Mastectomy or Margin Re-excision? A Nomogram for Close/Positive Margins After Lumpectomy for DCIS.

Abstract

BACKGROUND: Anatomic extent of ductal carcinoma in situ (DCIS) may be uncertain in spite of clinical, pathologic, and imaging data. Consequently close/positive margins are common with lumpectomy for DCIS and often lead to a challenge in deciding whether to perform a re-excision or mastectomy. PATIENTS AND METHODS: From a single health system, we identified cases of lumpectomy for DCIS with close/positive margins who underwent re-excision for the purpose of constructing a nomogram. In total, 289 patients were available for analysis. The patients were randomly divided into two sets allocating 70% to the modeling and 30% to the validation set. A multivariable logistic regression model was used to estimate the probability of overall positive margin status using multiple clinicopathologic predictors. Nomogram validation included internal tenfold cross-validation, internal bootstrap validation, and external validation for which a concordance index was calculated to assess the external validity. RESULTS: Significant predictors of persistent positive margins from regression modeling included necrosis at diagnosis (non-comedo or comedo); DCIS not associated with calcifications on core biopsy; high-grade DCIS; progesterone receptor positivity; and number of positive margins at initial surgery. When subjected to internal validation, the nomogram achieved an uncorrected concordance index of 0.7332, a tenfold cross-validation concordance index of 0.6795, and a bootstrap-corrected concordance index of 0.6881. External validation yielded an estimated concordance index of 0.7095. CONCLUSION: Using clinical and pathologic variables from initial diagnosis and surgery for DCIS, this nomogram predicts persistent positive margins with margin re-excision, and may be a valuable tool in surgical decision-making.