Background and purpose: The application of sentinel lymph node biopsy (SLNB) on breast ductal carcinoma in situ (DCIS) is still in controversy due to the unavoidable pathological underestimation. This study aimed to retrospectively investigate the risk factors of pathological underestimation in breast DCIS diagnosed by ultrasound-guided core needle biopsy (CNB) and discuss the possibility to avoid SLNB in pure DCIS patients with certain characteristics. Methods: Female patients who were diagnosed as DCIS (including microinvasion) by ultrasound-guided CNB with negative lymph node and treated in Peking University Cancer Hospital Institute from March 2005 to October 2014 were enrolled. All patients received formal operation

radiotherapy and systematic therapy for breast cancer. The correlation between clinicopathological characteristics and underestimation was counted with Chi-square test. The risk factors of underestimation in pure DCIS were analyzed by binary logistic regression. Results: The cases of Pure DCIS

DCIS with microinvasion and DCIS suspicious microinvasion were 360

63 and 31

respectively. 56.4% preoperative pure DCISs were diagnosed as DCIS postoperatively

of which 21.7% and 21.9% were pathologically upstaged to microinvasive and invasive cancer. The ratio of postoperatively microinvasive cancer was 30.2% and 35.5%

as well as invasive cancer was 66.7% and 61.3% in the latter two groups

respectively. The proportion of underestimation in pure DCIS was statistically lower compared with the other groups (P0.001). The risk of underestimation was 1.97 times (95% CI: 1.17-3.32

P=0.011) and 2.30 times (95%CI: 1.34-3.98

P=0.003) higher in tumor size3 cm and high grade when compared with tumor size ≤3 cm and non-high grade

respectively. However

the risk of human epidermal growth factor receptor 2 (HER2) undefined (OR=0.37

95%CI: 0.19-0.72

P=0.003) or positive (OR=0.38

95% CI: 0.20-0.73

P=0.004) was significantly lower than that of HER2 negative. The rate of underestimation of pure DCIS with tumor size3 cm

high grade and HER2 positive was the highest (73.1%)

while tumor size3 cm

high grade and HER2 undefined was the lowest (11.9%). Conclusion: The ratio of pathological underestimation in DCIS with microinvasion or DCIS suspicious microinvasion was far higher compared with pure DCIS diagnosed by ultrasound-guided CNB. Tumor size3 cm

high grade and negative HER2 were risk factors of underestimation in preoperative pure DCIS. SLNB could not be avoided in pure DCIS.