Context.—High–molecular weight cytokeratins, such as cytokeratin 5 (CK5), are helpful to distinguish usual ductal hyperplasia (UDH) from atypical ductal hyperplasia (ADH) or low-grade ductal carcinoma in situ (DCIS). Few studies have looked at combining CK5 with estrogen receptor (ER) to differentiate UDH from ADH.
Objective.—To evaluate the expression pattern of CK5 and ER as single or combined markers to separate UDH from ADH and low-grade DCIS.
Design.—A total of 23 ADH, 10 low-grade DCIS, and 32 UDH whole-tissue slides were stained for ER, CK5, progesterone receptor (PR), and Bcl-2. Nuclear staining of ER and PR was scored as diffuse (>80%), focal (10%–80%), or negative (<10%). Cytoplasmic staining of CK5 and Bcl-2 was scored as diffuse (>60%), focal (10%–60%), or negative (<10%). Differences in staining patterns were evaluated.
Results.—For ER staining: 94% of ADH/DCIS cases showed a diffuse staining pattern, whereas none of the 32 UDH cases showed diffuse staining. For CK5 staining: 96% of ADH/DCIS cases were negative or focally positive, whereas all 32 UDH cases had diffuse staining. The combination of ER and CK5 increased the sensitivity (94% to 97%). For PR staining: 11 of 23 ADH cases (48%), 6 of 10 DCIS cases (60%), and 4 of 32 UDH cases (13%) showed diffuse staining. Bcl-2 staining showed no statistical significance (P = .73).
Conclusions.—Although morphology remains the gold standard, ER and CK5 are useful makers to differentiate UDH from ADH. Progesterone receptor staining may have limited value, and Bcl-2 staining is not useful.
Breast cancer is the most common malignancy of women in the United States.1 The characterization and distinction of benign from atypical epithelial proliferative lesions of the breast are paramount because of the latter's increased relative risk for invasive carcinoma.1–6 Benign epithelial proliferative lesions, such as usual ductal hyperplasia (UDH), have a 1.5- to 2-fold relative risk of breast cancer.1,2 Atypical ductal hyperplasia (ADH) is associated with a 3- to 5-fold increase in relative risk of subsequent development of cancer, compared with women without ADH. Women who receive a diagnosis of low-grade ductal carcinoma in situ (DCIS) carry an 8- to 11-fold increased relative risk of developing invasive carcinoma subsequently.1–9
Distinguishing benign from atypical epithelial proliferative lesions can be challenging on hematoxylin-eosin (H&E) stain sections alone because of high interobserver variability.10,11 Previous studies have looked at the use of high–molecular weight cytokeratins, such as cytokeratin 5/6 (CK5/6), in distinguishing ADH or low-grade DCIS from UDH. Those results showed that UDH usually has a mosaic pattern of CK5/6, whereas ADH and low-grade DCIS tend to be negative or have only scattered positivity.12–15
More recently, the expression of estrogen receptor (ER) has been investigated in benign and atypical epithelial proliferative lesions of the breast.9,16,17 The ER expression in UDH is variable, with scattered positivity and few contiguous areas.9,17,18 In contrast, ADH and low-grade DCIS both tend to show diffuse ER expression with extensive contiguous areas.9,17,18 A limited number of studies, however, have looked at the utility of combining a high–molecular weight CK with ER to differentiate ADH or low-grade DCIS from UDH.
The goal of this study is to assess the utility of CK5 and ER each as a single stain or as combined markers in separating ADH and low-grade DCIS from UDH. In addition, the staining pattern of Bcl-2 and progesterone receptor (PR) expression was assessed. Bcl-2 was chosen because of its expression in columnar cell lesions, and PR was chosen because previous studies have shown that PR tends to have higher expression in ADH and low-grade DCIS than UDH.10,19
MATERIALS AND METHODS
Case Selection and Diagnostic Criteria
Archived pathology files identified 65 cases of atypical and benign epithelial proliferative lesions that consisted of the following: 23 ADH, 10 low-grade DCIS, and 32 UDH. The diagnosis of each case was confirmed and agreed upon by 2 separate pathologists.
Morphology was used as the gold standard for the diagnosis of ADH or low-grade DCIS and UDH. The distinguishing criteria used were based on those proposed by Page et al5 and Tavassoli and Norris.6 Briefly, ADH was diagnosed when there was a monotonous proliferation of low–nuclear grade cells that involved part of either the duct or lobule and had a small size, measuring less than 2 mm. The tumor cells often formed complex and rigid structures with solid or cribriform areas. The nuclei were more evenly spaced out in contrast to the overlapping nuclei of UDH. Low-grade DCIS was diagnosed when the entirety of the lobule or ducts was involved with a monotonous proliferation of low–nuclear grade epithelial cells, and the size of the area measured at least 2 mm. Usual ductal hyperplasia was diagnosed when there was an epithelial proliferation of mixed-cell populations distributed in a haphazard manner. The cells frequently showed streaming, overlapping, and haphazard distribution. No features of ADH or low-grade DCIS were allowed in the diagnosis of UDH.
