Liberman Laura, Mason Gary, Morris Elizabeth A, Dershaw D David
Department of Radiology, Breast Imaging Section, Memorial Sloan-Kettering Cancer Center, 1275 York Ave., New York, NY 10021, USA.
AJR Am J Roentgenol. 2006 Feb;186(2):426-30. doi: 10.2214/AJR.04.1707.
The purpose of this study was to determine the impact of lesion size on the positive predictive value (PPV) of biopsy in MRI-detected breast lesions.
A retrospective review was performed of 666 consecutive nonpalpable, mammographically occult lesions that had MRI-guided localization. MRI examinations were performed using a 1.5-T magnet. Lesions were measured by the interpreting radiologist before biopsy. Malignancy rate versus lesion size was determined.
The median MRI lesion size was 1 cm (range, 0.3-7.0 cm). Malignancy was present in 149/666 (22%) lesions, of which 80 (54%) were ductal carcinoma in situ (DCIS), 66 (44%) were invasive cancer, and three (2%) were lymphoma. The frequency of malignancy increased significantly (p = 0.0005) with lesion size, with malignancy found in one (3%) of 37 lesions less than 5 mm, 44 (17%) of 254 lesions 5-9 mm, 37 (25%) of 151 lesions 10-14 mm, 21 (28%) of 74 lesions 15-19 mm, and 46 (31%) of 150 lesions 20 mm or larger. Lesions less than 5 mm accounted for 37 (6%) of 666 lesions that had a biopsy and one (< 1%) of 149 cancers (one DCIS). Among lesions less than 10 mm, the likelihood of malignancy was highest in postmenopausal women (22% malignant) and in the extent of disease setting (22% malignant), and lowest in premenopausal women (10% malignant) and in the high-risk screening setting (10% malignant).
The PPV of biopsy for lesions identified at breast MRI using a 1.5-T magnet significantly increased with increasing lesion size. Biopsy is rarely necessary for lesions smaller than 5 mm because of their low (3%) likelihood of cancer. Further work is needed to develop an algorithm that uses size in addition to other patient and lesion factors to guide biopsy recommendations for MRI-detected breast lesions.
本研究旨在确定病变大小对MRI检测到的乳腺病变活检阳性预测值（PPV）的影响。
对666例连续的不可触及、乳腺X线摄影隐匿性病变且接受MRI引导定位的病例进行回顾性研究。使用1.5-T磁体进行MRI检查。活检前由解读影像的放射科医生测量病变大小。确定恶性率与病变大小的关系。
MRI病变大小的中位数为1 cm（范围0.3 - 7.0 cm）。666例病变中有149例（22%）为恶性，其中80例（54%）为导管原位癌（DCIS），66例（44%）为浸润性癌，3例（2%）为淋巴瘤。恶性频率随病变大小显著增加（p = 0.0005），小于5 mm的37例病变中有1例（3%）为恶性，5 - 9 mm的254例病变中有44例（17%）为恶性，10 - 14 mm的151例病变中有37例（25%）为恶性，15 - 19 mm的74例病变中有21例（28%）为恶性，20 mm或更大的150例病变中有46例（31%）为恶性。小于5 mm的病变占666例接受活检病变的37例（6%），占149例癌症（1例DCIS）的1例（< 1%）。在小于10 mm的病变中，绝经后女性（22%为恶性）和疾病范围情况（22%为恶性）中恶性可能性最高，绝经前女性（10%为恶性）和高危筛查情况（10%为恶性）中恶性可能性最低。
使用1.5-T磁体的乳腺MRI检查所发现病变的活检PPV随病变大小增加而显著升高。小于5 mm的病变因癌症可能性低（3%）很少需要活检。需要进一步开展工作来开发一种算法，该算法除了考虑其他患者和病变因素外，还利用病变大小来指导对MRI检测到的乳腺病变的活检建议。
