Patients
The study cohort included 702 patients with a mean age of 49.0 ± 10.9 years (range, 24–82 years) and a mean follow-up of 61 months (range, 6 - 93 months), with all the patients being Asian. The majority were originally diagnosed with invasive ductal carcinoma (637/702, 90.7%). There were 289 patients (289/702, 41.2%) with stage I disease and 413 patients (413/702, 58.8%) with stage II disease. Of the 702 patients, 621 received either adjuvant (548/702, 78.1%) or neoadjuvant chemotherapy (73/702, 10.4%). Most patients received radiation therapy (555/702, 79.1%) and breast conserving operation (557/702, 79.3%). The patient characteristics are summarized in Table 1.
Recurrence
Overall, 587 patients (587/702, 83.6%) had no evidence of recurrence, while 115 (115/702, 16.4%) had recurrence, including 38 (38/115, 33.0%) with locoregional recurrence, 42 (42/115, 36.5%) with distant recurrence, 19 (19/115, 16.5%) with contralateral breast cancer, and 16 (16/115, 13.9%) with recurrences in multiple sites.
Of the 38 locoregional recurrences, 25 were in the ipsilateral regional nodes, 12 in the ipsilateral breast, and 4 in the postmastectomy chest wall. In the 42 with distant recurrences, the lung was the most common site (n = 28), followed by bone (n = 10), brain (n = 9), and liver (n = 6). The mean time to recurrence was 26.8 months, with a range of 4–78 months. The 3 year recurrence rate was 12.4%, and the 5 year recurrence rate was 15.3% (Figure 1).
In the univariate analysis for recurrence, factors significantly associated with recurrence included increasing tumor size, positive nodal status, ki-67 index more than 14, presence of LVI, mastectomy, and neoadjuvant or no adjuvant chemotherapy. Compared with T1 patients, patients with an increased tumor size were more likely to have a recurrence, with hazard ratios (HR) of 1.92 and 6.75 for T2 and T3 tumors, respectively. Patients with positive axillary nodes also had a significantly higher risk of recurrence (HR = 1.92 for N1 tumors). Patients with recurrences were also more likely to have ki-67 index more than 14 (HR = 3.46), presence of LVI (HR = 3.27), received mastectomy (HR = 1.92), neoadjuvant chemotherapy (HR = 2.00), or to not have received adjuvant chemotherapy (HR = 1.79) (Table 1).
Imaging findings
Out of 702 patients, 692 (98.6%), 693 (98.7%), and 293 (41.7%) had available mammographic, US, and MRI examinations, respectively, for the radiologist to review. Imaging features of the subjects are summarized in Table 2. On mammography, tumors tended to present as a mass (586/692, 84.7%), with oval or round shape (295/586, 50.3%), with not circumscribed margins (477/586, 81.4%), and without microcalcifications (507/692, 73.3%). On US, tumors tended to present as an oval or round shaped mass (390/693, 56.3%), with not circumscribed margins (491/693, 70.9%), and with posterior acoustic enhancement (447/693, 68.8%). On MRI, tumors tended to present as a mass (276/293, 94.2%), with oval or round shape (179/276, 64.9%), not circumscribed margins (173/276, 62.7%), rim enhancement (140/293, 47.8%), washout kinetics (267/292, 91.4%), no intratumoral necrosis (176/293, 60.1%), and the presence of T2 peritumoral edema (191/293, 65.2%). In the univariate analysis for recurrence, imaging factors significantly associated with recurrence included moderate or marked BPE on MRI. Patients with moderate or marked BPE had a higher risk of recurrence, with a HR of 1.94 (Table 2, Figure 2,Figure 3). After controlling for all potential confounders in a multivariate anlalysis, increasing tumor size, positive nodal status, the presence of LVI, and not receiving adjuvant chemotherapy were independent associated with recurrence (Table 3).