In the current environment, wherein screening-detected, calcification-associated, and impalpable DCIS are more frequently encountered, the core biopsy has emerged as an excellent modality for the preoperative establishment of the diagnosis and for the optimization of therapeutic options [20, 21]. Nonetheless, the core biopsy is associated with a somewhat higher inaccuracy rate for the diagnosis of DCIS than for the diagnosis of invasive cancer [22, 23]. In the study of Dillon et al , for example, the initial core biopsy was not fully diagnostic in 33% and 26% of their cases of intermediate and high-grade DCIS, respectively. Previous studies have evaluated the significance of DCIS size, extent, and histologic features in core biopsies as a predictor of margin status in the subsequent excisions [24–27]. However, these studies were centered on cases of DCIS admixed with invasive carcinoma. There is very limited published data on a potential correlation between the pathologic features of pure DCIS in core biopsies and margin status. The goals of the present study are two-fold: 1) To determine if there are any core biopsy morphologic features that may predict a close or positive margin in the subsequent excision in a group of intermediate and high-grade DCIS cases, and 2) To determine how well the core biopsy predicts the maximal pathology in the associated excisions.
Although there is no consensus of what constitutes a "close" margin , there is a near consensus that the status of margins after an excision for DCIS is a strong predictor of residual disease and/or local recurrence [29–34]. As such, there have been a variety of proposals that are aimed at establishing the adequacy of the excision intraoperatively, including intraoperative specimen ultrasonography, intraoperative specimen mammography, gross specimen examination, the routine sampling of the "cavity margins" around the initial excision, and frozen section evaluation of margins [35–39]. While mammographic features, such as linear branching calcification, accurately identify many examples of high-grade DCIS as such [40, 41], this modality does not do so invariably , and may underestimate lesional size and extent, especially in non-comedo DCIS [43–45]. It may therefore be of clinical utility to have core biopsy features that may predict the necessity for a more generous excision during the initial surgery, which in addition to the aforementioned reasons, may also help stratify those patients that would qualify for specific therapies that require negative margins such as accelerated, hypofractionated whole breast radiotherapy  or partial breast intraoperative radiation . Furthermore, in addition to histologic grade and margin status, it has been shown that several other pathologic features in the excision specimens may be associated with local recurrence [48–51]. Among these are volume of DCIS near margins and the overall DCIS volume in the specimen - whether deduced from the number of slides with DCIS, number of DCIS ducts or DCIS-cancerized lobules, or percentage of blocks with DCIS [48–51]. If any core biopsy features can predict a large volume in the excision, this may also be of clinical utility.
Our study shows that of all the core biopsy features that were evaluated, including maximum nuclear grade, necrosis, cancerization of lobules, number of tissue cores with DCIS, number of DCIS ducts per tissue core, total DCIS ducts, comedo-pattern or other histologic patterns, only necrosis was significantly associated with a positive or close margin on univariate and multivariate analysis. Similarly, at the 2.7 cm size threshold, there was no significant correlation between lesional size in excisions and selected core biopsy features (number of DCIS ducts, DCIS ducts per tissue core, number of tissue cores with DCIS, and cancerization of lobules). Most of the core biopsy features evaluated herein are therefore of limited value in predicting close or positive margins, or large lesional size in intermediate and high-grade DCIS. Our findings, however, need to be confirmed in substantially larger datasets from multiple institutions.
The second goal of this study was to determine how well the core biopsy predicted the maximal pathology in the patient, as evaluated in the subsequent excisions in this specific dataset of intermediate and high-grade DCIS. Regarding the risk of invasive disease in a core biopsy showing DCIS alone, previous studies of pan-grade DCIS have shown that invasive disease is present in 8 to 36% of cases [52–57]. In high-risk subsets, i.e. those patients with large masses, numerous mammographic calcifications, DCIS of high histologic grade, or presentation with a palpable mass, this risk is approximately 41.5 to 48% [58, 59]. In our own dataset, we identified 4 examples of invasive carcinoma and 3 examples of microinvasive carcinoma out of the 49 excision specimens that we evaluated. Therefore, the risk of any kind of invasive disease in our patient population after a core biopsy diagnosis of intermediate or high-grade DCIS is somewhat low at 14% (7/49). There were too few cases of invasive disease to stratify for preoperative findings that may potentially predict invasion. The reverse corollary is that in 5 (10.2%) of 49 cases, no residual carcinoma was found. In another 4 cases, the changes were diagnostic only of atypical ductal hyperplasia. A recent report on 40,395 screening detected breast malignancies showed that in 174 cases no carcinoma was identified in the excision subsequent to the core biopsy . In 165 of these cases, the authors concluded that the lesion was entirely removed in the core biopsy, whereas the remaining 9 were considered to false positive diagnoses in the biopsies . Therefore, allowing for differences in sample size and composite of lesional types, our rate of finding no residual disease (5/49; 10.2%) is higher but largely comparable to the reported findings in that study (165/40,395; 4.3%).
The necessity for sentinel lymph node evaluation in patients with pure DCIS, and whether patient subsets can be preoperatively identified that would benefit the most from this evaluation, continues to be controversial [61, 62]. Sentinel nodes are found to be positive in 4.8% to 13% of patients with a core biopsy diagnosis of DCIS [57, 63–66]. Our findings provide insights into the frequency of lymph node involvement in this group of intermediate and high-grade DCIS. Amongst the subset of 28 patients that received any nodal evaluation, only 1 (3.6%) had sentinel lymph node involvement. Parenthetically, for this single patient, only touch preparations of the lymph node's cut section were performed intraoperatively, and did not reveal malignant cells even upon their detailed review in the aftermath. The malignant cells became morphologically evident only after subsequent additional node sectioning, when a "negative" intraoperative diagnosis had already been rendered. As such, she did not benefit from sentinel node evaluation. Our experience, therefore, is that the frequency of nodal involvement is low, which is concordant with most previously reported findings. However, our data is insufficient to make assertions about the validity of routine sentinel node evaluation in this group of patients.
We conclude that most of the core biopsy features evaluated herein are of limited value in predicting a close or positive margin in high and intermediate grade DCIS. A significant change [to invasive disease (14%) or to no residual disease (10%)] is seen in approximately 24% of excisions that follow a core biopsy diagnosis of intermediate or high-grade DCIS.