This retrospective observational cohort study aims to develop and externally validate an imaging-clinical multimodal fusion model for predicting postoperative prognosis in patients with endophytic renal cell carcinoma undergoing partial nephrectomy. Preoperative computed tomography imaging features, three-dimensional reconstruction-derived tumor characteristics, radiomics features, and clinical variables will be integrated using machine learning and deep learning approaches. The primary objective is to evaluate whether the multimodal model improves prediction of postoperative prognostic outcomes compared with single-modality models based on clinical or imaging features alone.
Partial nephrectomy is a standard nephron-sparing treatment for localized renal cell carcinoma. However, postoperative functional and oncologic outcomes remain heterogeneous, especially in patients with endophytic renal tumors, in whom tumor complexity may increase surgical difficulty and affect postoperative recovery. Conventional clinical variables and anatomical scoring systems may not fully capture the multidimensional risk profile of these patients. This study will retrospectively collect clinical, pathological, perioperative, and imaging data from patients with endophytic renal cell carcinoma who underwent partial nephrectomy. Preoperative multiphase computed tomography images will be used for radiomics feature extraction and deep learning-based image representation. Three-dimensional reconstruction-derived tumor features and conventional clinical variables will also be incorporated. The study will develop and validate multimodal prediction models, including clinical models, radiomics models, deep learning imaging models, and imaging-clinical fusion models. Model performance will be assessed using discrimination, calibration, and clinical utility metrics, including the area under the receiver operating characteristic curve, calibration curves, decision curve analysis, and external validation across independent cohorts.
Study Type
OBSERVATIONAL
Enrollment
406
Preoperative CT imaging features, radiomics features, three-dimensional reconstruction-derived features, and clinical variables will be retrospectively analyzed to develop and validate a multimodal model for predicting postoperative prognosis after partial nephrectomy. No intervention will be assigned to participants.
Tianjin Medical University Second Hospital
Tianjin, Tianjin Municipality, China
Modified Pentafecta Achievement After Partial Nephrectomy
Modified pentafecta achievement will be defined as the simultaneous fulfillment of predefined postoperative outcome criteria, including negative surgical margin, absence of major postoperative complications, preservation of renal function, absence of significant perioperative adverse events, and absence of early tumor recurrence or other prespecified unfavorable outcomes. Patients who do not meet all criteria will be classified as modified pentafecta failure.
Time frame: From the date of partial nephrectomy to the last available postoperative follow-up, up to 12 months after surgery.
Postoperative eGFR Decline Greater Than 20%
A clinically significant decline in renal function will be defined as a decrease in estimated glomerular filtration rate greater than 20% compared with the preoperative baseline value.
Time frame: From baseline to 3-12 months after partial nephrectomy
Major Postoperative Complications
Major postoperative complications will be defined as Clavien-Dindo grade III or higher complications occurring within the predefined postoperative period.
Time frame: Within 30 or 90 days after partial nephrectomy
Positive Surgical Margin
Positive surgical margin will be determined according to the postoperative pathological report.
Time frame: At final pathological evaluation after partial nephrectomy
Area Under the Receiver Operating Characteristic Curve of the Multimodal Model
The discriminatory performance of the multimodal model for predicting modified pentafecta achievement after partial nephrectomy will be evaluated using the area under the receiver operating characteristic curve.
Time frame: At completion of model development and external validation, using postoperative outcome data up to 12 months after surgery.
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.