Cytologic features of ductal carcinoma in situ in fine-needle aspiration of the breast mirror the histopathologic growth pattern heterogeneity and grading

Authors


Abstract

BACKGROUND

Approximately 20% of the breast carcinoma cases detected on mammography screening represent ductal carcinoma in situ (DCIS). Cytopathologists are exposed to cytologic material from DCIS when nonpalpable, mammographic lesions are aspirated during the workup of organized and opportunistic mammography screening.

METHODS

The material in the current study was comprised of 225 representative fine-needle aspiration cytology (FNAC) smears from histologically confirmed DCIS of the breast that were diagnosed between 1990–2003. Smears were rescreened to search for the following features: nuclear size (grading), monolayer sheets, solid and cribriform epithelial aggregates, micropapillary and true papillary structures, comedo-type necrosis, microcalcifications, myoepithelial cells, and discohesion.

RESULTS

There were 174 high-grade lesions (77% were Grade 3) and 51 nonhigh-grade lesions (Grades 1 and 2). The concordance between the cytologic and histologic grading was 97% in the Grade 3 lesions and 94% in the Grade 1/2 lesions. Smears from Grade 3 DCIS contained solid and/or cribriform epithelial aggregates in > 93%% of the cases, whereas smears from Grade 1/2 lesions were found to contain cribriform aggregates in 94% of the cases. Pure subtypes were virtually nonexistent. Monolayer sheets were found in 49% of nonhigh-grade DCIS and in 16% of high-grade DCIS. Myoepithelial cells were demonstrated in 51% of the nonhigh-grade DCIS lesions and 27% of the Grade 3 lesions. Microcalcifications were found on the smears from 96% of nonhigh-grade lesions and 84% of high-grade lesions. Approximately 47% of high-grade DCIS and 31% of nonhigh-grade DCIS were found to harbor a distinct single cell population.

CONCLUSIONS

The findings on FNAC from DCIS of the breast completely mirror the histologic heterogeneity of growth pattern subtypes. Primary cytologic grading can effectively separate the high-grade lesions from the nonhigh-grade lesions. Cancer (Cancer Cytopathol) 2005. © 2004 American Cancer Society.

Breast carcinoma is the most common malignancy in women worldwide. In Norway approximately 2500 new cases are diagnosed each year and 800 patients die of the disease.1 The treatment of breast carcinoma is centralized to optimize management. Ullevaal University Hospital is responsible for the diagnosis and treatment of breast carcinoma patients in Oslo and handles > 500 new cases each year.

In both organized and opportunistic mammography screening, ductal carcinoma in situ (DCIS) cases comprise approximately 20% of histologically confirmed carcinomas.1 There are several classification systems for DCIS based on histology.2 The majority use both nuclear atypia and growth patterns as well as the presence of necrosis as the basis for grading. In Norway, we use the Van Nuys3 classification system.

Organized mammography screening began in Oslo in 1996, but opportunistic screening provided cytologic material from mammographic microcalcifications several years earlier. Knowledge of the cytologic criteria of DCIS on fine-needle aspiration cytology (FNAC) has evolved gradually from our own experience as well as early reports in the literature.4–6

A number of articles regarding the cytologic features and diagnosis of DCIS have been published since 1992.4, 5, 7–15 The specificity of the cytologic criteria for high-grade DCIS is high,7, 8, 10, 11, 16 although the ability to recognize an additional invasive component is limited.7, 10–13, 15, 17 To our knowledge, experience with nonhigh-grade DCIS is limited5, 8, 18 and distinguishing this entity from proliferative lesions with or without atypia can be difficult.13, 18 The main focus in the descriptions of cytologic DCIS has been on solid and cribriform epithelial cell aggregates as well as nuclear atypia. However, with regard to histology, there are additional growth pattern subtypes such as micropapillary, papillary, and flat/“clinging” as well as signet ring, clear cell, and apocrine cell types.19 The first objective of the current study was to characterize the entire spectrum of growth pattern subtypes on FNAC smears from patients with histologically verified DCIS.

The grading of DCIS of the breast is an important prognostic marker. Surgical treatment may differ in patients with high-grade DCIS compared with those with nonhigh-grade DCIS20 and a preoperative, cytologic diagnosis of DCIS should include grading. Earlier studies have described cytologic features according to histologic grading,2, 16, 21 but the primary grading of DCIS based on cytologic material has not been reported to our knowledge. At Ullevaal University Hospital, we have adopted a slightly modified Van Nuys grading system (Table 1) for the primary grading of DCIS on FNAC smears. The second objective of the current study was to correlate the results of cytologic grading with the results of histologic grading.

Table 1. Van Nuys DCIS Grading, Modified for FNAC
  1. DCIS: ductal carcinoma in situ; FNAC: fine-needle aspiration cytology.

