Interpretable Machine Learning Model Based on Intra- and Peritumoral MRI Radiomics for Predicting Biochemical Recurrence After Radical Prostatectomy

Interpretable Machine Learning Model Based on Intra- and Peritumoral MRI Radiomics for Predicting Biochemical Recurrence After Radical Prostatectomy | 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 Interpretable Machine Learning Model Based on Intra- and Peritumoral MRI Radiomics for Predicting Biochemical Recurrence After Radical Prostatectomy Xuelian Zhao, Fei Jia, Hao Li, Ning Zhang, Xinlin Han, Yanli Jiang, and 6 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7173672/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 10 You are reading this latest preprint version Abstract Purpose To develop a predictive model for biochemical recurrence (BCR) after radical prostatectomy (RP) by integrating intratumoral and peritumoral MRI radiomics features with clinical independent risk factors. Methods This retrospective study analyzed 277 RP patients with complete follow-up data (≥1 year) from our institution, randomly divided into training (n=193) and test (n=84) cohorts. Regions of interest (ROIs) were manually delineated on T2-weighted imaging(T2WI) and apparent diffusion coefficient (ADC) maps. Peritumoral ROIs were expanded by 4 mm using Python and manually adjusted to exclude non-prostatic tissues. Radiomics models (Intra, Peri_4mm, IntraPeri_4mm), a clinical model (Clinic), and a combined radiomics-clinical model (Combined) were constructed. The predictive performance of these models was evaluated using different indexes. SHapley Additive exPlanations (SHAP) analysis was employed to visualize and interpret the decision-making process. Results Independent risk factors for BCR included extraprostatic extension (EPE), clinical N stage (N), seminal vesicle invasion (SVI), PI-RADS score, neutrophil-to-lymphocyte ratio (NLR), and maximum transverse diameter of prostate (MTD). The Clinic model achieved AUC of 0.897 (training) and 0.731 (test). The IntraPeri_4mm_ADC model showed AUC of 0.902 and 0.706, while the IntraPeri_4mm_T2 model yielded 0.842 and 0.662. The Combined model outperformed others (AUC: 0.978 and 0.810). DCA confirmed its higher net benefit. SHAP analysis revealed EPE as the top contributor to BCR prediction, followed by the ADC-derived radiomics score (ADC label_0). Conclusions The combined MRI radiomics-clinical model effectively predicts BCR post-RP. SHAP interpretability transforms "black-box" predictions into quantifiable feature contributions, aiding clinicians in risk stratification and personalized treatment planning. Prostate cancer Radiomics Machine learning Biochemical recurrence Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Prostate cancer (PCa) ranks as the second most prevalent malignancy in men worldwide [ 1 ], with rapidly increasing incidence and mortality rates in China. Radical prostatectomy (RP) [ 2 ] remains the primary treatment for organ-confined and locally advanced PCa. Biochemical recurrence (BCR) after RP is defined as prostate-specific antigen (PSA) levels rising above 0.1 ng/ml in two consecutive measurements following post-operative PSA nadir (< 0.1 ng/ml) [ 3 , 4 ]. Notably, 27%-53% of post-RP patients experience PSA elevation, with BCR strongly correlating with increased risks of local recurrence, distant metastasis, and overall mortality. Accurate BCR prediction therefore enables proactive surveillance of high-risk patients to improve outcomes. Previous investigations of BCR risk factors primarily focused on clinical parameters including preoperative PSA, biopsy Gleason score, clinical stage, age, and percentage of positive cores. However, with the widespread clinical application of MRI, conventional clinical-based assessment methods (such as D'Amico risk classification, Cancer of the Prostate Risk Assessment [CAPRA] score, and Kattan nomogram) [ 5 – 9 ] have become insufficiently comprehensive. Radiomics [ 10 ] enables extraction of high-throughput imaging features to evaluate tumor heterogeneity at microscopic scales, identifying quantitative imaging biomarkers for diagnosis, classification, and prognosis prediction. This non-invasive approach may reduce complications associated with biopsy [ 11 ]. Duenweg et al. [ 12 ] developed a BCR prediction model based on T2-weighted imaging (T2WI) features, achieving an AUC of 0.97, demonstrating excellent diagnostic performance. Similarly, Zhu et al. [ 13 ]employed comparable methodology to construct a radiomics model for preoperative prediction of post-radiotherapy BCR in PCa patients, with the model attaining an AUC of 0.83 and enabling accurate prognostic assessment. While intratumoral radiomic features have demonstrated significant value in tumor characterization, emerging research has revealed the critical role of peritumoral microenvironment in tumorigenesis and progression[ 14 ]. The peritumoral region, while macroscopically normal, exhibits microscopic heterogeneity at tumor-normal tissue interfaces. Studies confirm genetic, epigenetic, and transcriptomic alterations in peritumoral tissues of epithelial tumors (colorectal, bladder, prostate, and breast) [ 15 ]. Gu et al.[ 16 ] first proposed the peritumoral microenvironment (PME) concept in hepatocellular carcinoma, demonstrating PME heterogeneity's significance throughout tumor development and its diagnostic/therapeutic implications. Multiple studies [ 17 – 19 ]reveal superior diagnostic and predictive performance of peritumoral radiomics models compared to conventional intratumoral approaches, serving as crucial complementary tools. Some findings[ 20 ]suggest peritumoral features may outperform intratumoral features in certain cancers, though peritumoral radiomics exploration remains limited in PCa. Machine learning (ML) algorithms effectively uncover latent associations between radiomics features and clinical outcomes to build predictive models, but their clinical translation is hindered by the opaque 'black-box' decision-making process and lack of interpretability [ 21 – 23 ]. SHapley Additive exPlanations (SHAP) analysis [ 24 , 25 ] evaluates the contribution of individual features to model predictions through nonlinear association analysis, elucidating their roles in the predictive process. This approach not only enhances model interpretability but also identifies potential biomarkers driving BCR, thereby providing novel evidence for the correlation between imaging/clinical features and tumor biological behavior. Ultimately, SHAP analysis facilitates the translation of ML models into clinical decision-making tools. Methods Patients This study, conducted in accordance with the Declaration of Helsinki, was approved by the Medical Ethics Committee (No. 2024A-1253) with waived informed consent. We retrospectively analyzed 300 patients who underwent RP between January 2013 and February 2024. Inclusion criteria comprised: (1) preoperative 3.0T MRI (complete T2WI and DWI-derived apparent diffusion coefficient (ADC) maps); (2) pathologically confirmed prostatic adenocarcinoma; (3) BCR occurrence or ≥ 1-year follow-up if BCR-negative. Exclusion criteria included: MRI-unidentifiable tumors, non-adenocarcinoma pathology, incomplete data, preoperative radiotherapy/chemotherapy, or metastatic disease. Follow-up spanned from surgery to BCR date to March 31, 2025. Among the 277 included patients, 193 were assigned to the training cohort (37 with BCR and 156 without) and 84 to the test cohort (19 with BCR and 65 without). (Supplementary Fig. 1). Magnetic resonance imaging protocol All MRI examinations were performed on a 3.0T scanner using an integrated 18-channel phased-array body coil with patients in the supine position. The imaging protocol included axial T2WI (TR/TE = 6980/104 ms, slice thickness = 3 mm, slice spacing = 0 mm, FOV = 200×200 mm, matrix = 384×384) and axial DWI using single-shot echo-planar imaging (TR/TE = 6540/60 ms, slice thickness = 3 mm, slice spacing = 0 mm, FOV = 200×200 mm, matrix = 130×130) with b-values of 50, 1000, and 1500 s/mm² for ADC map generation, consistent with Prostate Imaging Reporting and Data System (PI-RADS v2.1) recommendations [ 26 ] that suggest using low (0-100 s/mm²) and intermediate (800–1000 s/mm²) b-values for ADC calculation while higher b-values (≥ 1400 s/mm²) improve PCa detection sensitivity. Tumor segmentation Region of Interests (ROIs) were delineated on both T2WI and ADC sequences. Prior to delineation, all images were resampled to an isotropic voxel spacing of 1 × 1 × 1 mm 3 and underwent grayscale normalization to compensate for voxel spatial heterogeneity and maintain intensity consistency. Three experienced radiologists collaboratively reviewed each patient's complete MRI dataset. Two radiologists independently outlined the tumor boundaries, while the third performed quality control by verifying all ROIs and correcting or redrawing questionable delineations. Tumor localization followed PI-RADS v2.1 criteria. Using ITK-SNAP software (version 4.0.0), manual slice-by-slice ROI delineation was performed along the tumor margins on T2WI and ADC maps. The Python package SimpleITK (version 3.7.12) was then employed to automatically expand each intratumoral ROI by 4 mm, followed by manual exclusion of periprostatic non-target tissues (e.g. fat, seminal vesicles, rectum, urethra, and bladder) to derive peritumoral regions. Representative examples of intratumoral and peritumoral ROI expansion are illustrated in Fig. 1 . Feature extraction and selection Radiomic features were extracted from both intratumoral (Intra) and peritumoral (Peri_4mm) ROIs using Python's pyradiomics library following established protocols [ 27 – 29 ]. Feature selection involved: Eliminating features with Spearman's rank correlation coefficient > 0.9 to remove highly correlated features; recursive feature elimination to remove redundancy and LASSO regression to construct the radiomics signature (Rad_Sig), with identical pipelines applied to both regions to ensure comparability. Normality was assessed using Student's t-test (normal distributions) or Mann-Whitney U test (non-normal distributions). Clinical variables were analyzed similarly (t-test/Mann-Whitney U for continuous variables; χ² test for categorical variables), with univariate and multivariate analyses identifying independent BCR risk factors. Construction and comparison of prediction models The Rad_Sig from both Intra and Peri_4mm regions were input into ExtraTrees machine learning models to generate individualized radiomics scores (Rad_Score), constructing three predictive models: Intra, Peri_4mm, and IntraPeri_4mm models, while clinical independent risk factors were used to build a Combined, which was further integrated with radiomics features to develop a clinic-radiomics (Combined) model. Model performance was rigorously evaluated through receiver operating characteristic (ROC) curve analysis (reporting AUC values and optimal cutoffs maximizing Youden's index), Delong's test for AUC comparisons, and decision curve analysis (DCA) to assess clinical utility, supplemented by SHAP analysis for both cohort-level and individual patient-level interpretation of feature importance in BCR. Statistical analysis All data analyses were performed using Python 3.7.12 on the OnekeyAI platform (version 4.9.1), with statistical evaluations conducted using statsmodels (version 0.13.2) and radiomics feature extraction implemented through PyRadiomics (version 3.0.1). Statistical significance was defined as a two-sided p-value ≤ 0.05. The intraclass correlation coefficient (ICC) was computed to evaluate feature stability, with features demonstrating ICC > 0.75 retained for subsequent analysis. The entire workflow from data collection through model construction, evaluation, and interpretation is illustrated in Supplementary Fig. 2 Results Basic characteristics of patients Among the 277 patients included in the study, biochemical recurrence (BCR) occurred in 56 cases (20.2%), with 37 cases (19.1%) in the training cohort and 19 cases (22.6%) in the test cohort. The baseline characteristics of the cohorts are presented in Table 1. The inter-group p-values exceeding 0.05 for all comparative analyses confirmed no statistically significant differences between cohorts, demonstrating unbiased group allocation. Development and validation of prediction models From T2WI and ADC sequences, we extracted 7336 radiomic features with identical numbers of Intra and Peri_4mm features (n=1834 each) per sequence, constructing Intra, Peri_4mm and IntraPeri_4mm models. For ADC sequences, 4 Intra- and 3 Peri_4mm-features correlated with BCR, yielding training and test AUC of 0.883 (95%CI:0.832-0.934)、0.694 (95%CI:0.582-0.805) for Intra_ADC and 0.889 (95%CI:0.841-0.938)、0.518 (95%CI:0.378-0.657) for Peri_4mm_ADC. T2WI sequences showed 4 Intra- and 9 Peri_4mm-feature correlations with training and test AUC of 0.732 (95%CI:0.646-0.846)、0.587 (95%CI:0.447-0.728) for Intra_T2 and 0.864 (95%CI:0.802-0.926)、0.637 (95%CI:0.483-0.791) for Peri_4mm_T2. The IntraPeri_4mm models demonstrated superior performance, ADC-based (5 features) achieved 0.902 (95%CI:0.853-0.951)、0.706 (95%CI:0.594-0.819), while T2WI-based (10 features) reached 0.842 (95%CI:0.778-0.906)、0.662 (95%CI:0.529-0.795), with feature distributions shown in Figure 2. Univariate logistic regression revealed that nearly all factors were significantly associated with biochemical recurrence (BCR). Multivariate analysis identified the following independent risk factors for BCR: extraprostatic extension (EPE), clinical nodal stage (N), seminal vesicle invasion (SVI), PI-RADS score, neutrophil-to-lymphocyte ratio (NLR), and maximum transverse diameter of the prostate (MTD) (Table 2). Based on these independent risk factors, we constructed the Clinic model, which achieved AUC of 0.897 (95% CI: 0.839-0.954) in the training set and 0.731 (95% CI: 0.602-0.861) in the test set (Table 3). By integrating the clinical independent risk factors with the Rad_Score from both ADC and T2WI models, we developed a superior diagnostic Combined model. In the training set, the Combined model achieved AUC of 0.978 (95% CI: 0.961–0.994), while in the test set, the AUC was 0.810 (95% CI: 0.707–0.912) (Table 3 and Figure 3A, 3B). DeLong’s test (Figure 3C, 3D) revealed that in the training set, the Combined model showed statistically significant differences compared to the Clinic, IntraPeri_4mm_ADC, and IntraPeri_4mm_T2 models. In the test set, the Combined model demonstrated significant differences versus the IntraPeri_4mm_ADC and IntraPeri_4mm_T2 models, but no significant difference compared to the Clinic model (p = 0.141). The Combined model exhibited significantly positive Net Reclassification Improvement (NRI) and Integrated Discrimination Improvement (IDI) values when compared to the IntraPeri_4mm_ADC, IntraPeri_4mm_T2, and Clinic models (Supplementary Figure 3.). DCA confirmed that the Combined model provided greater net clinical benefit than the other models (Figure 4). Model visualization and interpretation The SHAP analysis revealed the following order of feature importance for BCR prediction (Figure 5A). EPE demonstrated the highest contribution, followed by ADC negative prediction Rad_Score (ADC label_0), T2WI positive prediction Rad_Score (T2 label_1), MTD, ADC positive prediction Rad_Score (ADC label_1), T2WI negative prediction Rad_Score (T2 label_0), SVI, PI-RADS, NLR and N. The SHAP analysis demonstrated that EPE exhibited the strongest predictive influence, as evidenced by its widest band width, with high feature values (red dots) predominantly distributed on the right side. Individual case analysis using waterfall (Figure. 