|Home | About | Journals | Submit | Contact Us | Français|
A subset of patients with ductal carcinoma in situ (DCIS) will progress to invasive breast cancer. However, there are currently no markers to differentiate women at high risk from those at lower risk of developing invasive disease.
The association of two major tumor suppressor genes, retinoblastoma (RB) and phosphatase and tensin homolog (PTEN), with risk of any ipsilateral breast event (IBE) or progression to invasive breast cancer (IBC) was analyzed using data from 236 DCIS patients treated with breast conserving surgery with long-term follow-up. RB and PTEN expression was assessed with immunohistochemistry. The functional effects of RB and/or PTEN loss were modeled in MCF10A cells. Hazard ratios (HRs) were estimated with univariate and multivariable Cox regression models. All statistical tests were two-sided.
Loss of RB immunoreactivity in DCIS was strongly associated with risk of IBE occurrence (HR = 2.64; 95% confidence interval [CI] = 1.64 to 4.25) and IBC recurrence (HR = 4.66; 95% CI = 2.19 to 9.93). The prognostic power of RB loss remained statistically significant in multivariable analyses. PTEN loss occurred frequently in DCIS but was not associated with recurrence or progression. However, patients with DCIS lesions that were both RB and PTEN deficient were at further increased risk for IBEs (HR = 3.39; 95% CI = 1.92 to 5.99) and IBC recurrence (HR = 6.1, 95% CI = 2.5 to 14.76). Preclinical modeling in MCF10A cells demonstrated that loss of RB and PTEN impacted proliferation, motility, and invasive properties.
These studies indicate that RB and PTEN together have prognostic utility and could be used to target aggressive treatment for patients with the greatest probability of beneﬁt.
Ductal carcinoma in situ (DCIS) is a nonobligatory precursor to invasive breast cancer (IBC). With increased use of screening mammography to detect occult breast cancer, the incidence of DCIS has markedly increased. It is estimated that 1 million women will be living with this condition by 2020 (1,2). Left untreated, up to 53% of DCIS will progress to invasive breast cancer over a period of 10 or more years (3). Unfortunately, DCIS classifications used in clinical practice do not adequately predict the risk of DCIS recurrence and progression (4). Recently, a new pathologic grading system was proposed to improve the prediction of local recurrence (5). In this system, DCIS cases with high nuclear grade, predominantly solid architecture, and extensive (present in >50% of ducts) comedo-type necrosis had a particularly bad prognosis and were associated with recurrence or progression to IBC in less than 10 years. Sanders et al. examined the natural history of untreated, low-grade, noncomedo DCIS and showed that 39.3% of these patients developed invasive breast cancer in the same quadrant as the initial biopsy, with most events occurring within 10 to 15 years, but with some as late as 23 to 42 years after initial biopsy (6). Nearly half of the patients who developed invasive breast cancer died of metastatic disease 1 to 7 years after diagnosis. The results of this study suggest that a subset of patients with histologically low-grade DCIS will develop life-threatening invasive carcinoma. Stratifying DCIS patients using prognostic tumor markers might prevent both under and over treatment.
Active investigation of the biological processes responsible for progression of DCIS to invasive disease may facilitate development of better prognostic tests. The 2009 State of Science National Institutes of Health Conference on Diagnosis and Management of DCIS recommended development of risk stratification tests based on a comprehensive understanding of the clinical, radiological, pathological, and biological factors (2). Numerous biomarkers have been investigated for risk stratification of patients with DCIS. For example, elevated Ki67 levels, p53 mutations, and epidermal growth factor receptor 2 (HER2) amplification are known to be associated with increased nuclear grade and necrosis, histologic features that are associated with DCIS recurrence and progression (7–9). Cell cycle markers have also been studied, including p21, p27, and cyclin D1 (10–12). Despite these studies, no single biomarker has emerged to guide clinical practice.
