There is a paucity of data regarding molecular subtypes of pure ductal carcinoma in situ (pDCIS). We evaluated the expression of ER, PR, HER2, Ki67, and p53 and DNA ploidy in 118 pDCIS and 100 invasive breast carcinomas (IBCAs) by routine IHC and classified them according to molecular subtypes. Quantification of biomarkers and DNA ploidy was performed by image analysis. Expression of ER, PR, and high ki67 was more frequent in pDCIS compared to IBCA. High-grade tumors had lower ER and PR expression, high Ki67, overexpression of HER2 and p53, and DNA aneuploidy. Luminal A and HER2 subtypes were more common in pDCIS, and triple negative was more prevalent in IBCA. In both groups, HER2 and triple negative subtypes were characterized by high ki67, overexpression of p53, and DNA aneuploidy compared to luminal subtypes. Molecular subtypes of IBCA are distinct from those of pDCIS. Invasion is characterized by change in phenotype in some tumors.
Ductal carcinoma in situ (DCIS) is a complex disease with diverse clinical presentation, histologic subtypes, and biologic behavior [
Gene expression analysis has demonstrated several molecular subtypes of IBCA; these include the luminal A, luminal B, HER2, and triple negative/basal subtypes [
While the molecular subtypes of IBCA have been well characterized, far less is known about the biologic subtypes of pDCIS. A detailed analysis of biomarker expression in pDCIS has not been well documented, and there are very few studies that have compared biomarker expression and molecular subtypes in pDCIS with IBCA. There is a paucity of data regarding the relationship of the molecular subtypes with tumor grade, Ki67 and p53 expression, and DNA ploidy [
In the present study, we performed a comparative analysis of biomarkers (ER, PR, HER2, Ki67, and p53) expression and DNA ploidy in pDCIS and IBCA by automated image analysis using whole tumor sections and determined if there were quantitative differences in their expression in the two groups. We also attempt to classify pDCIS and IBCA according to molecular subtypes using IHC as proxy for gene expression and to determine if there were any differences in the prevalence of the subtypes in the two groups. We also investigated the relationship of Ki67 index, tumor grade, DNA ploidy, and p53 overexpression with molecular subtypes of pDCIS and IBCA.
This retrospective study consists of 118 consecutive patients with a diagnosis of pDCIS and 100 consecutive cases of IBCA on final surgical excision. Patient demographics, tumor size, and types of surgery were extracted from the electronic medical records of UT Southwestern Medical Center and Parkland Health and Hospital Systems after approval by the Institutional Review Board. All pathology data was obtained from electronic laboratory information system of the Department of Pathology. Grading of DCIS and IBCA was performed using the World Health Organization (WHO) criteria into low (grade 1), intermediate (grade 2), and high (grade 3). Size of DCIS was determined by the extent of the lesion in consecutive sections or by the number of involved sections and gross measurement. The architectural patterns of DCIS were also recorded, and comedo necrosis was evaluated as a separate parameter but was not quantitated. Grading of IBCA was done using the modified Bloom-Richardson Nottingham scoring system.
Biomarker expression and DNA ploidy were prospectively performed as part of the patient’s clinical workup. As a matter of routine, the most representative tumor section was selected for biomarker analysis. In patients with IBCA, paraffin block with predominantly invasive tumor was selected and image analysis was performed only on the invasive tumor, regardless of the presence or absence of in situ component.
