Construction of a prediction model for muscular invasion in upper urinary tract urothelial carcinoma based on preoperative MRI features
10.3760/cma.j.cn112330-20241123-00520
- VernacularTitle:基于术前MRI特征构建上尿路尿路上皮癌肌层浸润的预测模型
- Author:
Haonan CHEN
1
;
Lingkai CAI
;
Hongyuan DING
;
Hao JI
;
Tianxiao HONG
;
Hao YU
;
Qikai WU
;
Chaoran ZHAO
;
Xiao YANG
;
Qiang CAO
;
Xiancheng ZHAO
;
Pengchao LI
;
Qiang LYU
Author Information
1. 南京医科大学第一附属医院（江苏省人民医院）泌尿外科，南京 210029
- Publication Type:Journal Article
- Keywords:
Urothelial tumor;
Upper urinary tract urothelial carcinoma;
Magnetic resonance imaging;
Nomogram;
Muscle-invasion
- From:
Chinese Journal of Urology
2025;46(9):661-668
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To construct a nomogram based on preoperative MRI imaging features for the prediction of muscle-invasive upper urinary tract urothelial carcinoma（UTUC）and evaluate its performance.Methods:This retrospective cohort study analyzed the clinical data of 99 UTUC patients treated at the First Affiliated Hospital of Nanjing Medical University from April 2018 to May 2024. Among them，69（69.7%）were male and 30（30.3%）were female，with a median age of 67.0 years. All patients underwent preoperative MRI and radical nephroureterectomy. According to postoperative pathology，tumors staged ≥ T 2 were assigned to the muscle-invasive group，and those staged ≤ T 1 were assigned to the non-muscle-invasive group. Baseline data，pathological information，and imaging characteristics were collected and compared between the two groups. Logistic regression analysis was performed to identify risk factors for muscle-invasive UTUC，and a nomogram was constructed. The diagnostic performance of the model was assessed using receiver operating characteristic（ROC）curves，calibration curves，and decision curve analysis（DCA）. Results:Among the 99 patients，70（70.7%）were diagnosed with muscle-invasive UTUC，and 29（29.3%）with non-muscle-invasive UTUC. The muscle-invasive group had significantly larger tumor size［4.5（2.8，7.0）cm vs. 3.0（2.3，4.5）cm， P = 0.029］，a higher incidence of multifocal tumors［37.1%（26/70）vs. 3.5%（1/29）， P < 0.001］，patchy tumors［30.0%（21/70）vs. 6.9%（2/29）， P = 0.019］，spiculated tumor margins［52.9%（37/70）vs. 17.2%（5/29）， P = 0.001］，tumor compression on renal parenchyma or periureteral/peripelvic fat［68.6%（48/70）vs. 10.3%（3/29）， P < 0.001］，high-grade pathology［92.9%（65/70）vs. 75.9%（22/29）， P = 0.043］，lymph node metastasis［28.6%（20/70）vs. 0， P = 0.001］，and lymphovascular invasion［42.9%（30/70）vs. 10.3%（3/29）， P=0.002］. The apparent diffusion coefficient（ADC）values［0.9（0.8，1.1）× 10 -3 mm2/s vs. 1.1（1.0，1.4）× 10 -3 mm2/s， P < 0.001］and normalized ADC（NADC）values［0.8（0.7，1.0）vs. 0.9（0.8，1.1）， P = 0.002］were significantly lower in the muscle-invasive group. Univariate logistic regression identified multifocality，patchy tumor patterns，spiculated tumor margins，tumor compression on renal parenchyma or periureteral/peripelvic fat，and low NADC values as risk factors for muscle-invasive UTUC（all P < 0.05）. Multivariate analysis revealed multifocality（ OR = 17.903，95% CI 1.650 - 194.253， P = 0.018），tumor compression on renal parenchyma or perirenal / ureteral fat（ OR = 14.690，95% CI 3.069 - 70.323， P < 0.001），and low NADC value（ OR = 0.016，95% CI 0.001 - 0.471， P = 0.017）as independent risk factors. A nomogram was constructed based on these factors. The area under the ROC curve（AUC）of the model was 0.898（95% CI 0.838 - 0.957），with an optimal cutoff value of 0.639. The model showed an accuracy of 83.8%，sensitivity of 81.4%，and specificity of 89.7%. Calibration curves indicated good calibration，and DCA showed that the model provided substantial clinical net benefit. Conclusions:This study constructed a nomogram based on preoperative MRI features，including tumor multifocality，compression on renal parenchyma or periureteral/peripelvic fat and NADC value，which demonstrates good predictive performances for muscle-invasive UTUC.