Recently, retinoblastoma (RB) pathway disruption has been implicated in the invasive progression of DCIS. Gautier et al. reported that overexpression of p16ink4a and concomitant elevation of the proliferation marker Ki67 were observed in DCIS at risk for progressing to IBC (8). We confirmed this association in our DCIS cohort and demonstrated increased recurrence and invasive progression in tumors with elevated p16ink4a/Ki67 (13). Phosphatase and tensin homolog (PTEN) is another tumor suppressor that is commonly lost in breast cancer (14–16). No cancers rely on only a single tumor suppressor abnormality for invasive tumor development. However, in a number of tumor types, including breast cancer, there is a subtype of RB and PTEN doubly deficient tumors. Genetic modeling in glioblastoma and other cancers has suggested cooperation between these pathways in disease progression (17). Although PTEN expression has been associated with breast cancer metastasis and poor survival in invasive breast cancer, its role in DCIS progression has not been investigated. The goal of this study was to define the role of RB and PTEN as prognostic biomarkers for DCIS recurrence and progression.
DCIS breast tissue was obtained from the surgical pathology files at Thomas Jefferson University Hospital (Philadelphia, PA) with institutional review board approval. A total of 244 consecutive DCIS patients who underwent surgical resection from 1978 to 2008 and for whom tissue was available were included in this study. Clinical and treatment information was extracted from chart review. All patients were treated with breast conserving surgery only (no radiation or hormonal therapy) by the same surgeon (G F Schwartz). Negative margins (≥10 mm) were achieved at the conclusion of excision or re-excision, and removal of all suspicious calcifications was confirmed on postoperative mammography. Patients with DCIS involving more than one quadrant were treated by mastectomy and were excluded from this study. Patients who developed invasive ductal carcinoma within six months of a DCIS diagnosis were excluded because invasive ductal carcinoma was felt to be part of their original disease and not regarded as a recurrence. The date of diagnosis and recurrence were defined as the date of surgical procedure leading to the relevant pathologic diagnosis. The presence of DCIS, invasive cancer, or absence of disease at the last follow-up was established as the study endpoint. Median and mean follow up were 8.6 and 9.3 years, respectively, with 100 patients followed for more than 10 years and 51 patients followed for more than 15 years. For Kaplan–Meier analyses, 236 patients were at risk—165 at 5 years, 101 at 10 years, 53 at 15 years, and 15 at 20 years. For each case, size, the histological pattern (cribriform, solid, comedo, papillary, micropapillary), the presence or absence of necrosis, and nuclear grade were evaluated. For determination of DCIS size, the largest measurement of dimension from either the original report or review of histological sections was recorded. In some cases, size could not be assessed on review and was not recorded in the original pathology report. Nuclear grade was assigned using established criteria (18). Immunohistochemical stain for estrogen receptor (clone SP1), progesterone receptor (clone 1E2), Ki67 (clone 30- 9), and HER2 (clone 4B5) were performed on the BenchMark XT Slide Preparation System (Ventana Medical Systems, Tucson, AZ) using established clinical protocols and controls. Staining was scored using ASCO/CAP guidelines (19).
Expression of PTEN and RB was assessed by employing a standard immunoperoxidase method with primary PTEN antibody (dilution of 1:100, clone 138G6; Cell Signaling Technologies, Danvers, MA) and primary RB antibody (dilution of 1:50, clone 1F8; Thermoscientific, Waltham, MA). Methods for RB and PTEN immunohistochemical staining have been described (20,21). RB expression was scored semiquantitatively as negative (cancer cells showed no staining, whereas normal cells were positive), weak (staining intensity was less than adjacent normal cells), or strong (staining intensity was equal to adjacent normal cells). RB deficiency was defined as a score of either negative or weak. PTEN expression was scored as negative or positive using published scoring criteria (21). Evaluations were performed blinded to all clinical and biological variables. All stains were reviewed by two pathologists (A K Witkiewicz and M Qeenan), and disagreements were resolved by consensus.