Slides were stained on an automated immunostainer (Dako autostainer, Carpentaria, CA). Monoclonal antibodies from the Dako Medical Systems, Carpenteria, CA were used for ER (clone 1D5, 1 : 800), PR (PgR 636, 1 : 1000) and Ki67 (MIB-1, 1 : 300), HER2/neu (1 : 600), and p53 (DO-7, 1 : 2200, Dako, Carpenteria, Calif, USA). Scoring and quantification of ER, PR, Her2/neu, p53, and ki67 was performed on the most representative area of the tumor using the computerized Automated Cellular Imaging System (ACIS, Clarient, Inc. San Juan Capistrano, Calif, USA). The ACIS system consisted of an automated robotic bright-field microscope module, a computer, and a Windows-NT-based software interface. Subregions were selected from the digital images of the IHC-stained slides for analysis. Positive staining in 5% or more of the tumor cells for ER and PR in 10 selected subregions of the tumor was scored as positive for ER and PR expression. The results of ER and PR were reported as percent of positive staining nuclei, and staining intensity was graded from 1+ to 3+. To assess HER-2 overexpression, ACIS provided an average score for 5 selected subregions of the tumor with the highest cytoplasmic membrane staining intensity for HER-2. Tumors with more than 10% cells with an average score ≥2.0 were considered to have HER-2 overexpression, this was equivalent to a 3+ positive staining by manual scoring. Scores between 1.4 and 1.9 were reported as borderline (or 2+ by manual scoring), and <1.4 were reported negative. Computer-generated results were confirmed by manual review by pathologists with experience in image analysis. Tumors with ER and PR scores of less than 5% were considered negative expression. All cases with positive or borderline HER-2 results on IHC were confirmed by fluorescent in situ hybridization (FISH) using the FDA approved PathVysion kit (Abbott-Vysis Lab Abbott Park, Ill, USA) according to the manufacturer’s protocol. Briefly, dual-color FISH was performed with the HER2 probe labeled with spectrum red and chromosome 17 specific centromere (D17Z1) probe labeled with spectrum green on deparaffinized tumor sections cut from the same block. Fluorescent signals in at least 60 nonoverlapping interphase nuclei with intact morphology were scored with ×100 objective, using a fluorescence microscope. Only tumor cells from the area designated on the H&E slide by the pathologist were scored for the number of red (HER-2) and green (chromosome 17) signals. A ratio of the number of fluorescent signals of HER-2 to chromosome 17 greater than 2.0 was reported as HER2 amplified.
The Ki67 index of >10% was considered to be significant. P53 expression in >10% of the tumor cells were considered as overexpression. Bcl-2 immunostaining was performed in the pDCIS cases but not on the invasive group.
DNA analysis was performed by image analysis using Feulgen DNA stains on paraffin sections from the same tumor block that was used for biomarker analysis. DNA indices and ploidy were analyzed using the Autocyte Pathology Workstation (Tripath, Burlington, NC, USA). Briefly, a total of 200–300 nuclei were collected was and mean DNA index reported for each patient. DNA index was obtained by measuring the optical density of tumor cells in comparison with those of the nonneoplastic stromal cells in the sample using the latter as the diploid reference (value of 1.0). Tumors were classified into diploid and aneuploid/multiploid based on the DNA indices.
Pure DCIS and IBCA were classified according to the molecular subtypes. We used the Ki67 score of 14% as the cut-off for distinguishing luminal A (ER+, PR±, HER2−, Ki67 <14%) and luminal B (ER+, PR±, HER2−, Ki67 ≥14%) subtypes; luminal-Her2 (ER+, PR±, HER2+); HER2+ (ER−/PR−/HER2+) and triple negative (ER−/PR−/HER2−). The relationship of the molecular subtypes with tumor grade, Ki67 index, p53, and DNA ploidy was analyzed. Quantitative expression of ER, PR, HER2, Ki67, and p53 by image analysis was correlated with tumor grade in both the pDCIS and IBCA groups. We also compared tumor grade and ploidy with luminal (luminal A, B, lum-HER2) versus nonluminal (HER2 and triple negative) types.
Statistical analysis was performed by one-way ANOVA test followed by Tukey post hoc for comparison of continuous data and Fisher’s exact test and Chi-square for categorical data. The two-way ANOVA with Bonferroni post hoc test was used for comparing Ki67 in the different subtypes of pDCIS and IBCA using the Prism 5 software (Graphpad Software Inc, San Diego, Calif, USA).
The mean age, tumor size, and grade were similar in the two groups (Table
Comparison of clinicopathologic parameters in pDCIS versus IBCA.
Variables | pDCIS ( | IBCA ( | |
---|---|---|---|
Age (mean) | 61.4 ± 1.03 | 61.8 ± 0.87 | NS |
Tumor size (mean) | 2.95 ± 0.23 | 3.37 ± 0.32 | NS |
Tumor grade | |||
Low | 9 (7.6%) | 11 (11%) | NS |
Intermediate | 46 (38.9%) | 49 (49%) | NS |
High | 63 (53.3%) | 40 (40%) | 0.0570 |
*Chi-square test.
NS: not significant.
A comparison of biomarker expression in pDCIS versus IBCA.
