Biomarkers Predicting Progression and Prognosis of Ductal Carcinoma In Situ (DCIS)

Histopathological Features and Heterogeneity of DCIS

One of the primary histopathological characteristics of DCIS is the proliferation of neoplastic epithelial cells confined within the mammary ducts (12). As a non-obligate precursor for IBC, the major diagnostic criteria that differentiates it from IBC is the presence of intact basement membrane and myoepithelial cell layer in DCIS (13). DCIS displays heterogenous growth patterns including cribriform (irregular, glandular spaces formed by tumor cells), micropapillary (small finger-like projections of tumor cells within the ducts), solid/comedo (tightly packed tumor cells without differentiating glandular spaces with characteristic central duct space necrosis), and papillary (finger-like projections covered by tumor cells with a fibrovascular core) (13, 14). These patterns are used to facilitate diagnosis and determine the extent of DCIS. However, since there can be more than one pattern observed in a lesion, the more clinically significant system to classify DCIS is based on nuclear grade. Low-grade DCIS present as evenly spaced uniform cells with small regular nuclei, with infrequent mitosis and necrosis. Intermediate-grade DCIS exhibit some abnormal growth characteristics, with moderate cell size and shape variations. However, high-grade DCIS is composed of large, irregularly spaced pleomorphic cells with an irregular nuclear contour and high nuclear-to-cytoplasmic ratio (15). High-grade DCIS is more likely to recur or progress to IBC and in relation to the architecture, the presence of comedo pattern is associated with high-grade DCIS and therefore an increased risk of progression to IBC (16-18).

Overall, histopathological assessment is crucial for identifying and characterizing DCIS and it provides helpful prognostic and diagnostic information. Since these histopathologic characteristics are the basis of DCIS grading, which is one of the essential features used for predicting the possibility of disease progression, it guides the development of optimal treatment plans. However, these methods have limitations in predicting DCIS progression and are subject to differences in technical acuity of pathologists in analyzing samples. This challenge, along with the biological heterogeneity of DCIS grades, supports the need for better risk prediction markers and tailored treatment approaches.

DCIS Diagnosis and Upstaging Rates

The diagnosis of DCIS has dramatically increased due to the increased utilization of screening mammography (2, 3, 5, 19). One in every 1,300 mammography screenings leads to a diagnosis of DCIS (3). Multiple studies have estimated that progression rates of DCIS to IBC range from 14% to 53% over at least a 10-year period (1, 20). However, these rates are not uniform as progression is influenced by multiple factors, such as the grade, molecular characteristics, and size of the DCIS lesions, as well as individual patient factors (21). There is also considerable variation in the rate of progression of DCIS to invasive disease depending on the sensitivity of the diagnostic test and the type of the risk prediction model used. This can vary from 20% to 91% (21, 22). However, DCIS can also remain indolent and never progress to IBC despite being left untreated (15). The natural history of low-grade DCIS may extend more than four decades (23). Estimates of overdiagnosis range from 34% to 72% for patients aged 50-74 years who underwent screening every other year (21). To add complexity to the challenges in DCIS management, it is also important to consider that a significant number of DCIS lesions, ranging from 6-33%, are upstaged to invasive carcinoma upon excision (24-28). This corresponds to under-staging in approximately one in four DCIS diagnoses highlighting a potential risk of conservative management, with no surgery or radiation therapy. Herein lies the challenge in the management of DCIS. We must consider a significant number of DCIS cases that should be upstaged to IBC but also pay attention to concerns about overdiagnosis which can lead to unnecessarily aggressive treatment.

DCIS Treatment Options

The primary goal of DCIS treatment is to prevent progression to IBC. Currently, the management options for DCIS include breast-conserving surgery (BCS) or mastectomy, which can be followed by adjuvant radiation therapy (RT) and hormonal therapy in eligible patients. However, National Comprehensive Cancer Network (NCCN) cautions that while RT significantly decreased the risk of recurrence in patients with DCIS, the overall risk of recurrence is low. Furthermore, the choice of local treatment does not impact overall disease-related survival and therefore the individual’s preferences for risk reduction must be considered (29-31). In attempts to de-escalate treatment to avoid higher treatment costs and adverse effects, several trials have examined if BCS can be done alone without RT (32-34). This is especially for low-risk patients, wherein the benefits of RT do not outweigh the harms associated with radiation toxicity (e.g., myelosuppression, pneumonitis, fatigue, secondary cancer) and the risk of recurrence is unlikely to be significantly reduced by RT (35). Additionally, several trials such as the LORIS Trial [Surgery versus Active Monitoring for LOw RISk Ductal Carcinoma in Situ (DCIS); ISRCTN27544579] in the UK and COMET Trial (Comparing an Operation to Monitoring, With or Without Endocrine Therapy For Low Risk DCIS; NCT02926911) in the United States, and the LORD trial in the Netherlands (NCT02492607) are comparing the risks and benefits of standard surgery and active monitoring for low-risk DCIS with variable definition. Active monitoring of patients with low-risk DCIS includes mammograms every 6 to 12 months for at least 5 years and performing a magnetic resonance imaging (MRI) to determine if surgical intervention is necessary. The COMET trial is a randomized pragmatic non-inferiority clinical trial that enrolled 995 patients with low-risk DCIS defined as grade 1-2, hormone receptor (HR)-positive, human epidermal growth factor receptor 2 (HER2) negative DCIS randomized to surgery or active monitoring (36). After median follow up of 36.9 months, the authors reported noninferiority of both treatment modalities in the intention to treat and per protocol analysis. The primary endpoint of 2-year ipsilateral invasive cancer rate was 5.9% in the guideline-concordant arm compared to 4.2% in the active monitoring arm. Of note, nonacceptance of primary treatment in 29% favoring the active monitoring arm was observed highlighting significant patient interest in active monitoring (36). Patient reported outcomes in both groups showed similarities in quality of life regardless of treatment strategy (36). Taken together, results of the COMET trial demonstrate that in the short-term, active monitoring of low-risk DCIS is a reasonable approach. The LORIS trial has completed recruitment while the LORD trial is still ongoing.