MCF10A and PTEN-/- MCF10A cells were maintained in Dulbecco’s modified Eagle’s medium/F12 supplemented with 5% horse serum, 100 μg/mL of epidermal growth factor, 10 μg/mL of insulin, 100 μg/mL of hydrocortisone, 100U/mL of penicillin and streptomycin, and 2mM of L-glutamine. The MCF10A and PTEN-/- MCF10A cells were infected with retroviruses encoding nonspecific control (miNS) or RB specific (miRB) small hairpin RNAs. Cells were selected with puromycin, and the efficacy of the RB knockdown was confirmed by immunoblotting. For challenge with low serum, cells were grown in media as described above with the exception of 0.5% horse serum without epidermal growth factor. Cells were allowed to grow for three days and then processed for flow cytometry as described below. Additionally, cells were stained with 1% crystal violet to visualize outgrowth. For immunoblot analyses, equal total protein was separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis. Proteins were detected by standard immunoblotting procedure using the following primary antibodies: lamin B (M-20), pERK (E-4), ERK (K-32) (Santa Cruz Biotechnology, Santa Cruz, CA), PTEN (138G6) (Cell Signaling), and RB (G3–245) (Becton Dickson, Franklin Lakes, NJ).
For cell proliferation analysis, cells were incubated with BrdU (Amersham Pharmacia Biotech, Piscataway, NJ) for 1 hour before harvest. Cells were washed in phosphate-buffered saline and fixed in cold 70% ethanol. Bivariate flow cytometry was utilized for dual analysis of BrdU incorporation and total DNA content. All data is from three independent experiments.
Cells were treated with trypsin and resuspended in assay medium (Dulbecco’s modified Eagle’s medium/F12 supplemented with 2% horse serum, 10 μg/mL of insulin, and 100 μg/mL of hydrocortisone), 100U/mL of penicillin and streptomycin, and 2mM of L-glutamine at a concentration of 25,000 cells per mL. Eight-chambered glass slides (Nalgene Nunc, Naperville, IL) were coated with 40 μL of Matrigel (BD Bioscience, Bedford, MA) and solidified for 30 minutes. The cells were mixed 1:1 with assay medium containing 4% Matrigel and 10ng/mL of epidermal growth factor. Four hundred microliters of the cell mixture were added to each well, with 5000 cells per chamber. Assay media were replaced every four days. Acini growth was monitored by imaging MCF10A miNS (nonspecific knockdown), MCF10A miRB (RB knockdown), PTEN-/- miNS, and PTEN-/- miRB acini with a ×20 objective. Diameters were measured at the middle optical section of each acinus with the support of Image J software; more than 50 acini were measured per condition. Data show the mean and 95% confidence interval (CI). Ki67 staining to measure proliferation was performed, and more than 100 nuclei were scored. Data show the mean and 95% confidence interval.
MCF10A and PTEN-/- cells were seeded (5×104 cells) in Boyden Chambers (BioCoat 354578; Franklin Lakes, NJ) under low serum conditions. Complete growth medium was added to the wells as the chemoattractant. Chambers were placed in wells containing complete medium. The cells on the lower surface of the membrane were counterstained with 4,6-diamidino-2-phenylindole (Sigma-Aldrich, St. Louis, MO). Cells were scored with a fluorescent microscope (20). Three independent experiments were averaged, and the data presented represent the mean and 95% confidence interval. Wound healing assays were performed using standard protocols, and the invasion into the wounded area was quantified (20). Data shown are from at least three independent experiments and show the mean and 95% confidence interval.
