Fine-needle aspiration cytology plays an important role in the preoperative diagnosis of palpable masses as well as impalpable lesions that can only be sampled by stereotactic or ultrasound techniques. A further refinement of cytologic diagnosis would be the ability to distinguish among the different types of ductal carcinoma in situ (DCIS) also between in situ and invasive malignant disease.
Sixty-six cases of histologically proven, pure DCIS (39 high grade, 12 cribriform, and 15 low/intermediate grade) with a preoperative cytology report of carcinoma were retrieved from our files. All the cytology (wet-fixed and air-dried smears) was reviewed by G.M., and the histology sections were reviewed by G.T. Seven cytologic features, including cellularity, cell dissociation, nuclear size, cell uniformity, nucleoli, nuclear margins, and chromatin pattern, were assigned scores from 1 to 3. The presence of calcium, necrosis, and foamy macrophages was recorded. Cell clusters were examined for evidence of a cribriform pattern. Fat and stromal fragments were closely checked for infiltration by tumor cells.
The cell type was predominantly large and pleomorphic in high grade DCIS, whereas it was mainly small and well differentiated in the cribriform and low grade types. Calcium and necrosis were seen in most high grade lesions and less frequently in the cribriform and low grade ones. Macrophages were more common in high grade and cribriform DCIS than in low/intermediate grade DCIS. Cribriform spaces were noted only in cribriform DCIS. Stromal and fat infiltration by tumor cells was not present in any of the aspirates. Myoepithelial cells were rarely seen.
Fine-needle aspiration cytology is a well-accepted diagnostic procedure used in the investigation of palpable1 and impalpable2 breast lesions, the latter being performed under stereotactic or ultrasound-guided control. The role of cytology has expanded from distinguishing between benign and malignant lesions to preoperative typing3 and grading4 of breast carcinomas. With the introduction of stereotactic techniques, smaller and earlier-stage breast neoplasms are being sampled.5 The incidence of ductal carcinoma in situ (DCIS) has increased from about 2% to over 30%.6–10 Treatment options are available for women with DCIS, and the provision of this diagnosis prior to surgery would aid patient management decisions.
The classification of DCIS was traditionally based on architecture and included comedo, micropapillary, cribriform, and clinging types.11 New classifications have been proposed, which subdivide DCIS into high, intermediate, and low grade categories depending on the nuclear features.12–15 These differ from previous classifications, which did not take into account the nuclear pleomorphism. The importance of classifying DCIS by grade relates to the closer association of high grade (comedo) DCIS with invasive disease,16 and it has prognostic implications.17 The prognosis is also linked to the proportion of DCIS to invasive tumor, if present; the higher the proportion of DCIS, the more favorable the prognosis.18
The aim of this preliminary study was to ascertain whether there are features that would raise the possibility of DCIS and predict the grade on cytologic material, in view of its importance in patient management. Previous reports have indicated that distinguishing between DCIS and invasive carcinoma is not possible on cytology,19, 20 but a more recent study concluded that fine-needle aspiration cytology has a place in this diagnostic process (with limitations) and defines the authors' criteria for the diagnosis of DCIS.21 We selected only cases of pure DCIS of a single grade (high, low, or intermediate) for review in an effort to establish the diagnostic features of each grade. As cribriform DCIS may be of high or low grade,14 it was placed in a separate category. There was only one case of intermediate grade DCIS, so this was assimilated into the low grade group.
MATERIALS AND METHODS
The Department of Cytopathology at the Royal Surrey County Hospital in Guilford, United Kingdom, reports approximately 2500 breast cytology specimens annually. Approximately half of these are derived from the Jarvis Breast Screening and Assessment Center, and the rest from the surgical clinic. The cytologic material includes fine-needle aspirates, cyst fluids, nipple scrapings, and nipple discharge smears. The aspirates are performed by radiologists and clinical breast specialists at the Screening Center and by surgeons at the Breast Clinic. Not all of the subsequent histology is reported at the Royal Surrey County Hospital, as the Screening Center deals with women from five districts.
The histology files from 1992– 1996 were searched for cases of histologically proven DCIS without an invasive component for which there were preoperative malignant cytology reports. Sixty-six cases were retrieved: 39 cases of high grade (comedo) DCIS, 12 cases of cribriform DCIS, and 15 cases of low/intermediate grade DCIS. The cytology and histology slides were all reviewed. The ages and modes of presentation were noted, whether they were clinical masses, screen-detected high risk calcification, or nipple discharge specimens. The cytologic features reviewed included the cellularity of the sample, and the six features routinely assessed in the cytologic grading of breast carcinomas4 (namely, cell dissociation, nuclear size, cell uniformity, nucleoli, nuclear margins, and chromatin pattern). Each of these features was assigned a score of 1 to 3 on a wet-fixed, Papanicolaou-stained smear, which is more reliable than an air-dried smear for assessing subtle nuclear details.
