Large-Core Needle Biopsy of Nonpalpable Breast Lesions

Large-core needle biopsy (LCNB) of nonpalpable suspicious abnormalities found on mammography and/or breast ultrasound has become an alternative to surgical excision. Initial reports showed that LCNB had an accuracy comparable with surgical excision after wire localization.^{1 - 3} The recent introduction of the directional, vacuum-assisted biopsy (DVA) instrument has allowed harvesting of larger quantities of tissue with a single needle insertion.^{4 - 6} For example, the average weight of a biopsy sample obtained with a traditional automated 14-gauge needle has been reported to be 17 mg compared with 44 mg and 105 mg, respectively, with the 14- and 11-gauge needles using the DVA device.^{4 ,7}

We have carefully correlated the LCNB results with the imaging and pathologic findings to ensure that no presumed benign lesion would be subsequently shown to be malignant.⁸

We have analyzed the results of 1836 LCNBs performed in our radiology department from August 1, 1991, to December 31, 1997. Over this period the acquisition of digital imaging and improvements in needle technology have allowed for sampling of smaller and fainter particles of calcification and smaller masses.

A total of 1836 LCNBs were done in our radiology department during the study period. Nearly 76% of the LCNBs were performed from January 1995 to December 1997. The 1836 abnormalities were evaluated with either an automated 14-gauge needle (n=1333) or a DVA biopsy instrument using a 14 (n=372) or 11 (n=131) gauge needle. The abnormalities were found in 1643 women, 1476 women had 1 abnormality and biopsy was performed, 144 had 2 separate lesions sampled, 20 had 3 separate lesions, and 3 had an LCNB of 4 separate lesions. The patients ranged in age from 20 to 85 years with a mean age of 50 years. The masses ranged from 2 mm to 4 cm, with a median of 1.1 cm and mean of 1.3 cm. The calcification clusters measured 2 mm to 5.2 cm with a median of 8 mm and mean of 9 mm.

In 1388 of the 1836 procedures, stereotactic guidance was used and the biopsies were performed on a dedicated stereotactic core biopsy unit (Mammotest, Fischer Imaging, Denver, Colo; August 1991-1994; and DSM, Lorad, Danbury, Conn; September 1994-December 1997) equipped with digital imaging beginning in 1993. Of the 1388 procedures, there were 754 masses and 634 calcification clusters sampled using the stereotactic device.

From August 1, 1991, through April 30, 1996, all LCNBs for calcification clusters were performed using the multiple pass technique with an automated biopsy gun and a 14-gauge needle. A mean of 5 core samples were removed from each mass and 7 samples were removed from a calcification cluster. In calcification cluster sampling, if specimen radiography did not show any particles of calcification, additional samples were removed.⁹ From May 1, 1996, to December 31, 1997, all calcification cluster sampling was performed with a DVA instrument (Mammotome, Biopsys Medical, Irvine, Calif). With this device we used a 14-gauge needle in 372 procedures and an 11-gauge needle, which recently became available, in 131 procedures. The DVA device features a single needle insertion to retrieve multiple tissue samples.⁴ The tissue is vacuum aspirated into the notch of the needle and the sample is harvested by means of a high-speed rotating cutter and the specimen is then removed from the breast with the cutter assembly. The needle remains in the breast and may be rotated 360° to capture additional tissue. During the procedure, the biopsy site may be suctioned to minimize tissue displacement secondary to accumulating blood. A mean of 8 tissue samples were removed using either 14- or 11-gauge needles for calcification sampling. If no particles were visible on specimen radiography, additional core samples were removed.

For stereotactic biopsies using the 11-gauge needle in conjunction with the DVA device, the likelihood of complete excision of a mass or calcification cluster of less than 5 mm is high. We now use the 11-gauge needle exclusively for calcium cluster sampling and an increasing number of masses smaller than 1 cm. If radiography after the biopsy shows no residual particles of calcification or the likelihood of complete removal of a mass is high, a 2×2-mm radiopaque localizing clip (MicroMark, Biopsys Medical) is inserted into the biopsy site prior to the removal of the needle. This provides an accurate target for subsequent wire localization and excision if the lesion has atypical features or is malignant on pathologic examination.^{7 ,10}

The 448 noncalcified lesions sampled using ultrasound guidance were performed with a 7.5-MHz linear array transducer (Acoustic Imaging, Denver, Colo). A disposable automated 14-gauge needle was used in all procedures guided by sonography. An average of 5 core samples were obtained from each lesion. No calcification clusters were sampled with ultrasound.