DCIS Progression and Predictive Biomarkers

NCCN recommends the following work up for DCIS: history and physical examination, bilateral diagnostic mammography, pathology review, determination of tumor estrogen receptor (ER) status, and breast MRI when indicated. Among these, ER status may influence decision for disease management, as it is used to determine the benefit of adjuvant endocrine therapy for risk reduction (37). Evidence suggest that ER negativity does not significantly correlate with an incidence of a future ipsilateral breast event among patients (38). This highlights the scarcity of biomarkers that can predict DCIS progression in patients that can heavily impact the choice of management.

We will discuss what is currently known about genomic and transcriptomic changes, immune-related markers, and epigenetic factors that are associated with DCIS progression. We selected these biomarkers as they have been studied in the context of DCIS recurring or progressing to IBC especially in patient samples. While several biomarkers require further investigation and clinical validation, these studies will enrich our repertoire of potential biomarkers for DCIS progression and hopefully aid in making better clinical decisions when treating patients with DCIS.

Genomic changes in DCIS progression. Somatic genetic alterations include point mutations, copy number alterations (CNA), loss of heterozygosity (LOH), and chromosomal alterations. Genomic instability provides growth advantage to cells and is a fundamental characteristic of cancer’s heterogeneous nature. Understanding the stage-specific genomic differences may help find patterns that can predict DCIS progression.

Using comparative genomic hybridization (CGH), Buerger et al. (1999) found that well-differentiated and intermediately-differentiated DCIS (low-risk) had losses of 16q whereas poorly-differentiated DCIS (high-risk) had a higher frequency of amplifications in 17q12 and 11q13 and a higher average rate of genetic imbalances. Importantly, they first noted that adjacent IBC had a genetic pattern almost identical to DCIS and hence suggested that DCIS are genetically advanced, direct precursors of IBC (39). In support of these findings, a study by Vincent-Salomon et al. (2008) showed significant CNA overlap between DCIS and invasive ductal carcinoma (IDC), and these genomic changes correlated with modifications of the gene expression profiles in DCIS and IBC (40). Similarly, Johnson et al. (2012) performed genomic analysis of synchronous DCIS and IDC and noted that DCIS is an advanced, genomically unstable tumor that in many cases has already acquired most determinants of invasiveness. Matched IDC and DCIS samples had highly similar copy number profiles (41). However, they noted four chromosomal regions of loss (3q, 6q, 8p, and 11q) and four regions of gain in IDC (5q, 16p, 19q and 20) that were recurrently affected in IDC and not in DCIS. Gorringe et al. (2015) studied the genome-wide copy number and loss of heterozygosity in a cohort of patients with pure DCIS treated with wide local excision alone (42). Interestingly, their results showed that DCIS cases with recurrence had a higher fraction of the genome altered compared to non-recurrent cases (p=0.026), more copy number changes especially gain of 20p12 and 20q11 in DCIS recurrence as IBC (p<0.01), and chromosome 10 allelic imbalance (42). While there may be some conflicting results on the genomic profile alterations of DCIS as it progresses to IBC, it must be noted that these differences may be due to the limitations of available samples as well as the variations in the sequencing/profiling platforms used. A meta-analysis of 26 studies by Rane et al. (2015) characterized the somatic CNA in DCIS cases with and without IBC. They identified 339/851 cytobands that showed significant differences in the gain and loss frequencies (p≤0.05) with overrepresentation of genes in pathways known to be affected in breast carcinogenesis (41%) such as transmembrane signaling receptor activity, G-protein-coupled receptor activity, cadherin protein class, wingless-type MMTV integration site family (Wnt) pathway, and inflammation (43). From these analyses, they concluded that DCIS and IDC harbor similar numbers of genetic aberrations in more than 76% of the genome – suggesting that a significant part of the decision of whether disease will progress into IDC probably occurs early at the stage of DCIS (43). Similarly, Doebar et al. (2019) noted a high genomic concordance between synchronous DCIS and IBC (44). Supporting this, a recent genomic study on primary DCIS and their matched recurrences observed that most mutations and CNAs were already established in the primary DCIS, and they found no clear genomic markers for invasive progression. Frequency of CNA across DCIS and invasive recurrences also showed highly similar chromosomal gains and losses (10). Similar results were found by Kaplan et al. (2024) noting that most small variants and CNVs were shared between co-occurring DCIS and IBC lesions, unsurprisingly with IBC having a higher degree of additional mutations. In terms of chromosomal instability, pure DCIS also showed almost the same degree as invasive cancers but noted that low-grade DCIS had a higher 16q loss (p=0.002) and high-grade DCIS had higher 8p loss (p =0.005) (9).

With these studies, it appears that these stage-specific genomic differences between DCIS and IBC are limited (10, 40, 43, 45). While there are key alterations noted such as GATA-binding protein 3 (GATA3) mutation positivity [odds ratio (OR) for relapse=5.5; 95% confidence interval (CI)=0.63-76.7], phosphatidylinositol 3-kinase catalytic subunit alpha (PIK3CA), tumor protein p53 gene (TP53), and mitogen-activated protein kinase kinase kinase 1 (MAP3K1) mutations, amplification of cyclin D (CCND)1 and oncogene MYC, CNVs in myeloid cell leukemia-1 (MCL1), Cyclin-dependent kinases regulatory subunit 1 (CKSB1), and epidermal growth factor receptor-2 (ERRB2), and some patterns of loss and gain of chromosomal copy alterations among others, the genomic landscape between these two stages remains significantly similar (9, 41, 46, 47).

Gene Expression Markers

Molecular subtyping of DCIS. In attempts to transition from histopathological nuclear grading which demonstrated a significant amount of heterogeneity for DCIS, some studies focused on molecular subtyping of DCIS lesions similar to approach for IBC (48, 49). In breast cancer, these subtypes reflect the expression of the nuclear hormone receptors ER and progesterone receptor (PR), and epidermal growth factor receptor-2 (ERBB2/HER2) as the major regulators of cancer cell growth and survival. Zhou et al. (2010) investigated the prognosis of DCIS in relation to the molecular subtypes defined by immunohistochemistry (IHC). They studied a cohort of 458 women with a primary DCIS diagnosis, performed microarray gene-profiling, and determined who developed local DCIS vs. invasive recurrence. Their results showed that while DCIS with basal-like features had a higher risk of local recurrence [hazard ratio (HR)=1.8, 95%CI=0.9-4.2] and a higher risk of developing invasive cancer (HR=1.9; 95%CI=0.7-5.1) compared to other subtypes, the difference was not statistically significant (50). Furthermore, when they focused on a subgroup of women who only had BCS, the risk for progression of basal-like DCIS was doubled (HR=1.9; 95%CI=0.7-5.1) but still not significantly higher. Interestingly, they also analyzed histopathologic factors such as nuclear grade, size, necrosis and margins and found that the grade did not relate to prognosis (HR=1.0, 95%CI=0.3-3.9) and suggested that molecular subtyping by IHC might indeed be more useful (50).

