提高糖尿病胃癌患者的风险分层：基于光子计数CT的肌肉脂肪比作为术后并发症的预测指标

Enhancing risk stratification in diabetic gastric cancer: muscle-fat ratio from photon-counting CT as a predictor of postoperative complications.

作者信息

Lin Shuangxiang, Jin Yuchen, Xu Mengxi, Wang Shuyue, Yao Weisheng, Wu Jiaxing, Wang Xinhong, Sun Jianzhong

机构信息

Department of Radiology, The Second Affiliated Hospital Zhejiang, University School of Medicine, Hangzhou, 310000, China.

Department of Internal Medicine, Linhai Maternal and Child Health Care Hospital, Taizhou, 317000, China.

出版信息

BMC Med Imaging. 2025 Aug 19;25(1):334. doi: 10.1186/s12880-025-01872-1.

DOI:10.1186/s12880-025-01872-1

PMID:40830775

原文链接:https://pmc.ncbi.nlm.nih.gov/articles/PMC12363078/

Abstract

BACKGROUND

In diabetic gastric cancer patients, body composition (skeletal muscle–to–fat ratio, MFR) may influence surgical outcomes. We evaluated whether Photon-counting CT (PCD-CT) derived MFR predicts major postoperative complications, reflecting its value in perioperative risk stratification.

METHODS

A retrospective analysis of 134 gastric cancer patients with type 2 diabetes was conducted. Preoperative PCD-CT scans assessed body composition. Logistic regression models identified predictors of poor postoperative outcomes, defined by major postoperative complications. The predictive accuracy of models incorporating clinical variables and MFR was evaluated using receiver operating characteristic curves, integrated Discrimination Improvement (IDI), and net Reclassification Improvement (NRI).

RESULTS

Patients who developed major complications ( = 35) had significantly lower skeletal muscle area (45.5 vs. 56.2 cm²; < 0.01) and higher fat accumulation. Abnormal MFR (0.34–0.57)was a strong predictor of poor outcomes (OR = 1.94, 95% CI: 1.17–2.58, < 0.01) compared to patients without complications ( = 99). The model combining clinical variables with MFR had the best performance (AUC = 0.75, sensitivity = 0.74, specificity = 0.71) in predicting major complications, outperforming a model based solely on clinical factors. It also showed substantial improvements in predictive accuracy, with an NRI of 0.52 ( < 0.01) and an IDI of 0.09 ( < 0.01).

CONCLUSION

MFR, quantified by PCD-CT, is a reliable and accurate biomarker for identifying diabetic gastric cancer patients at higher risk of major postoperative complications. MFR demonstrates strong predictive value for adverse surgical outcomes, reinforcing its role in perioperative risk stratification.

CLINICAL TRIAL NUMBER

Not applicable.

SUPPLEMENTARY INFORMATION

The online version contains supplementary material available at 10.1186/s12880-025-01872-1.

摘要

背景

在糖尿病胃癌患者中，身体组成（骨骼肌与脂肪比率，MFR）可能会影响手术结果。我们评估了光子计数CT（PCD-CT）得出的MFR是否能预测术后主要并发症，以反映其在围手术期风险分层中的价值。

方法

对134例2型糖尿病胃癌患者进行回顾性分析。术前PCD-CT扫描评估身体组成。逻辑回归模型确定术后不良结局的预测因素，术后不良结局由术后主要并发症定义。使用受试者工作特征曲线、综合辨别改善（IDI）和净重新分类改善（NRI）评估纳入临床变量和MFR的模型的预测准确性。

结果

发生主要并发症的患者（n = 35）骨骼肌面积显著更低（45.5 vs. 56.2 cm²；P < 0.01）且脂肪堆积更多。与无并发症的患者（n = 99）相比，MFR异常（0.34 - 0.57）是不良结局的有力预测因素（OR = 1.94，95% CI：1.17 - 2.58，P < 0.01）。将临床变量与MFR相结合的模型在预测主要并发症方面表现最佳（AUC = 0.75，敏感性 = 0.74，特异性 = 0.71），优于仅基于临床因素的模型。其预测准确性也有显著提高NRI为0.52（P < 0.01），IDI为0.09（P < 0.01）。