5B) and force-directed (Figure. 5C) plots demonstrated that all examined factors except EPE contributed positively to BCR prediction. Among these ADC label_0 exhibited the strongest predictive influence, followed by EPE as the secondary determinant. Discussion The rapid proliferation of cancer cells induces relative hypoxia in the peritumoral region, where hypoxia-inducible factor-1 (HIF-1) and other hypoxia-related mediators remodel the tumor microenvironment by altering vascular and lymphatic networks [ 30 ]. Peritumoral inflammatory responses and immune cell infiltration further enhance microenvironmental heterogeneity [ 31 ]. Building upon our investigation of intratumoral heterogeneity, we integrated both intratumoral and peritumoral features, which resulted in improved model performance: the AUC increased from 0.883 to 0.902 for ADC-based models and from 0.732 to 0.842 for T2WI-based models. These enhancements confirm the value of peritumoral radiomic features in complementing intratumoral characteristics for BCR assessment. Peritumoral radiomics quantitatively evaluates tumor microenvironment heterogeneity through imaging biomarkers. When combined with intratumoral features, it provides a more comprehensive tumor profile [ 32 – 34 ], offering objective and multidimensional quantitative analysis for tumor characterization. Our findings align with Algohari et al. [ 27 ], who reported significantly improved prostate cancer risk stratification by incorporating peritumoral radiomic features. Similarly, Zhang et al. [ 35 ] demonstrated enhanced diagnostic performance for clinically significant prostate cancer using combined intratumoral-peritumoral radiomics and PSA levels, but their model lacked comprehensive clinical variables for optimal predictive performance. Most previous studies on BCR prediction models focused solely on either clinical data or radiomic features. Although the D'Amico risk classification, CAPRA score, and Kattan nomogram demonstrate satisfactory performance, they lack quantitative assessment of tumor heterogeneity. Hectors et al[ 36 ] and Ruan et al [ 37 ] investigated prostate cancer heterogeneity using multiparametric MRI-based radiomics, Piran et al [ 38 ] developed a post-radiotherapy biochemical recurrence (BCR) prediction model utilizing T2WI intratumoral and peritumoral radiomic features, achieving an AUC of 0.794. Although radiomics provides satisfactory characterization of tumor heterogeneity, its diagnostic performance shows limited stability, poor integration with current clinical guidelines, and suboptimal clinical applicability. In our study, we first established a Clinic model incorporating laboratory tests, PI-RADS scores, and pathological Gleason grading, which demonstrated AUC of 0.897 in the training set and 0.731 in the test set - comparable to established models (D'Amico score AUC 0.710; CAPRA score AUC 0.770). Furthermore, we developed an integrated model combining clinical independent risk factors with T2WI/ADC-derived Rad_Score for post-prostatectomy BCR prediction, showing superior performance with an AUC of 0.978 compared to standalone radiomic or clinical models, along with improved sensitivity, specificity, and accuracy. Bourbonne et al.[ 39 ]developed a prediction model based on clinical data and ADC maps, with AUC values of 0.68, 0.82, and 0.86 for the clinical, ADC radiomics, and combined models, respectively. The integration of clinical data and radiomics significantly improved diagnostic performance, along with enhanced sensitivity, specificity and accuracy, this result highly consistent with our results. Furthermore, our Combined model showed statistically significant improvements in both the NRI and IDI, demonstrating that the combined use of clinical and radiomic features provides a more comprehensive representation of tumor biological behavior. These results strongly support the necessity of data integration. We employed SHAP interpretability analysis to elucidate both global and local interpretability of the Combined model. The beeswarm plot reveals population-level impact patterns of clinical and imaging features on prediction outcomes through horizontal density distribution. Each point in the beeswarm plot represents the SHAP value of a feature for an individual patient, with color gradient from red to blue indicating feature values from high to low. Wider color regions in the plot indicate greater feature importance. In the beeswarm plot, EPE exhibited the widest color region, with red points predominantly distributed on the right side, indicating a positive correlation with BCR occurrence. This finding is highly consistent with the results of Merriman et al. [ 40 ].In both the Partin and MSKCC nomograms, EPE is consistently identified as a key prognostic parameter, further underscoring the importance of EPE detection in clinical practice. This study has several limitations that warrant consideration: (1) As a single-center retrospective investigation, potential selection bias may exist, necessitating test through multicenter prospective studies; (2) Manual lesion delineation introduces subjectivity, prompting our future adoption of semi-automated or fully automated segmentation methods to improve objectivity; (3) Currently focused solely on recurrence prediction, future work will incorporate salvage treatment outcomes (e.g. androgen deprivation therapy/adjuvant radiotherapy) to develop a comprehensive closed-loop system for personalized therapeutic decision-making. Conclusion our study developed a preoperative prediction model for post-radical prostatectomy BCR by integrating intratumoral and peritumoral MRI radiomic features with clinical independent risk factors, which facilitates precise individualized treatment planning and proactive monitoring of high-risk patients. Declarations Author Contribution G.Y.L. is the guarantor of integrity of the entire manuscript. G.Y.L.,X.L.Z.and J.Z. designed the study and searched literature. X.L.Z., F.J., H.L., N.Z.,X.L.H.,Y.L.J.and H.L. collected the clinical and imaging data. X.L.Z., F.J. and H.L. analyzed data. X.L.Z., F.J. and H.L. drafted the manuscript. All authors edited and approved the final manuscript. References Siegel RL, Giaquinto AN & Jemal A (2024) Cancer statistics, 2024. CA Cancer J Clin 74, 12-49. https://doi.org/[10.3322/caac.21820] Gandaglia G, Montorsi F, Karakiewicz PI & Sun M (2015) Robot-assisted radical prostatectomy in prostate cancer. Future Oncol 11, 2767-2773. https://doi.org/[10.2217/fon.15.169] Shee K, de la Calle CM, Chang AJ, et al (2022) Addition of Enzalutamide to Leuprolide and Definitive Radiation Therapy Is Tolerable and Effective in High-Risk Localized or Regional Nonmetastatic Prostate Cancer: Results From a Phase 2 Trial. Adv Radiat Oncol 7, 100941. https://doi.org/[10.1016/j.adro.2022.100941] Tilki D, Preisser F, Graefen M, Huland H & Pompe RS (2019) External Validation of the European Association of Urology Biochemical Recurrence Risk Groups to Predict Metastasis and Mortality After Radical Prostatectomy in a European Cohort. Eur Urol 75, 896-900. https://doi.org/[10.1016/j.eururo.2019.03.016] D'Amico AV, Whittington R, Malkowicz SB, et al (1998) Biochemical outcome after radical prostatectomy, external beam radiation therapy, or interstitial radiation therapy for clinically localized prostate cancer. Jama 280, 969-974. https://doi.org/[10.1001/jama.280.11.969] Kattan MW, Eastham JA, Stapleton AM, Wheeler TM & Scardino PT (1998) A preoperative nomogram for disease recurrence following radical prostatectomy for prostate cancer. J Natl Cancer Inst 90, 766-771. https://doi.org/[10.1093/jnci/90.10.766] Kattan MW, Wheeler TM & Scardino PT (1999) Postoperative nomogram for disease recurrence after radical prostatectomy for prostate cancer. J Clin Oncol 17, 1499-1507. https://doi.org/[10.1200/jco.1999.17.5.1499] Lemaitre L, Puech P, Poncelet E, et al (2009) Dynamic contrast-enhanced MRI of anterior prostate cancer: morphometric assessment and correlation with radical prostatectomy findings. Eur Radiol 19, 470-480. https://doi.org/[10.1007/s00330-008-1153-0] Punnen S, Freedland SJ, Presti JC, Jr., et al (2014) Multi-institutional validation of the CAPRA-S score to predict disease recurrence and mortality after radical prostatectomy. Eur Urol 65, 1171-1177. https://doi.org/[10.1016/j.eururo.2013.03.058] Bi WL, Hosny A, Schabath MB, et al (2019) Artificial intelligence in cancer imaging: Clinical challenges and applications. CA Cancer J Clin 69, 127-157. https://doi.org/[10.3322/caac.21552] Schroeder DW, Foster BR, Young DJ & Coakley FV (2024) Targeted biopsy of the prostate. Abdominal Radiology. https://doi.org/[10.1007/s00261-024-04452-z] Duenweg SR, Bobholz SA, Barrett MJ, et al (2023) T2-Weighted MRI Radiomic Features Predict Prostate Cancer Presence and Eventual Biochemical Recurrence. Cancers (Basel) 15. https://doi.org/[10.3390/cancers15184437] Zhu X, Liu Z, He J, et al (2023) MRI-Derived Radiomics Model to Predict the Biochemical Recurrence of Prostate Cancer Following Seed Brachytherapy. Arch Esp Urol 76, 264-269. https://doi.org/[10.56434/j.arch.esp.urol.20237604.30] Dieterich LC & Bikfalvi A (2020) The tumor organismal environment: Role in tumor development and cancer immunotherapy. Semin Cancer Biol 65, 197-206. https://doi.org/[10.1016/j.semcancer.2019.12.021] Hanahan D & Monje M (2023) Cancer hallmarks intersect with neuroscience in the tumor microenvironment. Cancer Cell 41, 573-580. https://doi.org/[10.1016/j.ccell.2023.02.012] Gu Y, Guo Y, Gao N, et al (2022) The proteomic characterization of the peritumor microenvironment in human hepatocellular carcinoma. Oncogene 41, 2480-2491. https://doi.org/[10.1038/s41388-022-02264-3] Li H, Chai L, Pu H, et al (2024) T2WI-based MRI radiomics for the prediction of preoperative extranodal extension and prognosis in resectable rectal cancer. Insights Imaging 15, 57. https://doi.org/[10.1186/s13244-024-01625-8] Lin P, Xie W, Li Y, et al (2024) Intratumoral and peritumoral radiomics of MRIs predicts pathologic complete response to neoadjuvant chemoimmunotherapy in patients with head and neck squamous cell carcinoma. J Immunother Cancer 12. https://doi.org/[10.1136/jitc-2024-009616] Zhang Z, Wan X, Lei X, et al (2023) Intra- and peri-tumoral MRI radiomics features for preoperative lymph node metastasis prediction in early-stage cervical cancer. Insights Imaging 14, 65. https://doi.org/[10.1186/s13244-023-01405-w] Wang X, Wei M, Chen Y, et al (2024) Intratumoral and peritumoral MRI-based radiomics for predicting extrapelvic peritoneal metastasis in epithelial ovarian cancer. Insights Imaging 15, 281. https://doi.org/[10.1186/s13244-024-01855-w] Gramegna A & Giudici P (2021) SHAP and LIME: An Evaluation of Discriminative Power in Credit Risk. Front Artif Intell 4, 752558. https://doi.org/[10.3389/frai.2021.752558] Li J, Liu S, Hu Y, et al (2022) Predicting Mortality in Intensive Care Unit Patients With Heart Failure Using an Interpretable Machine Learning Model: Retrospective Cohort Study. J Med Internet Res 24, e38082. https://doi.org/[10.2196/38082] Tang S, Zhang H, Liang J, et al (2024) Prostate cancer treatment recommendation study based on machine learning and SHAP interpreter. Cancer Sci 115, 3755-3766. https://doi.org/[10.1111/cas.16327] Qi X, Wang S, Fang C, et al (2025) Machine learning and SHAP value interpretation for predicting comorbidity of cardiovascular disease and cancer with dietary antioxidants. Redox Biol 79, 103470. https://doi.org/[10.1016/j.redox.2024.103470] Rodríguez-Pérez R & Bajorath J (2020) Interpretation of Compound Activity Predictions from Complex Machine Learning Models Using Local Approximations and Shapley Values. J Med Chem 63, 8761-8777. https://doi.org/[10.1021/acs.jmedchem.9b01101] Oerther B, Nedelcu A, Engel H, et al (2024) Update on PI-RADS Version 2.1 Diagnostic Performance Benchmarks for Prostate MRI: Systematic Review and Meta-Analysis. Radiology 312, e233337. https://doi.org/[10.1148/radiol.233337] Algohary A, Shiradkar R, Pahwa S, et al (2020) Combination of Peri-Tumoral and Intra-Tumoral Radiomic Features on Bi-Parametric MRI Accurately Stratifies Prostate Cancer Risk: A Multi-Site Study. Cancers (Basel) 12. https://doi.org/[10.3390/cancers12082200] Bai H, Xia W, Ji X, et al (2021) Multiparametric Magnetic Resonance Imaging-Based Peritumoral Radiomics for Preoperative Prediction of the Presence of Extracapsular Extension With Prostate Cancer. J Magn Reson Imaging 54, 1222-1230. https://doi.org/[10.1002/jmri.27678] Qiu Y, Liu YF, Shu X, et al (2023) Peritumoral Radiomics Strategy Based on Ensemble Learning for the Prediction of Gleason Grade Group of Prostate Cancer. Acad Radiol 30 Suppl 1, S1-s13. https://doi.org/[10.1016/j.acra.2023.06.011] Elhanani O, Ben-Uri R & Keren L (2023) Spatial profiling technologies illuminate the tumor microenvironment. Cancer Cell 41, 404-420. https://doi.org/[10.1016/j.ccell.2023.01.010] Ren X, Zhang L, Zhang Y, et al (2021) Insights Gained from Single-Cell Analysis of Immune Cells in the Tumor Microenvironment. Annu Rev Immunol 39, 583-609. https://doi.org/[10.1146/annurev-immunol-110519-071134] Han X, Gong Z, Guo Y, Tang W & Wei X (2024) Exploration of a noninvasive