Fine-needle aspiration biopsy (FNAB) is used as the first-line diagnostic test for lesions that require morphologic assessment in the authors' breast cancer screening program. A positive cytologic diagnosis is an indication to proceed to definitive surgery. Core biopsy is used if FNAB is not diagnostic. In the context of increased use of core biopsy at other centers, the authors reviewed their experience with the cytologic assessment of highly suspicious microcalcifications.
Between January 1996 and June 2000, the dominant radiologic abnormality was classified prospectively as high-grade microcalcifications in 182 lesions. Data were recorded on patient demographics, radiologic features, and the findings of FNAB and core biopsy, if performed. The results of the screening assessment were then compared with the final histologic findings.
Overall, 15.6% of all radiologically high-grade lesions were microcalcifications. The mean patient age was 58.76 years. The lesions had a mean size of 38.49 mm (range, 5–200 mm), and 92.31% of high-grade microcalcifications proved to be malignant. Among the cases evaluated by FNAB, a positive cytologic diagnosis of malignancy was made in 70.93% of lesions, without any false-positive diagnoses and obviating the need for diagnostic core biopsy. FNAB had a sensitivity of 77.22% and a positive predictive value (PPV) of 100%. When core biopsy was performed due to the absence of a positive cytologic diagnosis, it averted the need for open biopsy in 76% of lesions.
Microcalcifications constitute one of the principal categories of abnormalities detected by screening mammograms. Certain characteristics of microcalcifications, such as their size, pleomorphic shape, and linear distribution, have been correlated with the presence of ductal carcinoma in situ (DCIS).1 Such highly suspicious microcalcifications are classified as Grade 5 in the Tabar classification system for mammographic abnormalities1 and also in the classification scheme recently proposed by the American College of Radiology (Breast Imaging Reporting and Data System).2
Currently, impalpable breast lesions that require morphologic assessment are investigated by core biopsy in many centers. Since the early 1990s, these have been automated, 14-gauge core biopsies3; and, more recently, the 11-gauge, directional, vacuum-assisted devices have provided larger samples for tissue diagnosis.4–6 Core biopsy is a highly accurate method of investigating impalpable breast lesions.7–9 However, compared with fine-needle aspiration biopsy (FNAB), core biopsy has several drawbacks. These include higher cost, greater patient discomfort, longer delay in availability of results, greater distortion of the area by hemorrhage and tissue reaction that sometimes necessitates a delay in definitive surgery, epithelial displacement, and the risk of tumor implantation in the core biopsy track.10–13 The latter concern requires close consultation between radiologists and surgeons so that core biopsy sites are selected in locations that traverse the least amount of tissue and are amenable to complete excision of skin and track in case of malignant lesions. Despite these factors, the high reliability of the histologic results of core biopsy specimens14, 15 and the lack of access to cytopathologists in some centers have broadened the appeal of core biopsy for evaluating microcalcifications. Some centers in the United Kingdom breast cancer screening program have converted from cytology to core biopsy,16, 17 and in the Australian setting, several centers participating in the nationwide program of mammographic screening for breast cancer, BreastScreen Australia, also have made this change.
In our screening program, radiologic abnormalities that require pathologic evaluation are assessed primarily by FNAB with immediate on-site reporting. The cytologic diagnosis of malignancy is an indication for definitive surgery without confirmatory core biopsy or frozen section examination. Lesions in which the cytomorphology is not diagnostic of carcinoma are evaluated further by histology, including core and open surgical biopsy. Faced with the question of whether we should follow international trends and change our primary diagnostic modality from cytology to core biopsy, we reviewed our experience with the radiologic, cytologic, and histologic evaluation of Category 5 microcalcifications in the setting of a population-based screening program for breast cancer.
