Cystatin M loss is associated with the losses of estrogen receptor, progesterone receptor, and HER4 in invasive breast cancer

A total of 117 DCIS and 175 IBC patients participated in this study. Pure DCIS cases were included in this study, and DCIS lesions associated with invasive breast cancer were excluded. All specimens were obtained from patients who underwent surgical resection for DCIS and IBC between May 2001 and July 2006 at the Samsung Medical Center in Seoul, Korea. Written informed consent for the use of the surgically resected tumor tissues was provided by all of the DCIS and IBC patients prior to operation. Paired samples of primary breast cancer and matched noncancerous normal tissues were obtained from each patient. Surgically removed tissues were immediately snap-frozen in liquid nitrogen and stored at -80°C until use. All procedures used in this study were approved by the Institutional Review Board at the Samsung Medical Center. Information regarding demographics and lifestyle factors was obtained from an interviewer-administered questionnaire. Patient age ranged from 28 to 72 years for DCISs and from 25 to 83 years for IBCs. The pathologic stage of the IBC was classified according to the American Joint Committee on Cancer (AJCC) TNM criteria (sixth edition).

Tissue microarrays (TMAs) of DCISs and IBCs were prepared as described previously [15]. Briefly, tissue sections of four-micrometer-thickness were taken from the TMA blocks for immunohistochemistry and deparaffinized in xylene and rehydrated through a series of alcohols. The rehydrated sections were treated with 3% hydrogen peroxide in methanol for 10 minutes to block endogenous peroxidase. To unmask the antigens, sections were microwaved in citrate buffer (pH 6) for a total of 15 minutes. Sections were incubated with a primary antibody overnight at room temperature after blocking nonspecific proteins with 2% dried milk in phosphate-buffered saline (PBS) for 30 minutes and 5% goat serum in PBS for 1 h. The antibodies used were ER (clone 6F11, Novocastra, Vision Systems Inc., Norwell, MA, USA, diluted at 1:400), PR (clone 16, Novocastra, diluted at 1:800), C-erb2 (clone CB11, Novocastra, diluted at 1:400), C-erb4 (clone HFR-1, Neomarkers, Fremont, CA, USA, diluted at 1:30), and cystatin M (clone 211515, R&D Systems, Minneapolis, MN, USA, diluted at 1:20). Detection of immunoreactivity by each antibody was performed by the Vectastain Elite ABC reagent (Vector Laboratories, Burlingame, CA, USA) and 3.3'-diaminobenzidine tetrahydrochloride was used as a chromogen. All sections were counterstained with Mayer's hematoxylin, and negative controls were included in each staining sequence and obtained by omitting the primary antibodies. Cytokeratin 8/18 (Invitrogen, Carlsbad, CA, USA; Cat. No. 18-0213) was used as a positive control of staining to verify that tissues were stained. Representative stainings of ER, PR, HER2, HER4 and cystatin M expression in DCISs and IBCs are shown in Figure 1.

Cutoff values for positive or negative expression were determined considering widely accepted criteria among previously reported studies and expression patterns in our normal tissues (data not shown). The positivity of expression was assessed semiquantitatively by evaluating the percentage of positive staining cells and/or staining intensity. The expression of ER and PR was determined according to the Allred score. If the sum of both scores is more than three, the expression of ER and PR was considered positive. HER2 expression was considered as positive when moderate to strong membrane immunoreactivity was observed in more than 10% of tumor cells. The membrane or nuclear staining of HER4 was not considered in the final data analysis due to low frequency (< 5%) of the staining in IBCs. Only cytoplasmic reactivity was considered evidence of HER4 as well as cystatin M expression. An immunoreactive score (IS) of 0 to 7 for HER4 or cystatin M expression was obtained by adding the intensity score (0, none; 1, weak; 2, moderate; 3, strong) and a proportion score of positive staining tumor cells (0, absent; 1, 0 to 10%; 2, 10 to 50%; 3, 50 to 80%; 4, >80%). If the IS was less than two, the tumor was considered to exhibit loss of HER4 or cystatin M expression.