• High-grade DCIS—Grade 3, regardless of growth pattern as well as whether comedo-type necrosis was present (> 2 × the diameter of an erythrocyte)
• Nonhigh-grade DCIS (≤ 2 × the diameter of an erythrocyte) focal comedo-type necrosis—Grade 2

MATERIALS AND METHODS

The material was comprised of 225 representative FNAC smears from patients with histologically confirmed DCIS of the breast. The cases were retrieved from the files of the Department of Pathology, and the patients had been diagnosed between 1990–2003. The cytologic diagnoses had been classified as “carcinoma cells,” “suspicious,” “DCIS high nuclear grade, cannot evaluate invasiveness,” “papillary carcinoma in situ, cannot evaluate invasiveness,” or as “suspicious for ADH (atypical ductal hyperplasia)/DCIS non-high nuclear grade.”10 Between 1994–1995, a suggested or definitive diagnosis of DCIS was increasingly used. The vast majority of cases were nonpalpable, mammographically detected lesions that had been aspirated using radiologic guidance. A few cases had represented palpable lesions and had been aspirated directly.

All the smears were reviewed and evaluated according to the features listed in Tables 1 and 2. Reevaluation of the histologic grading22, 23 was performed independently of the cytologic evaluation.

Table 2. Features Evaluated on FNAC Smears from DCIS
  1. FNAC: fine-needle aspiration cytology; DCIS: ductal carcinoma in situ; NOS: not otherwise specified.

Grading
Microcalcification
Comedo-type necrosis
Debris
Myoepithelial cell nuclei
Small cell groups, NOS
Discohesion÷
 Mainly cohesive groups
 Single cells and groups
 Mainly single cells
Growth pattern÷
 Monolayer sheets (includes flat/“clinging” cell groups)
 Three-dimensional epithelial aggregates
  Solid
  Cribriform
 Micropapillary
 Papillary

RESULTS

The size of the DCIS lesions ranged from 1–90 mm with a mean diameter of 21.7 mm. The largest lesion was multifocal with 4 mammographically detected distinct foci; the total extension of Grade 3 DCIS into the breast was 90 mm. Foci of suspicious microinvasions were not found in a total of 12 representative sections. The lesions were not palpable and were aspirated using radiologic guidance. The smallest lesion represented a group of microcalcifications detected on mammography that had a total extension of 7 mm. However, only a minor focus of Grade 3 DCIS measuring 1 mm was found in the histologic specimens. There were no more microcalcifications detected in the remainder of the biopsy, a repeat biopsy, and on the control mammogram after the repeat biopsy was performed.

There were 174 high-grade lesions (77% = Grade 3) (mean diameter of 23.1 mm) and 51 nonhigh-grade lesions (Grades 1 and 2) (mean diameter of 20.1 mm) (Table 3). The concordance between cytologic and histologic grading was 97% in the Grade 3 lesions and 94% in the Grade 1/2 lesions. All results are shown in Tables 3–6. Eight cases were upgraded or downgraded based on histology. All lesions were heterogeneous with varying nuclear atypia. The discordance between cytology and histology was the result of suboptimal sampling of these heterogeneous lesions.

Table 3. Correlation between Cytologic and Histologic DCIS Grading according to the Van Nuys System
Histologic gradingCytologic gradingNonhigh-grade (Grades 1 and 2)High-grade (Grade 3)Subtotal
  1. DCIS: ductal carcinoma in situ.

1 29 (57%) 29 (13%)
2 19 (37%)5 (3%)24 (11%)
3 3 (6%)169 (97%)172 (76%)
Subtotal 51174225
Table 4. Growth Patterns and Other Features in High-Grade and Nonhigh-Grade DCIS in breast FNAC
Cytologic criteriaCytologic gradingNonhigh-grade (n = 51)High-grade (Grade 3) (n = 174)
  1. DCIS: ductal carcinoma in situ; FNAC: fine-needle aspiration cytology.

Comedo-type necrosis 8 (16%)106 (61%)
Microcalcification 46 (90%)147 (84%)
Myoepithelial cell nuclei 26 (51%)47 (27%)
Monolayer sheets 25 (49%)27 (16%)
Solid aggregates 30 (59%)161 (93%)
Cribriform aggregates 48 (94%)133 (76%)
Micropapillary groups 29 (57%)34 (20%)
True papillary structures 9 (18%)3 (2%)
Table 5. Growth Patterns in FNAC from Papillary DCIS of the Breast (n = 12)
Growth patterns (in addition to papillary)Cytologic gradeNonhigh-grade (n = 9)High-grade (Grade 3) (n = 3)
  1. FNAC: fine-needle aspiration cytology; DCIS: ductal carcinoma in situ.

