文献速递泌尿生殖腹部CT

光谱CT定量参数对膀胱尿路上皮癌病理分级的诊断价值再评估

Reassessing the diagnostic value of spectral CT-derived parameters for pathological grading of urothelial carcinoma of the bladder.
PubMed 文献 · Abdom Radiol (NY) · 2026-10-30

本研究回顾性纳入120例经病理证实为膀胱尿路上皮癌(UCB)的患者,评估光谱CT定量参数术前预测病理分级的诊断性能及其在临床变量基础上的增量价值。结果显示影像模型AUC为0.743,临床模型为0.746,联合模型为0.772,联合模型较临床模型的增量改善很小(ΔAUC=0.027)且无统计学意义(DeLong P=0.563)。结论认为光谱CT参数与UCB病理分级相关,但仅表现出中等程度的判别性能。

要点

  • 光谱CT参数与UCB病理分级相关,但判别性能仅中等(影像模型AUC 0.743)。
  • 联合模型较临床模型增量价值小且无统计学意义(ΔAUC=0.027,P=0.563)。
  • AEF是唯一的独立影像预测因子,钙化、肿瘤大小和数目在多变量分析中无统计学意义。

核心结论

  • 光谱CT定量参数与膀胱尿路上皮癌病理分级相关,但仅表现出中等程度的判别性能。
  • 联合模型较临床模型的增量改善很小(ΔAUC=0.027)且无统计学意义(DeLong P=0.563)。

研究设计与样本

  • 回顾性研究,纳入120例经病理证实的膀胱尿路上皮癌患者(65例高级别,55例低级别),术前接受增强光谱CT检查。
  • 提取光谱CT定量参数,采用组内相关系数(ICC)评估读者间一致性。
  • 单变量分析中P<0.10的变量进入多变量logistic回归,构建临床模型、影像模型和联合模型。
  • 采用LASSO和弹性网惩罚回归作为敏感性分析评估模型稳健性。
  • 使用ROC分析、DeLong检验和决策曲线分析(DCA)评估模型性能和临床净获益。

关键影像表现/征象

  • 光谱参数显示出良好至优秀的读者间一致性。
  • 单变量分析中多个影像特征与肿瘤分级相关(P<0.05)。
  • 多变量临床模型中,年龄仍与肿瘤分级独立相关,而肿瘤大小、肿瘤数目和钙化无统计学意义。
  • AEF是唯一的独立影像预测因子。

诊断标准与关键数值

  • 影像模型AUC为0.743,临床模型AUC为0.746,联合模型AUC为0.772。
  • 联合模型较临床模型的ΔAUC=0.027,DeLong检验P=0.563。
  • DCA显示联合模型在中等风险阈值下额外净获益有限。

临床应用与阅片要点

  • 光谱CT定量参数可作为术前预测膀胱尿路上皮癌病理分级的辅助参考,但判别性能中等,不宜单独用于分级决策。
  • AEF是独立影像预测因子,阅片时可关注该参数,但需结合临床变量综合评估。
  • 联合模型较临床模型的增量价值有限,提示光谱CT参数的附加诊断价值需进一步验证。

局限性

  • 回顾性研究设计。
  • 联合模型较临床模型的增量改善无统计学意义,光谱CT参数的附加价值有限。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

英文原文(摘要)· 点击展开

OBJECTIVE: To evaluate the diagnostic performance of spectral CT-derived quantitative parameters for preoperative prediction of pathologic grade in urothelial carcinoma of the bladder (UCB) and to assess their incremental value over clinical variables and model robustness.

METHODS: In this retrospective study, patients with pathologically confirmed UCB who underwent preoperative contrast-enhanced spectral CT were included. Quantitative spectral parameters were extracted. Inter-reader agreement was assessed using intraclass correlation coefficients (ICCs). Variables with P < 0.10 on univariable analysis were entered into multivariable logistic regression to construct clinical, imaging and combined models. Model robustness was evaluated using penalized regression (LASSO and elastic net) as sensitivity analyses. Model performance and clinical net benefit were assessed using receiver operating characteristic (ROC) analysis, DeLong testing, and decision curve analysis (DCA).

RESULTS: A total of 120 patients were included (65 high-grade tumours and 55 low-grade tumours). Spectral parameters showed good to excellent inter-reader agreement. Several imaging features were associated with tumor grade on univariable analysis (P < 0.05). After inclusion of calcification in the clinical analysis, age remained independently associated with tumor grade, whereas tumor size, tumor number, and calcification were not statistically significant in the multivariable clinical model. AEF remained the only independent imaging predictor. The AUC was 0.743 for the imaging model, 0.746 for the clinical model and 0.772 for the combined model. The incremental improvement of the combined model over the clinical model was small (ΔAUC = 0.027) and not statistically significant (DeLong P = 0.563). DCA showed limited additional net benefit of the combined model at intermediate risk thresholds.

CONCLUSION: Spectral CT-derived parameters are associated with pathologic grade in UCB but demonstrate only moderate discriminatory performance.

来源:Abdom Radiol (NY),查看原文。版权归原作者所有。

引用

  1. Wang, Du, Hu, et al. Reassessing the diagnostic value of spectral CT-derived parameters for pathological grading of urothelial carcinoma of the bladder. Abdom Radiol (NY). 2026. PMID: 42961642. doi:10.1007/s00261-026-05852-z

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。