radiomics classifier for breast cancer tumor microenvironment categorization and prognostic outcome prediction. Eur J Radiol 175, 111441. https://doi.org/[10.1016/j.ejrad.2024.111441] Hu Y, Cai Z, Aierken N, et al (2025) Intra- and peri-tumoral radiomics based on dynamic contrast-enhanced MRI for prediction of benign disease in BI-RADS 4 breast lesions: a multicentre study. Radiat Oncol 20, 27. https://doi.org/[10.1186/s13014-025-02605-y] Liu HF, Wang M, Wang Q, et al (2024) Multiparametric MRI-based intratumoral and peritumoral radiomics for predicting the pathological differentiation of hepatocellular carcinoma. Insights Imaging 15, 97. https://doi.org/[10.1186/s13244-024-01623-w] Zhang H, Li X, Zhang Y, et al (2021) Diagnostic nomogram based on intralesional and perilesional radiomics features and clinical factors of clinically significant prostate cancer. J Magn Reson Imaging 53, 1550-1558. https://doi.org/[10.1002/jmri.27486] Hectors SJ, Chen C, Chen J, et al (2021) Magnetic Resonance Imaging Radiomics-Based Machine Learning Prediction of Clinically Significant Prostate Cancer in Equivocal PI-RADS 3 Lesions. J Magn Reson Imaging 54, 1466-1473. https://doi.org/[10.1002/jmri.27692] Ruan M, Liu Y, Yao K, et al (2024) Development and Validation of Interpretable Machine Learning Models for Clinically Significant Prostate Cancer Diagnosis in Patients With Lesions of PI-RADS v2.1 Score ≥3. J Magn Reson Imaging 60, 2130-2141. https://doi.org/[10.1002/jmri.29275] Piran Nanekaran N, Felefly TH, Schieda N, et al (2024) Prediction of prostate cancer recurrence after radiotherapy using a fused machine learning approach: Utilizing radiomics from pretreatment T2W MRI images with clinical and pathological information. Biomed Phys Eng Express. https://doi.org/[10.1088/2057-1976/ad8201] Bourbonne V, Fournier G, Vallières M, et al (2020) External Validation of an MRI-Derived Radiomics Model to Predict Biochemical Recurrence after Surgery for High-Risk Prostate Cancer. Cancers (Basel) 12. https://doi.org/[10.3390/cancers12040814] Merriman KM, Harmon SA, Belue MJ, et al (2023) Comparison of MRI-Based Staging and Pathologic Staging for Predicting Biochemical Recurrence of Prostate Cancer After Radical Prostatectomy. AJR Am J Roentgenol 221, 773-787. https://doi.org/[10.2214/ajr.23.29609] Tables Table 1 Baseline Characteristics of Prostate Cancer Patients. Variable Total(n=277) Training cohort (n = 193) Test cohort(n=84) p value PLR 122.64 ± 57.10 120.89±57.93 126.65±55.27 0.312 LMR 3.89±1.58 4.00±1.60 3.63±1.50 0.093 NLR 3.06±2.92 3.09±3.28 3.00±1.84 0.185 SII 548.92±491.64 560.15±542.86 523.12±347.95 0.889 SIRI 15.29±7.04 15.91±7.33 13.86±6.12 0.036 Age (years) 68.84±7.06 68.40±7.19 69.85±6.70 0.119 tPSA (ng/mL) 47.05±45.99 45.29±47.58 51.08±42.12 0.331 fPSA (ng/mL) 8.97±13.83 8.48±13.31 10.11±14.97 0.471 Ki-67 0.14±0.14 0.14±0.14 0.14±0.13 0.383 PSA density (ng/mL 2 ) 1.06±1.16 1.05±1.22 1.07±1.01 0.682 MTD(cm) 4.78±0.85 4.77±0.84 4.79±0.86 0.741 MLD(cm) 5.03±1.11 5.03±1.11 5.05±1.12 0.529 MAPD(cm) 3.85±0.76 3.83±0.73 3.90±0.81 0.475 TDLV(cm 3 ) 52.00±28.04 51.46±28.05 53.24±28.14 0.491 VOS(cm 3 ) 49.76±26.91 49.19±26.65 51.08±27.60 0.568 TMD(cm) 3.81±2.36 3.81±2.57 3.84±1.82 0.262 TV(cm 3 ) 24.80±38.60 25.96±43.48 22.14±23.88 0.447 P504s 0.302 0 24(8.7) 14(7.3) 10(11.9) 1 253(91.3) 179(92.7) 74(88.1) CK8/18 0.002 0 25(9.0) 10(5.2) 15(17.9) 1 252(91.0) 183(94.8) 69(82.1) Clinical N stage 0.527 0 152(54.9) 103(53.4) 49(58.3) 1 125(45.1) 90(46.6) 35(41.7) Gleason 0.570 6 29(10.5) 23(11.9) 6(7.1) 7 94(33.9) 68(35.2) 26(31.0) 8 61(22.0) 40(20.7) 21(25.0) 9 81(29.2) 55(28.5) 26(31.0) 10 12(4.3) 7(3.6) 5(6.0) ISUP grade group 0.635 1 29(10.5) 23(11.9) 6(7.1) 2 34(12.3) 24(12.4) 10(11.9) 3 60(21.7) 44(22.8) 16(19.1) 4 61(22.0) 40(20.7) 21(25.0) 5 93(33.5) 62(32.1) 31(36.9) SVI 0.171 0 154(55.6) 113(58.5) 41(48.8) 1 123(44.4) 80(41.5) 43(51.2) EPE 0.548 0 151(54.5) 108(56.0) 43(51.2) 1 126(45.5) 85(44.0) 41(48.8) SM 0.514 0 258(93.1) 178(92.2) 80(95.2) 1 19(6.9) 15(7.8) 4(4.8) PI-RADS 0.703 2 30(10.8) 19(9.8) 11(13.1) 3 59(21.3) 44(22.8) 15(17.9) 4 62(22.4) 44(22.8) 18(21.4) 5 126(45.5) 86(44.6) 40(47.6) Abbreviations: PLR platelet-to-lymphocyte ratio; LMR lymphocyte-to-monocyte ratio; NLR neutrophil-to-lymphocyte ratio; SII systemic immune-inflammation index; SIRI systemic inflammation response index; tPSA total prostate-specific antigen; fPSA free prostate-specific antigen; Ki-67 antigen identified by monoclonal antibody Ki-67; MTD maximum transverse diameter; MLD maximum longitudinal diameter; MAPD maximum anteroposterior diameter; TDLV triaxial diameter-derived prostate volume; VOS volumetric segmentation-derived prostate volume; TMD tumor maximum diameter; TV tumor volume; p504s α-methylacyl-coA racemase; CK8/18 Cytokeratin 8/18; Gleason Prostate Cancer Gleason Grading System; ISUP International Society of Urological Pathology Grade Groups; SVI seminal vesicle invasion; EPE extraprostatic extension; SM positive surgical margins; PI-RADS Prostate Imaging Reporting and Data System Table 2 Univariate and multivariate analysis for BCR. Univariate analysis Multivariate analysis Variable OR 95% CI p value OR 95% CI p value Ki-67 0.014 0.002-0.08 ＜0.001 6.459 0.165-253.48 0.403 CK8/18 0.236 0.174-0.322 ＜0.001 1.75 0.152-20.086 0.706 P504s 0.252 0.185-0.342 ＜0.001 5.116 0.23-113.636 0.387 Clinical N stage 0.429 0.293-0.626 ＜0.001 7.335 2.385-22.579 0.004 PSA density (ng/mL 2 ) 0.571 0.456-0.716 ＜0.001 1.267 0.745-2.155 0.464 EPE 0.574 0.396-0.832 0.014 7.672 2.321-25.381 0.005 SVI 0.6 0.410-0.877 0.027 6.508 1.565-27.058 0.031 MAPD(cm) 0.685 0.633-0.742 ＜0.001 1.932 0.457-8.166 0.452 LMR 0.694 0.641-0.751 ＜0.001 1.151 0.702-1.887 0.640 MTD(cm) 0.724 0.678-0.773 ＜0.001 0.133 0.035-0.503 0.013 ISUP 0.737 0.682-0.796 ＜0.001 1.057 0.341-3.277 0.936 PI-RADS 0.738 0.688-0.792 ＜0.001 2.121 1.132-3.975 0.049 TMD(cm) 0.755 0.7-0.815 ＜0.001 0.696 0.418-1.158 0.242 MLD(cm) 0.759 0.715-0.806 ＜0.001 0.707 0.312-1.603 0.486 NLR 0.774 0.706-0.848 ＜0.001 2.187 1.157-4.137 0.043 Gleason 0.842 0.811-0.875 ＜0.001 1.141 0.327-3.979 0.862 SIRI 0.909 0.89-0.928 ＜0.001 0.97 0.834-1.127 0.740 fPSA (ng/mL) 0.966 0.947-0.984 0.003 1.024 0.975-1.076 0.425 VOS(cm 3 ) 0.971 0.965-0.978 ＜0.001 1.038 0.915-1.178 0.627 TDLV(cm 3 ) 0.973 0.967-0.979 ＜0.001 0.997 0.875-1.135 0.972 Age (years) 0.979 0.975-0.984 ＜0.001 0.948 0.885-1.014 0.192 tPSA (ng/mL) 0.984 0.979-0.989 ＜0.001 0.984 0.968-1.002 0.139 PLR 0.991 0.988-0.993 ＜0.001 1.014 0.998-1.03 0.160 TV(cm 3 ) 0.991 0.985-0.997 0.018 1.009 0.989-1.029 0.458 SII 0.998 0.998-0.999 ＜0.001 0.995 0.99-1 0.078 SM 1.5 0.631-3.568 0.442 Abbreviations: BCR biochemical recurrence; Ki-67 antigen identified by monoclonal antibody Ki-67; CK8/18 Cytokeratin 8/18; p504s α-methylacyl-coA racemase; EPE extraprostatic extension; SVI seminal vesicle invasion; MAPD maximum anteroposterior diameter; LMR lymphocyte-to-monocyte ratio; MTD maximum transverse diameter; ISUP International Society of Urological Pathology Grade Groups; PI-RADS Prostate Imaging Reporting and Data System; TMD tumor maximum diameter; MLD maximum longitudinal diameter; NLR neutrophil-to-lymphocyte ratio; Gleason Prostate Cancer Gleason Grading System; SIRI systemic inflammation response index; fPSA free prostate-specific antigen; VOS volumetric segmentation-derived prostate volume; TDLV triaxial diameter-derived prostate volume; tPSA total prostate-specific antigen; PLR platelet-to-lymphocyte ratio; TV tumor volume; SII systemic immune-inflammation index; SM positive surgical margins Table 3 Diagnostic performance of the different model for predicting BCR in the training and test cohorts. Models Accuracy AUC 95% CI Sensitivity Specificity Cohort Clinic 0.860 0.897 0.839 - 0.954 0.811 0.872 train IntraPeri4mm_ADC 0.824 0.902 0.853 - 0.951 0.730 0.846 train IntraPeri4mm_T2 0.710 0.842 0.778 - 0.906 0.865 0.673 train Combined 0.938 0.978 0.961 - 0.994 0.919 0.942 train Clinic 0.643 0.731 0.602 - 0.861 0.789 0.600 test IntraPeri4mm_ADC 0.571 0.706 0.594 - 0.819 0.789 0.508 test IntraPeri4mm_T2 0.488 0.662 0.529 - 0.795 0.895 0.369 test Combined 0.726 0.810 0.707 - 0.912 0.737 0.723 test Abbreviations: BCR biochemical recurrence; Clinic Clinical model constructed from independent risk factors; IntraPeri4mm_ADC ADC model with fusion of Intra- and Peritumoral radiomics features; IntraPeri4mm_T2 T2WI model with fusion of Intra- and Peritumoral radiomics features; Combined The model combining clinical risk factors and T2WI/ADC-based Intra- and Peritumoral radiomics signatures Additional Declarations No competing interests reported. Supplementary Files Supplementary.docx Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 10 Aug, 2025 Reviews received at journal 10 Aug, 2025 Reviewers agreed at journal 10 Aug, 2025 Reviews received at journal 01 Aug, 2025 Reviewers agreed at journal 26 Jul, 2025 Reviewers agreed at journal 24 Jul, 2025 Reviewers invited by journal 22 Jul, 2025 Editor assigned by journal 22 Jul, 2025 Submission checks completed at journal 22 Jul, 2025 First submitted to journal 21 Jul, 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-7173672","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":490433514,"identity":"45e1bbec-af37-433f-9fb5-8e6672b66395","order_by":0,"name":"Xuelian Zhao","email":"","orcid":"","institution":"Lanzhou University","correspondingAuthor":false,"prefix":"","firstName":"Xuelian","middleName":"","lastName":"Zhao","suffix":""},{"id":490433516,"identity":"9ab6f6f6-2039-4c12-8027-955b40078292","order_by":1,"name":"Fei Jia","email":"","orcid":"","institution":"Lanzhou University","correspondingAuthor":false,"prefix":"","firstName":"Fei","middleName":"","lastName":"Jia","suffix":""},{"id":490433518,"identity":"fd54141d-5d7f-48a0-8ca1-f6300b6e889b","order_by":2,"name":"Hao Li","email":"","orcid":"","institution":"Lanzhou University","correspondingAuthor":false,"prefix":"","firstName":"Hao","middleName":"","lastName":"Li","suffix":""},{"id":490433519,"identity":"61d82fbd-29a7-4761-8b37-94d40c7c51df","order_by":3,"name":"Ning Zhang","email":"","orcid":"","institution":"Lanzhou University","correspondingAuthor":false,"prefix":"","firstName":"Ning","middleName":"","lastName":"Zhang","suffix":""},{"id":490433520,"identity":"3d1df86d-58cf-4037-95a9-cf4e82357eba","order_by":4,"name":"Xinlin Han","email":"","orcid":"","institution":"Lanzhou University","correspondingAuthor":false,"prefix":"","firstName":"Xinlin","middleName":"","lastName":"Han","suffix":""},{"id":490433522,"identity":"96963487-76a2-4910-972c-a9711a2706cb","order_by":5,"name":"Yanli Jiang","email":"","orcid":"","institution":"Lanzhou University Second Hospital","correspondingAuthor":false,"prefix":"","firstName":"Yanli","middleName":"","lastName":"Jiang","suffix":""},{"id":490433523,"identity":"824325c8-4095-4d05-9598-48b0972efbc6","order_by":6,"name":"Hong Liu","email":"","orcid":"","institution":"Lanzhou University Second Hospital","correspondingAuthor":false,"prefix":"","firstName":"Hong","middleName":"","lastName":"Liu","suffix":""},{"id":490433524,"identity":"4d835377-7d64-423f-ab52-5e6b627bb783","order_by":7,"name":"Zhilong Dong","email":"","orcid":"","institution":"Lanzhou University Second Hospital, Key Laboratory of Urological Diseases in Gansu Province","correspondingAuthor":false,"prefix":"","firstName":"Zhilong","middleName":"","lastName":"Dong","suffix":""},{"id":490433525,"identity":"6e13de75-b769-4c55-bc49-892e35508a78","order_by":8,"name":"Yudong Zhang","email":"","orcid":"","institution":"the First Affiliated Hospital with Nanjing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yudong","middleName":"","lastName":"Zhang","suffix":""},{"id":490433527,"identity":"97bc06b2-b814-4f78-b9ed-64a6bc6c3320","order_by":9,"name":"Kai Ai","email":"","orcid":"","institution":"Philips healthcare","correspondingAuthor":false,"prefix":"","firstName":"Kai","middleName":"","lastName":"Ai","suffix":""},{"id":490433528,"identity":"2d672bd1-9c4c-4305-a0b2-6ca7b3e4960b","order_by":10,"name":"Jing Zhang","email":"","orcid":"","institution":"Lanzhou University Second Hospital","correspondingAuthor":false,"prefix":"","firstName":"Jing","middleName":"","lastName":"Zhang","suffix":""},{"id":490433529,"identity":"6cc02752-d67a-4bf3-ad94-a0d30a657b32","order_by":11,"name":"Guangyao Liu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABAUlEQVRIiWNgGAWjYLCCBwZQxocCMGWAWykMJEDVMM4wIFoLlGbmIUYLv/QZA4aEgsN2G473PpO2MahLbGBv3ibBUHMHpxbJvhygFoPDyRvOHDeTzjFgS2zgOVYmwXDsGU4tBmd4IFoMbqSxAbXwJDZI5JhJMDYcxqnFHq7l/jM2aQsDicQG+Tf4tRjwQLTYGdxgY5NmMDAA2sKDX4vEGbYCoJb0BMkzacyWPQYJxm08acUWCcdwa+HvYd7A8OGPtT3f8WOMN35U1Mn2sx/eeONDDW4tDAwc5j+AZOKCAwwsEiA+G4hIwKOBgYH9AYi0l29gYP6AV+EoGAWjYBSMWAAA2HhNcf48MeUAAAAASUVORK5CYII=","orcid":"","institution":"Lanzhou University Second Hospital","correspondingAuthor":true,"prefix":"","firstName":"Guangyao","middleName":"","lastName":"Liu","suffix":""}],"badges":[],"createdAt":"2025-07-21 06:08:08","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7173672/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7173672/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":87694299,"identity":"a0464f4c-9e0f-4210-bcc1-3d8864f5b265","added_by":"auto","created_at":"2025-07-28 05:40:32","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1794889,"visible":true,"origin":"","legend":"\u003cp\u003eTumor lesion delineation. \u003cstrong\u003eA \u003c/strong\u003eThe ADC sequence shows the lesion area. B The intratumoral region outlined by the radiologist. \u003cstrong\u003eC\u003c/strong\u003e The peritumoral region generated by expanding 4 mm outward from the tumor boundary using Python software.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-7173672/v1/ae13636fe313ceb8294a619d.png"},{"id":87693551,"identity":"184f7bf6-baa7-4795-85d3-97573493a52b","added_by":"auto","created_at":"2025-07-28 05:32:32","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":646013,"visible":true,"origin":"","legend":"\u003cp\u003eThe radiomic features selected from ADC \u003cstrong\u003e(A-C)\u003c/strong\u003e and T2WI \u003cstrong\u003e(D-F)\u003c/strong\u003eare displayed separately, including intratumoral features \u003cstrong\u003e(A, D)\u003c/strong\u003e, peritumoral features \u003cstrong\u003e(B, E)\u003c/strong\u003e, and combined intratumoral-peritumoral features \u003cstrong\u003e(C, F)\u003c/strong\u003e.