Monolayer sheets 8 (89%)2 (67%)
Solid aggregates 7 (78%)2 (67%)
Cribriform aggregates 6 (67%)3 (100%)
Micropapillary groups 8 (89%)2 (67%)
Table 6. Cell Cohesiveness in FNAC from High-Grade and Nonhigh-Grade DCIS of the Breast
Cohesion gradeCytologic gradeNonhigh-gradeHigh-grade (Grade 3)
  1. FNAC: fine-needle aspiration cytology; DCIS: ductal carcinoma in situ.

1 (mainly groups) 33 (65%)89 (51%)
2 (groups and single cells) 16 (31%)81 (47%)
3 (mainly single cells) 2 (4%)4 (2%)

Virtually all cases demonstrated a mixture of morphologic growth pattern subtypes. Table 5 shows the findings in the papillary carcinoma in situ group as an example of the growth pattern heterogeneity. Nevertheless, there were some clear trends. Smears from Grade 3 DCIS were found to contain solid and/or cribriform, 3-dimensional epithelial aggregates in > 93% of the cases, whereas smears from Grade 1/2 lesions contained cribriform, 3-dimensional aggregates in 94% of the cases. The smears from nonhigh-grade DCIS were found to be the most heterogeneous and demonstrated a significantly mixed growth pattern with cribriform, solid, and micropapillary aggregates (Figs. 1–3) and groups as well as monolayer sheets in approximately half of the cases. Myoepithelial cells were found in 32% of all cases, monolayer sheets were noted in 23% of all cases, and a loss of cell cohesion was observed in 55% of all cases.

Figure 1.

Three-dimensional, cribriform epithelial aggregate from a Grade 1 ductal carcinoma in situ (Giemsa stain, original magnification ×20).

Figure 2.

Three-dimensional solid epithelial aggregate from a Grade 3 ductal carcinoma in situ (Giemsa stain, original magnification ×20).

Figure 3.

Micropapillary cells group from a Grade 1 ductal carcinoma in situ (Giemsa stain, original magnification ×40).

DISCUSSION

The modified Van Nuys grading system demonstrated very good correlation with the histologic grading (Table 3) and appears to be able to distinguish high-grade DCIS from the nonhigh-grade tumors reliably. A few discordant cases occurred because of intratumoral heterogeneity. A preoperative diagnosis of high-grade DCIS10 allows the surgeon to plan definitive surgery. Primary surgery of the breast reportedly depends on the ultrasonographic and/or mammographic extension of the lesions, and not whether they are invasive, microinvasive, or pure high-grade DCIS.20 Since the advent of the sentinel lymph node technique in surgery for breast carcinoma, the distinction between invasive carcinoma and high-grade DCIS is not necessarily critical because removal of the sentinel lymph node(s) might be justified in both groups.20

In histology, there is often a distinct heterogeneity with regard to growth pattern subtypes within the same DCIS lesion. As expected, this heterogeneity is fully reflected in the FNAC smears from DCIS lesions. Even if solid (Fig. 1) and cribriform (Fig. 2) aggregates dominate, the smear patterns are virtually always a mixture of several growth pattern subtypes, all of which should be recognized as part of the spectrum of diagnostic features of DCIS. This is especially important in patients with nonhigh-grade DCIS, in which 50% of cases will demonstrate features of 4 different growth pattern subtypes. Proper recognition of the characteristic features may serve as an aid to diagnosis in the absence of definite nuclear atypia in many nonhigh-grade lesions. A cytologic diagnosis of nonhigh-grade DCIS should lead to a surgical excision biopsy, but no sentinel lymph node exploration.20

We agree with Cangiarella et al.16 that high-grade DCIS is easily identifiable on breast FNAC smears. Epithelial aggregates and single lying, high-grade carcinoma cells, comedo-type necrosis (Fig. 4), and microcalcifications are virtually pathognomonic of high-grade DCIS (Fig. 5) and may be diagnosed as such. In nonpalpable mammographic abnormalities diagnosed as DCIS on FNAC, 18% will demonstrate an additional invasive component on histology.11 Bonzanini et al.7 found no difference between pure DCIS and pT1a tumors. Whether criteria of invasion9, 17, 24 can be identified in the DCIS cases cited herein remains to be elucidated.

Figure 4.

Grade 3 ductal carcinoma in situ with comedo-type necrosis and microcalcifications (Giemsa stain, original magnification ×20).

Figure 5.

Ductal carcinoma in situ with abundant microcalcifications (Giemsa stain, original magnification ×40).

