Department of Diagnostic Imaging and Nuclear Medicine, Kyoto University Graduate School of Medicine, 54 Shogoin Kawaharacho, Sakyoku, Kyoto 606-8507, Japan.
Eur J Radiol. 2012 Oct;81(10):2627-31. doi: 10.1016/j.ejrad.2011.12.013. Epub 2012 Jan 4.
To evaluate inter-observer agreement and the predictive value of tumor size measurements using MRI for breast cancer under neoadjuvant chemotherapy (NAC) by comparing the measurements of the longest diameters (LD), total enhanced volumes (TEV) and washout volumes (WOV).
Thirty-seven female breast cancer patients were prospectively enrolled from August 2008 to October 2010. Two of these patients had locally advanced disease. MRI examinations were acquired within 2 weeks before and after NAC. Interim scans were also conducted in 30 patients. Tumor resection was undertaken within 2 weeks after the cessation of NAC. MRI images were independently measured for LD, TEV and WOV by two experienced radiologists. Inter-observer agreement was evaluated using concordance correlation coefficients (CCCs). Tumor sizes after NAC were evaluated relative to their initial sizes for early prediction of a pathological complete response (pCR).
The CCCs were 0.93 (CI: 0.90-0.95) for LD, 0.98 (CI: 0.97-0.98) for TEV and 0.99 (CI: 0.991-0.996) for WOV. All measurements had high inter-observer agreement, but the CCCs were significantly increased in the aforementioned order (P<0.0001). WOV measured after the completion of chemotherapy had significant discriminating ability (P=0.0056) when evaluated using receiver operating characteristic analysis, and was found to be superior to LD (P=0.045). The average WOV size was significantly smaller in pCR cases than in non-pCR cases (P=0.016).
Computer-aided detection-generated tumor volumes had significantly higher inter-observer concordance than conventional LD measurements. WOV measurements had the highest concordance, and WOV could better predict pCR after NAC at smaller tumor sizes.
通过比较最长径(LD)、总增强体积(TEV)和洗脱体积(WOV)的测量值,评估新辅助化疗(NAC)前后 MRI 测量乳腺癌肿瘤大小的观察者间一致性和预测价值。
2008 年 8 月至 2010 年 10 月前瞻性纳入 37 例女性乳腺癌患者,其中 2 例患者为局部晚期疾病。NAC 前 2 周和后进行 MRI 检查。30 例患者进行了中期扫描。NAC 停止后 2 周内行肿瘤切除术。由 2 名有经验的放射科医生独立测量 MRI 图像的 LD、TEV 和 WOV。采用一致性相关系数(CCCs)评估观察者间一致性。评估 NAC 后肿瘤大小相对于初始大小,以早期预测病理完全缓解(pCR)。
LD 的 CCC 为 0.93(置信区间:0.90-0.95),TEV 为 0.98(置信区间:0.97-0.98),WOV 为 0.99(置信区间:0.991-0.996)。所有测量值观察者间一致性均较高,但上述顺序 CCC 显著增加(P<0.0001)。化疗结束后测量的 WOV 在接受者操作特征分析评估时具有显著的鉴别能力(P=0.0056),且优于 LD(P=0.045)。pCR 病例的平均 WOV 大小明显小于非 pCR 病例(P=0.016)。
计算机辅助检测生成的肿瘤体积与传统 LD 测量相比具有更高的观察者间一致性。WOV 测量的一致性最高,WOV 可在更小的肿瘤体积下更好地预测 NAC 后的 pCR。