Genomic DNA was extracted from fresh-frozen tissues and formalin-fixed paraffin-embedded tissues using a DNeasy Tissue Kit (Qiagen, Valencia, CA, USA) according to the manufacturer's instructions. The methylation status of the CST6 gene in the formalin-fixed paraffin-embedded tissues was determined using methylation-specific PCR (MSP), as described by Herman et al. [16]. The MSP primers specific for the methylated CST6 gene were 5'-GTTTTTTGAATTTCGTAGGATTTC-3' (sense) and 5'- AACTTTTACCCGCTAAACCG-3' (antisense). The primers for the unmethylated CST6 gene were 5'-GGTTTTTTGAATTTTGTAGGATTTT-3' (sense) and 5'- CAACTTTTACCCACTAAACCACC -3' (antisense). The PCR conditions for DNA amplification consisted of 1 cycle at 95°C for 7 minutes; 35 cycles at 94°C for 30 s, 65°C (for M primer set) or 63°C (for U primer set) for 1 minute, 72°C for 1 minute; and final extension at 72°C for 10 minutes. DNA from the peripheral blood lymphocytes of healthy subjects was used as a negative control for the methylation-specific assays. Lymphocyte DNA from healthy volunteers was treated with Sss1 methyltransferase (New England BioLabs, Beverly, MA, USA), and then with sodium bisulfite. The resulting product was used as a positive control for methylated alleles. Bisulfite-modified DNA from normal lymphocytes was used as a positive control for unmethylated alleles, and the unconverted DNA from normal lymphocytes was used as a negative control for the methylated alleles.

The normality of continuous variables was tested using the Shapiro-Wilk test. The Wilcoxon rank-sum test (or t-test) and Fisher's exact test (or the Chi-squared test) were used for the univariate analysis of the continuous and categorical variables, respectively. A multivariate logistic regression analysis was conducted to determine the relationship between cystatin M loss and any covariates found to be statistically significant in the univariate analysis, and to calculate odds ratios (ORs). Covariates with a P-value of <0.25 in the univariate analysis or any variables that were considered to be biologically important were subjected to the multivariate analysis. All statistical analyses were two-sided with a type I error rate of 5%.

Heterogenous staining, defined in our study as when the staining pattern was different among cells within the same block, is often known to be seen in epigenetically silenced genes within advanced tumors. Heterogeneous staining was found in 4 (3%) of 117 DCISs and in 6 (3%) of 175 IBCs. When a slide showed heterogeneous staining, we grouped cells according to staining intensity and calculated the immunoreactive score (IS) of 0 to 7 for cystatin M expression in each group separately. We then calculated the mean IS in all groups. If the mean score was less than two, the tumor was considered to exhibit loss of cystatin M expression. The association between cystatin M loss and clinicopathological parameters in 117 DCISs and 175 IBCs is shown in Table 1. Cystatin M loss was found in 9 (8%) of the 117 DCISs and in 99 (56%) of 175 IBCs studied (P < 0.0001). The age of patients with DCIS or IBC was not significantly different according to cystatin M expression (P = 0.13 and P = 0.89, respectively). Cystatin M loss was found not to be associated with tumor size in patients with DCIS (P = 0.87) or IBC (P = 0.27). The mean tumor size was 2.9 cm and 2.6 cm for IBC patients with and without cystatin M loss, respectively. No association was found between cystatin M loss and the number of lymph nodes, family history of breast cancer in DCISs and IBCs. Furthermore, the histologic grade was not associated with cystatin M loss in DCISs (P = 0.90) or IBCs (P = 0.32). A statistically significant association was not found between cystatin M loss and pathologic stage in IBCs (P = 0.86).

The prevalence of ER, PR, HER2, or HER4 loss was significantly different between DCISs and IBCs (Table 1). ER loss occurred in 33 (28%) of 117 DCISs and in 74 (42%) of 175 IBCs, and this difference was statistically significant (P = 0.01). The losses of PR and HER2 also occurred at a significantly different prevalence between DCISs and IBCs (P = 0.02 and P = 0.008, respectively). In addition, the prevalence of HER4 loss was 38% (44 of 117) and 26% (45 of 175) in DCISs and IBCs, respectively, and this difference was also statistically significant (P = 0.03). The relationship between cystatin M loss and the loss of ER, PR, HER2, or HER4 was analyzed in DCISs and IBCs (Table 1). Cystatin M loss was found in 1 (3%) of 33 ER-negative DCISs and in 8 (10%) of 84 ER-positive DCISs, and this difference was not statistically significant (P = 0.44). However, the expression status of ER in IBCs was significantly associated with cystatin M loss (P = 0.01). Cystatin M loss occurred in 50 (68%) of 74 ER-negative IBCs and in 49 (49%) of 101 ER-positive IBCs. The relationships between cystatin M loss and PR loss in DCISs and IBCs were similar to those between cystatin M loss and ER loss, but HER2 loss was not significantly associated with cystatin M loss in DCISs and IBCs (P = 0.73 and P = 0.79, respectively). Cystatin M loss was found in 34 (76%) of the 45 HER4-negative IBCs and in 65 (50%) of the 130 HER4-positive IBCs and this difference was statistically significant (P = 0.003).