Tumor biomarkers | pDCIS | IBCA | |
---|---|---|---|
ER+ | 97 (82.2%) | 66 (66%) | |
ER− | 21 (17.7%) | 34 (34%) | |
PR+ | 81 (68.6%) | 50 (50%) | |
PR− | 37 (31.3%) | 50 (50%) | |
HER2+ (FISH) | 36 (30.5%) | 23 (23%) | |
HER2− | 82 (69.4%) | 74 (74%) | |
Ki67 > 10% | 78 (66.1%) | 84 (84%) | |
Ki67 < 10% | 40 (33.8%) | 16 (16%) | |
P53 >10% | 27 (22.8%) | 26 (26%) | |
P53 <10% | 91 (77.1%) | 74 (74%) | |
Total | 118 | 100 |
*Chi-square test;
The relationship of pDCIS grade with quantitative biomarker expression.
Variables | DCIS GRADE | |||
---|---|---|---|---|
No. of cases | Grade 1 (9) | Grade 2 (46) | Grade 3 (63) | |
ER score (%) | 89.7 ± 8.99 | 80.9 ± 4.78 | 58.6 ± 5.46 | |
PR score (%) | 62.0 ± 14.3 | 53.1 ± 5.42 | 31.9 ± 4.69 | |
Ki67 (%) | 12.7 ± 2.61 | 12.8 ± 1.38 | 25.6 ± 2.54 | |
P53 (%) | 10.0 ± 5.05 | 8.83 ± 3.11 | 21.0 ± 4.03 | |
BCL-2 (%) | 93.8 ± 6.25 | 75.7 ± 5.78 | 44.5 ± 6.78 |
Expression levels of ER and PR decreased significantly with increasing tumor grade (
Tumor DNA content was diploid in 38/55 (69%) of grade 1-2 pDCIS and 47/63 (74.6%) of grade 3 pDCIS were aneuploid (
Relationship of tumor grade with quantitative biomarker expression in IBCA.
Variables | IBCA grade | |||
---|---|---|---|---|
grade | Grade 1 (11) | Grade 2 (49) | Grade 3 (40) | |
ER-score (%) | 84.4 ± 8.80 | 75.3 ± 5.18 | 29.65 ± 6.69 | |
PR-score (%) | 49.1 ± 13.6 | 35.16 ± 5.34 | 14.30 ± 4.71 | |
Ki67 (%) | 14.7 ± 2.40 | 29.71 ± 3.55 | 61.2 ± 4.17 | |
P53 (%) | 11.8 ± 8.23 | 10.39 ± 3.26 | 27.95 ± 6.26 |
In the IBCA group, ER and PR scores decreased significantly with increasing tumor grade (
The prevalence of the molecular subtypes in pDCIS versus IBCA by IHC.
Molecular subtypes | Pure DCIS | IBCA | |
---|---|---|---|
Luminal-A | 44 (37.3%) | 21 (21%) | 0.0113 |
Luminal-B | 31 (26.3%) | 30 (30%) | 0.5490 |
HER2* | 36 (31%) | 23 (23%) | 0.2252 |
Triple negative | 7 (6%) | 26 (26%) | <0.0001 |
Total | 118 | 100 |
*Luminal-HER2 and HER2 subtypes are combined together as one group.
The prevalence of the subtypes differed significantly in the two groups. Luminal A was more common in pDCIS compared to IBCA
The luminal versus nonluminal pDCIS and IBCA showed significant association with tumor grade. In the pDCIS group, luminal tumors were frequently of lower grade (grade 1 and 2), 51/97 (52%), and nonluminal tumors were more likely to be grade 3, 17/21 (80.9%),
The pDCIS molecular subtypes showed significant association with DNA ploidy; luminal tumors were more likely to have diploid DNA, 53/97 (54.6%), and non luminal tumors were frequently aneuploid, 20/21 (95.2%),
A comparison of Ki67 index and p53 overexpression with the different subtypes is shown in Table
A comparison of Ki67 indices and p53 overexpression in the different subtypes of pDCIS and IBCAs.