MATERIALS AND METHODS
Breast Cancer Screening Program Design
Our program in South Australia, BreastScreen SA (BSSA), is part of a national breast cancer screening program and has been accredited to provide this service since 1991 after the completion of a 3-year pilot study. The screening efforts are aimed primarily at asymptomatic women ages 50–69 years. Screening mammograms are reviewed independently by two radiologists and a third reader arbitrates discordant findings. Women in whom the screening mammograms reveal Category 5 microcalcifications undergo further diagnostic imaging, including magnification views and ultrasound evaluation. During these assessments, stromal densities, stellate or circumscribed mass lesions and architectural distortions are sought specifically as signs of a possible invasive component. FNAB is then performed with stereotactic or ultrasound guidance or by palpation, as appropriate. The smears are prepared by a cytopathologist, are stained with a rapid staining technique (Diff-Quik), and are examined immediately. In light of the cytologic features, up to two further passes may be performed until the team is satisfied that a representative sample has been obtained. Drawing on published guidelines,18 the final cytologic diagnosis is classified into one of six categories: malignant, suspicious, atypical/equivocal, benign-specific, benign-nonspecific, or inadequate.
In this program, predetermined decision pathways are followed for the further management of each category of mammographic lesions and each possible cytologic diagnosis. Figure 1 indicates that, for Category 5 microcalcifications without a positive cytologic diagnosis of malignancy, our algorithms require tissue diagnosis of samples that are confirmed by specimen X-rays to contain the suspicious microcalcifications. This usually entails immediate core biopsy or, less frequently, open biopsy, subject to patient preferences and suitability of the lesion. The same pathologist examines the core biopsies the next day.
A breast surgeon is in attendance at the assessment clinic to discuss the results with the client and to help formulate management recommendations that are communicated with her and with her primary care provider. The clinical, radiologic and cytologic findings and the team's recommendations are documented by letter, and a copy of this letter is sent to the client's primary care provider, who coordinates her subsequent care.
All cytologically positive aspirates are reviewed independently by a second cytopathologist the next day. If there is doubt regarding the malignancy of the lesion, then the cytology is reclassified as suspicious. The surgeon then modifies the management recommendations on this basis, and these alterations are communicated promptly to the primary care physician.
In our program, core biopsy is performed for Category 5 microcalcifications under the following circumstances: 1) to establish invasion when the FNAB is positive and there is radiologic suspicion that the lesion may have an invasive component; 2) to establish the diagnosis when the FNAB result is not positive; and 3) on a few occasions, as the first-line investigation without FNAB, at the discretion of the duty radiologist. Results of all clinical, radiologic, cytologic and histologic findings are audited prospectively by designated specialists from each discipline and then entered into a large database by the program's data managers.
For the current study, we searched the database of BSSA between January 1996 and June 2000 for lesions in which the dominant mammographic abnormality was coded as Category 5 microcalcifications. In instances of multiple abnormal foci in the same patient, we focused on the main lesion, which was coded in the BSSA data base as Lesion 1. Lesions with Category 5 microcalcifications in which mammographic abnormalities of other types (mass, architectural distortion, stromal density) had been coded as constituting the dominant radiologic abnormalities were excluded from this study.
We collected data on patient age, year of assessment, round of screening, radiologic characteristics of the lesion (size of area involved, distribution of microcalcifications, associated abnormalities), method of obtaining a cytologic sample, and the results of FNAB. For lesions in which core biopsy was performed, we recorded the method of guidance for core biopsy, the type of core and the histologic diagnosis. The rates of performance of core biopsy and open biopsy also were recorded for each year under review. The results of the screening assessment (FNAB and core biopsy) then were compared with the final histologic findings, as documented in the surgical pathology report.
Patient and Lesion Characteristics
Between January 1996 and June 2000, among 1065 radiologically high-grade lesions, Category 5 microcalcifications had been recorded prospectively as the dominant radiologic category by the auditing radiologists in 182 cases (15.62%). These are the focus of the current study. The mean patient age was 58.8 years (range, 41–84 years). In 48.35% of patients, the lesion arose in the right breast, and the remaining 51.65% lesions affected the left breast. The mean round of screening at assessment was 2.56 (standard deviation, 1.4). The mean size of the area involved was 38.5 mm (range, 5–200 mm), as measured from the mammograms. The microcalcifications were disposed as a single cluster in 48.1% of lesions, as multiple clusters in 20.4% of lesions, and as dispersed calcifications in the remaining 31.5% of lesions.
