Introduction

The main aim of ductal carcinoma in situ management is to prevent local recurrence particularly invasive recurrence [1]. Despite a low mortality risk, ductal carcinoma in situ is often treated with mastectomy or breast conservative surgery with frequent re-excisions compared to invasive disease [2]. Mastectomy results in very few recurrences but is often considered as over-treatment for screen-detected precursor lesions. Conversely, breast conservative surgery alone increases the risk of recurrence, which may be considered as under-treatment in a proportion of high-risk ductal carcinoma in situ [3]. Radiotherapy can reduce the recurrence rate by approximately 50% but has significant side effects and its application should therefore be rationalized. Accurate risk stratification of ductal carcinoma in situ means not only predicting recurrence but also which patients are likely to progress to invasive disease.

Ductal carcinoma in situ is observed in more than half of invasive breast cancer [4] and in the majority of these cases, they share morphology, immunoprofile, and genetic features suggesting that the invasive component arises from the associated ductal carcinoma in situ [5,6,7]. However, there remains a lack of objective predictive markers of ductal carcinoma in situ behavior; not only of recurrence but also development of invasive disease that could aid in treatment decisions.

Clinicopathological characteristics are used as prognostic factors in guiding treatment decisions [3] but they are insufficient to reflect the molecular and clinical heterogeneity. Assessment of molecular markers can be used as surrogates of ductal carcinoma in situ biology and behavior and have the potential to predict patient outcome [8]. It is known that cancer cells continuously experience oxidative stress, resulting from heightened reactive oxygen species generation [9]. Enhanced oxidative stress has been implicated in the initiation and progression of cancer, promoting cell survival and drug resistance in some cases [10]. However, unresolved high levels of oxidative stress beyond the capacity of the cancer cell to manage causes critical damage to cellular DNA, proteins, and lipids ultimately leading to cell death [11]. A negative regulator of the major antioxidant thioredoxin (TRX), TRX-interacting protein (TXNIP), has potent growth suppressive, metastasis inhibitory and pro-apoptotic functions [12]. Loss of TXNIP expression was reported to be associated with the development of some solid tumors [12]. In the breast, high expression of TXNIP was reported in normal breast tissues [13] and showed reduced immunointensity in invasive tumors [14]. Ectopic expression of TXNIP in the breast cancer cell line MCF-7 drives cells to undergo senescence, accompanied by increased reactive oxygen species release. In a study of 98 locally advanced invasive breast cancer treated with anthracycline-based chemotherapy, Woolston et al. [15] have demonstrated that TXNIP is an independent predictor of outcome.

We hypothesized that oxidative stress could play a role in breast cancer progression and that TXNIP might be a potential prognostic marker in ductal carcinoma in situ. In this study, TXNIP expression was assessed in a large (n = 1015) and well-characterized cohort of ductal carcinoma in situ from a single institution with long-term follow-up. In addition, the prognostic significance of TXNIP gene copy number and gene expression was assessed in a large cohort of invasive breast cancer (n = 1980) as a molecular surrogate of ductal carcinoma in situ.

Patients and methods

Study cohort

This retrospective study was conducted on a consecutive series of 1059 primary pure ductal carcinoma in situ cases diagnosed, and treated in a uniform manner, between 1990 and 2012 at the Nottingham City Hospital, Nottingham, UK. Exclusion of referral, miscoded, and recurrent cases resulted in 776 pure primary ductal carcinoma in situ with available formalin-fixed paraffin-embedded tumor blocks for tissue microarray construction. A series of 239 cases diagnosed as synchronous ductal carcinoma in situ and invasive tumors (mixed ductal carcinoma in situ and invasive breast cancer) was also collected as a validation set. Patients’ demographic information, histopathological parameters, management, radiotherapy, and patient outcome data were collated. No patients were treated with hormonal therapy. Local recurrence-free survival was calculated based on the time (in months) from the date of primary surgical treatment to the time of ipsilateral local recurrence. The median follow-up period, in pure ductal carcinoma in situ cohort, was 118 months (range 2–240 months), during which 90 patients (14%) developed local ipsilateral recurrence, including invasive (60/90; 67%) or ductal carcinoma in situ (30/90; 33%). Recurrence following breast-conserving surgery and mastectomy occurred in 84 and 6 cases, respectively.