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-7173672/v1/cc21557f588fd261480cf064.png"},{"id":87693544,"identity":"ee9bc59c-a5b5-4ba5-bc08-5d3218e87de7","added_by":"auto","created_at":"2025-07-28 05:32:32","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":984264,"visible":true,"origin":"","legend":"\u003cp\u003eReceiver operating characteristic (ROC) curve analysis with DeLong test comparison in \u003cstrong\u003e(A, C)\u003c/strong\u003e training cohort, and \u003cstrong\u003e(B, D)\u003c/strong\u003e test cohort for BCR prediction.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAbbreviations: \u003c/strong\u003eBCR biochemical recurrence; Clinic Clinical model constructed from independent risk factors; IntraPeri4mm_ADC ADC model with fusion of Intra- and Peritumoral radiomics features; IntraPeri4mm_T2 T2WI model with fusion of Intra- and Peritumoral radiomics features; Combined The model combining clinical risk factors and T2WI/ADC-based Intra- and Peritumoral radiomics signatures.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-7173672/v1/bfb22991801a17d032476e62.png"},{"id":87693550,"identity":"59fcc4d9-98b6-407f-8750-15c96dce5c39","added_by":"auto","created_at":"2025-07-28 05:32:32","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":872692,"visible":true,"origin":"","legend":"\u003cp\u003eDecision curve analysis (DCA) of preoperative prediction models for BCR following radical prostatectomy in train \u003cstrong\u003e(A)\u003c/strong\u003e, and test \u003cstrong\u003e(B)\u003c/strong\u003ecohorts, demonstrating superior net benefit.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAbbreviations: \u003c/strong\u003eBCR biochemical recurrence; Clinic Clinical model constructed from independent risk factors; IntraPeri4mm_ADC ADC model with fusion of Intra- and Peritumoral radiomics features; IntraPeri4mm_T2 T2WI model with fusion of Intra- and Peritumoral radiomics features; Combined The model combining clinical risk factors and T2WI/ADC-based Intra- and Peritumoral radiomics signatures\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-7173672/v1/b669ea852e3fba25205f62a7.png"},{"id":87693530,"identity":"5ed26537-05f9-408c-8488-eed7151529e4","added_by":"auto","created_at":"2025-07-28 05:32:31","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1575531,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eA\u003c/strong\u003e Beeswarm plot of SHAP values in the combined model, showing the relative contribution of preoperative predictors to BCR. \u003cstrong\u003eB\u003c/strong\u003e Waterfall plot, and \u003cstrong\u003eC\u003c/strong\u003e force plot demonstrate correctly predicted BCR cases.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAbbreviations:\u003c/strong\u003e BCR biochemical recurrence; EPE extraprostatic extension; ADC label_0 ADC negative prediction Rad_Score; T2 label_1 T2WI positive prediction Rad_Score; MTD maximum transverse diameter; ADC label_1 ADC positive prediction Rad_Score; T2 label_0 T2WI negative prediction Rad_Score; SVI seminal vesicle invasion; PI-RADS Prostate Imaging Reporting and Data System; NLR neutrophil-to-lymphocyte ratio\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-7173672/v1/6d3ba1255dcf7d1cf3cf9de4.png"},{"id":87694842,"identity":"1e4f946e-2101-4368-a2df-fa280e7cff80","added_by":"auto","created_at":"2025-07-28 05:48:40","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5988831,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7173672/v1/9b7fb272-51cc-4b19-9085-586477784cd0.pdf"},{"id":87693531,"identity":"ef7ce41b-a91b-491f-8a61-d80dd8b7ce22","added_by":"auto","created_at":"2025-07-28 05:32:31","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":766739,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementary.docx","url":"https://assets-eu.researchsquare.com/files/rs-7173672/v1/38c4e585c4a03f59dccab23e.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Interpretable Machine Learning Model Based on Intra- and Peritumoral MRI Radiomics for Predicting Biochemical Recurrence After Radical Prostatectomy","fulltext":[{"header":"Introduction","content":"\u003cp\u003eProstate cancer (PCa) ranks as the second most prevalent malignancy in men worldwide [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e], with rapidly increasing incidence and mortality rates in China. Radical prostatectomy (RP) [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e] remains the primary treatment for organ-confined and locally advanced PCa. Biochemical recurrence (BCR) after RP is defined as prostate-specific antigen (PSA) levels rising above 0.1 ng/ml in two consecutive measurements following post-operative PSA nadir (\u0026lt; 0.1 ng/ml) [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Notably, 27%-53% of post-RP patients experience PSA elevation, with BCR strongly correlating with increased risks of local recurrence, distant metastasis, and overall mortality. Accurate BCR prediction therefore enables proactive surveillance of high-risk patients to improve outcomes. Previous investigations of BCR risk factors primarily focused on clinical parameters including preoperative PSA, biopsy Gleason score, clinical stage, age, and percentage of positive cores. However, with the widespread clinical application of MRI, conventional clinical-based assessment methods (such as D'Amico risk classification, Cancer of the Prostate Risk Assessment [CAPRA] score, and Kattan nomogram) [\u003cspan additionalcitationids=\"CR6 CR7 CR8\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e–\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e] have become insufficiently comprehensive. Radiomics [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e] enables extraction of high-throughput imaging features to evaluate tumor heterogeneity at microscopic scales, identifying quantitative imaging biomarkers for diagnosis, classification, and prognosis prediction. This non-invasive approach may reduce complications associated with biopsy [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Duenweg et al. [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e] developed a BCR prediction model based on T2-weighted imaging (T2WI) features, achieving an AUC of 0.97, demonstrating excellent diagnostic performance. Similarly, Zhu et al. [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]employed comparable methodology to construct a radiomics model for preoperative prediction of post-radiotherapy BCR in PCa patients, with the model attaining an AUC of 0.83 and enabling accurate prognostic assessment. While intratumoral radiomic features have demonstrated significant value in tumor characterization, emerging research has revealed the critical role of peritumoral microenvironment in tumorigenesis and progression[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. The peritumoral region, while macroscopically normal, exhibits microscopic heterogeneity at tumor-normal tissue interfaces. Studies confirm genetic, epigenetic, and transcriptomic alterations in peritumoral tissues of epithelial tumors (colorectal, bladder, prostate, and breast) [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Gu et al.[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e] first proposed the peritumoral microenvironment (PME) concept in hepatocellular carcinoma, demonstrating PME heterogeneity's significance throughout tumor development and its diagnostic/therapeutic implications. Multiple studies [\u003cspan additionalcitationids=\"CR18\" citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e–\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]reveal superior diagnostic and predictive performance of peritumoral radiomics models compared to conventional intratumoral approaches, serving as crucial complementary tools. Some findings[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]suggest peritumoral features may outperform intratumoral features in certain cancers, though peritumoral radiomics exploration remains limited in PCa.\u003c/p\u003e\u003cp\u003eMachine learning (ML) algorithms effectively uncover latent associations between radiomics features and clinical outcomes to build predictive models, but their clinical translation is hindered by the opaque 'black-box' decision-making process and lack of interpretability [\u003cspan additionalcitationids=\"CR22\" citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e–\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. SHapley Additive exPlanations (SHAP) analysis [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e] evaluates the contribution of individual features to model predictions through nonlinear association analysis, elucidating their roles in the predictive process. This approach not only enhances model interpretability but also identifies potential biomarkers driving BCR, thereby providing novel evidence for the correlation between imaging/clinical features and tumor biological behavior. Ultimately, SHAP analysis facilitates the translation of ML models into clinical decision-making tools.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cb\u003ePatients\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThis study, conducted in accordance with the Declaration of Helsinki, was approved by the Medical Ethics Committee (No. 2024A-1253) with waived informed consent. We retrospectively analyzed 300 patients who underwent RP between January 2013 and February 2024. Inclusion criteria comprised: (1) preoperative 3.0T MRI (complete T2WI and DWI-derived apparent diffusion coefficient (ADC) maps); (2) pathologically confirmed prostatic adenocarcinoma; (3) BCR occurrence or ≥ 1-year follow-up if BCR-negative. Exclusion criteria included: MRI-unidentifiable tumors, non-adenocarcinoma pathology, incomplete data, preoperative radiotherapy/chemotherapy, or metastatic disease. Follow-up spanned from surgery to BCR date to March 31, 2025. Among the 277 included patients, 193 were assigned to the training cohort (37 with BCR and 156 without) and 84 to the test cohort (19 with BCR and 65 without). (Supplementary Fig.\u0026nbsp;1).\u003c/p\u003e\u003cp\u003e\u003cb\u003eMagnetic resonance imaging protocol\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAll MRI examinations were performed on a 3.0T scanner using an integrated 18-channel phased-array body coil with patients in the supine position. The imaging protocol included axial T2WI (TR/TE = 6980/104 ms, slice thickness = 3 mm, slice spacing = 0 mm, FOV = 200×200 mm, matrix = 384×384) and axial DWI using single-shot echo-planar imaging (TR/TE = 6540/60 ms, slice thickness = 3 mm, slice spacing = 0 mm, FOV = 200×200 mm, matrix = 130×130) with b-values of 50, 1000, and 1500 s/mm² for ADC map generation, consistent with Prostate Imaging Reporting and Data System (PI-RADS v2.1) recommendations [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e] that suggest using low (0-100 s/mm²) and intermediate (800–1000 s/mm²) b-values for ADC calculation while higher b-values (≥ 1400 s/mm²) improve PCa detection sensitivity.\u003c/p\u003e\u003cp\u003e\u003cb\u003eTumor segmentation\u003c/b\u003e\u003c/p\u003e\u003cp\u003eRegion of Interests (ROIs) were delineated on both T2WI and ADC sequences. Prior to delineation, all images were resampled to an isotropic voxel spacing of 1 × 1 × 1 mm\u003csup\u003e3\u003c/sup\u003e and underwent grayscale normalization to compensate for voxel spatial heterogeneity and maintain intensity consistency.\u003c/p\u003e\u003cp\u003eThree experienced radiologists collaboratively reviewed each patient's complete MRI dataset. Two radiologists independently outlined the tumor boundaries, while the third performed quality control by verifying all ROIs and correcting or redrawing questionable delineations. Tumor localization followed PI-RADS v2.1 criteria. Using ITK-SNAP software (version 4.0.0), manual slice-by-slice ROI delineation was performed along the tumor margins on T2WI and ADC maps. The Python package SimpleITK (version 3.7.12) was then employed to automatically expand each intratumoral ROI by 4 mm, followed by manual exclusion of periprostatic non-target tissues (e.g. fat, seminal vesicles, rectum, urethra, and bladder) to derive peritumoral regions. Representative examples of intratumoral and peritumoral ROI expansion are illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eFeature extraction and selection\u003c/b\u003e\u003c/p\u003e\u003cp\u003eRadiomic features were extracted from both intratumoral (Intra) and peritumoral (Peri_4mm) ROIs using Python's pyradiomics library following established protocols [\u003cspan additionalcitationids=\"CR28\" citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e–\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. Feature selection involved: Eliminating features with Spearman's rank correlation coefficient \u0026gt; 0.9 to remove highly correlated features; recursive feature elimination to remove redundancy and LASSO regression to construct the radiomics signature (Rad_Sig), with identical pipelines applied to both regions to ensure comparability. Normality was assessed using Student's t-test (normal distributions) or Mann-Whitney U test (non-normal distributions). Clinical variables were analyzed similarly (t-test/Mann-Whitney U for continuous variables; χ² test for categorical variables), with univariate and multivariate analyses identifying independent BCR risk factors.\u003c/p\u003e\u003cp\u003e\u003cb\u003eConstruction and comparison of prediction models\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe Rad_Sig from both Intra and Peri_4mm regions were input into ExtraTrees machine learning models to generate individualized radiomics scores (Rad_Score), constructing three predictive models: Intra, Peri_4mm, and IntraPeri_4mm models, while clinical independent risk factors were used to build a Combined, which was further integrated with radiomics features to develop a clinic-radiomics (Combined) model. Model performance was rigorously evaluated through receiver operating characteristic (ROC) curve analysis (reporting AUC values and optimal cutoffs maximizing Youden's index), Delong's test for AUC comparisons, and decision curve analysis (DCA) to assess clinical utility, supplemented by SHAP analysis for both cohort-level and individual patient-level interpretation of feature importance in BCR.