09-06T10:46:07+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Cancer","date":"2025-09-06T10:42:47+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcan","sideBox":"Learn more about [BMC Cancer](http://bmccancer.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcan/default.aspx","title":"BMC Cancer","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"ec10c7b8-61ee-4800-b3c3-4207c1068f48","owner":[],"postedDate":"October 27th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-10-27T14:26:47+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-27 14:26:47","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7461232","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7461232","identity":"rs-7461232","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}