Patients who developed contralateral disease following ductal carcinoma in situ diagnosis were censored at the time of diagnosis of the contralateral cancer.

Tissue microarray and immunohistochemistry

Tissue microarrays were prepared from representative ductal carcinoma in situ lesions of the pure cases and from ductal carcinoma in situ and invasive tumors from the mixed cases. The tissue microarray was constructed using three-dimensional Histech® Grand Master®, whereby cores of 1 and 0.6 mm were taken from ductal carcinoma in situ and invasive tumor samples, respectively. In addition, a set of whole tissue sections from 10 cases containing ductal carcinoma in situ and invasive tumors were assessed to evaluate heterogeneity and the pattern of TXNIP expression in normal and malignant breast lesions.

Validation of TXNIP antibody (Abcam; clone EPR14774) specificity was performed using western blotting on whole-cell lysates of MCF-7, SKBr3, and MDA-MB-231 human breast cancer cell lines (obtained from the American Type Culture Collection; Rockville, MD, USA). TXNIP antibody was used at 1:3000 dilution, which showed a single specific band at the predicted size of 44 kDa (Fig. 1a).

Fig. 1
figure 1

Characterization of TXNIP expression in ductal carcinoma in situ. a Western blot analysis of the anti-TXNIP antibody confirmed a single specific band at 44 kDa in a variety of cell lines, image for MCF-7 shown, b TXNIP cytoplasmic expression in normal breast tissue showing expression in the epithelium and a few scattered myoepithelial cells. Occasional membrane and/or nuclear staining is also seen. TXNIP immunohistochemistry expression in ductal carcinoma in situ showing variable intensities from negative (c), weak (d), moderate (e), and strong (f) staining (×20 magnification). g Full-face section of a ductal carcinoma in situ case mixed with invasive carcinoma and show moderate TXNIP expression in ductal carcinoma in situ but reduced/negative expression in the adjacent invasive carcinoma component

Full size image

Expression of TXNIP protein in ductal carcinoma in situ was assessed by immunohistochemistry using the Novocastra Novolink polymer detection system (Leica, Newcastle, UK). Four-micrometer tissue microarray and full-face sections were stained with mouse monoclonal TXNIP antibody (1:3000) for 30 min. 3,30-Diaminobenzidine tetrahydrochloride (Novolink DAB substrate buffer) was used as a chromogenic substance. Sections were counterstained with hematoxylin. Positive staining controls (human tonsil and kidney tissue) were included while a negative control was achieved by omitting the application of the primary antibody. Expression of estrogen receptor (ER) in ductal carcinoma in situ was carried out using ER clone SP1, Ventana Benchmark® ULTRA system (Tucson, Arizona, USA), as per the recommended protocol (Supplementary material 1). A cutoff of ≥1% for ER staining positivity was used [16]. HER2 status was assessed using immunohistochemistry staining (1:400, DAKO, no antigen retrieval) with HercepTest scoring method as previously published [16, 17] (Supplementary material 1). In equivocal cases (HER2 immunohistochemistry score 2+), HER2 gene amplification was determined by chromogenic in situ hybridization using ZytoDot 2CSPEC ERBB2/CEN 17 Probe Kit (Supplementary material 2) [17, 18].

Cytoplasmic TXNIP staining was assessed using the semiquantitative H-score taking into consideration the intensity of staining and the percentage of stained tumor cells within each tissue core [19]. All cases were scored blinded to clinicopathological and outcome data. Cases with multiple cores (n = 210) were scored and the average was used as the final score.

This work obtained ethics approval by the North West—Greater Manchester Central Research Ethics Committee under the title; Nottingham Health Science Biobank, reference number 15/NW/0685.

Analysis of TXNIP in invasive breast cancer

To confirm the prognostic and clinical significance of TXNIP in invasive breast cancer, as a molecular surrogate of ductal carcinoma in situ in the breast, TXNIP gene copy number and normalized gene expression (mRNA) data were analyzed using the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) dataset that comprises 1980 tumors with long-term follow-up [20].