\u003c/p\u003e\u003ch2\u003eStatistical analysis\u003c/h2\u003e\u003cp\u003eAll data analyses were performed using Python 3.7.12 on the OnekeyAI platform (version 4.9.1), with statistical evaluations conducted using statsmodels (version 0.13.2) and radiomics feature extraction implemented through PyRadiomics (version 3.0.1). Statistical significance was defined as a two-sided p-value ≤ 0.05. The intraclass correlation coefficient (ICC) was computed to evaluate feature stability, with features demonstrating ICC \u0026gt; 0.75 retained for subsequent analysis. The entire workflow from data collection through model construction, evaluation, and interpretation is illustrated in Supplementary Fig.\u0026nbsp;2\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eBasic characteristics of patients\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAmong the 277 patients included in the study, biochemical recurrence (BCR) occurred in 56 cases (20.2%), with 37 cases (19.1%) in the training cohort and 19 cases (22.6%) in the test cohort. The baseline characteristics of the cohorts are presented in Table 1. The inter-group p-values exceeding 0.05 for all comparative analyses confirmed no statistically significant differences between cohorts, demonstrating unbiased group allocation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDevelopment and validation of prediction models\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFrom T2WI and ADC sequences, we extracted 7336 radiomic features with identical numbers of Intra and Peri_4mm features (n=1834 each) per sequence, constructing Intra, Peri_4mm and IntraPeri_4mm models. For ADC sequences, 4 Intra- and 3 Peri_4mm-features correlated with BCR, yielding training and test AUC of 0.883 (95%CI:0.832-0.934)、0.694 (95%CI:0.582-0.805) for Intra_ADC and 0.889 (95%CI:0.841-0.938)、0.518 (95%CI:0.378-0.657) for Peri_4mm_ADC. T2WI sequences showed 4 Intra- and 9 Peri_4mm-feature correlations with training and test AUC of 0.732 (95%CI:0.646-0.846)、0.587 (95%CI:0.447-0.728) for Intra_T2 and 0.864 (95%CI:0.802-0.926)、0.637 (95%CI:0.483-0.791) for Peri_4mm_T2. The IntraPeri_4mm models demonstrated superior performance, ADC-based (5 features) achieved 0.902 (95%CI:0.853-0.951)、0.706 (95%CI:0.594-0.819), while T2WI-based (10 features) reached 0.842 (95%CI:0.778-0.906)、0.662 (95%CI:0.529-0.795), with feature distributions shown in Figure 2.\u003c/p\u003e\n\u003cp\u003eUnivariate logistic regression revealed that nearly all factors were significantly associated with biochemical recurrence (BCR). Multivariate analysis identified the following independent risk factors for BCR: extraprostatic extension (EPE), clinical nodal stage (N), seminal vesicle invasion (SVI), PI-RADS score, neutrophil-to-lymphocyte ratio (NLR), and maximum transverse diameter of the prostate (MTD) (Table 2). Based on these independent risk factors, we constructed the Clinic model, which achieved AUC of 0.897 (95% CI: 0.839-0.954) in the training set and 0.731 (95% CI: 0.602-0.861) in the test set (Table 3).\u003c/p\u003e\n\u003cp\u003eBy integrating the clinical independent risk factors with the Rad_Score from both ADC and T2WI models, we developed a superior diagnostic Combined model. In the training set, the Combined model achieved AUC of 0.978 (95% CI: 0.961\u0026ndash;0.994), while in the test set, the AUC was 0.810 (95% CI: 0.707\u0026ndash;0.912) (Table 3 and Figure 3A, 3B).\u003c/p\u003e\n\u003cp\u003eDeLong\u0026rsquo;s test (Figure 3C, 3D) revealed that in the training set, the Combined model showed statistically significant differences compared to the Clinic, IntraPeri_4mm_ADC, and IntraPeri_4mm_T2 models. In the test set, the Combined model demonstrated significant differences versus the IntraPeri_4mm_ADC and IntraPeri_4mm_T2 models, but no significant difference compared to the Clinic model (p = 0.141). The Combined model exhibited significantly positive Net Reclassification Improvement (NRI) and Integrated Discrimination Improvement (IDI) values when compared to the IntraPeri_4mm_ADC, IntraPeri_4mm_T2, and Clinic models (Supplementary Figure 3.). DCA confirmed that the Combined model provided greater net clinical benefit than the other models (Figure 4).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eModel visualization and interpretation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe SHAP analysis revealed the following order of feature importance for BCR prediction (Figure 5A). EPE demonstrated the highest contribution, followed by ADC negative prediction Rad_Score (ADC label_0), T2WI positive prediction Rad_Score (T2 label_1), MTD, ADC positive prediction Rad_Score (ADC label_1), T2WI negative prediction Rad_Score (T2 label_0), SVI, PI-RADS, NLR and N. The SHAP analysis demonstrated that EPE exhibited the strongest predictive influence, as evidenced by its widest band width, with high feature values (red dots) predominantly distributed on the right side. Individual case analysis using waterfall (Figure. 5B) and force-directed (Figure. 5C) plots demonstrated that all examined factors except EPE contributed positively to BCR prediction. Among these ADC label_0 exhibited the strongest predictive influence, followed by EPE as the secondary determinant.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe rapid proliferation of cancer cells induces relative hypoxia in the peritumoral region, where hypoxia-inducible factor-1 (HIF-1) and other hypoxia-related mediators remodel the tumor microenvironment by altering vascular and lymphatic networks [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. Peritumoral inflammatory responses and immune cell infiltration further enhance microenvironmental heterogeneity [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. Building upon our investigation of intratumoral heterogeneity, we integrated both intratumoral and peritumoral features, which resulted in improved model performance: the AUC increased from 0.883 to 0.902 for ADC-based models and from 0.732 to 0.842 for T2WI-based models. These enhancements confirm the value of peritumoral radiomic features in complementing intratumoral characteristics for BCR assessment. Peritumoral radiomics quantitatively evaluates tumor microenvironment heterogeneity through imaging biomarkers. When combined with intratumoral features, it provides a more comprehensive tumor profile [\u003cspan additionalcitationids=\"CR33\" citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e], offering objective and multidimensional quantitative analysis for tumor characterization. Our findings align with Algohari et al. [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e], who reported significantly improved prostate cancer risk stratification by incorporating peritumoral radiomic features. Similarly, Zhang et al. [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e] demonstrated enhanced diagnostic performance for clinically significant prostate cancer using combined intratumoral-peritumoral radiomics and PSA levels, but their model lacked comprehensive clinical variables for optimal predictive performance.\u003c/p\u003e\u003cp\u003eMost previous studies on BCR prediction models focused solely on either clinical data or radiomic features. Although the D'Amico risk classification, CAPRA score, and Kattan nomogram demonstrate satisfactory performance, they lack quantitative assessment of tumor heterogeneity. Hectors et al[\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e] and Ruan et al [\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e] investigated prostate cancer heterogeneity using multiparametric MRI-based radiomics, Piran et al [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e] developed a post-radiotherapy biochemical recurrence (BCR) prediction model utilizing T2WI intratumoral and peritumoral radiomic features, achieving an AUC of 0.794. Although radiomics provides satisfactory characterization of tumor heterogeneity, its diagnostic performance shows limited stability, poor integration with current clinical guidelines, and suboptimal clinical applicability.\u003c/p\u003e\u003cp\u003eIn our study, we first established a Clinic model incorporating laboratory tests, PI-RADS scores, and pathological Gleason grading, which demonstrated AUC of 0.897 in the training set and 0.731 in the test set - comparable to established models (D'Amico score AUC 0.710; CAPRA score AUC 0.770). Furthermore, we developed an integrated model combining clinical independent risk factors with T2WI/ADC-derived Rad_Score for post-prostatectomy BCR prediction, showing superior performance with an AUC of 0.978 compared to standalone radiomic or clinical models, along with improved sensitivity, specificity, and accuracy. Bourbonne et al.[\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]developed a prediction model based on clinical data and ADC maps, with AUC values of 0.68, 0.82, and 0.86 for the clinical, ADC radiomics, and combined models, respectively. The integration of clinical data and radiomics significantly improved diagnostic performance, along with enhanced sensitivity, specificity and accuracy, this result highly consistent with our results. Furthermore, our Combined model showed statistically significant improvements in both the NRI and IDI, demonstrating that the combined use of clinical and radiomic features provides a more comprehensive representation of tumor biological behavior. These results strongly support the necessity of data integration. We employed SHAP interpretability analysis to elucidate both global and local interpretability of the Combined model. The beeswarm plot reveals population-level impact patterns of clinical and imaging features on prediction outcomes through horizontal density distribution. Each point in the beeswarm plot represents the SHAP value of a feature for an individual patient, with color gradient from red to blue indicating feature values from high to low. Wider color regions in the plot indicate greater feature importance. In the beeswarm plot, EPE exhibited the widest color region, with red points predominantly distributed on the right side, indicating a positive correlation with BCR occurrence. This finding is highly consistent with the results of Merriman et al. [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e].In both the Partin and MSKCC nomograms, EPE is consistently identified as a key prognostic parameter, further underscoring the importance of EPE detection in clinical practice.\u003c/p\u003e\u003cp\u003eThis study has several limitations that warrant consideration: (1) As a single-center retrospective investigation, potential selection bias may exist, necessitating test through multicenter prospective studies; (2) Manual lesion delineation introduces subjectivity, prompting our future adoption of semi-automated or fully automated segmentation methods to improve objectivity; (3) Currently focused solely on recurrence prediction, future work will incorporate salvage treatment outcomes (e.g. androgen deprivation therapy/adjuvant radiotherapy) to develop a comprehensive closed-loop system for personalized therapeutic decision-making.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eour study developed a preoperative prediction model for post-radical prostatectomy BCR by integrating intratumoral and peritumoral MRI radiomic features with clinical independent risk factors, which facilitates precise individualized treatment planning and proactive monitoring of high-risk patients.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\n\u003cp\u003eG.Y.L. is the guarantor of integrity of the entire manuscript. G.Y.L.,X.L.Z.and J.Z. designed the study and searched literature. X.L.Z., F.J., H.L., N.Z.,X.L.H.,Y.L.J.and H.L. collected the clinical and imaging data. X.L.Z., F.J. and H.L. analyzed data. X.L.Z., F.J. and H.L. drafted the manuscript. All authors edited and approved the final manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eSiegel RL, Giaquinto AN \u0026amp; Jemal A (2024) Cancer statistics, 2024. CA Cancer J Clin 74, 12-49. https://doi.org/[10.3322/caac.21820]\u003c/li\u003e\n\u003cli\u003eGandaglia G, Montorsi F, Karakiewicz PI \u0026amp; Sun M (2015) Robot-assisted radical prostatectomy in prostate cancer. Future Oncol 11, 2767-2773. https://doi.org/[10.2217/fon.15.169]\u003c/li\u003e\n\u003cli\u003eShee K, de la Calle CM, Chang AJ, et al (2022) Addition of Enzalutamide to Leuprolide and Definitive Radiation Therapy Is Tolerable and Effective in High-Risk Localized or Regional Nonmetastatic Prostate Cancer: Results From a Phase 2 Trial. Adv Radiat Oncol 7, 100941. https://doi.org/[10.1016/j.adro.2022.100941]\u003c/li\u003e\n\u003cli\u003eTilki D, Preisser F, Graefen M, Huland H \u0026amp; Pompe RS (2019) External Validation of the European Association of Urology Biochemical Recurrence Risk Groups to Predict Metastasis and Mortality After Radical Prostatectomy in a European Cohort. Eur Urol 75, 896-900. https://doi.org/[10.1016/j.eururo.2019.03.016]\u003c/li\u003e\n\u003cli\u003eD\u0026apos;Amico AV, Whittington R, Malkowicz SB, et al (1998) Biochemical outcome after radical prostatectomy, external beam radiation therapy, or interstitial radiation therapy for clinically localized prostate cancer. Jama 280, 969-974. https://doi.org/[10.1001/jama.280.11.969]\u003c/li\u003e\n\u003cli\u003eKattan MW, Eastham JA, Stapleton AM, Wheeler TM \u0026amp; Scardino PT (1998) A preoperative nomogram for disease recurrence following radical prostatectomy for prostate cancer. J Natl Cancer Inst 90, 766-771. https://doi.org/[10.1093/jnci/90.10.766]\u003c/li\u003e\n\u003cli\u003eKattan MW, Wheeler TM \u0026amp; Scardino PT (1999) Postoperative nomogram for disease recurrence after radical prostatectomy for prostate cancer. J Clin Oncol 17, 1499-1507. https://doi.org/[10.1200/jco.1999.17.5.1499]\u003c/li\u003e\n\u003cli\u003eLemaitre L, Puech P, Poncelet E, et al (2009) Dynamic contrast-enhanced MRI of anterior prostate cancer: morphometric assessment and correlation with radical prostatectomy findings. Eur Radiol 19, 470-480. https://doi.org/[10.1007/s00330-008-1153-0]\u003c/li\u003e\n\u003cli\u003ePunnen S, Freedland SJ, Presti JC, Jr., et al (2014) Multi-institutional validation of the CAPRA-S score to predict disease recurrence and mortality after radical prostatectomy. Eur Urol 65, 1171-1177. https://doi.org/[10.1016/j.eururo.2013.03.058]\u003c/li\u003e\n\u003cli\u003eBi WL, Hosny A, Schabath MB, et al (2019) Artificial intelligence in cancer imaging: Clinical challenges and applications. CA Cancer J Clin 69, 127-157. https://doi.org/[10.3322/caac.21552]\u003c/li\u003e\n\u003cli\u003eSchroeder DW, Foster BR, Young DJ \u0026amp; Coakley FV (2024) Targeted