Cancer Research-UK and EPSRC Cancer Imaging Centre, Research Data Management and Statistics Unit, and Department of Gynaecological Oncology, Royal Marsden NHS Foundation Trust and Institute of Cancer Research, Sutton, Surrey SM2 5PT, England.
Radiology. 2011 Oct;261(1):182-92. doi: 10.1148/radiol.11110577. Epub 2011 Aug 9.
To prospectively evaluate apparent diffusion coefficient (ADC) histograms in the prediction of chemotherapy response in patients with metastatic ovarian or primary peritoneal cancer.
Research ethics committee approval and patient written informed consent were obtained. Diffusion-weighted (DW) magnetic resonance (MR) imaging was performed through the abdomen and pelvis before and after one and three cycles of chemotherapy in 42 women (mean age, 63.0 years ± 11.4 [standard deviation]) with newly diagnosed or recurrent disease. Reproducibility and intra- and interobserver agreement of ADC calculations were assessed. Per-patient weighted ADC histograms were generated at each time point from pixel ADCs from five or fewer target lesions. Mean ADC, percentiles (10th, 25th, 50th, 75th, 90th), skew, kurtosis, and their change were analyzed according to histologic grade, primary versus recurrent disease status, and response, determined with integrated biochemical and morphologic criteria, with a linear mixed model. Areas under receiver operating characteristic curve (AUCs) for combinations of parameters were calculated with linear discriminant analysis.
Coefficients of variation for repeat measurements and for within and between observers were 4.8%, 11.4%, and 13.7%, respectively. Grade and disease status did not significantly affect histogram parameters. Pretreatment ADCs were not predictive of response. In responders, all ADCs increased after the first and third cycle (P < .001), while skew and kurtosis decreased after the third (P < .001 and P = .006, respectively); however, in nonresponders, no parameter changed significantly. Percentage change of the 25th percentile performed best in identifying response (AUC = 0.82 and 0.83 after first and third cycle, respectively), whereas combination of parameters did not improve accuracy.
An early increase of ADCs and later decrease of skew and kurtosis characterize chemotherapy response. Quantitative DW MR imaging can aid in early monitoring of treatment efficacy in patients with advanced ovarian cancer.
前瞻性评估表观扩散系数（ADC）直方图在预测转移性卵巢癌或原发性腹膜癌患者化疗反应中的作用。
本研究获得了伦理委员会的批准和患者的书面知情同意。在 42 名新诊断或复发性疾病的女性患者（平均年龄 63.0 岁±11.4[标准差]）中，在接受一周期和三周期化疗前后，通过腹部和骨盆进行了弥散加权（DW）磁共振（MR）成像。评估了 ADC 计算的可重复性以及观察者内和观察者间的一致性。在每个时间点，从 5 个或更少的目标病变的像素 ADC 生成每个患者的加权 ADC 直方图。根据组织学分级、原发性与复发性疾病状态以及根据综合生化和形态学标准确定的反应，分析平均 ADC、百分位数（第 10、25、50、75、90 位）、偏度、峰度及其变化，采用线性混合模型。采用线性判别分析计算参数组合的受试者工作特征曲线（AUC）下面积。
重复测量的变异系数、观察者内和观察者间的变异系数分别为 4.8%、11.4%和 13.7%。分级和疾病状态均未显著影响直方图参数。治疗前 ADC 不能预测反应。在反应者中，所有 ADC 在第一周期和第三周期后均增加（P<.001），而偏度和峰度在第三周期后降低（P<.001 和 P=.006）；然而，在无反应者中，没有参数明显改变。第 25 百分位数的百分比变化在识别反应方面表现最佳（第一周期和第三周期的 AUC 分别为 0.82 和 0.83），而参数组合并不能提高准确性。
ADC 的早期增加以及偏度和峰度的后期降低可用于描绘化疗反应。定量 DW MR 成像有助于早期监测晚期卵巢癌患者的治疗效果。