Statistical analysis

Statistical analyses were performed using SPSS v23 (Chicago, IL, USA) for Windows. X-tile software program (Yale University, version 3.6.1) was utilized to define the optimal cutoff point for TXNIP expression (H-score 120) with corrected p value and relative risk against local recurrence-free survival [21, 22]. Association between TXNIP expression and clinicopathological parameters using categorized data was evaluated using Chi-squared test. Survival rates were determined using the Kaplan–Meier method and compared by the log-rank test. Multivariate analysis using Cox proportional hazard regression model determined the influence of TXNIP expression, when adjusted to other variables, on local recurrence-free survival. All tests were two-tailed and a p value of <0.05 was considered as statistically significant.

Results

Analysis of TXNIP mRNA levels in invasive breast cancer revealed significant association between reduced levels/loss of expression and high histological grade (p = 0.004), large tumor size (p = 0.003), advanced-stage tumors (p = 0.01), and with ER-negative and HER2-positive phenotype (p = 0.003 and 0.011, respectively) and luminal B and basal intrinsic subtypes according to the PAM50 classification (p = 6.7 × 10−27) (Supplementary Table 1). Loss of copy number and reduced TXNIP mRNA expression were associated with worse outcome in terms of shorter breast cancer-specific survival (p = 0.003 and 0.002, respectively) (Supplementary Figure 1). Reduced TXNIP mRNA expression was also associated with loss of copy number (p = 0.049). These findings confirmed the tumor suppressor function of TXNP in breast cancer and its potential prognostic value and supported investigating its role in ductal carcinoma in situ.

Frequency and localization pattern of TXNIP in ductal carcinoma in situ

Assessment of whole tissue sections revealed cytoplasmic expression of TXNIP in a homogenous distribution pattern with occasional nuclear staining in normal luminal and myoepithelial cells. The homogenous pattern of expression confirmed the validity of using tissue microarray technology to assess its expression. Variable staining intensities and localization was detected in the different morphological components, with occasional strong nuclear staining in normal breast epithelial cells, strong to moderate cytoplasmic staining in ductal carcinoma in situ (with occasional membrane accentuation), and mild to negative cytoplasmic staining in invasive tumors (with no nuclear staining). TXNIP-positive pure ductal carcinoma in situ tumor cells exhibited distinct cytoplasmic staining with occasional peripheral/perimembrane accentuation and infrequent nuclear staining in some low-grade tumors (Fig. 1). After exclusion of uninformative cores (i.e., loss and/or folding of cores), a total of 636 pure ductal carcinoma in situ and 231 tumors with mixed ductal carcinoma in situ and invasive components were included in the analysis. TXNIP expression showed a unimodal distribution with a median H-score of 185 (range 0–300) in the pure ductal carcinoma in situ.

In mixed ductal carcinoma in situ and matched invasive cancer, both components were consistently positive in 27 (12%) cases and negative in 65 (28%) cases. High ductal carcinoma in situ/low invasive breast cancer expression for TXNIP was present in 134 (58%) cases however, there were 5 (2%) cases in which invasive breast cancer was positive/high expressing and ductal carcinoma in situ low expressing for TXNIP. When pure ductal carcinoma in situ was compared to ductal carcinoma in situ mixed with invasive breast cancer, a statistically significant difference was detected, with the TXNIP cytoplasmic mean H-score being higher in pure ductal carcinoma in situ tumors (\(\bar X\) ± SD = 185 ± 76 and 112 ± 88, respectively; t-test = 15.8 and p < 0.0001). In the mixed cases, a significant difference was also observed whereby the TXNIP cytoplasmic mean H-score was higher in the ductal carcinoma in situ component when compared with the adjacent invasive component (\(\bar X\) ± SD = 112 ± 88 and 83 ± 78, respectively; Z = 16.6 and p < 0.0001) (Supplementary Figure 2). Similar results were observed when TXNIP was analyzed as a categorical variable (Chi-squared p < 0.001 for pure vs mixed ductal carcinoma in situ and p < 0.001 for ductal carcinoma in situ vs adjacent invasive components). No nuclear expression was seen in tumor cells of the cases mixed with invasion.

