General Surgery 3, Department of Surgical, Oncological and Gastroenterological Sciences, University of Padova, Padova.
Department of Mathematics, University of Genova, Genova.
Nucl Med Commun. 2022 Jul 1;43(7):815-822. doi: 10.1097/MNM.0000000000001570. Epub 2022 Apr 26.
Reliable markers to predict the response to neoadjuvant chemoradiotherapy (nCRT) in locally advanced rectal cancer (LARC) are lacking. We aimed to assess the ability of 18F-FDG PET/MRI to predict response to nCRT among patients undergoing curative-intent surgery.
Patients with histological-confirmed LARC who underwent curative-intent surgery following nCRT and restaging with 18F-FDG PET/MRI were included. Statistical correlation between radiomic features extracted in PET, apparent diffusion coefficient (ADC) and T2w images and patients' histopathologic response to chemoradiotherapy using a multivariable logistic regression model ROC-analysis.
Overall, 50 patients were included in the study. A pathological complete response was achieved in 28.0% of patients. Considering second-order textural features, nine parameters showed a statistically significant difference between the two groups in ADC images, six parameters in PET images and four parameters in T2w images. Combining all the features selected for the three techniques in the same multivariate ROC curve analysis, we obtained an area under ROC curve of 0.863 (95% CI, 0.760-0.966), showing a sensitivity, specificity and accuracy at the Youden's index of 100% (14/14), 64% (23/36) and 74% (37/50), respectively.
PET/MRI texture analysis seems to represent a valuable tool in the identification of rectal cancer patients with a complete pathological response to nCRT.
局部晚期直肠癌（LARC）缺乏可靠的标志物来预测新辅助放化疗（nCRT）的反应。我们旨在评估 18F-FDG PET/MRI 预测接受根治性手术的患者对 nCRT 反应的能力。
纳入接受 nCRT 并进行 18F-FDG PET/MRI 分期后接受根治性手术且组织学证实为 LARC 的患者。使用多变量逻辑回归模型 ROC 分析，评估在 PET、表观扩散系数（ADC）和 T2w 图像中提取的放射组学特征与患者对放化疗的组织病理学反应之间的统计学相关性。
总体而言，本研究共纳入 50 例患者。28.0%的患者达到病理完全缓解。考虑到二阶纹理特征，ADC 图像中有 9 个参数在两组之间存在统计学差异，PET 图像中有 6 个参数，T2w 图像中有 4 个参数。将三种技术中选择的所有特征结合到同一个多变量 ROC 曲线分析中，我们得到 ROC 曲线下面积为 0.863（95%CI，0.760-0.966），在 Youden 指数处的灵敏度、特异性和准确度分别为 100%（14/14）、64%（23/36）和 74%（37/50）。
PET/MRI 纹理分析似乎是一种有价值的工具，可以识别出对 nCRT 具有完全病理缓解的直肠癌患者。