FNAB was performed in 172 lesions. The outcome of the cytologic assessment is shown in Table 1. The cytology was interpreted as positive in 70.93% of lesions, suspicious in 8.14% of lesions, atypical in 4.65% of lesions, benign in 4.65% of lesions, and inadequate in 11.62% of lesions. All lesions (122 of 122) with Category 5 microcalcifications and malignant cytology proved to be malignant after histologic examination of the surgical specimen. Malignancy ultimately was found in 12 of 14 lesions (85.7%) with suspicious cytology, in 5 of 8 lesions (62.5%) with atypical cytology, in 6 of 8 lesions (75%) with benign cytology, and in 13 of 20 lesions (65%) with inadequate cytology.
Table 1. The Cytologic-Histologic Correlation in 172 Category 5 Microcalcifications Assessed by Fine-Needle Aspiration Biopsy
FNAB status (%)
FNAB: fine-needle aspiration biopsy.
Proportion of all lesions
The FNAB results were interpreted as positive in 67.07% of lesions with a single cluster of microcalcifications, in 69.44% of lesions with multiple microcalcifications, and in 77.36% of lesions with dispersed microcalcifications (Fig. 2). The difference in the rate of a positive cytologic diagnosis between these groups was not statistically significant (P = 0.42).
Reasons for Nonmalignant Cytology in Aspirates of Lesions with High-Grade Microcalcifications
A malignant cytologic diagnosis was not made in 29.07% of the high-grade microcalcifications assessed by FNAB (50 lesions). There are three reasons for this result, each accounting for approximately one-third of the lesions with nonmalignant cytologic results. These reasons are 1) benignity of lesion, 2) inadequate or nonrepresentative samples, and 3) difficult smears with indefinite features.
In the current series, 8.2% of lesions that presented as Category 5 microcalcifications were, in fact, benign. These lesions included stromal calcifications, ductal hyperplasia, adenosis, calcified fibroadenoma, mucocele-like lesions, lobular carcinoma in situ, and atypical lobular hyperplasia. None of these lesions were diagnosed as malignant.
Together, inadequate and clearly benign smears accounted for 16.27% of all lesions assessed by FNAB. In 75.7% of these samples, the lesions proved to be malignant. The sample obtained for FNAB was most likely nonrepresentative in the lesions that had a benign pattern, because high-grade DCIS was the most common final diagnosis, and the marked nuclear atypia that characterizes high-grade DCIS is unlikely to be overlooked if it is present on the smear. In another 34% of lesions without a positive FNAB diagnosis, the cytologic features were not definite for malignancy for a confident, positive diagnosis by the reporting pathologist.
Further Assessment of Lesions without Positive FNAB Results
According to our protocols, the 50 lesions in which the result of FNAB assessment was other than malignant required further investigation. The method used to reach a final diagnosis is summarized in Table 2. FNAB was repeated outside the program in one lesion, resulting in a positive diagnosis that was confirmed by definitive surgery. Twenty-four patients underwent open biopsy, and the remaining 25 patients were evaluated by core biopsy. Among the 24 patients who underwent open biopsy, the lesion ultimately was identified as malignant in 17 patients (70.83%) and nonmalignant in 7 patients.
Table 2. Method of Reaching a Final Diagnosis for 182 Category 5 Microcalcifications
FNAB: fine-needle aspiration biopsy.
FNAB outside program
Of 25 lesions that were evaluated by core biopsy after nonmalignant cytology, 18 lesions (72%) finally were proven malignant, and 7 lesions were benign. Core biopsy successfully identified 15 of these 18 malignant lesions (83.3%); and, in another 4 lesions, provided a definitive benign diagnosis that accounted for the mammographic findings, so that open biopsy was not required. Considering lesions assessed by FNAB in the program, FNAB had a sensitivity of 77.22% (122 of 158 biopsies) positive predictive value (PPV) of 100%.