Immunohistochemical Studies and Evaluation
Immunohistochemical studies were performed on 5-μm formalin-fixed, paraffin-embedded tissue sections using the following antibodies: ER (1:50; Dako, Carpenteria, California), PR (1:400; Dako), Bcl-2 (Bond, Ready to use; Dako), and CK5 (Bond, Ready to use). Primary antibody detection was performed using the Bond Polymer Refine Detection Kit (3,3′-diaminobenzidine chromogen; Leica Microsystems, Bannockburn, Illinois). The detection system avoids the use of streptavidin and biotin, and therefore eliminates nonspecific staining as a result of endogenous biotin. All steps were performed on the Leica Bond Maxx III automated system. Specimens were deparaffinized and underwent antigen retrieval. Slides were incubated with the primary antibody for 15 minutes, with postprimary polymer for 8 minutes, and were blocked with 3% hydrogen peroxide for 5 minutes and 3,3′-diaminobenzidine (brown chromogen) for 10 minutes, followed by hematoxylin counterstain for 5 minutes. These incubations were performed at room temperature. Between incubations, sections were washed with Tris-buffered saline (Bond wash solution). Coverslipping was performed using the Tissue-Tek SCA (Sakura Finetek USA Inc, Torrance, California) coverslipper.
Only nuclear staining for ER and PR was regarded as positive staining and scored as diffuse (>80%), focal (10–80%), or negative (<10%). These cutoff values are different from the American Society of Clinical Oncology/College of American Pathologists (ASCO/CAP) recommendation for ER and PR, which is used for prognostic purpose. Cytoplasmic staining for CK5 and Bcl-2 was regarded as positive staining and scored as diffuse (>60%), focal (10%–60%), or negative (<10%). The immunohistochemical studies were evaluated blinded to the H&E diagnoses.
Statistical Analysis
A student 2-tailed t test was used to examine the differences in staining among the study groups. The sensitivities and specificities were calculated to assess the ability of each stain to correctly identify each study group by immunohistochemistry, using morphology as the gold standard.
RESULTS
ER, CK5, and PR Staining Patterns Were Significantly Different in ADH and Low-Grade DCIS Versus UDH
Immunohistochemical studies with ER, CK5, and PR showed a significant difference in staining pattern between UDH versus ADH and low-grade DCIS (P < .001; Figure). Table 1 shows the different immunostaining patterns in UDH, ADH, and low-grade DCIS cases included in the study.
Comparison of hematoxylin-eosin (A, D, and G), estrogen receptor (B, E, and H), and cytokeratin 5 (C, F, and I) staining in atypical ductal hyperplasia (A through C), low-grade ductal carcinoma in situ (D through F), and usual ductal hyperplasia (G through I). The patchy staining pattern of estrogen receptor in usual ductal hyperplasia (H) is similar to the staining pattern in nonproliferative normal breast tissue with variable staining intensity (original magnification ×200).
Comparison of hematoxylin-eosin (A, D, and G), estrogen receptor (B, E, and H), and cytokeratin 5 (C, F, and I) staining in atypical ductal hyperplasia (A through C), low-grade ductal carcinoma in situ (D through F), and usual ductal hyperplasia (G through I). The patchy staining pattern of estrogen receptor in usual ductal hyperplasia (H) is similar to the staining pattern in nonproliferative normal breast tissue with variable staining intensity (original magnification ×200).
Immunohistochemical Staining Patterns of Estrogen Receptor (ER), Cytokeratin 5 (CK5), Progesterone Receptor (PR), and Bcl-2 in Atypical Ductal Hyperplasia (ADH), Low-Grade Ductal Carcinoma In Situ (DCIS), and Usual Ductal Hyperplasia (UDH)a
For ER staining: 21 of 23 ADH cases (91%) and 10 of 10 DCIS cases (100%) showed diffuse staining, whereas none of the 32 UDH cases showed diffuse staining. The different staining pattern between ADH or low-grade DCIS and UDH is statistically significant (P < .001). The sensitivity of diffuse ER staining in ADH or low-grade DCIS is 94%, and the specificity is 100%. Both of the ADH cases with focal ER staining showed negative CK5 staining (Tables 1 and 2).
