Nougaret Stephanie, Vargas Hebert Alberto, Lakhman Yulia, Sudre Romain, Do Richard K G, Bibeau Frederic, Azria David, Assenat Eric, Molinari Nicolas, Pierredon Marie-Ange, Rouanet Philippe, Guiu Boris
From the Department of Abdominal Imaging, Saint Eloi University Hospital, CHU Montpellier, 80 Avenue Augustin Fliche, 34295 Montpellier Cedex 5, France (S.N., M.A.P., B.G.); Institute of Cancer Research of Montpellier (IRCM-U1194), Montpellier, France (S.N., F.B., D.A., B.G.); INSERM, Unit 896, Montpellier, France (S.N., B.G.); Department of Radiology, Memorial Sloan Kettering Cancer Center, New York, NY (H.A.V., Y.L., R.K.G.D.); Departments of Radiology (R.S.), Histopathology (F.B.), Oncology and Radiation Therapy (D.A., E.A.), and Surgery (P.R.), Montpellier Cancer Institute, Montpellier, France; and Department of Statistics-IMAG, Montpellier University, CHU Montpellier, Montpellier, France (N.M.).
Radiology. 2016 Aug;280(2):446-54. doi: 10.1148/radiol.2016150702. Epub 2016 Feb 26.
Purpose To determine the diagnostic performance of intravoxel incoherent motion (IVIM) parameters and apparent diffusion coefficient (ADC) to assess response to combined chemotherapy and radiation therapy (CRT) in patients with rectal cancer by using histogram analysis derived from whole-tumor volumes and single-section regions of interest (ROIs). Materials and Methods The institutional review board approved this retrospective study of 31 patients with rectal cancer who underwent magnetic resonance (MR) imaging before and after CRT, including diffusion-weighted imaging with 34 b values prior to surgery. Patient consent was not required. ADC, perfusion-related diffusion fraction (f), slow diffusion coefficient (D), and fast diffusion coefficient (D*) were calculated on MR images acquired before and after CRT by using biexponential fitting. ADC and IVIM histogram metrics and median values were obtained by using whole-tumor volume and single-section ROI analyses. All ADC and IVIM parameters obtained before and after CRT were compared with histopathologic findings by using t tests with Holm-Sidak correction. Receiver operating characteristic curves were generated to evaluate the diagnostic performance of IVIM parameters derived from whole-tumor volume and single-section ROIs for prediction of histopathologic response. Results Extreme values aside, results of histogram analysis of ADC and IVIM were equivalent to median values for tumor response assessment (P > .06). Prior to CRT, none of the median ADC and IVIM diffusion metrics correlated with subsequent tumor response (P > .36). Median D and ADC values derived from either whole-volume or single-section analysis increased significantly after CRT (P ≤ .01) and were significantly higher in good versus poor responders (P ≤ .02). Median IVIM f and D* values did not significantly change after CRT and were not associated with tumor response to CRT (P > .36). Interobserver agreement was excellent for whole-tumor volume analysis (range, 0.91-0.95) but was only moderate for single-section ROI analysis (range, 0.50-0.63). Conclusion Median D and ADC values obtained after CRT were useful for discrimination between good and poor responders. Histogram metrics did not add to the median values for assessment of tumor response. Volumetric analysis demonstrated better interobserver reproducibility when compared with single-section ROI analysis. (©) RSNA, 2016 Online supplemental material is available for this article.
目的 通过使用源自全肿瘤体积和单层面感兴趣区（ROI）的直方图分析，确定体素内不相干运动（IVIM）参数和表观扩散系数（ADC）在评估直肠癌患者联合化疗和放疗（CRT）疗效方面的诊断性能。材料与方法 机构审查委员会批准了这项对31例直肠癌患者的回顾性研究，这些患者在CRT前后接受了磁共振（MR）成像检查，包括术前具有34个b值的扩散加权成像。无需患者同意。通过双指数拟合在CRT前后采集的MR图像上计算ADC、灌注相关扩散分数（f）、慢扩散系数（D）和快扩散系数（D*）。通过全肿瘤体积和单层面ROI分析获得ADC和IVIM直方图指标及中位数。使用经Holm-Sidak校正的t检验将CRT前后获得的所有ADC和IVIM参数与组织病理学结果进行比较。生成受试者操作特征曲线以评估源自全肿瘤体积和单层面ROI的IVIM参数对组织病理学反应预测的诊断性能。结果 除极值外，ADC和IVIM的直方图分析结果与肿瘤反应评估的中位数相当（P > 0.06）。在CRT之前，中位数ADC和IVIM扩散指标均与随后的肿瘤反应无关（P > 0.36）。全容积或单层面分析得出的中位数D和ADC值在CRT后显著增加（P≤0.01），且在反应良好组与反应不良组中显著更高（P≤0.02）。中位数IVIM f和D*值在CRT后无显著变化，且与CRT的肿瘤反应无关（P > 0.36）。观察者间一致性对于全肿瘤体积分析非常好（范围为0.91 - 0.95），但对于单层面ROI分析仅为中等（范围为0.50 - 0.63）。结论 CRT后获得的中位数D和ADC值有助于区分反应良好组与反应不良组。直方图指标在评估肿瘤反应方面并未补充中位数的价值。与单层面ROI分析相比，容积分析显示出更好的观察者间可重复性。（©）RSNA，2016 本文有在线补充材料。
