Clinicopathologic Significance of Ductal Carcinoma in situ in Breast Core Needle Biopsies With Invasive Cancer

To assess whether the presence and amount of intraductal component (IC) in diagnostic needle core biopsies (NCB) is predictive of an extensive IC (EIC), the authors evaluated 50 invasive ductal carcinomas diagnosed with NCB, and then excised via lumpectomy, with regard to the extent of IC in both the NCB and subsequent lumpectomy specimen. These parameters were compared with each other and with the lumpectomy margin status. Extent of IC in the NCB was evaluated by dividing the number of ducts that contained IC by the total number of tissue cores. A ratio of more than 0.5 was considered EIC (EIC_C). IC extent in the lumpectomy was established by estimating the percentage of the tumor corresponding to IC and was considered extensive (EIC_L) if more than 25% and if there was presence of IC away from the invasive tumor. The mean size of resected tumors was 1.6 ± 0.7 cm. In 29 cases (58%) there was no IC in the NCB (NegIC_C), 11 cases (22%) exhibited nonextensive IC (NEIC_C), and 10 cases (20%) demonstrated EIC_C. A total of 7%, 36%, and 70% of the NegIC_C, NEIC_C, and EIC_C cases respectively had EIC_L (p < 0.0001). The presence of EIC_L correlated significantly with close or positive margin status for in situ disease (EIC_L positive, 12 of 13 [92%] vs EIC_L negative, 11 of 37 [30%];p = 0.004). None of the NegIC_C, 27% of NEIC_C, and 40% of EIC_C had a positive margin for in situ neoplasm in the lumpectomy specimen (p = 0.004), and 24%, 18%, and 50% had positive margins for invasive neoplasm (p = not significant). The authors conclude that EIC_C predicts EIC_L and constitutes a risk factor for positive lumpectomy margin status-particularly for in situ tumor. EIC_C may thus be of clinical value in identifying a subset of patients that requires a wider local excision.

Ultrasound-guided or stereotactic needle core biopsy (NCB) has proved to be an accurate and reliable method to diagnose breast carcinoma. ^3,4,9 However, beyond establishing the presence of neoplasm and hormone receptor status, ⁸ it is yet unclear whether other information derived from the NCB could be useful in patient management. One pathologic parameter with clinical relevance is the relative amount of the intraductal component (IC; i.e., in situ neoplasm). Extensive intraductal component (EIC) in breast carcinoma has been associated previously with increased risk of recurrence in patients treated with limited excision and radiotherapy. ^1,2,6 Because lumpectomy margin status is a more powerful predictor of local recurrence than presence or absence of EIC, ¹⁰ the ability of EIC to predict local recurrence most likely reflects a correlation with unresected neoplasm.

We hypothesized that if in situ carcinoma comprises a substantial proportion of the neoplasm that is identified in the NCB, then it may correlate with the presence of EIC in the lumpectomy specimen, and hence with a higher risk for inadequate excision. The relevance of such a relationship, if present, is that it may be useful in planning surgical therapy, possibly avoiding reexcisions due to histologically involved margins. We were thus prompted to retrospectively study a group of patients with invasive breast carcinoma, diagnosed with stereotactic or ultrasound-guided NCB, and to correlate the presence or degree of in situ neoplasm in the NCB with the extent of IC and margin status in the subsequent lumpectomy.

MATERIALS AND METHODS

A consecutive series of invasive breast ductal carcinomas, diagnosed during 1996, 1997, and 1998 at Harper Hospital and the Karmanos Cancer Institute with manual stereotactic or ultrasound-guided NCB, were reviewed. Those patients who had a subsequent lumpectomy as surgical treatment were considered further, resulting in a study population of 50 patients. All lumpectomies contained invasive ductal carcinoma. Four patients also had a minor (<20%) invasive lobular carcinoma component. The mean age of the patients included in the study was 58.5 ± 13.6 years.

Ultrasound-guided or stereotactic NCB was performed in all patients by a radiologist who targeted suspicious mammographic findings, whereas manual core biopsies were performed by a surgeon to evaluate a palpable mass. Thirty-five biopsies were ultrasound guided, two were stereotactic, three were manual, and in ten patients the mode of obtaining the biopsy was unspecified. The mean number of cores per biopsy was 5.5 (range, 1-36 cores per biopsy). Extent of IC was evaluated by dividing the number of ducts or lobules involved with carcinoma in situ (CIS) by the number of cores in the biopsy. A ratio of more than 0.5 was arbitrarily considered EIC in the NCB (EIC_C). Special care was taken not to count two or more planes of section of the same duct as different ducts.