To investigate the use of HER2 for molecular subtyping of DCIS, Roses et al. (2009) studied 106 patients diagnosed with DCIS by either core needle or excisional biopsies and performed IHC staining for ER, PR, and HER2 (51). Their results showed that HER2 was the only significant predictor for the presence of invasive disease after multivariate adjustment (OR=6.4; p=0.01) (51). Nagasawa et al. (2021) also proposed that HER2 amplification is critical marker (46). Yang et al. (2022) showed that HER2 over-expression had a 15.15% 5-year overall secondary breast events (SBEs) and is the only independent prognostic factor among the subtypes (p=0.012) (52). Another study showed that there are more high-risk cases and higher risk of IBE recurrence in HR-HER2+ subtype than in HR+HER2− subtype (53). However, these are not consistent with other studies hence the importance of HER2 status as a prognostic indicator of DCIS has been unclear (50, 54, 55). Furthermore, there was no statistically significant benefit to using trastuzumab with RT in HER2 amplified DCIS (56). Since the results do not affect management, NCCN guidelines do not recommend HER2 testing for DCIS (57).

These findings provide insufficient evidence and clinical significance to support the use of molecular subtyping as a marker of DCIS progression. However, they can be used as candidates for biomarker studies, and a guiding tool to subcategorize heterogenous DCIS.

Cancer associated genes in DCIS progression. Cancer is a remarkably complex disease that results from cells acquiring capabilities that allow them to proliferate uncontrollably and invade surrounding tissues. These characteristics were described by Hanahan and Weinberg (2000) as the hallmarks of cancer (58). Here, we discuss several gene expression changes that have been shown to have an association with DCIS progression and are key mediators of cancer hallmarks, such as cell proliferation and survival, DNA repair, and invasion (58).

Tumor suppressor gene p16/ cyclin-dependent kinase inhibitor 2A (CDKN2A) which regulates cell cycle through inhibition of cyclin-dependent kinases, is co-expressed with cell proliferation marker Ki67 and cyclooxygenase-2 (COX-2), and associates with basal-like behavior in DCIS cases (59). In agreement, Kerlikowske et al., (2010) also found that the eight-year risk of DCIS progressing to a subsequent IBC was higher (p=0.018) for women with initial DCIS lesions that were detected by palpation (HR=2.7; 95%CI=1.4-5.5), or were p16, COX-2, and Ki-67 triple positive (HR=1.9; 95%CI=0.8-4.3) than for women with lesions detected by mammography and were p16, COX-2, and Ki-67 negative (54). These observations were made in a nested case-control study in a population-based cohort of 1162 women who were diagnosed with DCIS and treated with lumpectomy alone (54) and ER, PR, Ki-67, p53, p16, HER2, and COX-2 were assayed by IHC staining of paraffin-embedded DCIS tissues to identify factors or combinations associated with progression to IBC.

In support of these findings, COX-2 has been associated with tumor survival, invasion, and metastasis in cancer and has been noted to correlate with nuclear grade in pure DCIS, albeit with limited samples (60, 61). Similarly, Boland et al. (2004) showed that the median COX-2 expression in DCIS and invasive cancer are significantly greater than in normal breast ducts (p<0.0001), although there was no difference between DCIS and invasive cancer in median expression (p=0.59) (62). However, Ki-67, a proliferation marker already established as a marker of aggression in IBC, did not show any association in DCIS associated with IBC compared to pure DCIS (63). While these markers show limited power individually, combining these factors may have a better prognostic power for predicting DCIS progression (54).

The most fundamental hallmark of cancer cells is their ability to sustain cell proliferation and survival. Several genes and pathways may influence DCIS progression by affecting cell proliferation in high-risk cases, including Aurora Kinase A (AURKA/STK15). This serine-threonine kinase plays a crucial role in centrosome duplication and mitotic regulation. In a 2022 study, Miligy et al. investigated the levels of Aurora Kinase A in a cohort consisting of pure DCIS (n=776) and DCIS associated with invasive breast cancer (n=239) (64). For patients with pure DCIS treated with breast conserving surgery only, high AURKA expression was associated with both local and invasive recurrence (HR=6.9, 95%CI=2.7-7.0, p<0.0001) and even when restricting to invasive recurrence only (HR=3.9, 95%CI=1.7-7.1, p=0.001) (64).

Dettogni et al. (2020) found that genes related to the cell cycle, Hedgehog signaling, and Wnt pathways were associated with the aggressiveness of DCIS through gene expression profiling of normal breast tissue, pure DCIS, and DCIS occurring alongside invasive breast cancer (65). Supporting the notion of early gene expression changes, pure DCIS showed the most altered gene expression profile compared to normal breast tissue. Additionally, three differentially expressed genes, fibroblast growth factor 2 (FGF-2 ; fold change 1.5, p=0.004), growth arrest-specific 1 (GAS1 ; fold change 2.67, p=0.007), and secreted frizzled related protein 1 (SFRP1; fold change 2.61, p=0.001), were identified by comparing pure DCIS with DCIS co-occurring with IBC, suggesting their possible role in the acquisition of invasiveness in DCIS (65). In another study, Eslarraj et al. (2020) found that the RNA and protein levels of B cell lymphoma 9 (BCL9), which plays a role in the canonical Wnt signaling, was significantly elevated during the time of DCIS progression to invasive disease (66). Furthermore, they showed that BCL9 interacts with Signal transducer and activator of transcription 3 (STAT3), affecting the expression of STAT3-regulated gene targets such as integrin β3 (ITGB3) and matrix metallopeptidase 16 (MMP16). To demonstrate the role of BCL9 in DCIS progression, they showed that rosemary extract, which attenuates BCL9/β-catenin dependent transcription, resulted in loss of BCL9 expression and subsequent down-regulation of targets integrin β3 and MMP16, causing a dose-dependent reduction in the number of invasive lesions (p=0.00924) (66).

