点击显示 收起
1 From the Department of Radiology, Palo Alto Medical Clinic, 795 El Camino Real, Palo Alto, CA 94301. From the 1999 RSNA scientific assembly. Received January 4, 2001; revision requested February 28; revision received May 1; accepted May 22. Partially supported by an educational grant from Biopsys to the Palo Alto Medical Foundation.
| ABSTRACT |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
MATERIALS AND METHODS: Thirty-one (1.3%) of 2,399 consecutive lesions on which stereotactic histologic biopsy was performed were in breasts containing implants. Biopsy difficulties were evaluated for lesions in breasts with and breasts without implants. Biopsy was performed on lesions in patients with implants prone on a dedicated table, with automated large-core (n = 13) or directional vacuum-assisted (n = 18) devices. Follow-up was surgical (11 of 11 malignancies and two of three high-risk lesions) and mammographic (one of three high-risk lesions and 17 of 17 benign lesions).
RESULTS: There were no implant ruptures, hematomas requiring drainage, infections requiring treatment, false-negative findings, or histologic underestimations. Difficulties with stereotactic histologic biopsy were more prevalent in breasts with implants and included positioning problems in 10 (50%) of 20 lesions in breasts with subglandular implants and zero (0%) of 10 with subpectoral implants, lesions seen on only one view in four (13%) of 31 lesions, specimen radiographs negative for calcifications in two (10%) of 20 lesions, prominent bleeding in two (6%) of 31 lesions, and suboptimally small tissue samples in three (10%) of 31 lesions.
CONCLUSION: Stereotactic histologic biopsy is safe in breasts with implants. Compared with that in breasts without implants, biopsy is often technically more difficult and may eventually prove less accurate.
Index terms: Biopsies, technology • Breast, biopsy, 00.1261, 00.1262, 00.1267 • Breast, prostheses • Breast neoplasms, diagnosis, 00.1261, 00.1262, 00.1267, 00.30
| INTRODUCTION |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
Any type of biopsy can potentially damage a breast implant. Cytologic diagnosis with fine-needle aspiration (FNA) of lesions in breasts with implants by using upright stereotactic (7), palpation (8), and ultrasonographic (US) (9) guidance has been reported. Histologic diagnosis of mammographically detected lesions in breasts with implants has been reported, with prelumpectomy needle localization by using a combination of implant displacement, as described by Eklund et al (10) for diagnostic mammography, and hook-wire insertion (11,12).
Histologic diagnosis of nonpalpable breast lesions with image-guided needle biopsy (13–16) is a progressively popular alternative to needle-localized breast biopsy (NLBB). Tissue acquisition has been described as automated large-core biopsy (hereafter, large-core biopsy) (13–17) and directional vacuum-assisted biopsy (hereafter, vacuum-assisted biopsy) (13,14,16,18). To our knowledge, these methods have not been described in patients with breast implants. The purpose of our retrospective study was to report our experience with large-core and vacuum-assisted biopsy with prone stereotactic guidance in patients with breast implants.
| MATERIALS AND METHODS |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
In the study group, the 31 biopsies were performed on 26 breasts in 25 patients (age range, 35–75 years; median age, 58 years). In five of the patients, biopsy was performed on a second lesion in the ipsilateral breast the same day, and in one of these five patients, biopsy was performed on a third lesion in the contralateral breast. The only discretely palpable lesion was the one lesion in the study on which FNA was performed prior to stereotactic biopsy; the procedure was performed at an outside facility by using palpation guidance and was nondiagnostic. In no lesion in the breasts with implants was the initial diagnostic biopsy performed with NLBB or US guidance.
The 2,368 biopsies performed outside our study group were in 2,149 breasts without implants in 2,089 patients (age range, 29–94 years; median age, 55 years). Patients underwent biopsy of multiple lesions (two [n = 186], three [n = 15], or four [n = 1]) in the same breast on the same day. In no lesions in the breasts without implants was initial diagnostic biopsy performed with US guidance at our institution. We do not have data regarding initial diagnostic biopsy with FNA or NLBB in breasts without implants.
Data were evaluated for most variables in the study and nonstudy (without implants) groups. Prospective analysis was performed by the radiologist performing biopsy (including R.J.J. and R.L.L.)(for prebiopsy mammographic and biopsy factors), by the clinician referring the patient for biopsy (for lesion palpability), by the surgeon performing postbiopsy surgery (for implant rupture), by the radiologist performing needle localization for NLBB (for implant rupture), and by the radiologist interpreting postbiopsy mammograms (for implant rupture and the postbiopsy status of lesions for which surgery was not performed). All retrospective analysis was performed by one of the authors (R.J.J.).