Of the 1836 lesions sampled, 412 (22%) were found to be malignant on the initial LCNB. Careful correlation of pathologic results, imaging findings, and the technical quality of the procedure resulted in a recommendation for a repeat biopsy in 202 lesions yielding 32 additional malignancies. The pathologic findings on surgical excision of these 444 breast malignancies are outlined in Table 1.

Since core biopsy is a sampling procedure, it is imperative to recommend surgical excision when the pathologist discovers atypia, most commonly, atypical ductal hyperplasia (ADH).^{11 - 12} Other atypical features that prompted a recommendation for excision included (1) increased cellularity in a fibroadenoma indicating a possible phyllodes tumor, (2) the presence of necrosis or extravasated mucin, and (3) differentiation of lobular carcinoma in situ from possible ductal carcinoma in situ (DCIS). Of the 88 lesions, in which 27 additional malignancies were diagnosed, 9 were fibroadenoma lesions and possibly phyllodes tumors (7 fibroadenomas and 2 malignant [phyllodes tumors]), 10 were masses with atypical features (all benign), 14 were calcification clusters with atypical features (10 benign and 4 malignant [DCIS]), and 55 were ADHs (34 benign, 15 DCIS, and 6 infiltrating ductal carcinomas [IDCs]).

When only normal breast tissue is diagnosed by LCNB or no calcifications are seen on specimen radiography, the likelihood of a miss is high. In this situation a repeat biopsy is imperative. There were 58 possibly missed lesions. Of these 58, a total of 13 benign and 2 IDC lesions were found when no calcification on specimen radiograph was present and 40 benign and 3 IDC lesions were found in 43 mass lesions. In this group, 5 the malignancies found were during repeat breast biopsy.

The final group of abnormalities carefully scrutinized are those with a strongly suspicious mammographic and/or sonographic appearance, despite a specific benign histologic diagnosis on core biopsy. These included masses and clusters of calcification that the radiologist was confident were accurately sampled. We recommend surgical excision of such lesions to confirm the LCNB diagnosis, as well as to obviate the difficult task of following these abnormalities with mammography and/or ultrasound. There were 43 lesions in this category and in each instance the surgical excision confirmed the core biopsy diagnosis and no additional malignancies were discovered. Surgical excision was performed on 13 lesions for a variety of other reasons and the LCNB diagnosis was confirmed in each instance.

Surgical excision after the diagnosis of DCIS was made on LCNB yielded the following results: no residual DCIS (4 calcification clusters and 1 mass), DCIS (88 calcification clusters and 20 masses), and IDC (16 calcification clusters and 4 masses). Of the 133 cases of DCIS, 20 (15%) were upgraded to IDC. The DVA instrument was used in the 5 cases in which no residual DCIS was found on surgical excision.

Complications of LCNB were minor and infrequent. One superficial infection was successfully treated with antibiotics and 1 pneumothorax requiring no treatment was diagnosed 2 days after the LCNB. There were no hematomas necessitating surgical evacuation.

Table 2 outlines the age distribution and rate of positive biopsy results for both calcifications and masses. The positive predictive value (PPV) is defined as the percentage of women whose mammographic findings were classified as "suspicious biopsy recommended" who had pathologically diagnosed cancer. Overall there were 444 (24% PPV) malignancies diagnosed in 1836 biopsies.

Our follow-up recommendation is a 1-year reexamination for nearly all benign abnormalities. Of the 1222 lesions assigned to follow-up, 70% have been reimaged a year or more after the LCNB. We have yet to discover a malignant lesion that has been assigned to follow-up.

A concern raised since the introduction of LCNB as an alternative to surgical excision is the potential for decreased accuracy resulting in the delay in diagnosis of a breast cancer. Although needle-localized breast biopsy is considered the criterion standard by which other biopsy procedures of nonpalpable abnormalities should be judged, it is not infallible. Correlation of the preoperative mammographic findings, specimen image, and pathology must be made to ensure that the abnormality is successfully removed.¹³ A recent review of needle-localized breast biopsy showed failure to remove a nonpalpable abnormality in 0% to 17.9% (mean, 2.6%) of cases. This study included 48 series of 100 or more needle-localized breast biopsy procedures. The range for missed cancers was 0% to 7.9% (mean, 2.0%).¹⁴

To achieve the highest possible accuracy using the LCNB tissue sampling technique, similar thorough correlations must be made. For example, in our series, after review of pathologic imaging studies and the technical quality of the LCNB, a recommendation for repeat biopsy was made in 202 patients, which represents 14% (202/1424) of the lesions initially diagnosed as nonmalignant. This safety net resulted in the discovery of an additional 32 breast cancers.