Triple-negative breast cancers (TNBCs) of ER, PR, and HER2, are known to be typically associated with poor prognosis due to aggressive behavior of the tumor and lack of targeted therapies. Therefore, we analyzed the association of triple-negative status of ER, PR, and HER2 with cystatin M loss. The triple-negative status of ER, PR, and HER2 occurred in 2 (2%) of 117 DCISs and in 35 (20%) of 175 IBCs. However, the triple-negative status was not associated with cystatin M loss in DCISs (P = 1.00) and IBCs (P = 0.08), and therefore the triple-negative status of ER, PR, and HER2 was not considered in the further analysis. Another triple-negative status of ER, PR, and HER4 occurred in 11 (9%) of 117 DCISs and in 23 (13%) of 175 IBCs. For the DCIS cases, cystatin M loss was not associated with the triple-negative status of ER, PR, and HER4 (P = 1.00). However, cystatin M loss in IBCs was significantly associated with the triple-negative status of ER, PR, and HER4 (P = 0.001; Figure 2A): cystatin M loss occurred in 20 (87%) of 23 triple-negative IBCs of ER, PR, and HER4 and in 79 (52%) of 152 other subtypes. To rule out the possibility of a confounding effect of HER2 in the relationship between cystatin M loss and the triple-negative status of ER, PR, and HER4, we stratified the data according to the expression status of HER2 and reanalyzed the relationship according to HER2. However, no confounding effect of HER2 was found in this study: cystatin M loss occurred at a higher prevalence in the triple-negative IBCs than in other subtypes irrespective of HER2 expression (Figure 2B).

In univariate logistic regression analysis (Table 2A), cystatin M loss occurred at 2.20 times (95% confidence interval (CI) = 1.19 to 4.06; P = 0.02) higher prevalence in the ER-negative IBCs than in the ER-positive IBCs. Cystatin M loss also occurred at a high prevalence in the PR-negative IBCs (OR = 2.41, 95% CI = 1.32 to 4.39; P = 0.007) as well as in the HER4-negative IBCs (OR = 3.09, 95% CI = 1.44 to 6.62; P = 0.003) than in the PR- or HER4-positive cases, respectively. A multivariate logistic regression analysis was performed to control for the potential confounding effects of variables, such as age, and to calculate odds ratio (Table 2B). The coefficient for age variable in IBCs was not statistically significant in our univariate analysis (P = 0.89), but age was considered to be biologically important and, therefore, it was included in the multivariate analysis in order to better construct a parsimonious model. The losses of ER, PR, and HER4 were associated with each other (data not shown), and therefore the triple-negative status of ER, PR, and HER4 included in the multivariate logistic regression as a single covariate. Cystatin M loss occurred at a 3.57 times (95% CI = 1.28 to 9.98; P = 0.01) higher prevalence in the triple-negative IBCs than in other subtypes, after adjusting age. These observations suggest that the losses of ER, PR, and HER4 may be significantly associated with cystatin M loss in IBCs.

Epigenetic silencing of CST6 gene in breast cancer has been reported by several groups. To investigate whether the loss of ER, PR, or HER4 is associated with the hypermethylation of CpG island at the promoter region of the CST6 gene, we analyzed the methylation status of the CST6 gene using MSP. The primers designed for MSP yielded an amplicon spanning from nucleotides -108 to -2 (Figure 3A). Primers for MSP were validated using 100% methylated and 100% unmethylated DNA (Figure 3B). The hypermethylation of the CST6 gene by MSP was found in 22 (20%) of 108 DCISs studied and in 59 (34%) of 173 IBCs (Figure 3C). No association was also found between CST6 methylation and cystatin M loss in DCISs (P = 0.38) or IBCs (P = 0.50) (Figure 3D). The lack of association between cystatin M loss and hypermethylation of the CST6 gene may be due to the density dependence of methylated CpGs on transcriptional silencing by CST6 methylation or the inconsistence of methylation status between primer binding sites and other CpG sites within PCR product.

The methylation status of 13 CpG sites (Figure 3A) in the region spanning from nucleotides -108 to -2 were, therefore, reassessed quantitatively using pyro-sequencing in 51 fresh-frozen tissues and matched normal tissues from IBC patients. The quantities of methylated CpGs between normal and tumor tissues were usually different (Figure 4A). The association between the quantity of CST6 methylation and the expression statuses of cystatin M, ER, PR, and HER4 in tumor tissues was analyzed (Figure 4B). The methylation level at each CpG in tumor tissues was adjusted by subtracting the quantity of methylated CpGs in matched normal tissue from that of methylated CpGs in tumor tissue. Average quantities of methylated CpGs in tumor tissues were significantly associated with cystatin M loss (P = 0.008; Wilcoxon rank sum test) or ER loss (P = 0.0002; Wilcoxon rank-sum test). The average quantities of methylated CpGs between groups with and without ER loss were 16.4% and 8.7%, respectively. However, the quantities of methylated CpGs were not associated with the expression status of PR (P = 0.64) or HER4 (P = 0.87).