Lesions were prospectively categorized according to lesion type (microcalcifications and/or masses), maximum mammographic lesion diameter, and the Breast Imaging Reporting and Data System (BI-RADS) lexicon of the American College of Radiology (19) (categories 2–5); BI-RADS classification started prospectively in February 1995. Lesion palpability was prospectively categorized as discrete, vague, or not present.
Biopsy was performed by using a prone-biopsy table (Mammotest; Fischer Imaging, Denver, Colo) and three successive techniques. We started with large-core biopsy with a long-throw (2.3-cm-excursion) biopsy gun (Biopty; Bard Urological, Covington, Ga), using a variety of 14-gauge cutting needles, and switched to vacuum-assisted biopsy with a Mammotome device (Biopsys Medical/Ethicon Endo-Surgery, Cincinnati, Ohio) when that technology became available. Vacuum-assisted biopsy was performed initially with 14-gauge probes and later with 11-gauge probes. During the 1-month overlap between the various biopsy techniques, the availability of the new probe determined which method was used. No lesion or patient variables were used to determine the method used. Biopsy was performed on lesions in patients with implants by using large-core (n = 13) or vacuum-assisted (n = 18) devices. The number of tissue samples obtained with each biopsy technique was recorded.
All lesions were prospectively evaluated to determine if they were evident on one or two mammographic views. All patients were examined prospectively by the radiologist performing the biopsy for complications hampering biopsy completion (ie, bleeding or pain) and for complications requiring unusual postbiopsy care (ie, bleeding or infection). Starting in March 1993, specimen radiographs were prospectively evaluated for calcifications. This was done routinely for microcalcification lesions and at the discretion of the radiologist performing biopsy for calcified masses.
Percutaneous histologic diagnoses were prospectively recorded for all lesions as malignant, high risk, or benign. Malignant lesions included invasive carcinoma, ductal carcinoma in situ (DCIS), lymphoma, and sarcoma. The type and number of surgeries required for therapy of the malignant lesions were recorded. We consider atypical ductal hyperplasia (ADH), atypical lobular hyperplasia, lobular carcinoma in situ, and radial scar to be high-risk lesions in which the associated presence of carcinoma can be underestimated with percutaneous biopsy; we recommend that high-risk lesions are surgically excised. The surgical histologic diagnoses were recorded for those lesions. Lesions not categorized as histologically malignant or high risk were classified as benign.
We recommend 12-month postbiopsy bilateral mammography if the benign diagnosis is fibroadenoma or a lymph node. For all other benign diagnoses thought to be concordant with the mammographic findings, we recommend 6-month postbiopsy mammography in the breasts in which biopsy is performed and 12-month postbiopsy bilateral mammography. Subsequently, all patients with benign lesions for which repeat biopsy is not thought necessary are asked to undergo annual diagnostic mammographic follow-up for at least 3 years after biopsy. Postbiopsy mammograms were evaluated prospectively, and the lesion on which biopsy was performed was categorized as not present, decreased in size, stable, or increased in size. Repeat biopsy, with a more aggressive percutaneous approach or surgical excision, is recommended if there is discordance between mammographic and histologic findings or substantial lesion progression at postbiopsy mammographic follow-up. Follow-up in 31 study lesions was surgical (11 of 11 malignancies and two of three high-risk lesions) and mammographic (one of three high-risk lesions and 17 of 17 benign lesions).
All lesions for which patients were referred for stereotactic biopsy that was not completed had prospective reasons recorded by the radiologist attempting the biopsy. Reasons were categorized as inadequate lesion suspicion to warrant biopsy or as various difficulties precluding biopsy of a worrisome lesion.
Data for some variables were evaluated only for lesions in the study group. BI-RADS classification was performed retrospectively for study lesions on which biopsy was performed before February 1995. Prebiopsy mammograms were used to retrospectively categorize the implant position (subglandular or subpectoral) and type (silicone filled and/or saline filled). Stereotactic images, available for 30 of the 31 lesions, obtained with the patient prone during biopsy were retrospectively evaluated for difficulties with implant displacement, lesion visualization, and separation of the lesion from the implant. Prospective comments were recorded for difficulties in achieving firm compression and adequate thickness in the z-axis direction of the breast tissue on which biopsy was being performed. The preceding factors were retrospectively combined into a subjective "positioning problem" category described as absent, mild, moderate, marked, or impossible to overcome.