Retrospective examination of all original hematoxylin-eosin-stained slides from lumpectomy specimens was performed independently and without knowledge of the NCB findings. Pathologic examination included assessment of the histologic type and grade of tumor, presence of multifocality, extent of invasive and intraductal components of the tumor, and margin status with respect to both invasive and in situ neoplasm. Tumor differentiation in ductal tumors was determined using the modified Bloom-Richardson grading system. ⁵ Multifocal growth was considered to be present when at least two foci of invasive tumor were separated completely by at least one low-power field of benign tissue in the same section. Extent of IC was estimated as the percentage of tumor comprised of in situ neoplasm. EIC in the lumpectomy specimen (EIC_L) was defined as the presence of more than 25% of IC and the presence of in situ disease beyond the edges of the invasive tumor. ²

Routine processing of a lumpectomy specimen at Harper Hospital includes marking of the apparent margin of resection with India ink and extensive sampling to assess margin status. The mean number of slides examined per patient in the lumpectomy specimens was 20.9 (range, 3-63 slides per patient); 21 specimens (42%) were submitted entirely for histologic examination. The margin was defined as positive if neoplastic cells were found in contact with India ink at the margin of the tissue. A close margin was defined as the presence of neoplastic cells within 1 mm of the inked margin. A negative margin was defined as the presence of tumor more than 1 mm away from the inked margin. Positive margins were classified further as involved by invasive or in situ neoplasm. The length of the involved margin was measured by determining the number of diameters of a 10× high-power field (Olympus BX40 microscope, Tokyo, Japan) that fit in the margin involved by tumor. If the sum of all positive margins measured more than 10 diameters, the margin was consider extensively involved. An attempt was made to establish whether the positive margin in invasive tumors was due to the presence of the main tumor at the margin, or, on the contrary, was the result of the presence of a satellite tumor nodule at the margin (multifocality).

RESULTS

The mean size of the resected tumors was 1.6 ± 0.7 cm. Six tumors were well differentiated, 17 were moderately differentiated, and 27 were poorly differentiated. In 20 of 45 patients (44%), axillary nodes were positive for metastatic carcinoma. Thirteen patients (26%) demonstrated EIC_L. With respect to in situ tumor, 27 samples (54%) had a negative margin, 16 samples (32%) had a close margin, and 7 samples (14%) had a positive margin. With respect to the invasive component, 25 samples (50%) had a negative margin, 11 samples (22%) had a close margin, and 14 samples (28%) had a positive margin. Four samples (8%) demonstrated margins positive for both components, and 17 samples (34%) for either.

In 29 patients (58%), there was no in situ neoplasm identified in the NCB (NegIC_C), in 11 patients (22%) the extent of CIS was considered nonextensive (i.e., the number of ducts per number cores was less than 0.5, NEIC_C), and in 10 patients (20%) the extent of CIS was considered extensive (i.e., EIC_C). The relationship between the extent of IC in the NCB and the lumpectomy specimen is presented in Table 1. EIC_L in the subsequent resection was observed in 7% of the NegIC_C cases, 36% of the NEIC_C cases, and 70% of the EIC_C cases (p = 0.0001). Thus, EIC in the NCB predicts EIC_L with a sensitivity of 54%, a specificity of 92%, a negative predictive value of 70%, and a positive predictive value of 85%. The relationship between the extent of in situ tumor in the lumpectomy, and the margin status is presented in Table 2. In summary, 4 of 13 EIC_L-positive cases (31%) had a positive margin, 8 of 13 (62%) had a close margin, and 12 of 13 (92%) had either a positive or close margin for CIS, compared with 3 of 37 (8%), 8 of 37 (22%), and 11 of 37 (30%) of EIC_L-negative cases respectively (p = 0.04, 0.008, and 0.0001 respectively). A total of 7 of 13 patients (54%) with EIC_L had a positive (46%) or close (8%) margin for invasive tumor, compared with 49% of patients without EIC_L (p = not significant).

Table 1

Table 2

Extent of IC in the diagnostic core biopsy also correlated with the margin status of the subsequent lumpectomy (Table 3). None of the 29 NegIC_C cases, 3 of the 11 (27%) NEIC_C cases, and 4 of the 10 (40%) EIC_C cases had a positive margin for in situ tumor in the lumpectomy (p = 0.004). A total of 7 of 29 (24%) NegIC_C cases, 2 of 11 (18%) NEIC_C cases, and 5 of 10 (50%) EIC_C cases had a margin positive for invasive neoplasm (p = not significant). If positive and close margins are combined, then 28% (8 of 29), 55% (6 of 11), and 90% (9 of 10) of the NegIC_C, NEIC_C, and EIC_C cases respectively were positive for in situ neoplasm (p = 0.007). Thus, EIC_C predicts intraductal carcinoma margin positivity with a sensitivity of 39%, a specificity of 96%, a positive predictive value of 90%, and a negative predictive value of 65%. Forty-five percent (13 of 29), 45% (5 of 11), and 70% (7 of 10) of IC_C, NEIC_C, and EIC_C cases respectively had a positive or close margin for invasive tumor in the subsequent lumpectomy (p = not significant).