Tumor type | Luminal-A | Luminal-B | Lum-HER2 | HER2 | Triple negative | |
---|---|---|---|---|---|---|
DCIS | 44 (37.3%) | 31 (26.3%) | 22 (8.6%) | 14 (11.8%) | 7 (6%) | |
Ki67 (%) | 6.98 ± 0.52 | 23.8 ± 2.22 | 23.27 ± 2.72 | 33.57 ± 4.87 | 39.57 ± 13.6 | |
P53 (%) | 7.02 ± 2.89 | 12.3 ± 3.92 | 17.8 ± 6.10 | 24.4 ± 8.53 | 52.3 ± 18.6 | |
IBCA | 21 (21%) | 30 (30%) | 15 (15%) | 8 (8%) | 26 (26%) | |
Ki67 (%) | 7.05 ± 0.75 | 33.7 ± 2.53 | 41.4 ± 6.50 | 58.0 ± 9.46 | 67.5 ± 6.04 | |
P53 (%) | 2.15 ± 0.81 | 6.87 ± 2.12 | 36.3 ± 10.1 | 23.1 ± 15.1 | 31.0 ± 8.04 |
Comparison of mean Ki67 indices in molecular subtypes of pDCIS and IBCA.
subtypes | Ki67 | Ki67 | | |
DCIS | IBCA | |||
LUMINAL-A | 6.98 ± 0.52 ( | 7.05 ± 0.75 ( | ||
LUMINAL-B | 23.8 ± 2.22 ( | 33.76 ± 2.53 ( | ||
LUM-HER2 | 23.27 ± 2.72 ( | 41.67 ± 6.50 ( | ||
HER2 | 33.57 ± 4.87 ( | 58.00 ± 9.46 ( | ||
TRIPLE NEG | 39.57 ± 13.64 ( | 67.52 ± 6.04 ( |
*
In the current study, biomarker expression was performed prospectively on whole tumor sections at the time of the patient’s diagnosis. Previous studies on biomarker expression were performed retrospectively on tissue microarray obtained from archived tumor blocks [
There are very few studies that have compared molecular subtypes of pDCIS and IBCA [
The prevalence of luminal B subtype was similar in both pDCIS and IBCA. Tamimi et al. showed that luminal B subtype was more frequent in pDCIS compared to IBCA [
In the current study, there was a trend towards higher prevalence of HER2-positive pDCIS (31%) compared to IBCA (23%). Several studies have demonstrated significantly higher HER2 overexpression in pDCIS compared to IBCA [
The clinical significance of HER2 overexpression in DCIS is not known at this time. Some suggests that HER2 overexpression in DCIS predicted a more rapid progression to invasive carcinoma [
Although we did not examine the expression of basal markers in the triple negative tumors, the majority of them have been shown to have basal phenotypes [
The lower ER, PR and HER2 expression in some IBCA compared to pDCIS as shown in this study may explain the higher prevalence of triple negative tumors in the former. This may support the theory that triple negative IBCA may be an acquired phenotype that evolved from either a luminal B or HER2 subtype of DCIS [
Ki67 index has been shown to have prognostic significance in breast cancer [
The highly significant association of Ki67 index and molecular subtypes of pDCIS with IBCA has not been shown previously. Proliferation was highest in the triple negative tumors followed by HER2 and luminal B subtypes. Luminal A had the lowest proliferation (<10%). Others have shown only a modest increase in Ki67 index using tissue microarray [
The differences in proliferative activity among the molecular subtypes of invasive carcinoma have been shown in gene microarray studies [
The association between increased proliferation and DCIS progression has been elucidated by Ma et al., where genes involved in cell proliferation and DNA repair were expressed at higher level in high-grade DCIS, which were further elevated in IBCA revealing a link between proliferation, tumor grade, and invasion [
One limitation of this study is that we did not analyze the DCIS component of the IBCA for comparison with pDCIS since this is a retrospective analysis. Several studies have shown no differences in the biomarker expression in the DCIS component of IBCA versus IBCA only [
In conclusion, pDCIS is heterogeneous and like IBCA can be classified into distinct subtypes. Since DCIS is a direct precursor of IBCA, it can be inferred that there are distinct pathways to tumor progression, based on the molecular subtypes. Contrary to previous observation, we have shown that invasion is associated with significant increase in Ki67 index and decrease in ER, PR, and HER2 expression. Although most breast carcinomas maintain their phenotype during tumor progression, in some there is a change in phenotype possibly as a result of clonal evolution. Stromal factors may also play a role in influencing tumor growth and biomarker expression.
Abstracts of this study were presented at the 98th USCAP annual meeting in March, 2009 in Boston, Mass, USA and the 99th USCAP annual meeting in March, 2010 in Washington, DC, USA.