Two time-related endpoints were analyzed. The first endpoint was ipsilateral breast event (IBE) occurrence, where failure is defined as the first reported DCIS recurrence or invasive breast carcinoma (IBC) progression. If the patient did not experience an event, she was classified as a censored observation on date of last follow-up. Second endpoint was invasive breast carcinoma (IBC) progresssion, where failure is defined as an invasive progression reported at any time. Follow-up time was measured from the date of first DCIS surgery. The failures rates were estimated by Kaplan–Meier method and compared by log-rank test. The Cox proportional hazards models were utilized to determine the univariate and multivariable hazard ratios (HRs) for standard clinical and pathological variables. The validity of the proportional hazards assumption was verified by the inclusion of time-dependent variable(s) in the multivariable models. A forward stepwise procedure with the criterion of P less than .05 was used to select individual variables for subsequent multivariable analysis. To test for interaction between PTEN and RB, a Cox model was used with the two main effects and the interaction term. P values less than .05 were considered statistically significant and were not adjusted for multiple testing. All analyses were performed with the use of SAS software (version 9.2; SAS Institute, Cary, NC). The statistical tests performed were two-sided.
Prior studies have implicated the RB-pathway as a determinant of DCIS recurrence (8,20,22). These studies utilized p16ink4a and Ki67 as surrogates for loss of RB function (8,20,22). Surprisingly, no published studies have directly investigated the expression of the RB tumor suppressor protein in a cohort of DCIS cases with long-term follow-up. We employed a cohort of 236 patients with DCIS that were treated by surgical resection with wide margins and long-term follow-up (median = 8.3 years) and had tissue sufficient to perform staining for two markers, RB and PTEN. In this cohort, the frequency of any IBE was 32% (n = 75), and the frequency of recurrent invasive (IBC recurrence) disease was 11% (n = 27). Multiple standard clinical and pathological variables were evaluated for association with these two events (Supplementary Table 1, available online). As shown, age, necrosis, nuclear grade, comedo histology, estrogen receptor status, and progesterone receptor status were not statistically significantly associated with IBE or IBC occurrence (Supplementary Table 1 and Supplementary Figure 1, available online). High levels of HER2 were weakly associated with an IBE (HR = 1.66; P = .048) but were not associated with IBC recurrence. These findings are largely consistent with other studies (23). Because of the absence of confounding effects of adjuvant therapy, this cohort is ideal for evaluating prognostic determinants of disease recurrence.
Immunohistochemical analysis of RB was optimized in a clinical laboratory, and all staining was performed with positive and negative controls. Staining for RB was considered positive (RB proficient) when the cells of the DCIS lesion retained levels of nuclear staining comparable with that of stromal cells or tissue lymphocytes (Figure 1, ,A).A). Staining was defined as RB deficient if there was negative or very weak nuclear staining within the tumor epithelia but surrounding cells were positive, serving as “built-in” internal positive control. In the analyzed cohort, 19.5% percent of cases were RB deficient, with 17.8% exhibiting negative and 1.7% exhibiting very weak staining in reference to the internal positive control. Kaplan–Meier analysis was performed on recurrence as a function of RB status. The data revealed that decreased RB expression was statistically significantly associated with IBEs (HR = 2.64; 95% CI = 1.64 to 4.25; P < .001 (Table 1, Table 2 and Figure 1, ,B).B). The estimated 10-year IBE occurrence rate in RB-deficient DCIS was 58% vs 25% for RB-proficient DCIS (Table 1). In the analyses of IBC, there was an even stronger effect (Figure 1, ,C),C), with RB-proficient cases exhibiting a relatively low risk of undergoing invasive progression (8% at 10 years), whereas RB-deficient cases exhibited 39% recurrence as invasive lesions (HR = 4.66; 95% CI = 2.19 to 9.93, P < .001) (Table 1). These data remained statistically significant in multivariable analyses against standard clinicopathological features of DCIS (Table 3).