Association of TXNIP with clinicopathological parameters in pure ductal carcinoma in situ

High/positive TXNIP cytoplasmic expression was significantly associated with good prognostic factors in ductal carcinoma in situ, including low nuclear grade (p = 1.6 × 10−5), absence of comedo-type necrosis (p = 0.001), positive ER status (p = 2 × 10−6), negative HER2 status (p = 0.007), and with the luminal (ER+/HER2−) phenotype (p = 4.6 × 10−5) (Table 1).

Table 1 Correlation between TXNIP expression and the clinicopathologic variables of pure ductal carcinoma in situ cases
Full size table

Association with patient outcome of pure ductal carcinoma in situ

In univariate analysis, patients with positive TXNIP expression had a significantly longer local recurrence-free survival (log rank (LR) = 7.99, 95% confidence interval (CI): 0.34–0.83, and p = 0.009; Fig. 2a). When the analysis was restricted to patients who underwent breast conservative surgery, to avoid the impact of surgical management on outcome prediction, TXNIP expression retained its significant association with improved outcome (LR = 6.67, 95% CI: 0.39–0.88, and p = 0.04; Fig. 2b). No significant association was observed between TXNIP expression and the type of local recurrence.

Fig. 2
figure 2

Kaplan–Meier curves show that high expression of TXNIP is associated with (a) longer local recurrence-free survival in the whole series (p = 0.009), b improved survival in patients treated with breast-conserving surgery (p = 0.04), c improved outcome in non-radiotherapy-treated patients (p = 0.02), and d local recurrence-free survival in radiotherapy-treated patients (p = 0.50). e TXNIP is also associated with local recurrence-free survival when ductal carcinoma in situ stratified into luminal ER+/HER2− (p = 0.040)

Full size image

Positive TXNIP expression was associated with improved outcome in the group of patients treated with breast conservative surgery without postoperative radiotherapy (94/220) (p = 0.02) but not in those who were offered radiotherapy (21/93) (p = 0.5) (Fig. 2c, d).

In ductal carcinoma in situ cases treated with breast conservative surgery without postoperative radiotherapy, loss/reduced TXNIP expression was observed in 126 cases and those showed approximately 40% recurrence rate (51/126), such patients could be candidates for radiotherapy to improve local recurrence-free survival. Low TXNIP expression was also associated with recurrence of 17% (12/72) of cases in the group of patients treated with breast conservative surgery and postoperative radiotherapy, which points out the importance of TXNIP as a predictor marker. Overall proportion of recurrences after breast conservative surgery that showed low TXNIP expression were 75% (63/84).

When molecular classes were considered, high TXNIP was associated with improved outcome in the luminal (ER+/HER2−) class (p = 0.04; Fig. 2e).

Multivariate Cox regression model, including ductal carcinoma in situ size, nuclear grade, and presence of comedo necrosis showed that high TXNIP expression was an independent predictor of good outcome in pure ductal carcinoma in situ (p = 0.005, hazard ratio (HR) 0.51, 95% CI: 0.32–0.81) (Table 2).

Table 2 Multivariate Cox proportional hazards for predictors of local recurrence in pure ductal carcinoma in situ patients
Full size table

Discussion

The treatment of ductal carcinoma in situ remains a challenge as the clinicopathologic features of the disease do not reliably stratify patients into distinct risk groups to guide treatment decision [23]. For this reason, some studies have attempted to risk stratify ductal carcinoma in situ based on genetic and molecular factors, including Oncotype Dx® ductal carcinoma in situ score [24,25,26,27,28]. Although ductal carcinoma in situ score minimized the proportion of patients undergoing radiotherapy, the assay seems expensive and not cost-effective [29]. In fact, none of the investigated parameters has been translated to clinical practice. Therefore, there is a pressing need to identify cost-effective and robust biomarker(s) to predict outcome for ductal carcinoma in situ patients.

The role of TXNIP expression was investigated in different tissues [30,31,32,33,34], including locally advanced invasive breast cancer [15] and we observed that loss of TXNIP expression was associated with features of increasing aggressiveness. It is likely that the effect of TXNIP on cell growth and proliferation are cell-context-dependent and might be circumvented via activation of alternate mitogenic pathways [9], nevertheless the data are consistent with TXNIP being a tumor suppressor [30]. TXNIP expression in cancer might be downregulated through epigenetic, transcriptional, post-transcriptional, or translational mechanisms [35].

