De-escalating indications for excision when breast core needle biopsy returns fibroepithelial lesion-not further characterized

Purpose Surgical excision is often performed to exclude phyllodes tumor (PT) when Core Needle Biopsy (CNB) of the breast returns fibroepithelial lesion-not further characterized (FEL-NFC). If imaging or CNB pathology features can be identified that predict a very low probability of borderline/malignant PT, thousands of women could be spared the expense and morbidity of surgical excisions. Methods This retrospective cohort study includes 180 FEL-NFC from 164 patients who underwent surgical excisional biopsy. Results The upgrade rate from FEL-NFC to benign PT was 15%, and to borderline/malignant PT 7%. Imaging features predicting upgrade to borderline/malignant PT included greater size (p = 0.0002) and heterogeneous echo pattern on sonography (p = 0.117). Histologic features of CNB predicting upgrade to borderline/malignant PT included "pathologist favors PT" (p = 0.012), mitoses (p = 0.014), stromal overgrowth (p = 0.006), increased cellularity (p = 0.0001) and leaf-like architecture (p = 0.077). A three-component score including size > 4.5 cm (Size), heterogeneous echo pattern on sonography (Heterogeneity), and stromal overgrowth on CNB (Overgrowth) maximized the product of sensitivity x specificity for the prediction of borderline/malignant PT. When the SHO score was 0 (72% of FEL-NFC) the probability of borderline/malignant PT on excision was only 1%. Conclusion The combination of size <= 4.5 cm, homogeneous echo pattern, and absence of stromal overgrowth is highly predictive of a benign excision potentially sparing most patients diagnosed with FEL-NFC the expense and morbidity of a surgical excision.