biopsy of the prostate. Abdominal Radiology. https://doi.org/[10.1007/s00261-024-04452-z]\u003c/li\u003e\n\u003cli\u003eDuenweg SR, Bobholz SA, Barrett MJ, et al (2023) T2-Weighted MRI Radiomic Features Predict Prostate Cancer Presence and Eventual Biochemical Recurrence. Cancers (Basel) 15. https://doi.org/[10.3390/cancers15184437]\u003c/li\u003e\n\u003cli\u003eZhu X, Liu Z, He J, et al (2023) MRI-Derived Radiomics Model to Predict the Biochemical Recurrence of Prostate Cancer Following Seed Brachytherapy. Arch Esp Urol 76, 264-269. https://doi.org/[10.56434/j.arch.esp.urol.20237604.30]\u003c/li\u003e\n\u003cli\u003eDieterich LC \u0026amp; Bikfalvi A (2020) The tumor organismal environment: Role in tumor development and cancer immunotherapy. Semin Cancer Biol 65, 197-206. https://doi.org/[10.1016/j.semcancer.2019.12.021]\u003c/li\u003e\n\u003cli\u003eHanahan D \u0026amp; Monje M (2023) Cancer hallmarks intersect with neuroscience in the tumor microenvironment. Cancer Cell 41, 573-580. https://doi.org/[10.1016/j.ccell.2023.02.012]\u003c/li\u003e\n\u003cli\u003eGu Y, Guo Y, Gao N, et al (2022) The proteomic characterization of the peritumor microenvironment in human hepatocellular carcinoma. Oncogene 41, 2480-2491. https://doi.org/[10.1038/s41388-022-02264-3]\u003c/li\u003e\n\u003cli\u003eLi H, Chai L, Pu H, et al (2024) T2WI-based MRI radiomics for the prediction of preoperative extranodal extension and prognosis in resectable rectal cancer. Insights Imaging 15, 57. https://doi.org/[10.1186/s13244-024-01625-8]\u003c/li\u003e\n\u003cli\u003eLin P, Xie W, Li Y, et al (2024) Intratumoral and peritumoral radiomics of MRIs predicts pathologic complete response to neoadjuvant chemoimmunotherapy in patients with head and neck squamous cell carcinoma. J Immunother Cancer 12. https://doi.org/[10.1136/jitc-2024-009616]\u003c/li\u003e\n\u003cli\u003eZhang Z, Wan X, Lei X, et al (2023) Intra- and peri-tumoral MRI radiomics features for preoperative lymph node metastasis prediction in early-stage cervical cancer. Insights Imaging 14, 65. https://doi.org/[10.1186/s13244-023-01405-w]\u003c/li\u003e\n\u003cli\u003eWang X, Wei M, Chen Y, et al (2024) Intratumoral and peritumoral MRI-based radiomics for predicting extrapelvic peritoneal metastasis in epithelial ovarian cancer. Insights Imaging 15, 281. https://doi.org/[10.1186/s13244-024-01855-w]\u003c/li\u003e\n\u003cli\u003eGramegna A \u0026amp; Giudici P (2021) SHAP and LIME: An Evaluation of Discriminative Power in Credit Risk. Front Artif Intell 4, 752558. https://doi.org/[10.3389/frai.2021.752558]\u003c/li\u003e\n\u003cli\u003eLi J, Liu S, Hu Y, et al (2022) Predicting Mortality in Intensive Care Unit Patients With Heart Failure Using an Interpretable Machine Learning Model: Retrospective Cohort Study. J Med Internet Res 24, e38082. https://doi.org/[10.2196/38082]\u003c/li\u003e\n\u003cli\u003eTang S, Zhang H, Liang J, et al (2024) Prostate cancer treatment recommendation study based on machine learning and SHAP interpreter. Cancer Sci 115, 3755-3766. https://doi.org/[10.1111/cas.16327]\u003c/li\u003e\n\u003cli\u003eQi X, Wang S, Fang C, et al (2025) Machine learning and SHAP value interpretation for predicting comorbidity of cardiovascular disease and cancer with dietary antioxidants. Redox Biol 79, 103470. https://doi.org/[10.1016/j.redox.2024.103470]\u003c/li\u003e\n\u003cli\u003eRodr\u0026iacute;guez-P\u0026eacute;rez R \u0026amp; Bajorath J (2020) Interpretation of Compound Activity Predictions from Complex Machine Learning Models Using Local Approximations and Shapley Values. J Med Chem 63, 8761-8777. https://doi.org/[10.1021/acs.jmedchem.9b01101]\u003c/li\u003e\n\u003cli\u003eOerther B, Nedelcu A, Engel H, et al (2024) Update on PI-RADS Version 2.1 Diagnostic Performance Benchmarks for Prostate MRI: Systematic Review and Meta-Analysis. Radiology 312, e233337. https://doi.org/[10.1148/radiol.233337]\u003c/li\u003e\n\u003cli\u003eAlgohary A, Shiradkar R, Pahwa S, et al (2020) Combination of Peri-Tumoral and Intra-Tumoral Radiomic Features on Bi-Parametric MRI Accurately Stratifies Prostate Cancer Risk: A Multi-Site Study. Cancers (Basel) 12. https://doi.org/[10.3390/cancers12082200]\u003c/li\u003e\n\u003cli\u003eBai H, Xia W, Ji X, et al (2021) Multiparametric Magnetic Resonance Imaging-Based Peritumoral Radiomics for Preoperative Prediction of the Presence of Extracapsular Extension With Prostate Cancer. J Magn Reson Imaging 54, 1222-1230. https://doi.org/[10.1002/jmri.27678]\u003c/li\u003e\n\u003cli\u003eQiu Y, Liu YF, Shu X, et al (2023) Peritumoral Radiomics Strategy Based on Ensemble Learning for the Prediction of Gleason Grade Group of Prostate Cancer. Acad Radiol 30 Suppl 1, S1-s13. https://doi.org/[10.1016/j.acra.2023.06.011]\u003c/li\u003e\n\u003cli\u003eElhanani O, Ben-Uri R \u0026amp; Keren L (2023) Spatial profiling technologies illuminate the tumor microenvironment. Cancer Cell 41, 404-420. https://doi.org/[10.1016/j.ccell.2023.01.010]\u003c/li\u003e\n\u003cli\u003eRen X, Zhang L, Zhang Y, et al (2021) Insights Gained from Single-Cell Analysis of Immune Cells in the Tumor Microenvironment. Annu Rev Immunol 39, 583-609. https://doi.org/[10.1146/annurev-immunol-110519-071134]\u003c/li\u003e\n\u003cli\u003eHan X, Gong Z, Guo Y, Tang W \u0026amp; Wei X (2024) Exploration of a noninvasive radiomics classifier for breast cancer tumor microenvironment categorization and prognostic outcome prediction. Eur J Radiol 175, 111441. https://doi.org/[10.1016/j.ejrad.2024.111441]\u003c/li\u003e\n\u003cli\u003eHu Y, Cai Z, Aierken N, et al (2025) Intra- and peri-tumoral radiomics based on dynamic contrast-enhanced MRI for prediction of benign disease in BI-RADS 4 breast lesions: a multicentre study. Radiat Oncol 20, 27. https://doi.org/[10.1186/s13014-025-02605-y]\u003c/li\u003e\n\u003cli\u003eLiu HF, Wang M, Wang Q, et al (2024) Multiparametric MRI-based intratumoral and peritumoral radiomics for predicting the pathological differentiation of hepatocellular carcinoma. Insights Imaging 15, 97. https://doi.org/[10.1186/s13244-024-01623-w]\u003c/li\u003e\n\u003cli\u003eZhang H, Li X, Zhang Y, et al (2021) Diagnostic nomogram based on intralesional and perilesional radiomics features and clinical factors of clinically significant prostate cancer. J Magn Reson Imaging 53, 1550-1558. https://doi.org/[10.1002/jmri.27486]\u003c/li\u003e\n\u003cli\u003eHectors SJ, Chen C, Chen J, et al (2021) Magnetic Resonance Imaging Radiomics-Based Machine Learning Prediction of Clinically Significant Prostate Cancer in Equivocal PI-RADS 3 Lesions. J Magn Reson Imaging 54, 1466-1473. https://doi.org/[10.1002/jmri.27692]\u003c/li\u003e\n\u003cli\u003eRuan M, Liu Y, Yao K, et al (2024) Development and Validation of Interpretable Machine Learning Models for Clinically Significant Prostate Cancer Diagnosis in Patients With Lesions of PI-RADS v2.1 Score\u0026thinsp;\u0026ge;3. J Magn Reson Imaging 60, 2130-2141. https://doi.org/[10.1002/jmri.29275]\u003c/li\u003e\n\u003cli\u003ePiran Nanekaran N, Felefly TH, Schieda N, et al (2024) Prediction of prostate cancer recurrence after radiotherapy using a fused machine learning approach: Utilizing radiomics from pretreatment T2W MRI images with clinical and pathological information. Biomed Phys Eng Express. https://doi.org/[10.1088/2057-1976/ad8201]\u003c/li\u003e\n\u003cli\u003eBourbonne V, Fournier G, Valli\u0026egrave;res M, et al (2020) External Validation of an MRI-Derived Radiomics Model to Predict Biochemical Recurrence after Surgery for High-Risk Prostate Cancer. Cancers (Basel) 12. https://doi.org/[10.3390/cancers12040814]\u003c/li\u003e\n\u003cli\u003eMerriman KM, Harmon SA, Belue MJ, et al (2023) Comparison of MRI-Based Staging and Pathologic Staging for Predicting Biochemical Recurrence of Prostate Cancer After Radical Prostatectomy. AJR Am J Roentgenol 221, 773-787. https://doi.org/[10.2214/ajr.23.29609]\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003e\u003cstrong\u003eTable 1\u0026nbsp;\u003c/strong\u003eBaseline Characteristics of Prostate Cancer Patients.\u003c/p\u003e\n\u003cdiv align=\"\"\u003e\n \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"663\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVariable\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTotal(n=277)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTraining cohort (n = 193)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTest cohort(n=84)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ep value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePLR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e122.64 \u0026plusmn; 57.10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e120.89\u0026plusmn;57.93\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e126.65\u0026plusmn;55.27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.312\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLMR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e3.89\u0026plusmn;1.58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e4.00\u0026plusmn;1.60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e3.63\u0026plusmn;1.50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.093\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNLR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e3.06\u0026plusmn;2.92\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e3.09\u0026plusmn;3.28\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e3.00\u0026plusmn;1.84\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.185\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSII\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e548.92\u0026plusmn;491.64\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e560.15\u0026plusmn;542.86\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e523.12\u0026plusmn;347.95\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.889\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSIRI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e15.29\u0026plusmn;7.04\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e15.91\u0026plusmn;7.33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e13.86\u0026plusmn;6.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.036\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge (years)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e68.84\u0026plusmn;7.06\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e68.40\u0026plusmn;7.19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e69.85\u0026plusmn;6.70\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.119\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003etPSA (ng/mL)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e47.05\u0026plusmn;45.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e45.29\u0026plusmn;47.58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e51.08\u0026plusmn;42.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.331\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003efPSA (ng/mL)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e8.97\u0026plusmn;13.83\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e8.48\u0026plusmn;13.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e10.11\u0026plusmn;14.97\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.471\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eKi-67\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e0.14\u0026plusmn;0.14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e0.14\u0026plusmn;0.14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e0.14\u0026plusmn;0.13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.383\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePSA density (ng/mL\u003csup\u003e2\u003c/sup\u003e)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e1.06\u0026plusmn;1.16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e1.05\u0026plusmn;1.22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e1.07\u0026plusmn;1.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.682\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMTD(cm)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e4.78\u0026plusmn;0.85\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e4.77\u0026plusmn;0.84\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e4.79\u0026plusmn;0.86\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.741\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMLD(cm)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e5.03\u0026plusmn;1.11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e5.03\u0026plusmn;1.11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e5.05\u0026plusmn;1.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.529\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMAPD(cm)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e3.85\u0026plusmn;0.76\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e3.83\u0026plusmn;0.73\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e3.90\u0026plusmn;0.81\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.475\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTDLV(cm\u003csup\u003e3\u003c/sup\u003e)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e52.00\u0026plusmn;28.04\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e51.46\u0026plusmn;28.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e53.24\u0026plusmn;28.14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.491\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVOS(cm\u003csup\u003e3\u003c/sup\u003e)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e49.76\u0026plusmn;26.91\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e49.19\u0026plusmn;26.65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e51.08\u0026plusmn;27.60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.568\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTMD(cm)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e3.81\u0026plusmn;2.36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e3.81\u0026plusmn;2.57\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e3.84\u0026plusmn;1.82\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.262\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTV(cm\u003csup\u003e3\u003c/sup\u003e)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e24.80\u0026plusmn;38.60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e25.96\u0026plusmn;43.48\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e22.14\u0026plusmn;23.88\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.447\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eP504s\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.302\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e24(8.