Although the association of RB status with recurrence and progression was statistically significant, there were clearly RB-deficient tumors that failed to progress. Our analyses of RB-negative invasive breast cancer cases (in preparation) showed concurrent PTEN loss at relatively high frequency, suggesting that cooperation between RB and PTEN loss may be particularly germane to aggressive disease. As with RB, the staining for PTEN was rigorously optimized with positive and negative controls (Figure 2, ,A).A). In our cohort, 44% of DCIS cases were PTEN deficient. PTEN loss is generally believed to be associated with aggressive tumorigenic behavior. However, in the DCIS cohort analyzed, PTEN status was not associated with risk of IBEs (P = .41) or risk of IBC recurrence (P = .31) (Figure 2, ,BB and Table 1). Thus, PTEN loss is not a prognostic variable in DCIS in this cohort.
To specifically investigate the interaction between RB and PTEN, the cases were stratified into four groups based on combined RB and PTEN status. Kaplan–Meier recurrence-free survival curves revealed that cases that were deficient for both RB and PTEN were at statistically significantly increased risk of failure relative to all other groupings (Figure 3, A–D). In univariate analysis, combined RB and PTEN loss was strongly associated with risk of IBE occurrence (HR = 3.39; 95% CI = 1.92 to 5.99; P < .001) (Table 2). In contrast, RB-deficient but PTEN-positive lesions were not at statistically significantly increased risk of IBE occurrence (HR = 1.28). Thus, PTEN status was highly relevant for defining RB-deficient tumors that would recur. Combined deficiency for RB and PTEN was even more strongly associated with risk of IBC recurrence (HR = 6.10; 95% CI = 2.5 to 14.76; P < .001) (Table 2). Statistical modeling demonstrated that there was a statistically significant interaction between RB and PTEN, and the combination of RB and PTEN as a determinant of recurrence was statistically significant in multivariable analyses (Table 3). Thus, lesions with combined RB and PTEN loss are at high risk of DCIS recurrence and progression. At 2 and 5 years, their estimated rates of IBEs were 36% and 61%, respectively, and their rates of IBC occurrence were 16% and 38%, respectively (Table 2). Because HER2 was found to be statistically significant in the univariate IBE analysis, it was further evaluated in the Cox model with RB, PTEN, and their interaction. HER2 was found to be not statistically significant (HR = 1.55; P = .09), whereas the interaction of RB and PTEN continued to be highly statistically significant (HR = 3.55; P = .03). Together, these data indicate that the combination of RB and PTEN can be employed effectively to define DCIS at high risk of recurrence and progression to invasive disease (Figure 3, ,EE and andFF).
To determine the functional effect of RB and PTEN alone and in combination on DCIS progression, we modeled the loss of the tumor suppressors in MCF10A cells. The MCF10A model is an immortalized breast epithelial line that has been extensively utilized to dissect determinants of breast carcinogenesis. To model the inactivation of PTEN, matched MCF10A cell lines with intact PTEN or PTEN deleted by homologous recombination were employed (24). In these cells, RB was depleted using established knockdown methodology (miRB) (20). As shown in Figure 4, ,A,A, effective deletion of PTEN and knockdown of RB was confirmed in these cells by immunoblotting. Furthermore, elevated ERK activity was observed with PTEN deletion consistent with published data that have also demonstrated expected AKT activation (24). Thus, the four lines recapitulate the loss of RB and PTEN, as is observed in DCIS. Because both RB and PTEN are implicated in cell-cycle control, we evaluated proliferation by monitoring BrdU incorporation (Figure 4, ,B).B). In the context of MCF10A cells, PTEN deficiency had no statistically significant effect on BrdU incorporation in full serum (Figure 4, ,B).B). In contrast, RB loss led to a modest, yet statistically significant, increase in BrdU incorporation. Because recurrence and progression likely reflect the ability of cells to disseminate away from the primary lesion, we evaluated the motility and invasive properties of the cell populations. Although PTEN has been implicated in invasion, it had little effect on the ability of cells to invade through Matrigel in modified Boyden chamber assays (not shown). However, RB promoted a more invasive phenotype in both Boyden chamber and wound-healing assays (Figure 4, ,CC and andD),D), which is consistent with prior studies (20,25). It is well known that RB-deficient cells retain a number of dependencies that limit the development of an invasive cancer. Interestingly, when we tested the ability of the populations to grow under conditions of low growth factor (Figure 4, ,EE and andF)F) or lack of