Most studies have regarded myoepithelial cell nuclei as a feature of benign or “non-malignant” lesions.18, 25, 26 However, we identified them on the epithelial cell aggregates in 51% of nonhigh-grade DCIS and 27% of Grade 3 lesions (Fig. 6). The nuclei were usually rather few, but nevertheless they were clearly identifiable on morphology as myoepithelial. Their location was always at the periphery of the epithelial clusters and mainly as single-cell nuclei lying in a plane that appeared to be above the epithelial cells on the smears and not within the epithelial clusters. The latter would have been an indication of an atypical ductal hyperplasia, and not a monoclonal proliferation of atypical cells. This finding is in accordance with McKee et al.9 who found them in 86% of their cases. Cangiarella et al.16 found them in approximately 30% of low-grade and intermediate-grade DCIS, but not in high-grade lesions. All these findings demonstrate that the myoepithelial cell nuclei do not completely disappear in lesions diagnosed as DCIS, but rather that the number decreases gradually, and usually, but not invariably, the nuclei are no longer present in the invasive carcinomas. If we want to be able to diagnose DCIS based on cytologic material, we have to modify our interpretation concerning the presence or absence of myoepithelial cells on epithelial aggregates and monolayer sheets, especially in nonhigh-grade lesions.

Figure 6.

Grade 3 ductal carcinoma in situ with remnants of myoepithelial cells (Giemsa stain, original magnification ×40).

To our knowledge, monolayer sheets have not been identified as a diagnostic feature of DCIS previously. In the current study, they were found in 49% of nonhigh-grade DCIS (Fig. 7) and in 16% of high-grade DCIS (Fig. 8), respectively. They are the equivalent of a flat/“clinging” growth pattern (Figs. 9,10) in histologic specimens. They showed the same nuclear characteristics as the three-dimensional epithelial aggregates and therefore could be identified as part of the tumor cell population on the FNAC smears. Although DCIS is a proliferative lesion, areas of flat, single layers or a pseudostratified cell growth pattern is not uncommon in histologic specimens. The monolayer sheets in the FNAC smears mirror these histologic findings.

Figure 7.

Palisading (flat) Grade 1 ductal carcinoma in situ (Giemsa stain, original magnification ×40). The corresponding histological specimen is shown in Figure 10.

Figure 8.

Monolayer sheet from a Grade 3 ductal carcinoma in situ, corresponding to a flat/“clinging” growth pattern subtype (Giemsa stain, original magnification ×40). The corresponding histological specimen in is shown in Figure 9.

Figure 9.

Flat/“clinging” growth pattern of a Grade 3 ductal carcinoma in situ. The same case is shown in Figure 8 (H & E, original magnification ×20).

Figure 10.

Flat/“clinging” growth pattern in a Grade 1 ductal carcinoma in situ. The same case is shown in Figure 7 (H & E, original magnification ×20).

DCIS with a component of true papillary growth pattern comprised a subgroup of cases (12 of 225 cases [5%]). The growth pattern was always mixed (Table 5). The majority of papillary DCIS cases were nonhigh-grade lesions. However, high-grade papillary DCIS may occur (1.3% of cases). A number of lesions, both benign and malignant, may mimic papillary-type epithelial aggregates.27 Fragments from a fibrovascular core were demonstrated in all the cases in the current study and were paramount in identifying a true papillary growth pattern. This finding also represents an important differential diagnostic feature compared with a micropapillary, invasive carcinoma, which is always a high-grade lesion (Grade 3).

The radiologic features of DCIS almost always were found to be comprised of microcalcifications and were almost always identifiable on the FNAC smears (96% of nonhigh-grade lesions and 84% of high-grade lesions, respectively). They often were abundant and in general easy to identify. Papillary (i.e., intracystic) DCIS, however, is an exception to the above. Radiologically, these tumors often present as round or lobulated tumors. However, a radiologically probable DCIS with mammographically detected microcalcifications may well harbor one or more true papillary lesions within a “classic” DCIS and the cytologic findings may reflect this.

Cell cohesiveness in general is reduced in malignant lesions. Dissociation and the demonstration of single, well preserved, atypical cells is one of the features of malignancy. Discohesion of breast carcinoma cells is evident before the lesion becomes invasive. Therefore, in the current study, 47% of high-grade DCIS lesions and 31% of nonhigh-grade DCIS lesions were found to harbor a distinct single cell population in addition to groups, sheets, and aggregates (Table 6). A small fraction of DCIS presented as a mainly single cell population on the smears. Loss of cell cohesiveness and single lying carcinoma cells should not be interpreted as a feature of invasiveness.

The cytologic findings on FNAC from DCIS lesions of the breast appear to completely mirror the histologic heterogeneity of growth pattern subtypes. All these lesions can be easily and reliably identified on the FNAC smears. The modified Van Nuys grading system can effectively separate the high-grade lesions from the nonhigh-grade lesions. Loss of cell cohesion can be found in > 40% of all cases of DCIS; care should be taken not to interpret these cases as invasive lesions.

Ancillary