The size of the tissue samples extracted from each study lesion was retrospectively evaluated by using prospective comments and, when available, retrospective evaluation of the size of the tissue samples on the specimen radiographs. The sample size was subjectively categorized as adequate or suboptimally small relative to the size of samples in other lesions on which biopsy was performed with the same technique.
All patients with study lesions were prospectively evaluated for implant rupture caused by stereotactic biopsy. For lesions on which postbiopsy surgery was performed, this was done by using the prospective comments in the surgical report (n = 13) and the needle-localization report (for lesions on which NLBB was performed [n = 6]). For lesions for which postbiopsy mammographic follow-up without surgery was performed (n = 18), this was done by using prospective comments in the mammographic report.
Data were analyzed with statistical software (STAT-VIEW; Abacus Concepts, Berkeley, Calif). A 2 test–based P value of less than .05 was considered to indicate a significant difference. Age was a per-patient variable. Implant rupture was a per-breast variable. All other variables were per lesion; therefore, the lesion was the unit of analysis. To eliminate possible confounding variables of multiple lesions per breast, the inability to see the lesion on both mammographic views for patients with implants was compared with that for patients without implants, for breasts with one or two lesions.
| RESULTS |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
|
Positioning problems were present in 10 (33%) of the 30 lesions for which stereotactic images obtained with the patient prone during biopsy were available . The problems occurred in 10 (50%) of the 20 lesions in breasts with subglandular implants and in none (0%) of the 10 lesions in breasts with subpectoral implants. Positioning problems were associated with four (25%) of 16 silicone gel–filled implants and six (55%) of 11 saline-filled implants; no problems were associated with the three double-lumen implants. The positioning problems were thought to be clinically unimportant in six lesions and important in four lesions—rated in three as moderate (Fig 1) and in one as marked (Fig 2). None of the positioning problems was rated as impossible to overcome.
|
|
|
|
Four other types of percutaneous biopsy difficulties occurred . First, despite vigorous prebiopsy imaging efforts, four (13%) of the 31 study lesions and 34 (1%) of the 2,368 nonstudy lesions could be depicted on only one mammographic view (P < .001). Inability to see the lesion on both mammographic views in breasts with one lesion occurred in three (14%) of 21 study lesions and 31 (1.6%) of 1,947 nonstudy lesions (P < .001). With two lesions per breast, the inability occurred in one (10%) of 10 study lesions and three (0.8%) of 372 nonstudy lesions (P < .007).
Second, prominent bleeding occurred in two (6%) of the 31 study lesions and 50 (2%) of the 2,368 nonstudy lesions (P > .09). In the study group, both lesions (13%) were part of the 15 lesions on which 11-gauge vacuum-assisted biopsy was performed. One patient required 3 hours of vigorous postbiopsy breast compression to stop prominent venous bleeding. The other bleeding problem, a large hematoma developing early in the course of the biopsy, is discussed subsequently. Biopsy completion was not hampered by any other bleeding problems or by pain.
Third, specimen radiographs were negative for calcifications in two (10%) of the 20 microcalcification study lesions and 55 (5%) of the 1,195 microcalcification nonstudy lesions for which such radiographs were available (P > .25). In the study group, one negative specimen radiograph was obtained for the lesion with marked positioning problems (poor implant displacement and poor compression of breast tissue) and suboptimally small tissue samples from large-core biopsy (Fig 2); the other was from the lesion with a large hematoma that obscured the calcific lesion and thus prevented retargeting of the lesion after specimen radiographic findings in the initial five specimens were negative . Both calcified masses in the study group had specimen radiographs positive for calcifications.
|
|
Percutaneous histopathologic results in the study group were malignant in 11 (35%), high-risk in three (10%), and benign in 17 (55%) of the 31 lesions. The malignancies were invasive ductal carcinoma with associated DCIS (n = 2), invasive lobular carcinoma (n = 4), and DCIS without associated invasive carcinoma (n = 5). Seven of the carcinomas in four breasts in four patients (including the three patients who each had two carcinomas in the same breast) were treated with initial mastectomy. The other four carcinomas were initially treated with therapeutic lumpectomy; three of those four lumpectomy specimens were found to have positive margins, and the three patients with those specimens subsequently underwent mastectomy. None of the five DCIS lesions was upgraded to invasive carcinoma at surgery.
Two of the three high-risk study lesions diagnosed at percutaneous biopsy were ADH; neither was upgraded to carcinoma at subsequent lumpectomy. The third high-risk lesion, a combination of ADH and atypical lobular hyperplasia, was in a patient who at the time this article was written had declined surgery but had undergone mammographic follow-up; the 7-mm calcific cluster was not present on mammograms obtained 7 and 21 months after biopsy.