Guo et al. (2019) described the MAPK-interacting serine/threonine-protein kinase 1 (MNK1)/nodal growth differentiation factor (NODAL) signaling axis as a promoter of DCIS progression through the promotion of cell proliferation, induction of cancer stem cell properties, and invasion (67). They found that MNK1 activity is elevated in high-grade DCIS and IDC compared with low-grade DCIS patient samples. Knockout of MNK1 using CRISPR/Cas9 suppresses NODAL expression, which in turn prevents IDC conversion and reduces tumor relapse and metastasis (67). Additional evidence for the role of the MNK/NODAL signaling pathway in DCIS progression was provided by the finding that the MNK1/2 inhibitor SEL201 effectively blocked DCIS progression in vivo (67). NRAS, an upstream regulator of MAPK signaling that promotes cell proliferation and survival, is also associated with low ER and high Ki-67 levels, markers indicative of a higher risk of DCIS progression (68). Over-expression of oncogene NRAS induces a basal-like gene signature in DCIS cells and may drive the emergence of basal IBC, highlighting its potential as a therapeutic target to prevent DCIS progression (68). One of the ways RAS signaling pathway has been a critical player in the growth-promoting pathways of breast tumorigenesis is through its role in the HER2/Neu/ epidermal growth factor receptor (EGFR)/RAS signaling. Behling et al. (2013) showed that expression of Siah E3 ubiquitin ligase 1 (SIAH1), a “gatekeeper” for the HER2/EGFR/RAS signal transduction, is significantly higher in DCIS cases associated with recurrence compared to those without recurrence (p <0.0001) and also higher in primary DCIS with invasive recurrence (p=0.036) (69). In addition, SIAH1 expression in DCIS (OR=1.31, 95%CI=1.06-1.62, p=0.013) and adjacent normal breast tissue (OR=1.1, 95%CI=1.03-1.16, p=0.003) were both significant predictors of disease recurrence with a sensitivity of 76% and specificity of 79% (69).

Impaired DNA repair results in genomic instability, which is another of the hallmarks of cancer. Al-Kawaz et al. (2021) found high expression of DNA repair protein flap structure-specific endonuclease 1 (FEN1) to correlate with cancer progression (70, 71). Normal breast tissues express low levels of FEN1, pure DCIS shows relatively higher expression, and DCIS co-occurring IBC exhibit high levels of FEN1. Moreover, high expression of FEN1 associates with aggressive and high-risk features such as higher nuclear grade, proliferation index, HR negativity, and triple-negative phenotype. Conversely, DNA polymerase β (POLβ), another key player in DNA repair, shows a gradual reduction in expression from normal breast tissue to DCIS and lowest expression in IBC. Low POLβ associate with aggressive DCIS features and serves as an independent predictor of recurrence (HR=0.490, 95%CI=0.256-0.936, p=0.031) (70, 71).

Finally, epithelial-to-mesenchymal transition (EMT) is one of the major events in cancer tumorigenesis that allows invasion and metastasis, a hallmark of cancer. A study by Treekitkarnmongkol et al. showed that epigenetic activation of SRY-box transcription factor 11 (SOX11) is associated with recurrence and progression of DCIS to IBC (72). SOX11 was noted to promote an epithelial/mesenchymal tropism, and it is associated with increased distant metastasis in IBC (73). First, they showed that SOX11 was detected only in DCIS lesions compared to matched normal breast tissue, and this was correlated with high DCIS grade and recurrence score. Further, they showed that activation of a serine/threonine kinase AKT, an upstream regulator of SOX11, correlated with chromatin accessibility and enrichment of histone modifying enzyme enhancer of zeste homolog 2 (EZH2) on SOX11 promoter thereby up-regulating SOX11 expression. They used human cell lines and mouse models of DCIS disease progression and validated it in DCIS cohorts, which showed a statistically significant difference in overall relapse-free survival (RFS) between SOX11+ and SOX11− DCIS (HR=1.9, 95%CI=1.2-2.9; p=0.003) (72).

Gene expression changes associated with myoepithelial layer and stromal alterations. The myoepithelial layer serves not only as a contracting unit for milk secretion but as a structural barrier between mammary epithelial cells and the surrounding stroma. There is no surprise that in DCIS, a confined neoplastic lesion, the integrity of myoepithelial layer prevents it from being a full-blown invasive lesion. There have been several studies that provide evidence for the role of the myoepithelial layer in preventing in situ tumors from invasion. Myoepithelial cells lose differentiation markers such as tumor protein 63 (p63), calponin-1, and α-smooth muscle actin (α-SMA) prior to losing their structural integrity. Understanding the molecular signatures of normal myoepithelial cell differentiation can provide useful information about predicting high risk DCIS.

An early histopathologic study by Man et al. (2003) noted that 42.7% of DCIS cases contained disrupted myoepithelial cell layer and these were associated with higher proliferation rates and loss of ER expression on adjacent cells in the ducts (74). They also noted different patterns of loss of heterozygosity and microsatellite instability in the ER-negative compared to ER-positive epithelial cells in the same ducts overlying the disrupted myoepithelial cells. These suggest interaction changes causing degradation of the myoepithelial layer and clonal expansion and invasion of cells in these disrupted foci (74). Hu et al. (2008) also studied the role of myoepithelial cells and surrounding fibroblasts in the progression of in situ lesions using human DCIS and primary breast tumors and their results further support the role of myoepithelial layer in DCIS progression (75). Based on their immunohistochemical analysis and gene expression profiling, they identified extensive crosstalk among tumor growth factor-β (TGF-β), hedgehog (Hh), cell adhesion, and p63 signaling pathways in DCIS cells and demonstrated critical role of TGF-β and Hh pathways in DCIS progression since a decrease in these pathways resulted in loss of myoepithelial cells and accelerated progression to invasion (75).