Table 3

There was minimal correlation between tumor size and lumpectomy margin involvement by in situ or invasive tumor (Table 4). For in situ tumor, 67%, 44%, 48%, 36%, and 50% of tumors measuring <0.5 cm, 0.6 to 1.0 cm, 1.1 to 2.0 cm, 2.1 to 3.0 cm, and >3.0 cm respectively had positive or close margins (p = not significant). Similarly, 67%, 33%, 56%, 45%, and 50% of tumors in the same size categories respectively had positive or close margins for invasive neoplasm (p = not significant). Fifty-five percent (6 of 11) of cases with multifocal invasive neoplasm had a positive margin for invasive tumor compared with 21% (8 of 39) of cases with unifocal disease (Table 5;p = 0.03). Of the 14 cases with positive margins for invasive tumor, only 4 cases had more than 10 10× fields of margin involved by tumor. Of these, two represented cases in which the main tumor was grossly transected. The other two had extensive involvement of the margin due to multifocal tumor nodules. One additional case with nonextensive margin involvement was secondary to a multifocal tumor nodule.

Table 4

Table 5

Nine patients underwent local reexcision after lumpectomy, and four patients underwent mastectomy (Table 6). Of these 13 patients, 11 had a positive margin for invasive tumor, 2 had a close margin for invasive tumor, and 8 had concurrent close or positive margins for DCIS. Twelve cases had a positive margin for either type of tumor in the preceding lumpectomy and one case had only a close margin for invasive tumor. Three cases had multifocal disease in the lumpectomy. Of the four mastectomies performed due to incomplete initial excision, one had no residual invasive or in situ tumor, two had residual DCIS, and one had residual invasive and in situ neoplasm. Of the nine reexcisions, seven had no residual tumor and two had residual DCIS with negative margins. All cases with residual tumor identified during the reexcision procedure had a positive margin for invasive neoplasm in the lumpectomy, two also had a positive margin for in situ neoplasm, and two had a close margin for in situ neoplasm. Thus, of the 14 cases with a positive margin for invasive tumor, 3 cases have not had a third procedure performed, 1 of 11 (9%) had residual invasive tumor in the third procedure, and 5 cases (45%) had residual DCIS. Only two of the cases with a close invasive tumor margin in the lumpectomy (one also had a positive margin for in situ tumor) had a third procedure performed, and neither had residual tumor. Similarly, of the seven cases with a positive margin for in situ tumor, three have not had a third procedure performed, and two of four (50%) had residual DCIS in the third procedure. Of the 16 cases with a close margin for DCIS in the lumpectomy, only 4 cases underwent a third procedure (all 4 cases had a positive margin for invasive tumor), of which 2 cases (50%) had residual DCIS in the third procedure. Three of the four cases (75%) with extensive margin involvement had residual disease in the subsequent procedure (two in situ, one invasive and in situ) compared with two of seven cases (28%) with nonextensive margin involvement. Of the five cases that had residual DCIS in the reexcision/mastectomy, three had EIC_L, and the same three had multifocal tumors in the lumpectomy.

Table 6

DISCUSSION

NCB has become widely accepted as a useful method for the diagnosis of nonpalpable breast lesions. Sensitivity ranges from 71% to 100% for malignant processes, with a specificity of 90% to 100%. ⁹ Dahlstrom et al. ⁴ have reported a 96% concordance between stereotactic core biopsy and excision biopsy for the diagnosis of invasive or in situ cancer. Recently, Jacobs et al. ⁸ reported 100% concordance between immunostaining results for bcl-2, estrogen receptors, c-erbB-2 and p53 performed on NCB when compared with those performed in the excision specimens, implying that NCB provides a representative sampling of a malignancy. Despite this high accuracy, some pathologic aspects of breast neoplasia are not always defined reproducibly by NCB. For example, Dahlstrom et al. ⁴ found that concordance between NCB and the lumpectomy for the type of tumor (ductal vs lobular) was only 78%, and that NCB may underestimate nuclear grade. In several studies, it has also been demonstrated that whereas a benign or malignant diagnosis is highly reliable, a diagnosis of atypical lesions should warrant open surgical biopsy because as many as 75% of subsequent open biopsies in such cases show malignant neoplasm. ^3,4,9

