Dipartimento di Scienze Medico-Chirurgiche, Università degli Studi di Milano, IRCCS Policlinico San Donato, San Donato Milanese, Milan, Italy.
Invest Radiol. 2010 Mar;45(3):114-20. doi: 10.1097/RLI.0b013e3181cc2929.
To increase specificity of gadobenate dimeglumine-enhanced breast magnetic resonance (MR) imaging without losing sensitivity.
IRB approval was obtained for this retrospective study and informed consent was waived. Sixty-eight patients with 73 breast lesions underwent dynamic breast 1.5-T MR imaging with 0.1 mmol/kg of gadobenate dimeglumine and 103 or 111-second temporal resolution. A score system based on shape (round/oval/lobular = 0, linear/dendritic/stellate = 1), margins (defined = 0, undefined = 1), pattern (homogeneous = 0, inhomogeneous = 1, rim = 2), kinetics (continuous = 0, plateau = 1, washout = 2), and initial enhancement was used. Initial enhancement was evaluated with standard (<50% = 0, 50%-100% = 1, >100% = 2) or adjusted (<180% = 0, 180%-240% = 1, >240% = 2) thresholds. Scores 0-2, 3, 4 or 5, and from 6 to 8 were considered as equivalent to Breast Imaging Reporting and Data System 2, 3, 4, and 5, respectively. Pathologic examination (core-, vacuum-assisted, or surgical biopsy) served as a reference standard. Nonparametric tests were used.
At histopathology, 52 lesions were malignant, 21 were benign. The initial enhancement for benign lesions was 141% +/- 65% (mean +/- SD), whereas for malignant lesions it was 210% +/- 80% (P = 0.001). The median multifactorial score for benign lesions was 4 using the standard thresholds and 3 using the adjusted thresholds (P < 0.001) whereas for malignant lesions it was 7 and 6, respectively (P < 0.001). Specificity increased from 38% with standard thresholds to 71% with adjusted thresholds (gain of 33%; P = 0.016); positive predictive value 80% and 90%, negative predictive value 100% and 100%, and accuracy 82% and 92%, respectively. Sensitivity was 100% for both thresholds.
Specificity of gadobenate dimeglumine-enhanced breast MR imaging can be increased without limiting its high sensitivity.
在不降低敏感性的情况下提高钆贝葡胺增强乳腺磁共振（MR）成像的特异性。
本回顾性研究获得了机构审查委员会的批准，并豁免了知情同意。68 例 73 个乳腺病变患者行 1.5-T 乳腺动态 MR 成像，对比剂用量为 0.1mmol/kg 钆贝葡胺，时间分辨率为 103 秒或 111 秒。采用基于形态（圆形/椭圆形/分叶状=0，线性/树枝状/星状=1）、边缘（边界清楚=0，边界不清楚=1）、模式（均匀=0，不均匀=1，边缘强化=2）、时间-信号强度曲线（连续型=0，平台型=1，廓清型=2）和初始强化程度的评分系统。采用标准（<50%=0，50%-100%=1，>100%=2）或调整（<180%=0，180%-240%=1，>240%=2）阈值评估初始强化程度。评分 0-2、3、4 或 5 和 6-8 分别相当于乳腺影像报告和数据系统（BI-RADS）2、3、4 和 5。以病理检查（核心活检、真空辅助活检或手术活检）为参考标准。采用非参数检验。
在组织病理学检查中，52 个病变为恶性，21 个为良性。良性病变的初始强化程度为 141%±65%（均值±标准差），而恶性病变为 210%±80%（P=0.001）。使用标准阈值时，良性病变的多因素评分中位数为 4，使用调整阈值时为 3（P<0.001）；而恶性病变的评分中位数分别为 7 和 6（P<0.001）。使用标准阈值时，特异性从 38%提高到 71%（提高 33%；P=0.016）；阳性预测值分别为 80%和 90%，阴性预测值分别为 100%和 100%，准确率分别为 82%和 92%。两种阈值的敏感性均为 100%。
在不降低高敏感性的情况下，可提高钆贝葡胺增强乳腺 MR 成像的特异性。