Loss of smooth muscle actin (SMA)-related protein calponin has also been reported to be associated with DCIS progression in mouse models (76). In line with these findings, p63+TCF7+ myoepithelial cells are also decreased in the normal breasts of breast cancer gene (BRCA)1 and BRCA2 germline mutation carriers compared to non-carriers suggesting these molecular alterations are indicative of higher risk of breast cancer (77). In agreement with the role of p63 and transcription factor (TCF)7 in higher breast cancer risk in general, high-risk DCIS also exhibit lower numbers of p63+TCF7+ myoepithelial cells (77). These findings support the role of p63 and TCF7 in normal differentiated myoepithelial cell phenotype and their loss may contribute to the increased breast cancer risk of BRCA mutation carriers. This can also potentially lead to the loss of differentiated myoepithelial cells in DCIS promoting progression to invasion. As DCIS progresses to an invasive phenotype, breast myoepithelial layer stretches and leads to increase in duct size. This mechanical expansion of myoepithelium is characterized by myoepithelial expression of cell integrin β6 and periductal expression of extracellular matrix protein fibronectin (78, 79). These integrin β6 positive aberrant myoepithelial cells have been shown to promote the invasive progression of DCIS into IDC via TGF-β and matrix metalloproteinase 9 (MMP9) signaling activation (80). Approximately 96% of DCIS cases that had concurrent IBC have been found to express αvβ6-integrin, whereas 52% of the non-high grade DCIS and 69% of high grade DCIS express αvβ6-integrin (81).

Loss of caveolin 1 (Cav-1) in the fibroblasts, a key component of the surrounding stromal microenvironment of myoepithelial layer, predicts early DCIS progression to invasive breast cancer (82). When down-regulated, Cav-1 is associated with DCIS recurrence and progression to IBC (HR=3.569, p=0.000013) (82). Martinez-Outschoorn et al. (2010) demonstrated that breast cancer cells induced Cav-1 down-regulation in adjacent stromal fibroblasts via a paracrine mechanism by causing autophagic/lysosomal degradation (83). Furthermore, co-culture also caused normal fibroblasts to acquire a cancer-associated fibroblast phenotype characterized by Cav-1 down-regulation, increased expression of myofibroblast markers and extracellular matrix proteins, and constitutive activation of TGF-β/Smad2 signaling. Interestingly, chloroquine was found to inhibit the degradation of Cav-1 and thus seems like a promising drug for the prevention and treatment of DCIS. Clinical studies that are testing the efficacy of chloroquine such as the Preventing Invasive Breast Neoplasia with Chloroquine (PINC) Trial (ClinicalTrials.gov ID NCT01023477) showed measurable reduction in proliferation of DCIS lesions and enhanced immune cell migration into the duct. Loss of stromal Cav-1 is often accompanied by the gain of stromal monocarboxylate transporter 4 (MCT4) when matched in situ and invasive breast cancer samples are compared (75%, p<0.0001) (84). Progression of DCIS to invasive state is characterized by loss of Cav-1, which causes metabolic reprogramming of stromal cells to enhance the expression of MCT4 that plays a role in aerobic glycolysis and energy metabolism. This is needed to support the increased cell proliferation and growth of surrounding epithelial cells and in turn increased need for energy (84).

A large-scale transcriptomic study by Rebbeck et al. (85) modelled a continuous trajectory of the gene expression patterns of DCIS lesions and noted a progressive loss in basal layer integrity heading towards IDC supporting the study of Man et al. (74). This was coupled with expression of genes associated with EMT not only at the transition to invasive disease but even much earlier in the process, and finally, followed by an increase in cell proliferation. They identified calcium/calmodulin dependent protein kinase II inhibitor 1 (CAMK2N1), motor neuron and pancreas homeobox 1 (MNX1), adenylyl cyclase 5 (ADCY5), homeobox C11 (HOXC11) and ankyrin repeat domain 22 (ANKRD22) as potential biomarkers, whose reduced expression is associated with the progression of DCIS to invasive breast cancer (85).

Gene expression panels and prognostic scoring systems in predicting DCIS progression. In the recent years, there have been several studies that investigated panels of genetic markers for DCIS progression, in part due to the advent of improved technologies for multi-omic studies. With this, the focus has shifted to finding the panels of genetic or gene signatures to stratify DCIS recurrence and progression as compared to individual biomarkers as these panels may better account for the heterogeneity of breast cancer pathogenesis. Currently, there are DCIS scoring systems that have been clinically validated in patients with DCIS: Oncotype DX DCIS score, DCISion RT Decision score, and Oncotype 21-gene recurrence score. Oncotype DX uses a panel of 12 genes including seven cancer related genes corresponding to a cell proliferation group and progesterone receptor (PR), with five reference genes. In the ECOG E5194 study group population, the DCIS Score was significantly associated with developing an ipsilateral breast event (IBE) (HR=2.31, 95%CI=1.15-4.49; p=0.02) and an even higher HR for invasive breast disease (HR=3.68, 95%CI=1.34-9.62; p=0.01) (86, 87). This was also validated in an Ontario population-based DCIS cohort where an HR of 1.78 (95%CI=1.31-4.41, p=0.04) (88) was noted. DCISionRT (PreludeDx) uses markers including COX-2, FOXA1, HER2, Ki-67, PR, and SIAH2 with four clinicopathologic features (age, size, margin status, palpability). This was validated in a study population of 526 and showed that a biological signature driven decision score was correlated with 10-year IBC (HR=3.1, 95%CI=1.5-6.5; p=0.003) and IBE risks (HR=1.9, 95%CI=1.1-3.1, p=0.016) (89). Finally, the Oncotype 21-gene recurrence score (RS), which is a panel of 21 different genes related to proliferation, invasion, HER2, and estrogen with 5 reference genes that have been validated in invasive breast cancer, was also tested in DCIS (90). Their results showed that women with high recurrence score treated by breast conserving surgery alone had a 1.8-fold increase in risk of invasive breast recurrence and a 21.7% cumulative 20-year risk of invasive recurrence compared with a low RS; hence helping identify which individuals RT can be given to reduce risk of death from IBC (88). These genomic scoring tools represent a significant advancement in the personalization of treatment decision-making for patients with DCIS.

Immune System in DCIS Progression

The interplay between cancer cells and immune cells plays a pivotal role in cancer development and progression. According to the cancer immunoediting theory that describes the role of immune system in cancer development, the initial elimination phase of immunosurveillance involves pro-inflammatory immune cells eliminating aberrant preneoplastic cells to prevent cancer development. During the equilibrium phase, characterized by impaired immunosurveillance, new genetic variants of cancer cells evolve, allowing these cells to survive. This impaired immune environment fails to completely eradicate the tumor cells but keeps tumor growth in check. Finally, in the escape phase, tumor cells form a palpable mass, create an immunosuppressive environment, and completely evade immunosurveillance (91).