None of the patients with 17 benign study lesions underwent surgery; all 17 lesions underwent mammographic follow-up for 14–97 months (mean, 49 months; median, 40 months) after biopsy. Mammographic follow-up was 14–25 months in seven lesions and 37–97 months in 10 lesions. At initial follow-up mammography, eight lesions were not present (including lesions at 14, 19, 21, and 25 months follow-up), seven had decreased in size, two were unchanged, and none had increased in size.
| DISCUSSION |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
NLBB, in past years the standard method for tissue diagnosis in mammographically detected lesions, was reviewed in a study by Jackman and Marzoni (21). There was failure to remove at least part of a nonpalpable lesion in 0%–17.9% (mean, 2.6%) of cases and a cancer miss rate of 0%–7.9% (mean, 2.0%).
Percutaneous biopsy as an alternative to NLBB for nonpalpable breast lesions can be performed with a variety of techniques. The Radiologic Diagnostic Oncology Group 5 multicenter trial (22) was designed to test the relative accuracy of image-guided biopsy of nonpalpable breast lesions with FNA versus large-core biopsy. Because FNA produced insufficient samples in 128 (34%) of 377 lesions, the FNA arm of the study was terminated early without waiting for the accuracy comparison (22).
Large-core biopsy (13–17) performed in standardized fashion (with stereotactic guidance, a prone table, a 14-gauge needle, a long-throw gun, and extraction of at least five specimens per lesion) (17) is a progressively popular alternative to NLBB, with a cancer miss rate of 0.3%–8.2% (mean, 4.0%) (23).
Stereotactic biopsy with a vacuum-assisted technique has resulted in extraction of larger specimens, more specimens per unit time, and more contiguous tissue than has the large-core technique, with no increase in complications (18,24). By averaging the results of some studies (18,24,25), the mean weight per individual biopsy specimen has been reported as 17 mg for 14-gauge large-core biopsy, 37 mg for 14-gauge vacuum-assisted biopsy, and 95 mg for 11-gauge vacuum-assisted biopsy. It is interesting that those weights are paralleled by percutaneous removal rates of histologically benign lesions. As judged by using initial postbiopsy mammograms, usually obtained 6–12 months after biopsy, no residual lesion was identified in 9% (40 of 422) of the 14-gauge large-core biopsies, 22% (21 of 95) of the 14-gauge vacuum-assisted biopsies, and 64% (96 of 150) of the 11-gauge vacuum-assisted biopsies (26).
Although the cancer miss rate for vacuum-assisted biopsy is unknown at this time, the prominent increases in weight per specimen and in removal rates of histologically benign lesions are encouraging. We speculate that there will be a significant decrease in cancer misses at vacuum-assisted biopsy, especially with the 11-gauge probe.
US-guided biopsy of nonpalpable lesions has been reported with large-core (27–29) and vacuum-assisted techniques (30,31). We are unaware of any US-guided histologic biopsy series (vs stereotactic histologic biopsy series) in which all lesions resulted in either correlative surgical biopsy or a minimum of 2–3 years of imaging follow-up with stable findings, making US-guided biopsy accuracy more difficult to evaluate. US-guided biopsy of lesions in breasts with implants could theoretically decrease biopsy difficulties and increase accuracy by allowing one to visualize both the lesion and the implant during biopsy. We have no experience with that technique and will be interested to see the results of future studies.
Lesions in breasts containing implants have been diagnosed cytologically with FNA (7–9) and histologically with NLBB (11,12) . In those series, there were no reported cancer misses, implant ruptures, or other clinically important complications.
|
Because the study group of 31 lesions was just 1.3% of the 2,399 lesions on which biopsy was performed, it would require large differences between the two groups to achieve statistical significance. The data insuggest that there may be more biopsy-related problems in breasts with implants than in breasts without implants when comparing lesions depicted on only one mammographic view (13% vs 1%, respectively), specimen radiographs negative for calcifications (10% vs 5%), and prominent bleeding (6% vs 2%). Whereas a statistically significant difference was found only for seeing the lesions on one view, the other findings may be clinically important. Prominent bleeding in the study group occurred only in lesions in which 11-gauge vacuum-assisted biopsy was performed (two [13%] of 15 lesions), which may be meaningful. In some implant cases, it is impossible to maximally compress the breast tissue and not push the lesion out of mammographic view. Suboptimally compressed breast tissue may have more tendency to bleed during biopsy with any kind of device.