The development of the DVA device has improved the accurate sampling of calcification clusters as well as masses. This instrument is designed for a single needle insertion and has a high-speed rotating cutter that can be positioned to sample the tissue adjacent to the needle over 360°. We compared the success of the traditional multiple pass technique for calcification sampling with that of the DVA device using a 14-gauge needle. The miss rate in 130 calcification clusters with the multiple pass technique was 9.2% compared with 100% successful particle removal in 106 clusters using the DVA instrument.⁶

For the past year we have been using an 11-gauge needle with the DVA instrument for all calcification clusters. This needle removes larger core samples (105 mg) compared with 44 mg for the 14-gauge DVA and 17 mg with the automated 14-gauge needle.^{4 ,7} The use of the 11-gauge needle has not increased the complication rate and has allowed accurate sampling of smaller lesions. There is a potential to remove the entire abnormality with both the 11-and 14-gauge needles. Therefore, a 2-mm radiopaque clip is placed at the biopsy site through the needle when lesion removal is considered likely.^{7 ,10} If the pathologic findings indicate a surgical excision is necessary, the clip is used as a marker for subsequent wire localization. Complete lesion removal with the DVA device and a 14-gauge needle has been reported to be from 13% to 48% of all lesions and 53% to 93% of those measuring 5 mm or less.^{4 ,15 - 16}

Of the lesions diagnosed as ADH with LCNB, an average of 5 samples were removed from masses and 8 from each calcification cluster when a 14-gauge automated needle was used. Ten (56%) of 18 lesions were malignant (7 DCIS and 3 IDC) on surgical excision. This compares with other series using an automated 14-gauge needle reporting malignancy rates of 52%¹¹ and 56%¹² after surgical excision of ADH lesions. Acquisition of a larger volume of tissue with the DVA instrument allows for more frequent differentiation of ADH from DCIS.^{17 - 18} In our experience with the DVA instrument using a 14-gauge needle, ADH was diagnosed in 24 calcification cluster samplings. On surgical excision, there were 9 upgrades (38%) with 7 to DCIS and 2 to IDC. With the 11-gauge needle there was 1 ADH upgrade to DCIS in 9 patients (11%).

When DCIS was diagnosed by LCNB using an automated 14-gauge needle, reported upgrades to IDC on surgical excision were 19% and 20%.^{11 - 12} In our experience, with both the 14-gauge automated and DVA needles, there was a 19% incidence of DCIS upgrade to IDC. However, using the 11-gauge needle with the DVA instrument to sample calcifications, there was only 1 DCIS upgrade to IDC in 28 cases (4%). It is clear that the larger the volume of tissue removed, the less likely there will be an upgrade of ADH or DCIS diagnosed on the LCNB.

One of the criticisms of LCNB is the potential for unnecessary sampling of probably benign lesions that would result in an unacceptably low PPV. There has also been concern that LCNB would replace the 6-month imaging follow-up, which has been the cornerstone of managing benign-appearing solitary masses or clusters of calcification.¹⁹ In the evaluation of suspicious nonpalpable breast lesions in the US medical literature, the PPV of mammography ranges from 15% to 30%.²⁰ In a series from our hospital, 237 (18.8%) of 1261 nonpalpable breast abnormalities were found to be malignant on excision after wire localization.¹³ Over a subsequent 3-year period from 1990 to 1992, the PPV of 2246 wire localizations of nonpalpable suspicious breast lesions was 25%. The PPV of 24% achieved in our LCNB patient population suggests that we are not overusing this diagnostic tool.

Another advantage of LCNB over surgical biopsy is reduced cost, which by many estimates is one half to one third.^{21 - 22} In addition, when a carcinoma is initially diagnosed with LCNB, significantly fewer surgical procedures are required to achieve clear margins when breast conservation is the therapeutic goal.²³ In addition to cost saving, LCNB is more easily scheduled and takes less time to perform.

In summary, we have found LCNB of suspicious breast abnormalities accurate, safe, and well accepted by patients and referring physicians. Overuse is not apparent in our experience.