Combined Staining Pattern of Estrogen Receptor (ER) and Cytokeratin 5 (CK5) in Atypical Ductal Hyperplasia (ADH), Low-Grade Ductal Carcinoma In Situ (DCIS), and Usual Ductal Hyperplasia (UDH)
For CK5 staining: 22 of 23 ADH cases (96%) and 7 of 10 low-grade DCIS cases (70%) were negative for CK5 staining. The other 3 DCIS cases showed focal CK5 staining. Overall, 22 of 23 ADH cases and 10 of 10 DCIS cases showed negative or focal staining for CK5. All 32 UDH cases had diffuse CK5 staining. The sensitivity and specificity of diffuse CK5 staining in UDH are 100% and 97%, respectively. Of the 23 ADH cases, 1 case showed diffuse staining with CK5. This case also showed diffuse ER staining. Three low-grade DCIS cases showed focal ER staining (Table 1).
When comparing the combination of ER and CK5 staining versus ER stain alone, the sensitivity of identifying low-grade DCIS or ADH increased from 94% to 97%. The expression patterns of either CK5 or ER between ADH and low-grade DCIS were not significant (P = .52 and .35, respectively; Table 2).
For PR staining: 11 of 23 ADH cases (48%), 6 of 10 DCIS cases (60%), and 4 of 32 UDH cases (13%) showed diffuse staining. A total of 9 of 23 ADH cases (39%), 2 of 10 DCIS cases (20%), and 25 of 32 UDH cases (78%) showed focal staining. A total of 3 of 23 ADH cases (13%), 2 of 10 DCIS cases (20%), and 3 of 32 UDH cases (9%) were negative for PR (Table 2).
Bcl-2 Staining in UDH Is Not Significantly Different From That in ADH or Low-Grade DCIS
In evaluation of Bcl-2 staining, there was no statistical significance in staining pattern between UDH versus ADH and DCIS (P = .73). A total of 13 of 23 ADH cases (57%), 4 of 10 DCIS cases (40%), and 13 of 32 UDH cases (41%) showed diffuse staining for Bcl-2. A total of 10 of 23 ADH cases (43%), 6 of 10 DCIS cases (60%), and 18 of 32 UDH cases (56%) stained focally for Bcl-2. In addition, 1 UDH case was negative for Bcl-2 (Table 2).
COMMENT
Usual ductal hyperplasia is a benign lesion thought to derive from a proliferation of luminal cells with stem cell properties, whereas ADH and low-grade DCIS are thought to derive from more differentiated luminal cells.20,21 Both ADH and low-grade DCIS are characterized by a monotonous proliferation of low-grade tumor cells. Although most ADH or low-grade DCIS can be diagnosed with confidence, some lesions show overlapping features with UDH. Interobserver variability in distinguishing benign from atypical proliferative lesions can be problematic and has a significant impact on patient management.10,11 Immunohistochemical studies may help in such situations.
Cytokeratin 5/6 has been shown to be present in the luminal epithelial proliferating cells of usual ductal hyperplasia in 88% to 100% of cases.12,13,15 Conversely, the luminal cells of ADH and low-grade DCIS tend to be negative or show only scattered positivity.12–15 Our results also support these observations. Estrogen receptor expression in ADH and low-grade DCIS tends to be diffuse, whereas ER expression tends to be variable in UDH.9,17,18,22 By combining the use of CK5 with ER, the sensitivity of identifying low-grade atypical epithelial proliferative lesions increased to 97%.
Estrogen receptor staining showed a strong, diffuse staining pattern with many contiguous areas in ADH and low-grade DCIS. Conversely, ER was patchy and without contiguous areas in UDH. The patchy staining pattern in UDH is similar to the staining pattern in nonproliferative normal breast tissue with variable staining intensity. One advantage of ER staining is that ER is a nuclear stain and thus easier to evaluate than the cytoplasmic staining of CK5. Consistent with other studies, our results showed CK5 staining was absent or only focally positive in 97% of ADH and all low-grade DCIS cases. Additionally, CK5 staining exhibited a strong, mosaic pattern in all UDH cases (Figure and Tables 1 and 2).
One ADH case showed diffuse staining with ER and CK5. This is unusual because high–molecular weight CK positivity has been previously reported in ADH, where it tends to represent less than 10% of the luminal cells, and therefore was interpreted as negative.12,13,16 This case was independently reviewed by 2 pathologists with concurring diagnoses of ADH.
Our results suggest that PR may have a role in distinguishing atypical proliferative lesions from UDH. Previous studies have shown that PR expression tends to be higher in ADH and low-grade DCIS than in UDH.10,19 Our results show a similar trend; however, these results must be interpreted with caution, because some ADH cases can be negative for PR and some UDH cases can be diffusely positive for PR staining.
Bcl-2, when used alone, does not seem to have any value in distinguishing between ADH and low-grade DCIS from UDH.
In summary, although morphology remains the gold standard for the diagnosis of UDH, ADH, and low-grade DCIS, the combined use of ER with CK5 increases the sensitivity of distinguishing ADH and low-grade DCIS from UDH compared with either stain alone. Progesterone receptor staining may play a role in the differential diagnosis, but its sensitivity is low and there is overlapping staining pattern. Bcl-2 does not appear to be helpful.