Axillary Nodal Metastases in Invasive Lobular Carcinoma Versus Invasive Ductal Carcinoma: Comparison of Node Detection and Morphology by Ultrasound

BACKGROUND. Invasive lobular carcinoma is more subtle on imaging compared with invasive ductal carcinoma; nodal metastases may also differ on imaging between these two. OBJECTIVE. The purpose of this study was to determine whether invasive lobular carcinoma and invasive ductal carcinoma differ in the detection rate by ultrasound (US) of metastatic axillary nodes and in the metastatic nodes' US characteristics. METHODS. This retrospective study included 695 women (median age, 53 years) who had breast cancer in a total of 723 breasts (76 lobular, 586 ductal, 61 mixed ductal and lobular histology) with biopsy-proven axillary nodal metastases and who underwent pretreatment US. A single breast radiologist reviewed US images in patients with suspicious nodes on US and classified number of nodes, size, and morphology. Morphologic assessment used a previously described classification according to the relationship between node cortex and hilum. Nodal findings were compared between lobular and ductal carcinoma. A second radiologist independently classified node morphology in 241 cancers to assess interreader agreement. RESULTS. A total of 99 metastatic axillary nodes (15 lobular, 66 ductal, 18 mixed histology) were not visualized on US and were diagnosed by surgical biopsy. The remaining 624 metastatic nodes (61 lobular, 520 ductal, 43 mixed ductal and lobular histology) were visualized on US and diagnosed by US-guided fine-needle aspiration. US detected the metastatic nodes in 80.3% for lobular carcinoma versus 88.7% for ductal carcinoma (p = .04). Among metastatic nodes detected by US, retrospective review identified three or more abnormal nodes in 50.8% of lobular carcinoma versus 69.2% of ductal carcinoma (p = .003); node size was 2.0 cm or smaller in 65.6% for lobular carcinoma versus 47.3% for ductal carcinoma (p = .03); morphology was type III or IV (diffuse cortical thickening without hilar mass effect) rather than type V or VI (marked cortical thickening with hilar mass effect) in 68.9% for lobular carcinoma versus 28.8% for ductal carcinoma (p < .001). Interreader agreement assessment for morphology exhibited a kappa coefficient of 0.63 (95% CI, 0.54-0.73). CONCLUSION. US detects a lower percentage of nodal metastases in lobular than in ductal carcinoma. Nodal metastases in lobular carcinoma more commonly show diffuse cortical thickening and with less hilar mass effect.