The presence of calcium, necrosis, foamy macrophages, and a cribriform pattern was also noted using both the wet-fixed and the air-dried, May-Grünwald Giemsa–stained smears. The calcium particles in the smears were identified and polarized for birefringence, and the shape was noted (round and laminated or irregular). A careful search was made for apparent infiltration of any stromal or fat fragments by neoplastic cells.
There were no significant differences in the mean ages for the different diagnostic groups. The clinical and cytologic findings are shown in Table 1. Goodness-of-fit tests were carried out to determine the equivalence between groups. There were no statistically significant differences identified, despite the apparent elevation in mammographic calcification in the cribriform group relative to the high grade and low/intermediate DCIS groups. One case in each of the high grade and low grade groups had nipple discharge as the sole presenting symptom. The diagnosis of carcinoma in these two cases had been made based on the nipple discharge smears and were confirmed on fine-needle aspirates following mammography.
Table 1. Clinical and Cytologic Findings for Ductal Carcinoma In Situ
Four of the cytologic features are compared in Table 1. There was no significant difference between the percentage of cases that showed calcium in the aspirate, whether laminated (Fig. 1) or in the form of irregular particles (Fig. 2), and none of the calcium particles were birefringent. However, there was a statistically significant difference among groups in the incidence of cellular necrosis. It was clearly shown in this analysis that high grade DCIS cases are significantly more likely to demonstrate necrosis than those in the other two groups. Foamy macrophages were usually found adjacent to necrotic debris and were commonly noted in high grade and cribriform DCIS, but they were seen to a much lesser extent in low/intermediate grade groups. This difference, however, did not reach statistical significance at the 5% level. Every case of cribriform DCIS demonstrated rounded, clearly defined spaces within cell clusters (Fig. 3), but these were not seen in any of the high grade or low/intermediate grade DCIS aspirates.
None of the 66 aspirates showed infiltration of fat cell clusters or stromal fragments by carcinoma cells. It is essential to differentiate between the appearance of neoplastic cells superimposed on connective tissue or fat fragments due to spreading artefact and true infiltration of tumor cells between fat cells; this is done by following the boundaries of the adipocytes, as illustrated in Figure 4. It is more difficult to assess stromal invasion, as the fragments are usually quite thick. We did not find the presence of single malignant cells to be of particular significance, as they were seen in all our cases of DCIS and indeed were useful in making the diagnosis of malignancy in the low grade tumors.
The scores calculated for the rest of the cytologic features assessed are summarized and shown in Table 2. A nonparametric test of group differences, the Kruskal–Wallis test, was performed for each feature. This test was used in preference to the more typical parametric approach because the metric of the scoring scheme was not felt to support the assumptions of such an approach. Five of the features in question were able to discriminate among the groups to a statistically significant degree. The cribriform group manifested higher cellularity. The high grade DCIS cases showed a significantly higher proportion of carcinoma cells with very large nuclei and marked pleomorphism (Fig. 5) compared with the small, mildly pleomorphic cells seen in the two other groups (Fig. 6). The high grade group also showed significantly greater nucleolar and chromatin abnormality.
Table 2. Summary (Mean Ranks) of Cytologic Feature Ratings
Table 3 shows the group percentages for each feature falling into upper score category (score 3). This demonstrates the pattern equivalence of feature incidence across the three groups. The cellularity feature represents the only departure from the prevailing trend, with its greatest incidence in the cribriform group.
Table 3. Percentages of Each Feature Falling into the Upper Score Category (>2)
The pathology of DCIS has been discussed in detail by Bobrow and Millis22 in a review article emphasizing the prognostic importance of classification. The treatment recommended for invasive carcinoma may not be appropriate for in situ lesions; hence, it would be useful to identify any possible features that distinguish among the various types of DCIS. Cytologic criteria for the diagnosis of DCIS have been studied by various authors. Sneige et al., in an early study of 12 cases of DCIS,20 concluded, as did Wang et al.,19 that aspiration cytology appearances of DCIS are identical to those of invasive ductal carcinoma. However, further work by Shin and Sneige23 identified two indicators of invasion, namely, tubular structures (although these were found in only 24% of cases) and stromal fragments (which are seen in both DCIS and invasive ductal carcinoma). Bondeson and Lindholm24 found it helpful if at least two of the following four significant features were present: tubular structures, cytoplasmic lumina, fibroblast proliferation, and elastoid stroma. Their study emphasized the stromal characteristics and concluded that aspiration cytology can predict invasiveness with a high degree of accuracy in many impalpable breast carcinomas.