adhesion (not shown), only PTEN loss provided an advantage to the cell populations. These data indicate differential and complementary effects of RB and PTEN loss on proliferative control in MCF10A cells. One means to evaluate multiple aspects of mammary cell biology is the analyses of three-dimensional growth in Matrigel (26,27). In this assay, RB loss leads to an acceleration of acinar growth and enhanced Ki67 staining (Figure 5). However, RB-deficient cells are still prone to cell death and are cleared from the lumen, resulting in formation of hollow acinar structures. As such, RB-deficient MCF10A populations retain a relatively normal overall morphology and organization (Figure 5, ,A).A). In contrast, PTEN loss facilitates a degree of acinar disorganization not observed in RB-deficient cells (Figure 5, ,A).A). These effects were largely additive in reference to acinar size and proliferation (Figure 5, ,BB and andC).C). Although individual loss of RB or PTEN has distinct effects on mammary epithelia, the combined loss results in a rapidly proliferating and invasive population. Together, these aspects likely contribute to the pronounced aggressiveness of RB- and PTEN-deficient DCIS and may explain their prognostic significance in clinical samples.
Our findings indicate that histological loss of RB protein expression is a strong independent marker of DCIS recurrence and progression to invasive cancer. Although PTEN loss is frequently observed in DCIS, it had no statistically significant association with disease outcome as a single variable. However, PTEN status served to effectively stratify RB-deficient cases that were either relatively indolent (PTEN positive) or prone to recurrence or invasive progression (PTEN negative). Together, these analyses indicate the importance of multimarker analyses and suggest a relatively simple means to detect high-risk DCIS.
The clinical management of DCIS changed substantially over the last three decades. Initially, DCIS lesions were treated by mastectomy (1). Subsequently, breast-conserving surgery with radiation therapy and hormonal interventions became a standard treatment (1,28,29). However, it is unclear whether patients who have DCIS uniformly benefit from these interventions. In several studies, radiotherapy reduced in situ or invasive recurrences by about 50% (30–32). Although radiotherapy is associated with substantial reductions in local recurrence, no differences have been reported in overall survival. Furthermore, because only 10%–15% of cases recur as invasive disease without radiation therapy, clearly not all DCIS patients require radiation. Similarly, effects of adjuvant tamoxifen are difficult to evaluate. In a recent analysis, adjuvant tamoxifen treatment conferred a decreased risk of contralateral, but not ipsilateral, invasive breast cancer, which opens the question of whether tamoxifen works through suppression of recurrence or second primary disease (33). Therefore, identifying DCIS cases that can be cured by surgery alone is critical for effectively managing disease and mitigating overtreatment of patients.
Currently, no clinicopathologic features or biomarkers allow for personalized treatment of DCIS (22,34). Because of the difficulty of obtaining sufficient material from banked DCIS tissue, gene expression profiling has substantially lagged behind that of invasive breast cancer. As such, most markers that have been associated with disease recurrence or invasive progression of DCIS have emerged from histological analyses of factors implicated in the pathogenesis of invasive breast cancer. For example, HER2 overexpression has been extensively investigated as a marker of recurrence and progression to invasive disease. The results from such studies remain in debate, and, as shown here, HER2 overexpression alone is a relatively weak marker for invasive progression of DCIS. It has been previously demonstrated that loss of RB tumor suppressor pathway function, as measured by high levels of p16ink4a and Ki67, is associated with recurrence and progression to invasive breast cancer (8,13,22). Here we used a highly optimized protocol for detecting loss of RB tumor suppressor protein expression in DCIS. As a single variable, loss of RB was statistically significantly associated with overall recurrence and invasive progression in DCIS. Interestingly, although there is strong concordance between p16ink4a/Ki67 high expression and RB loss, there is not an absolute relationship. In our cohort, RB loss was a stronger prognostic marker than the combination of p16ink4a and Ki67 high expression (not shown). Although the hazard ratios for RB loss were relatively large, it was clear that a substantial fraction of women harboring RB-deficient tumors could be cured by surgery alone. This finding led us to investigate other markers that could be used to further stratify RB-deficient cases.