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e14(7.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e10(11.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e253(91.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e179(92.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e74(88.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCK8/18\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.002\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e25(9.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e10(5.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e15(17.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e252(91.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e183(94.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e69(82.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eClinical N stage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.527\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e152(54.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e103(53.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e49(58.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e125(45.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e90(46.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e35(41.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eGleason\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.570\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e6\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e29(10.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e23(11.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e6(7.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e7\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e94(33.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e68(35.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e26(31.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e8\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e61(22.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e40(20.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e21(25.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e9\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e81(29.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e55(28.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e26(31.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e10\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e12(4.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e7(3.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e5(6.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eISUP grade group\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.635\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e29(10.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e23(11.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e6(7.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e34(12.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e24(12.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e10(11.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e3\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e60(21.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e44(22.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e16(19.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e4\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e61(22.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e40(20.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e21(25.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e5\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e93(33.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e62(32.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e31(36.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSVI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.171\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e154(55.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e113(58.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e41(48.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e123(44.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e80(41.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e43(51.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEPE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.548\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e151(54.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e108(56.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e43(51.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e126(45.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e85(44.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e41(48.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSM\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.514\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e258(93.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e178(92.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e80(95.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e19(6.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e15(7.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e4(4.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePI-RADS\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e0.703\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e30(10.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e19(9.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e11(13.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e3\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e59(21.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e44(22.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e15(17.9)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e4\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e62(22.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e44(22.8)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e18(21.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e5\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e126(45.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 165px;\"\u003e\n \u003cp\u003e86(44.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e40(47.6)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 91px;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u003cstrong\u003eAbbreviations:\u0026nbsp;\u003c/strong\u003ePLR platelet-to-lymphocyte ratio; LMR lymphocyte-to-monocyte ratio; NLR neutrophil-to-lymphocyte ratio; SII systemic immune-inflammation index; SIRI systemic inflammation response index; tPSA total prostate-specific antigen; fPSA free prostate-specific antigen; Ki-67 antigen identified by monoclonal antibody Ki-67; MTD maximum transverse diameter; MLD maximum longitudinal diameter; MAPD maximum anteroposterior diameter; TDLV triaxial diameter-derived prostate volume; VOS volumetric segmentation-derived prostate volume; TMD tumor maximum diameter; TV tumor volume; p504s \u0026alpha;-methylacyl-coA racemase; CK8/18 Cytokeratin 8/18; Gleason Prostate Cancer Gleason Grading System; ISUP International Society of Urological Pathology Grade Groups; SVI seminal vesicle invasion; EPE extraprostatic extension; SM positive surgical margins; PI-RADS Prostate Imaging Reporting and Data System\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2\u0026nbsp;\u003c/strong\u003eUnivariate and multivariate analysis for BCR.\u003c/p\u003e\n\u003cdiv align=\"\"\u003e\n \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"635\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e \u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"bottom\" style=\"width: 35.0975%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eUnivariate analysis\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"3\" valign=\"bottom\" style=\"width: 34.9415%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMultivariate analysis\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVariable\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eOR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e95% CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e\u003cstrong\u003ep value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eOR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e\u003cstrong\u003e95% CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e\u003cstrong\u003ep value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eKi-67\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.002-0.08\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e6.459\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.165-253.48\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.403\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCK8/18\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.236\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.174-0.322\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e1.75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.152-20.086\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.706\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eP504s\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.252\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.185-0.342\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e5.116\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.23-113.636\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.387\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eClinical N stage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.429\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.293-0.626\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e7.335\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e2.385-22.579\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.004\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePSA density (ng/mL\u003csup\u003e2\u003c/sup\u003e)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.571\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.456-0.716\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e1.267\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.745-2.155\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.464\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEPE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.574\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.396-0.832\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e0.014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e7.672\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e2.321-25.381\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.005\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSVI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.410-0.877\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e0.027\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e6.508\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e1.565-27.058\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.031\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMAPD(cm)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.685\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.633-0.742\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e1.932\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.457-8.166\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.452\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLMR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.694\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.641-0.751\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e1.151\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.702-1.887\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.640\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMTD(cm)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.724\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.678-0.773\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e0.133\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.035-0.503\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.013\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eISUP\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.737\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.682-0.796\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e1.057\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.341-3.277\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.936\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePI-RADS\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.738\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.688-0.792\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e2.121\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e1.132-3.975\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.049\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTMD(cm)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.755\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.7-0.815\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e0.696\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.418-1.158\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.242\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eMLD(cm)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.759\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.715-0.806\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e0.707\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.312-1.603\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.486\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNLR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.774\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.706-0.848\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e2.187\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e1.157-4.137\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.043\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eGleason\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.842\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.811-0.875\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e1.141\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.327-3.979\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.862\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSIRI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.909\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.89-0.928\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e0.97\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.834-1.127\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.740\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003efPSA (ng/mL)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.966\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.947-0.984\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e1.024\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.975-1.076\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.425\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eVOS(cm\u003csup\u003e3\u003c/sup\u003e)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.971\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.965-0.978\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e1.038\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.915-1.178\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.627\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTDLV(cm\u003csup\u003e3\u003c/sup\u003e)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.973\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.967-0.979\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e0.997\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.875-1.135\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.972\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge (years)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.979\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.975-0.984\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e0.948\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.885-1.014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.192\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003etPSA (ng/mL)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.984\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.979-0.989\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e0.984\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.968-1.002\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.139\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003ePLR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.991\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.988-0.993\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e1.014\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.998-1.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.160\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTV(cm\u003csup\u003e3\u003c/sup\u003e)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.991\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.985-0.997\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e0.018\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e1.009\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.989-1.029\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.458\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSII\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e0.998\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.998-0.999\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e＜0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e0.995\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e0.99-1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e0.078\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"bottom\" style=\"width: 24.9582%;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSM\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.9554%;\"\u003e\n \u003cp\u003e1.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 16.0669%;\"\u003e\n \u003cp\u003e0.631-3.568\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 11.0752%;\"\u003e\n \u003cp\u003e0.442\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 7.6435%;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 17.1588%;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"bottom\" style=\"width: 10.1393%;\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u003cstrong\u003eAbbreviations:\u003c/strong\u003e BCR biochemical recurrence; Ki-67 antigen identified by monoclonal antibody Ki-67; CK8/18 Cytokeratin 8/18; p504s \u0026alpha;-methylacyl-coA racemase; EPE extraprostatic extension; SVI seminal vesicle invasion; MAPD maximum anteroposterior diameter; LMR lymphocyte-to-monocyte ratio; MTD maximum transverse diameter; ISUP International Society of Urological Pathology Grade Groups; PI-RADS Prostate Imaging Reporting and Data System; TMD tumor maximum diameter; MLD maximum longitudinal diameter; NLR neutrophil-to-lymphocyte ratio; Gleason Prostate Cancer Gleason Grading System; SIRI systemic inflammation response index; fPSA free prostate-specific antigen; VOS volumetric segmentation-derived prostate volume; TDLV triaxial diameter-derived prostate volume; tPSA total prostate-specific antigen; PLR platelet-to-lymphocyte ratio; TV tumor volume; SII systemic immune-inflammation index; SM positive surgical margins\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3\u0026nbsp;\u003c/strong\u003eDiagnostic performance of the different model for predicting BCR in the training and test cohorts.\u003c/p\u003e\n\u003cdiv align=\"\"\u003e\n \u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"687\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eModels\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAccuracy\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAUC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e95% CI\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSensitivity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSpecificity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCohort\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eClinic\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.860\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.897\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e0.839 - 0.954\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.811\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.872\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003etrain\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eIntraPeri4mm_ADC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.824\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.902\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e0.853 - 0.951\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.730\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.846\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003etrain\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eIntraPeri4mm_T2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.710\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.842\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e0.778 - 0.906\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.865\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.673\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003etrain\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCombined\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.938\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.978\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e0.961 - 0.994\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.919\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.942\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003etrain\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eClinic\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.643\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.731\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e0.602 - 0.861\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.789\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.600\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003etest\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eIntraPeri4mm_ADC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.571\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.706\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e0.594 - 0.819\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.789\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.508\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003etest\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eIntraPeri4mm_T2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.488\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.662\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e0.529 - 0.795\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.895\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.369\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003etest\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 144px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCombined\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.726\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.810\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 136px;\"\u003e\n \u003cp\u003e0.707 - 0.912\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.737\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 83px;\"\u003e\n \u003cp\u003e0.723\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 76px;\"\u003e\n \u003cp\u003etest\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003e\u003cstrong\u003eAbbreviations:\u0026nbsp;\u003c/strong\u003eBCR biochemical recurrence; Clinic Clinical model constructed from independent risk factors; IntraPeri4mm_ADC ADC model with fusion of Intra- and Peritumoral radiomics features; IntraPeri4mm_T2 T2WI model with fusion of Intra- and Peritumoral radiomics features; Combined The model combining clinical risk factors and T2WI/ADC-based Intra- and Peritumoral radiomics signatures\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"abdominal-radiology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"aima","sideBox":"Learn more about [Abdominal Radiology](http://link.springer.com/journal/261)","snPcode":"261","submissionUrl":"https://submission.springernature.com/new-submission/261/3","title":"Abdominal Radiology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Prostate cancer, Radiomics, Machine learning, Biochemical recurrence","lastPublishedDoi":"10.21203/rs.3.rs-7173672/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7173672/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003ePurpose\u003c/strong\u003eTo develop a predictive model for biochemical recurrence (BCR) after radical prostatectomy (RP) by integrating intratumoral and peritumoral MRI radiomics features with clinical independent risk factors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003eThis retrospective study analyzed 277 RP patients with complete follow-up data (≥1 year) from our institution, randomly divided into training (n=193) and test (n=84) cohorts. Regions of interest (ROIs) were manually delineated on T2-weighted imaging(T2WI) and apparent diffusion coefficient (ADC) maps. Peritumoral ROIs were expanded by 4 mm using Python and manually adjusted to exclude non-prostatic tissues. Radiomics models (Intra, Peri_4mm, IntraPeri_4mm), a clinical model (Clinic), and a combined radiomics-clinical model (Combined) were constructed. The predictive performance of these models was evaluated using different indexes. SHapley Additive exPlanations (SHAP) analysis was employed to visualize and interpret the decision-making process.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e Independent risk factors for BCR included extraprostatic extension (EPE), clinical N stage (N), seminal vesicle invasion (SVI), PI-RADS score, neutrophil-to-lymphocyte ratio (NLR), and maximum transverse diameter of prostate (MTD). The Clinic model achieved AUC of 0.897 (training) and 0.731 (test). The IntraPeri_4mm_ADC model showed AUC of 0.902 and 0.706, while the IntraPeri_4mm_T2 model yielded 0.842 and 0.662. The Combined model outperformed others (AUC: 0.978 and 0.810). DCA confirmed its higher net benefit. SHAP analysis revealed EPE as the top contributor to BCR prediction, followed by the ADC-derived radiomics score (ADC label_0).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions\u003c/strong\u003e The combined MRI radiomics-clinical model effectively predicts BCR post-RP. SHAP interpretability transforms \"black-box\" predictions into quantifiable feature contributions, aiding clinicians in risk stratification and personalized treatment planning.\u003c/p\u003e","manuscriptTitle":"Interpretable Machine Learning Model Based on Intra- and Peritumoral MRI Radiomics for Predicting Biochemical Recurrence After Radical Prostatectomy","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-28 05:32:25","doi":"10.21203/rs.3.rs-7173672/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-08-10T13:37:59+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-10T07:49:44+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"290591745903675490867258900002923151745","date":"2025-08-10T07:46:38+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-08-01T16:52:46+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"322768350963001145976692246537824466040","date":"2025-07-26T15:38:13+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"237111709780260448555311875172560165765","date":"2025-07-24T15:44:28+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-07-22T14:02:35+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-07-22T12:04:28+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-07-22T09:16:53+00:00","index":"","fulltext":""},{"type":"submitted","content":"Abdominal Radiology","date":"2025-07-21T05:52:05+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"abdominal-radiology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"aima","sideBox":"Learn more about [Abdominal Radiology](http://link.springer.com/journal/261)","snPcode":"261","submissionUrl":"https://submission.springernature.com/new-submission/261/3","title":"Abdominal Radiology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"54764570-3d68-4853-a318-96d13a724f21","owner":[],"postedDate":"July 28th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-10-27T13:01:19+00:00","versionOfRecord":[],"versionCreatedAt":"2025-07-28 05:32:25","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7173672","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7173672","identity":"rs-7173672","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}