To date, there is no well-established robust publically available database for ductal carcinoma in situ mRNA or copy number aberrations. To confirm the prognostic significance of TXNIP mRNA in breast cancer and to assess the role of copy number alteration on its expression and function, we have used the METABRIC cohort. This analysis showed that loss of copy number is one of the mechanisms for downregulation of TXNIP, which is associated with aggressive behavior and poor outcome. Utilizing a large annotated series of ductal carcinoma in situ treated at a single institution, we have demonstrated that TXNIP downregulation is associated with increased aggressiveness and poor outcome in terms of shorter local recurrence-free survival. We also found downregulation of TXNIP expression in invasive disease compared to the coexisting ductal carcinoma in situ. These results support our hypothesis that TXNIP is a tumor suppressor in breast cancer.

Our results demonstrate that loss of expression of TXNIP in ductal carcinoma in situ is not only associated with parameters characteristic of poor prognosis but is also an independent predictor of recurrence. These results are in line with TXNIP functioning as a tumor suppressor that is reported to be commonly silenced by genetic or epigenetic mechanisms in cancer cells [11, 36].

In this study, a significant negative association between the expression of TXNIP and local recurrence in ductal carcinoma in situ was identified. A previous study indicated that TXNIP stabilizes cyclin-dependent kinase inhibitor p27, which plays a pivotal role in inhibiting cell proliferation and apoptosis resulting in G1 arrest and inhibition of cell cycle [37]. This observation may explain our finding that high TXNIP expression was associated with low-grade, low-proliferating ER-positive and HER2-negative ductal carcinoma in situ and with longer local recurrence-free survival. Independent of its interaction with TRX, TXNIP has the ability to inhibit cell cycle progression by indirectly inhibiting mammalian target of rapamycin, a regulator of cell growth and metabolism [38]. Therefore, in addition to inducing a metabolic shift important to tumor biology, downregulation of TXNIP is likely to promote cell survival, growth, invasion, and metastasis. TXNIP can also reduce tumor invasion and angiogenesis through inhibition of TRX and can directly have an impact on cell survival by promoting a pro-apoptotic environment [39, 40]. The exact mechanisms by which TXNIP exerts its tumor suppressive functions in ductal carcinoma in situ cells are not yet fully clear. Future studies of the mechanisms by which TXNIP is expressed and functions in ductal carcinoma in situ will improve our understanding of the progression to invasive disease.

In this study, a significant correlation between TXNIP expression and ER+/HER2− status was identified. This finding is consistent with previous reports indicating that TXNIP may interrupt HER2-mediated oncogenic effect [36]. To our knowledge, few studies have reported an explicit impact of HER2 signaling on TXNIP expression. It might suggest that there is a direct link between ERBB2 and TXNIP or could be explained by the glucose theory. It is well known that cancer cells experience energetic stress and glucose deprivation especially in HER2-positive cancer cells [37]. Glucose induces TXNIP expression, and reactive oxygen species triggers the dissociation of TXNIP from TRX, leading to increased TXNIP availability [41]. Therefore, TXNIP may be induced by hyperglycemia regulated TRX− reactive oxygen species activity in the HER2 signaling pathway. Information regarding this mechanism will require further investigation in the future.

The current study demonstrated differential expression of TXNIP between pure ductal carcinoma in situ and ductal carcinoma in situ associated with invasive disease, which may have potential application for predicting invasive disease in ductal carcinoma in situ diagnosed on preoperative biopsy samples, after further validation. In addition, the ability of TXNIP to inhibit the major reactive oxygen species scavenger TRX, further regulate cell cycle progression and metastasis and promote apoptosis, clearly identifies TXNIP as an important therapeutic target.

In conclusion, the difficulty in implementing predictive markers for ductal carcinoma in situ remains the availability of large cohorts with consistent treatment for validation, and our study is an exception. We report that TXNIP is an independent prognostic factor and a potential tumor suppressor in breast cancer. Overexpression of TXNIP in ductal carcinoma in situ is therefore a good prognostic factor that can potentially improve ductal carcinoma in situ risk stratification for management purpose. Further functional studies are recommended to extensively validate the role of TXNIP as a predictor of progression to invasive disease.