The 10% incidence of suboptimally small sample size in the implant cases is subjectively thought to be much higher than that in the nonimplant cases, but we have no data from the nonimplant cases for comparison. The suboptimally small tissue samples in the study group occurred only in cases in which 14-gauge large-core biopsy was performed (three [23%] of 13 lesions), which may be meaningful. Obtaining tissue samples of adequate size from suboptimally compressed breast tissue may be less problematic with the vacuum-assisted devices. Tissue mobility should not hamper vacuuming tissue into the collection chamber. With the large-core devices, needle firing and the tissue acquisition are intimately related; mobile tissue might easily be pushed by the needle movement and not enter the collection chamber.
Although positioning problems, especially adequate implant displacement, were present in 10 (33%) of 30 lesions, they were clinically important in only four (13%) of the 30 lesions in which evaluation was possible. Although positioning problems did not prevent us from completing any biopsy, they were thought to be responsible for two of the three cases in which suboptimally small cores were obtained. Positioning problems occurred, as one might expect, with only subglandular implants (10 [50%] of 20 lesions). With subpectoral implants, the intervening pectoralis muscle increased the distance between the breast lesion and the implant. Whereas positioning problems were more frequent with saline-filled implants (55%) than with silicone-filled implants (25%), we think this is related to the position and not the content of the implant .
The combination of a large subglandular implant, a small volume of breast tissue, and the proximity of a small lesion to the implant created the most challenging positioning situation. Achievement of complete implant displacement from the visualized field and optimal firm compression of the adjacent breast tissue can also displace the lesion from the field of view. Vigorous efforts and compromises in positioning are required to displace the implant, see the lesion, and have relatively firm compression of adequately thick breast tissue to proceed with biopsy.
Implant rupture is a concern during any kind of breast biopsy, particularly if the lesion is positioned close to the implant after maximal implant displacement and breast compression. We much prefer the vacuum-assisted device in those circumstances because we can rotate the collection chamber away from the implant during tissue acquisition (Fig 1). With the large-core device, the collection chamber is always in the twelve o’clock position, which is often occupied by the implant. In addition, when the lesion is in close proximity to the implant, we prefer to manually push rather than mechanically fire the probe to its final biopsy position. This allows us to progressively monitor the proximity of the probe to the implant and might decrease the risk of implant rupture (Fig 1). This can be done with only the vacuum-assisted device, since mechanical firing is a necessary part of tissue extraction with the large-core device.
The 11-gauge vacuum-assisted device is strongly preferred by the authors for most stereotactic biopsies of lesions in breasts with implants, particularly if the breast tissue is suboptimally compressed and/or the lesion is close to the implant. As suggested in this study, however, the 11-gauge device may have an increased incidence of bleeding problems.
At the time this article was written, there were no known false-negative biopsy findings in the 17 histologically benign lesions, and none resulted in correlative surgical excision. These numbers are too small to be meaningful. The two patients with specimen radiographs negative for calcifications each had small clusters of amorphous calcifications that met our criteria for BI-RADS category 4 . Those two lesions were of less concern than usual, however, because the calcifications had shown just equivocal mammographic progression for 9.0 and 3.5 years prior to biopsy. After postbiopsy discussion with their referring physicians, both patients with the lesions elected for mammographic follow-up (which has shown postbiopsy stability for 70 and 20 months) rather than repeat biopsy. Mammographic follow-up in the 17 benign lesions for 14–97 months after biopsy has shown no lesion progression. Neither of the two ADH lesions and none of the five DCIS lesions proved to be histologic underestimations at subsequent surgery, but these numbers are too small to be meaningful. One patient with ADH and atypical lobular hyperplasia has declined surgery but has undergone mammographic follow-up, which, at 7 and 21 months, has shown the calcifications on which biopsy was performed to be absent.
Stereotactic histologic biopsy in this small series of 31 lesions in breasts with implants appears safe (with no implant ruptures, hematomas requiring drainage, or infections requiring treatment) and may be accurate (with no false-negative findings or histologic underestimations), but technical difficulties (with positioning problems and relatively high rates of lesions being seen on only one mammographic view, specimen radiographs being negative for calcifications, biopsy-induced bleeding, and suboptimally small tissue samples) suggest that the eventual accuracy may be less than in breasts without implants. Larger studies are needed for further evaluation. Because biopsy of lesions in breasts with implants is uncommon (representing only 1.3% of the stereotactic biopsies at our institution), it may require a multiinstitutional study to acquire sufficient data.
| ACKNOWLEDGMENTS |
|---|
| REFERENCES |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|