Flat epithelial atypia (FEA) is an alteration of terminal duct-lobular units by a proliferation of ductal epithelium with low grade atypia. No consensus exists on whether the diagnosis of FEA in core needle biopsy (CNB) requires excision (EXC). We retrospectively identified all in-house CNBs obtained between 1/2012-7/2018 with FEA. We reviewed all CNB slides and assessed radiologic-pathologic concordance. An upgrade was defined as invasive carcinoma (IC) and/or ductal carcinoma in situ in the EXC. The EXC slides of all upgraded cases were re-reviewed. Out of ~15,700 consecutive CNBs in the study period, 106 CNBs from 106 patients yielded FEA alone or with classic lobular neoplasia (LN). We excluded 52 CNBs (40 patients with prior/concurrent carcinoma and 12 without EXC). After re-review, we reclassified 14 cases (2 marked nuclear atypia, 10 focal ADH, 2 benign). The final FEA study cohort consisted of 40 CNBs from 40 women. The CNB targeted mammographic calcifications in 36 (90%) cases, MRI non-mass enhancement in 3 (8%), and 1 (2%) sonographic mass. All CNBs were deemed radiologicpathologic concordant. FEA was present alone in 34 CNBs and with LN in 6. EXC yielded 2 low grade IC, each spanning less than 2 mm, identified in tissue sections without biopsy site changes. The remaining 38 cases had no upgrade. Classic LN did not affect the upgrade. The upgrade rate of FEA was 5%; both minute, low-grade "incidental" IC. We conclude that non-surgical management may be considered in patients without prior/concurrent carcinoma and radiologicpathologic concordant CNB diagnosis of FEA.

Background Although the rate of carcinoma upgrade for atypical ductal hyperplasia (ADH) diagnosed on core needle biopsy (CNB) is variable, current standard treatment consists of surgical excision (SE) for all ADH CNB diagnoses. Our objective was to identify features of ADH on CNB that may stratify carcinoma upgrade risk on SE. Methods We retrospectively analyzed cases diagnosed as ADH on CNB. An independent slide review and detailed analysis of radiological and clinical data was performed. Statistical analyses were used to identify predictors for upgrade. Using variables predictive of upgrade, a model to stratify the probability of upgrade of ADH diagnosed on CNB was constructed. Results We identified 124 ADH cases with subsequent SE. Of these, 62 cases (50%) were upgraded to carcinoma. Features predictive of upgrade were as follows: diagnosis of “At least ADH”, percentage of cores involved by ADH, radiologic lesion size, presence of ipsilateral carcinoma, and patient age. A 4‐tiered predictive model using percentage of cores involved by ADH, histologic extent of ADH, radiologic lesion size, and patient age was constructed. This predictive model has a fair accuracy, with an area under the ROC curve of 0.76. Conclusion We have identified several predictors of carcinoma upgrade for ADH diagnosed on CNB. Our predictive model may be used to stratify the risk of carcinoma upgrade on SE.