The extent to which fine-needle aspiration cytology is utilized preoperatively depends largely on the confidence placed in the cytology report, which in turn relates to the volume of breast aspirates examined and the expertise of the reporting cytopathologists. At our institution, definitive surgery is performed on the basis of a “malignant” cytology report (the positive predictive value being 100%), providing that all three parameters—clinical, mammographic, and cytologic—concur. If there is the slightest doubt about malignancy, the cytology report is classified as “suspicious” rather than “malignant.” The clinicians expect typing and grading of breast carcinomas as part of the routine report. This preliminary study to assess the features of in situ disease was undertaken in an effort to determine whether the service we provide could be improved even more. A second study is being undertaken with a larger, mixed sample of invasive and in situ cases, in which the cytology will be reviewed “blind” to validate our findings.
A constant finding in our study was the absence of tumor cells infiltrating fat and stromal fragments. When this feature is present, it is almost invariably associated with invasive carcinoma in our experience, although this has not been confirmed by other authors.25 However, the absence of these fragments cannot be used as definite evidence of in situ disease. Other features that have been mentioned as needing further study in possibly diagnosing DCIS are clusters of atypical ductal cells with single malignant cells or necrosis, hyperplastic ductal cells with single malignant cells or necrosis, and tissue fragments with a cribriform pattern.26
Previous work has focused on distinguishing among the different types of DCIS. In a study by Malamud et al. of 13 pure DCIS cases that were confirmed on surgical biopsy, there were differences in the cytologic features of individual cells in comedo DCIS and noncomedo DCIS.27 The cytologic diagnosis of low grade cribriform DCIS has been shown to be difficult, with one of three cases studied diagnosed as infiltrating ductal carcinoma and the other two as irregular hyperplasia.28 We found in our study that most of the histologic features of the three types of DCIS could be detected in cytologic preparations, provided careful assessment was made of all the material aspirated. The aspirates from the high grade DCIS cases were all originally graded as Grade 2 or Grade 3 carcinoma, with 23 of the 39 additionally reported as consistent with high grade DCIS. The 12 cases in the cribriform DCIS category had all been reported as low grade carcinoma, and in 9 of these cases cribriform features were mentioned. All the carcinomas in the low/intermediate group were originally diagnosed as low grade or Grade 1 tumors. Two were further defined as being consistent with intracystic papillary carcinoma, and these were confirmed on the surgical specimens.
Information about the type of calcification present on the mammograms was available for the aspirates that originated from the Screening Center. Descriptions of the typical high risk type of calcification associated with comedo DCIS should prompt the cytopathologist to look more closely for calcium, macrophages, and necrotic cellular debris in an effort to confirm the diagnosis. Calcium is also present in aspirates of DCIS lesions that present as palpable masses, but it may be overlooked if the mammogram result is not available to the pathologist. Calcium is a feature that is often not noticed or is misinterpreted as debris or stain artefact on aspirates, especially if the particles are irregular rather than laminated. The calcium may be seen in the background or in the center of cell clusters, the latter being more common in the cribriform and low/intermediate grade varieties. Calcium was not present in the intracystic papillary tumors, but these contained large numbers of foamy macrophages and single tumor cells that were columnar in shape. Myoepithelial cells were very rarely seen in the aspirates and were therefore of no significance in this study.
We conclude from our findings that cases with a mammographic picture of comedo-type calcification, in which the aspirates contain large, pleomorphic carcinoma cells of cytologic Grade 2 or 3 accompanied by necrotic cellular debris, calcium, and foamy macrophages but no stromal infiltration, may be reported as consistent with high grade (comedo) DCIS; however, invasion cannot be definitely ruled out. Aspirates that contain cytologic Grade 1 carcinoma cells in clusters exhibiting a cribriform pattern, unaccompanied by stromal infiltration, are consistent with (though not diagnostic of) cribriform DCIS. Low/intermediate grade DCIS has no features that distinguish it from low grade invasive carcinoma, apart from the lack of stromal infiltration.
Fine-needle aspiration cytology has much to offer in the field of preoperative diagnosis and is in danger of being replaced by core biopsies if it is not utilized to its fullest extent. Because of wider sampling of the lesion, fine-needle aspirates contain most of the information necessary not only for the diagnosis of malignancy but also for grading and, in conjunction with the mammographic findings, for raising the possibility of an in situ lesion. However, it cannot rule out invasion elsewhere in the lesion.
The authors thank Dr. Julie Cooke and Dr. Barbara Thomas, The Jarvis Breast Screening Center, and Mr. Mark Kissin, Consultant Breast Surgeon, The Royal Surrey County Hospital, for providing the cytologic material used in this study.