Lesions Assessed without First-Line FNAB
Table 2 shows that there were 10 lesions in the database without recorded results of FNAB. Six of these lesions were evaluated only by core biopsy, and all were diagnosed as DCIS. Of the four other lesions, the final diagnosis was reached by open biopsy in two lesions and FNAB performed outside the program in two other lesions. All of these lesions were malignant.
Summary of Our Experience with Core Biopsy for Category 5 Microcalcifications
Our experience with the use of core biopsy in Category 5 microcalcifications is summarized in Table 3. Overall, core biopsy was performed in 54 of 182 lesions (29.67%), 19 of which were vacuum-assisted cores, and the remaining samples were from automated core biopsies. The mean number of cores per lesion was 4.17 ± 1.58 cores (standard deviation). Core biopsy was performed to evaluate the presence of invasion in samples with positive cytology in 23 lesions (42.6%), to establish the diagnosis when the cytology was not positive in 25 lesions (46.3%), and as the first-line investigation without a prior FNAB in 6 lesions (11.1%).
Table 3. Results of Core Biopsy in the Assessment of Category 5 Microcalcifications
Among the 23 cases in which core biopsy was performed to evaluate the presence of invasion, invasion was identified in the final surgical specimen in 10 cases. Core biopsy included the invasive component in 5 of 10 lesions (50%). Core biopsy was inconclusive or unsuccessful in sampling the lesion in 3 of 23 cases (13.0%).
When core biopsy was performed because a malignant cytologic diagnosis was not available, a definite diagnosis that accounted for the radiologic abnormalities was achieved in 19 of 25 lesions (76%). This was malignant disease in 15 lesions and a nonmalignant process in 4 lesions. The core biopsies were nondiagnostic in 6 of 25 lesions (24%), necessitating open biopsy to reach a final diagnosis. Considering this group of lesions, in which FNAB had not been positive and core biopsy was being used as a diagnostic test, its sensitivity was 83.33% and its PPV was 100%. It was found that all 6 lesions in which core biopsy was the primary method of investigation of Category 5 microcalcifications constituted in situ carcinoma.
Final Outcome of Assessment
After final pathologic examination, 168 of 182 lesions (92.3%) were malignant. Among these, DCIS was the final diagnosis in 65.3% of lesions, whereas an invasive component was documented in 34.7% of lesions. The DCIS was high grade in 77.2% of lesions, intermediate grade in 15.4% of lesions, and low grade in 7.4% of lesions.
Among the 58 lesions with foci of invasion, the mean greatest dimension of the invasive component was 12.8 mm (range, 1–54 mm). The tumors were Grade 1 in 28% of lesions, Grade 2 in 58% of lesions, and Grade 3 in 7% of lesions. Lymph node metastases were found in 28.1% of invasive carcinomas. The mean number of positive lymph nodes was 1.4.
The 14 instances of nonmalignant diagnoses in this series included stromal calcifications, ductal hyperplasia, adenosis, calcified fibroadenoma, mucocele-like lesions, lobular carcinoma in situ, and atypical lobular hyperplasia. Overall, the final diagnosis was reached by FNAB in 68.1% of lesions, by core biopsy in 13.7% of lesions, and by open biopsy in 18.1% of lesions.
High-grade microcalcifications constitute an important class of mammographic abnormalities, because they often are associated with DCIS. Given these patients' heightened risk of developing invasive cancer, the detection of DCIS and its appropriate management represent ideal points for medical intervention.
In our setting, imaging assessment has a high positive predictive value (PPV) for malignancy. The PPV for Category 5 microcalcifications was 92.31%. Although this figure represents a very high rate of reliability of radiologic assessment, it also reinforces the need for morphologic assessment of all such lesions prior to definitive therapy, because approximately 1 in 10 lesions with highly suspicious microcalcifications, in fact, proved to be benign.