A key feature of DCIS progression is the loss of the surrounding ductal epithelium, which limits the exposure of ductal cancer cells to surrounding immune cells. Understanding these critical immune alterations that occur during the progression of preinvasive DCIS to invasive breast cancer will help differentiate high-risk DCIS from those with a low risk of progression. DCIS and IBC are characterized by distinct immune cell compositions and their functional states. Numerous studies have shown that the immune environment is largely suppressed in invasive breast cancer. However, studies focused on characterizing the immune environment in DCIS are rather limited. Invasive breast cancer is enriched in T helper cells, regulatory T cells, macrophages, B cells, and PD-L1+ T cells compared to DCIS (92, 93). This suppressive immune environment in invasive breast cancer is associated with more aggressive subtypes like TNBC and is a predictor of poor prognosis. Conversely, anti-tumor CD8+ cytotoxic T cells, Type 1 helper T cells (Th1), M1 macrophages, and natural killer (NK) cells are associated with better breast cancer prognosis. Research efforts aimed at determining whether the immune environment of DCIS is associated with disease progression have yielded conflicting results, as outlined below.

Tumor-infiltrating lymphocytes (TILs) in DCIS progression. Multiple studies have measured the levels of tumor-infiltrating lymphocytes (TILs) in H&E-stained archived tissue sections of DCIS to assess their power in predicting the high risk of DCIS progression. Despite their favorable prognosis in IBC, the significance of TILs in DCIS progression is less clear, with studies displaying contradictory results.

A number of studies reported that high levels of TILs in DCIS are largely associated with a higher chance of recurrence and progression to IBC (94-100). Recently, Li et al. analyzed a large dataset with 718 patients and found that higher TIL density is associated with several adverse prognostic indicators in DCIS, such as larger lesion size, higher cytonuclear grade, the presence of necrosis, ER negativity, and HER2 over-expression (100). DCIS with high TIL numbers were also found to have a three-fold higher risk of developing ipsilateral IBC than DCIS with low TIL levels (100). Thompson et al. also discovered that more aggressive ER-negative DCIS exhibited higher numbers of all TIL subsets compared to ER-positive DCIS (101). In ER-positive DCIS, the immune environment was more tumor-protective, as indicated by a higher ratio of CD8/FoxP3 than in ER-negative DCIS. Although TILs are more prevalent in the more aggressive ER− and HER2+ DCIS, this increase in TILs has failed to significantly associate with DCIS recurrence in several studies (102, 103). Some of these conflicting results could be due to the simplistic nature of TIL scoring on H&E slides without consideration of the TIL type and functional state.

Recent studies have used techniques like multiplexed immunofluorescence and a combination of global profiling with single-cell methods to distinguish various cell types within tissue-infiltrating lymphocytes. In one such study performed by Alcazar et al., aimed at delineating the co-evolution of cancer cells and the immune microenvironment during tumor progression using matched DCIS and recurrent IBC samples, it was shown that there was a decrease in activated GZMB+CD8+ T cells in HER2+ and triple-negative invasive breast cancer compared to DCIS (104). This was particularly evident in DCIS cases that recurred locally as IBC, implying that decreased immune activity, particularly of activated T cells, may be necessary for invasive progression (104). Furthermore, T-cell receptor clonotype diversity was significantly higher in DCIS than in invasive breast cancer, indicating a more robust immune system in DCIS relative to IBC (104). In addition to the lower levels of protective cytotoxic T cells in recurrent IBC, there was an increase in co-inhibitory immune checkpoint receptor programmed death-ligand 1(PD-L1) and cytotoxic T-lymphocyte associated protein 4 (CTLA4)-expressing T cells in triple-negative subtype IBCs (104). Although DCIS tumor epithelial cells express low levels of PD-L1, high proportions of PD-L1+ TILs have been observed in high-grade DCIS (101). Levels of immunosuppressive T cells, such as Tregs, vary between DCIS and IBC (92). Regulatory T cells (Tregs), characterized by the presence of forkhead box P3 (FOXP3), are suppressive immune cells that increase during breast tumor progression, suggesting that this could be used to predict the risk of invasive progression. However, FOXP3+ T cells have not been consistently detected in DCIS. However, T cell immunoreceptor with Ig and ITIM domains (TIGIT)-expressing T cells are more frequent in triple-negative DCIS (104).

Macrophages. Chronic inflammation plays a pivotal role in cancer development and often associates with a higher risk of breast cancer progression. A wide variety of secretory molecules and immune cell types including macrophages form this inflammatory tumor microenvironment. Tumor-associated macrophages (TAMs) make up the majority of tumor cell contents and correlate with poor cancer prognosis and low efficacy of checkpoint inhibitor therapy in cancer patients (50, 105-108). TAMs are immunosuppressive and functionally similar to M2 macrophages. M1 macrophages, however, are pro-inflammatory and tumor inhibitory. Current understanding in the field points towards a theory that an immunosuppressive environment in the tumor leads to the generation of tumor-associated macrophages (TAMs), which in turn secrete factors that further promote cancer progression (109).

An increase in both pro- and anti-tumor macrophages has been reported in a p53-null mouse model of early breast cancer progression (110). Macrophages were found to be recruited to high-risk ductal hyperplasias, where they are differentiated and polarized toward a tumor-promoting phenotype (110). Similarly in human DCIS, macrophages are more abundant in high-grade and cribriform DCIS than in low/intermediate-grade DCIS (111). The prognostic relevance of macrophages in DCIS is also highlighted by studies that report higher macrophage density in DCIS to be associated with poorer prognostic parameters (112). Importantly, higher stromal cluster of differentiation (CD) 163 + macrophage density, indicative of M2 phenotype, predicted both recurrence and ipsilateral invasive recurrence (112). In contrast, Hoskoppal et al. did not find any correlation between intratumoral CD163 content, as measured by IHC staining, and DCIS grade (113). In this study, although a trend for higher stromal CD163 expression was observed with higher-grade DCIS, it was not statistically significant. The authors point towards the lack of uniform criteria for the interpretation of high vs. low CD163 staining and low interobserver correlation as a possible limitation of data analysis (113).