One of the pathways implicated in breast cancer etiology and progression is the PTEN/PI3K/AKT pathway. Additionally, germline PTEN mutations are found in the autosomal-dominant Cowden syndrome, which is characterized by multiple hamartomas as well as an increased risk of breast cancer. PTEN loss can result from mutation, loss of heterozygosity, and epigenetic down-modulation, and has been reported in nearly 50% of human cancers, including breast cancer (35). Studies evaluating PTEN expression by immunohistochemistry demonstrated loss or weak expression in 33% of invasive breast cancer (36–38). Furthermore, PTEN loss correlated with high tumor grade, larger tumor size, negative hormone receptor status, and poor prognosis (36). Although PTEN influences prognosis of established malignancy, its role in the early stages of cancer development is less established. In one study investigating frequency of loss of heterozygosity in DCIS, loss of heterozygosity at the PTEN locus was only present in DCIS lesions coexisting with IBC and not in pure DCIS (16). A recent study investigating PTEN promoter methylation status in DCIS demonstrated presence of PTEN hypermethylation in the subset of pure DCIS and DCIS associated with IBC. Surprisingly, here we found that, although the histological loss of PTEN was relatively common in DCIS, it was not associated with recurrence or progression to invasive breast cancer. Therefore, although lost at relatively high-frequency, PTEN as a single marker has little prognostic significance in DCIS treated by surgery.
For clinical utilization it is critical to develop panels of markers that have high specificity and sensitivity. By employing PTEN status, the overall specificity of RB loss for IBE occurrence and IBC recurrence was substantially enhanced. This relationship may have been anticipated given the cooperation of RB and PTEN loss in specific genetic models (17), although there is also evidence that RB and PTEN function in the same pathway and thus could be viewed as redundant (39,40). In our functional studies, it was apparent that RB loss contributes to invasiveness and aberrant proliferation, whereas PTEN loss enhanced cell survival and the ability to proliferate under low exogenous growth factor concentration. We have recently shown that the epithelial-to-mesenchymal transition is associated with progression to invasive disease, and RB loss has been associated with this process (25,41,42). This finding suggests that invasive properties associated with disease progression could be an especially relevant consequence of RB loss in breast cancer. Presumably, the combination of deregulated proliferation, aberrant response to stress signals, and invasive properties underlie the risk for DCIS recurrence and progression. Given the specificity of combined RB and PTEN loss for invasive progression and recurrence, it suggests that this “subtype” of DCIS would warrant more aggressive therapeutic intervention.
This study also had some limitations. Although the cohort employed is ideal for evaluating IBE and IBC endpoints after breast-conserving surgery, it is important to realize that this represents a single cohort, and validation in other similar cohorts will be required to further interrogate the veracity of the findings. Importantly, most patients with a DCIS diagnosis are treated by surgery and adjuvant radiation therapy. The key question of whether patients with RB- or PTEN-deficient DCIS will benefit from radiation therapy has not been addressed in our study. Both RB and PTEN deficiency are associated with increased sensitivity to DNA-damaging agents (43–45). Thus, our findings herein suggest that directing radiation therapy against the RB- or PTEN-deficient DCIS, which harbors poor prognosis, may be beneficial. Additional investigation and prospective validation will be required to define the ideal clinical utilization of these markers in the treatment of DCIS. However, our study, in combination with the work of others, provides rationale for development of a recurrence test for DCIS based on the evaluation of the RB and PTEN tumor suppressor pathways.
This work was supported by the National Cancer Institute, National Institutes of Health (CA163863 to AKW; CA129134 to ESK).