A positive cytologic diagnosis of malignancy was achieved in 70.93% of high-grade microcalcifications assessed by FNAB without any false-positive diagnoses. This rate of positive cytologic diagnosis is similar to the 73.12% rate documented for the entire group of Category 5 lesions in our program (data not presented). The distribution of the microcalcifications, whether as single clusters, multiple clusters, or dispersed microcalcifications, made little difference in the likelihood of obtaining a positive cytologic diagnosis. This is noteworthy, because it may have been expected that dispersed microcalcifications would produce a greater proportion of nonrepresentative samples. However, previous studies have shown that high-grade DCIS most commonly is a unicentric process,19 and our observations suggest that, even though the microcalcifications may be dispersed in these lesions, the intervening regions are likely to be involved by DCIS, increasing the chances of obtaining a diagnostic sample.
Some authors have expressed concern about proceeding to definitive breast surgery on the basis of positive cytology alone, without confirmation by core biopsy or frozen section examination.20 If all positive cytologic diagnoses had to be confirmed by core biopsy, then the value of FNAB in the assessment of Category 5 lesions would be abolished, because they would require core biopsy whatever the results of FNAB. In our program, a positive cytologic diagnosis of malignancy in Category 5 microcalcifications is an indication for definitive breast surgery. In the current series, all 122 lesions that were designated malignant by FNAB proved to be malignant. The 100% PPV of a cytologic diagnosis of malignancy in Category 5 microcalcifications supports this approach in our setting.
Reasons for Nonmalignant Cytology—The Sensitivity of FNA Cytology
A positive cytologic diagnosis of malignancy was not made in 29.07% of Category 5 microcalcifications assessed by FNAB. An important reason for this outcome was benign lesions that presented as highly suspicious microcalcifications. Nonrepresentative samples and indefinite cytology accounted for the remaining lesions without positive cytology. Because definitive cancer surgery results from a positive cytologic diagnosis in our program, caution and adherence to strict cytologic criteria for malignancy are of paramount importance. Although this may result in the classification of a greater proportion of lesions as atypical or suspicious rather than positive, we believe that the absence of false-positive cancer diagnoses among our patients justified this cautious approach. It is in the further assessment of the group of lesions with highly suspicious imaging characteristics but without a positive cytologic diagnosis that we found core biopsy particularly useful.
The Place of Core Biopsy in the Preoperative Diagnosis of High-Grade Microcalcifications
In the current series, among lesions that were assessed by both FNAB and core biopsy, core biopsy was useful in planning management in 24 of 48 lesions (50%), either by demonstrating invasion when FNA results already were positive (5 of 10 lesions), by providing a malignant diagnosis when FNA results were not positive (15 lesions), or by averting the need for open biopsy of a benign lesion (4 lesions).
It would be ideal if all lesions with an invasive component could be identified preoperatively, because axillary surgery could be performed at the same session as the surgery for the primary lesion. However, even with the use of core biopsy, when the dominant radiologic category was high-grade microcalcifications, in this series, invasion was demonstrated in only 50% of lesions that were selected for further assessment by core biopsy. This is because stromal invasion may be a rather focal feature in DCIS. Although core biopsy of high-grade microcalcifications frequently confirms the diagnosis of DCIS, it will be found on final histology that between 10% and 20% of these lesions have an invasive component that was not included in the core samples.21, 22 In these lesions, apart from the presence of a mass on imaging, few characteristics of the lesion on core biopsy are predictive of invasion.
Core biopsy was most helpful where the FNA results were not positive. In 76% of these lesions, it provided a definite diagnosis that accounted for the radiology, obviating the need for an open biopsy. This has advantages in terms of patient morbidity and cost savings. The increased use of core biopsy in our program has coincided with a diminishing need for open biopsy. In the final year of this series, 45.5% of all category 5 microcalcifications underwent core biopsy, whereas open biopsy was performed in less than 5% of lesions. Because little else has changed in the design of our program during this period, we attribute this trend to more precise preoperative diagnosis made possible, in turn, through the appropriate use of core biopsy.