A multi-omics study that inferred immune cell composition from RNA sequencing of young-onset DCIS samples and paired normal breast tissue/blood samples by applying deconvolution analyses revealed that the majority of these high-risk DCIS were “immune hot” (114). These immune-hot DCIS were characterized by features such as an increase of M1 macrophages, CD4+ memory resting T cells, CD8+ T cells, memory B cells, CD4+ memory activated T cells, naïve B cells, and M2 macrophages compared to the “cold” group. In addition, there was higher expression of Programmed cell death protein 1 (PDCD1) and cytotoxic T-lymphocyte antigen 4 (CTLA4), which suggests that immune escape mechanisms may contribute to the malignant characteristics of “immune hot” DCIS. Overall, “immune hot” DCIS that were HER2 or basal-like subtypes exhibited a combination of activated but immunosuppressive gene signatures indicating an active yet evasive immune response (114).

Understanding the molecular mechanisms that cause macrophage homing and their polarization to a protumor phenotype will help differentiate high-risk DCIS and potentially prevent DCIS progression. Chemokines are secretory molecules that play a role in the recruitment of immune cells during inflammation and cancer. The interplay of chemokine C-C motif chemokine ligand 2 (CCL2) and its receptor, C-C motif chemokine receptor 2 (CCR2) signaling in breast cancer cells has been found to regulate macrophage recruitment and promote breast cancer progression. CCL2 mediates the progression of DCIS through the phosphorylation of MET proto-oncogene, receptor tyrosine kinase (MET), enhancing cancer cell proliferation, survival, migration, and glycolysis (115).

Fibroblasts. Since cancer is considered a wound that does not heal, similar to a wound, tumor cells release and express several factors that promote clotting. This procoagulant environment, created by fibroblast-dependent mechanisms, further supports cancer progression. Normal fibroblasts, which are crucial for maintaining the structural and functional integrity of normal breast tissue, are replaced by cancer-associated fibroblasts during cancer progression. This phenotypic switch occurs at the preinvasive stage, as evidenced by increased fibroblast expression of tumor-promoting clotting activation markers such as tissue factor (TF), thrombin, and protease-activated receptor (PAR)1 and PAR2 (116), in DCIS and IBC compared to normal breast tissue. The fibroblast procoagulant phenotype also correlates with aggressive breast cancer subtypes and reduced survival (116). Over-expression of fibroblast activation protein-alpha (FAP-α) (alone or in combination with Golgi phosphoprotein 3 in carcinoma cells) in stromal fibroblasts is highly predictive of DCIS recurrence and progression into invasive breast cancer (117). The immunophenotype conversion of stromal fibroblasts from CD34(+) α-SMA(−) FAP-α(−) in DCIS to CD34(−) α-SMA(+) FAP-α(+) in IBC may also serve as an important determinant of DCIS with microinvasion (118).

Cancer-associated fibroblasts constitute an important component of the tumor immune microenvironment in solid tumors, modulating the transition from DCIS to IDC through their secretion of factors that modify the surrounding stromal matrix (119, 120). An interplay between cancer cells and cancer- associated fibroblasts (CAFs) seems to be involved in breast cancer progression. Croizer et al. reported that cancer cells drive the transition of immuno-protective CAFs towards immunosuppressive extracellular matrix (ECM)-producing CAFs via a dipeptidyl peptidase 4 (DPP4)- and a transcriptional coactivator Yes-associated protein (YAP)-dependent mechanism (121). In turn, these immunosuppressive ECM-CAFs polarize triggering receptor expressed on myeloid cells 2 (TREM2)+ macrophages and regulatory NK and T cells to induce an immunosuppressive cell state. Similarly, immuno-protective CAFs are associated with folate receptor beta (FOLR2)+ macrophages. In nutshell, inflammatory and myofibroblastic CAF clusters are differentially localized within breast tumors and may potentially predict invasive recurrence of DCIS (121).

CCL2/CCR2 signaling between fibroblasts and breast epithelial cells has emerged as another mechanism of DCIS progression. Brummer et al. found that fibroblasts derived from DCIS patient samples accelerate progression from DCIS to IDC through CCR2-dependent mechanisms, mediated by up-regulation of aldehyde dehydrogenase 1 family member A1 (ALDH1A1) and down-regulation of high-temperature requirement protein A2 (HTRA2) in DCIS cells (122). There is also evidence of interleukin-6 (IL-6)-mediated paracrine crosstalk between preinvasive human DCIS cells and human breast CAFs in the progression to invasive breast carcinoma in 3D cultures (123).

To test whether the overall state of tumor microenvironment can predict DCIS cases that are likely to metastasize, formalin-fixed, paraffin-embedded (FFPE) sections of 117 high-grade DCIS cases with a history of recurrence and non-high-grade DCIS cases were analyzed (124). The study revealed that high-risk DCIS features were associated with high levels of FoxP3+ T cells, CD68+ and CD68+ proliferating cell nuclear antigen (PCNA+) macrophages, human leukocyte antigen – DR isotype (HLA-DR)+ cells, CD4+ T cells, CD20+ B cells, and total TILs compared to non-high-grade DCIS. Additionally, CD8+HLADR+ T cells, CD8+HLADR- T cells, and CD115+ immune cell populations were associated with the risk of DCIS recurrence (124).

A comprehensive study by Almekinders et al. screened more than 10,000 patients and investigated the prognostic value of immune cell characteristics for predicting the risk of ipsilateral IBC (iIBC) in 141 patients with pure DCIS and matched iIBC, followed up over a median of 12 years (125). Of these, 77 women went on to develop iIBC. The study reported that high stromal immune cell density (T cells, activated T cells, Tregs, B cells, and macrophages) correlates with factors (ER negativity, HER2 positivity, Ki-67, grade, and periductal fibrosis) known to be associated with invasive recurrences. However, none of these immune factors were found to be independent predictors of the risk of subsequent iIBC after the diagnosis of primary, pure DCIS (125).

Interplay Between Myoepithelium and Immune Environment

Myoepithelium is a gate keeper that restricts the tumor cell escape in DCIS and their exposure to surrounding immune cells. Disruption of the myoepithelial layer, that forms a physical barrier, is a hallmark of progression of DCIS to invasive breast carcinoma.