EIC has been implicated as an independent predictor of recurrence in breast carcinomas treated with conservative surgery and radiotherapy. In 1990, Boyages et al. ¹ reported a 5-year incidence of parenchymal recurrence for EIC-positive carcinomas of 26% compared with 7% in EIC-negative cases. Furthermore, the presence of EIC in a lumpectomy specimen correlates with extensive residual disease in unresected breast tissue. In the series by Holland et al., ⁷ 71% of cases with EIC in the diagnostic biopsy had residual DCIS in the subsequent mastectomy compared with 28% of EIC-negative cases. Similarly, Schnitt et al. ¹¹ found that 88% of EIC-positive cases had residual DCIS in a subsequent reexcision compared with 48% of EIC-negative cases. This suggests that the distribution of EIC-positive tumors may be underappreciated by surgical assessments at the time of excision. This same group of investigators has reported a high correlation between EIC and margin status in lumpectomy specimens. ¹⁰ Moreover, it was found that margin status was a stronger predictor of recurrence, independent of whether the tumor is EIC positive or negative. This suggests that the ability of EIC to predict recurrence relies, at least in part, in its association with unresected disease. Accordingly, our data show that the presence or extent of IC in the NCB correlates not only with EIC in the resection specimen, but also with a higher risk for involved margins.

We assessed the amount of IC in a core biopsy by dividing the number of ducts involved by DCIS by the number of cores present in the biopsy. This yielded a numeric value that we designated arbitrarily as EIC if more than 0.5. These criteria, in our opinion, are not only objective and reproducible, but are required to account for variable, partial sampling in core biopsies. They would be inappropriate to employ with surgical excision, in which sampling is, presumably, complete. Although we concede that other criteria for quantification of IC may be more specific/sensitive, our findings nonetheless imply that the presence or extent of IC in the NCB reflects tumor pathology accurately, and thus helps to identify a subset of patients that would be at risk for margin positivity in subsequent excision. It could thus be theorized that this subset of patients may benefit from planning a wider excision to avoid reexcision. Possibly, close correlation with mammographic findings or utilization of intraoperative pathologic consultation may facilitate complete tumor excision. This preliminary conclusion, based on our retrospective analysis, should prompt the additional study of this issue in a prospective manner.

As expected, the extent of IC in NCB was less effective in predicting the incidence of margin positivity for invasive tumor. We thus tried to identify other variables that could correlate with invasive tumor margin positivity, such as tumor size and multifocality. Tumor size, which in lumpectomy series tends to be relatively small, did not correlate significantly with invasive tumor margin status. However, multifocality appears to confer significant risk of positive margin for invasive neoplasm, likely reflecting the presence of occult (surgically inapparent) deposits of tumor. Unfortunately, multifocality is difficult to assess in NCB due first to the small sample size and second because separate cores may not be reconstructed to infer neoplastic architecture.

Our study performs a careful analysis of margins in the lumpectomy specimens. Specimens were sampled generously (mean number of slides, 20.9; 42% of cases embedded completely), and margin status was analyzed for extent (if positive) as well as proximity. We further correlated margin status with histologic findings in those patients who underwent subsequent reexcision or mastectomy. It is interesting to note that of the 11 cases that initially had a positive margin for invasive neoplasm, only 1 case had residual invasive tumor, but 5 cases had residual DCIS evident in the reexcision. On the other hand, four of the eight cases (50%) that had a positive margin for in situ tumor harbored residual DCIS. Similarly, of the four cases with a close margin for in situ tumor, 50% also had residual DCIS. Our data show that EIC_C not only predicts likelihood of positive margins but also of close margins for DCIS in the subsequent resection. Because clinical handling of resection specimens with close margins is variable, some may question the relevance of the observed correlation between EIC and close lumpectomy margin status (i.e., for DCIS). Although the number of cases is relatively small, our data suggest that DCIS constitutes the preponderance of unresected disease whether close or at the margin of resection, and that both scenarios have a similar chance of having residual DCIS in the remaining breast. This may be accounted for by the anatomically discontinuous growth pattern characteristic of in situ ductal lesions. For in situ disease, others have shown that the presence of close margins is clinically relevant. Holland et al. ⁶ found a 36% incidence of local recurrence in cases with close margins (<1 mm) in the lumpectomy specimen compared with 2% in cases with negative (>1 mm) margins. One could also theorize that DCIS may be more resistant than invasive tumor to the rim of tissue necrosis that ensues in the remaining breast after a surgical procedure ² because we only found residual DCIS and no invasive tumor in the reexcised material, even with extensive positive margins for invasive tumor. Thus, the data suggest that the pathologic and possibly clinical significance of margin status for DCIS is likely different from invasive disease.

In summary, we conclude that NCB, besides establishing the diagnosis of malignancy, may also provide useful information for the therapeutic decision-making process in the management of breast cancer. We found that a significant amount of IC in the core correlates with EIC in the lumpectomy and with margin positivity, especially for DCIS. This suggests that patients with EIC in NCB may benefit from a wider excision.