In a small number of lesions with Category 5 microcalcifications, the cytology was not positive and core biopsy demonstrated a benign lesion that accounted for the radiologic findings. Open biopsy then was not performed. In these instances, it is important to make the decision to forgo open biopsy only after establishing the existence of a close correlation between the radiologic and morphologic features of the lesion to ensure representative cores. Mammographic surveillance should continue.
Optimal Strategies for the Assessment of Category 5 Microcalcifications
Histologic examination of core biopsies of breast lesions is highly reliable, with few reports of false-positive diagnoses. Furthermore, whereas cytology is a highly operator dependent skill,23 there is a larger pool of well-trained surgical pathologists available to interpret core biopsy samples of breast tissue. Consequently, many centers are using core biopsy as the first-line diagnostic modality for the assessment of impalpable breast lesions. The choice of method of assessment for impalpable breast lesions should not be between FNAB or core biopsy; rather, where access to both techniques is available, for each category of mammographic abnormalities, an optimal strategy for reaching a reliable diagnosis should be determined. We have demonstrated that FNAB is particularly suitable for the investigation of highly suspicious microcalcifications, because high-grade DCIS is the most likely underlying process in these lesions, and the extreme nuclear pleomorphism that characterizes this disease contrasts with the cytologic features of many alternative causes for microcalcifications. We note that others even have reported superior performance of FNAB over core biopsy in the assessment of microcalcifications.24
In our program, experienced breast cytopathologists could diagnose > 70% of these lesions as malignant, without any false-positive diagnoses, substantially reducing the rate of core biopsy. This high rate of positive cytologic diagnosis has significant economic implications, because FNAB is far less costly than core biopsy. Considering consumables alone, the cost of FNAB has been estimated at $6.50 compared with $25.00 for stereotactic core biopsy and $285.00 for Mammotome biopsy.25 That group further estimated that the Medicare reimbursement fee was $311.62 for FNAB, $653.05 for automated core biopsy, and $978.05 for vacuum-assisted core biopsy. Such economic considerations are highly relevant in the design of population-based screening programs that assess large numbers of clients. Our results suggest that where adequate cytologic expertise is available, FNAB can be used as a rational and cost-effective system of triage for selecting patients who would benefit the most from investigation by the more expensive modality of core biopsy.
The combined use of cytology and core biopsy also has been advocated by others as a means of increasing the diagnostic yield and accuracy of assessment of impalpable breast lesions.26, 27 However, rather than using the two techniques in parallel, as suggested by those authors, our program employs them sequentially, and only as required, thereby rationalizing the use of core biopsies. For programs largely reliant on core biopsy, the combined use of FNAB and core biopsy is advantageous in cases of nonrepresentative or false-negative core biopsy results. This is estimated to be in the region of 1.2–20.5% in different series28–32 and was illustrated by finding unsuccessful core biopsy or inconclusive results in 13% of lesions in which core biopsy was performed after a positive cytologic diagnosis to assess the presence of stromal invasion. Had the lesion been assessed only by means of core biopsy, a definitive diagnosis of malignancy would not have been available in these tumors, necessitating recall of the patient for further investigation.
Both FNAB and core biopsy are valuable in the preoperative diagnosis of impalpable breast lesions, and although FNAB is far less costly, either technique is preferable to open biopsy in terms of morbidity and cost. Where there is access to experienced breast cytopathologists, the sequential use of the two techniques serves as a reliable and cost-effective means of triage for selecting lesions that would benefit most from assessment by core biopsy. Core biopsy supplements FNAB and improves the rate of preoperative diagnosis of highly suspicious mammographic microcalcifications.
The authors acknowledge the expertise and cooperation of all the radiologists, pathologists, and surgeons who have been associated with BreastScreen South Australia (Wayville, Australia). The authors also thank Ms. Jill Rogers, Ms. Prue Playford, and Ms. Ann Dunn for their assistance with the collection of the data and the anonymous reviewers whose comments have improved the presentation of this study.