Risom et al. constructed a 2D spatial atlas of breast cancer progression to identify features in primary DCIS associated with the risk of invasive relapse using human breast tissues that captured the full spectrum of breast cancer progression, from normal tissue to DCIS and matched ipsilateral IBC recurrences (126). The authors found that coordinated transformation of ductal myoepithelium and surrounding stroma plays a central role in cancer progression by establishing a tumor-permissive niche that favors local invasion. Surprisingly, contrary to the widely accepted role of myoepithelium in breast cancer, Risom et al. found DCIS samples with more intact myoepithelium and high E-cadherin expression to be of higher risk of ipsilateral invasive recurrence following primary DCIS resection (126). Conversely, thin, broken, low-E-cadherin expressing myoepithelium present in non-progressor tumors was correlated with a more reactive desmoplastic stroma with more immune cells, CAFs, and collagen remodeling. The authors acknowledge that the 2D nature of the analyses and the small dataset of matched DCIS/IBC samples are key limitations of the study, pending further direct cause-effect causal studies on myoepithelial integrity and changes in the immune microenvironment (126).

Nonetheless, a compromise in myoepithelial barrier function appears to be enough to expose tumor cells to immune surveillance, eventually leading to immune cell exhaustion. Supporting this concept, Mitchell et al. found that a subset of clinically and pathologically defined low-risk DCIS lesions had lost calponin-1 (a marker of differentiated myoepithelium) and had exhausted CD8+ T cells, similar to the majority of high-risk DCIS lesions (127). These observations suggest that this subset of DCIS lesions has a high risk of progression (127).

Epigenetic Factors

Environmental factors and lifestyle increase the risk of breast cancer, as suggested by 5-fold higher incidence rate of breast cancer in women in Western countries relative to Asian women. The impact of these epigenetic factors becomes clearer with a subsequent increase in breast cancer rate of Asian women living in Western countries (128). Environment can shape gene expression and health outcomes through heritable epigenetic mechanisms such as DNA methylation. DNA methylation along with polycomb repressor complex and histone modifications determine the overall state of chromatin (repressive or activated) and ultimately determine the state of cell plasticity in normal cellular development and cancer. Global hypomethylation is a general feature of cancers including breast cancer (129), however, hypermethylation of multiple gene promoters has been reported. These defects in DNA methylation lead to silencing of tumor suppressors and other genes involved in DNA damage repair, cell survival, and cell proliferation through hypermethylation of their promoters (130).

Numerous studies provide evidence that DNA methylation of gene promoters is an early event, and a majority of DNA methylation changes have already occurred prior to invasive breast carcinoma (131-134). In support of this model, Fleischer et al. found DNA methylation profiles of DCIS to be radically different compared with normal breast tissue (more than 5,000 genes) while the changes between DCIS and IBC were comparably modest (about 1,000 genes) (135). Interestingly, the affected pathways suggested that many of the changes may not occur in the tumor, but in infiltrating cells suggesting epigenetic reprogramming of immune cells. Moelans et al. also found DCIS to be epigenetically as advanced as IDC as no additional changes in promoter methylation between DCIS and IDC were found in a set of 25 analyzed genes (136). Although these findings point towards DCIS to be as advanced as IBC in term of epigenetic changes, higher grade DCIS was not found to have significantly higher methylation levels of glutathione s-transferase pi 1 (GSTP1) and CDKN2A (p16) compared to low grade DCIS (136). In support of the role of DNA methylation in high DCIS grade and in predicting of its malignant potential, Hoesel et al. also found hypermethylation of MINT31 to be associated with unfavorable lesion characteristics such as ER negativity in DCIS (134). In addition, hypermethylation of adenomatous polyposis coli (APC), CCND2, cadherin 1(CDH1), retinoic acid receptor beta (RARB) have been associated with higher nuclear grade of DCIS. Additionally, hypermethylation of ATP binding cassette subfamily B member 1 (ABCB1), forkhead box C1 (FOXC1), Ras association domain family member 1 A (RASSF1A), cilia and flagella associated protein (DLEC1), protein phosphatase 2 regulatory subunit B beta (PPP2R2B), and phosphatase and tensin homolog (PTEN) have been linked with the unfavorable characteristics of DCIS such as HER2 amplification, Ki67 index, and TP53 mutation (137-140). Fleischer et al. also found a prognostic signature based on methylation level of 18 cytosine-phosphate-guanine (CpGs) to be associated with DCIS patient prognosis (135). To determine which DCIS is likely to progress, it is also imperative to consider its molecular subtype and take a subtype specific approach to obtain prognostic information. Bergholtz et al. studied gene expression, DNA methylation, and copy number analysis in 57 pure DCIS and 313 IBC cases and found basal-like DCIS tumors to be highly similar to IBC of Luminal A subtype, in terms of higher degree of differentiation, lower proliferation and lower genomic instability than basal-like IBC (141).

Although these studies suggest correlation between methylated genes and DCIS aggressiveness, studies that assessed the prognostic power of methylation to predict clinical outcome are limited. In one such longitudinal study Johnson et al. profiled 40 ER positive patient with DCIS of which 13 patients progressed to IBC in order to identify methylation alterations that contribute to progression of DCIS (142). Homeobox-containing genes such as polycomb group gene targets and genes involved with limb morphogenesis were found to be altered (142). Interestingly, these methylation alterations correlated strongly with their gene expression in an independent set of early stage ER positive breast cancers suggesting that differential methylation of these sites may directly contribute to an increased risk of invasion through aberrations in gene expression (142).

LNCRNAs

LncRNAs have been widely reported to play an instrumental role in gene regulation and cancer progression. Multiple lncRNAs have been reported to be associated with breast cancer and correlate with aggressiveness of DCIS such as HOX transcript antisense RNA (HOTAIR), HOTAIRM1, long intergenic non-protein coding RNA (LINC)00885, LINC01011 and LINC01024. Laboratory evidence suggests that LINC00885 and HOTAIR over-expression induces premalignant phenotypic changes in normal breast epithelial and DCIS cells by increasing cell proliferation, motility, and migration, and altering 3D growth (143, 144). Imprinted gene in the Prader-Willi syndrome region (IPW) is another lncRNA that is down-regulated in DCIS and its ectopic expression inhibits DCIS growth (145). However, it remains to be tested whether the levels of these lncRNA can predict the prognosis of patients with DCIS.