1. Role of Genetics and Genomics in Primary Breast Cancer
1.1. Overview
Current knowledge indicates that only a select number of driver genes are routinely implicated in primary BC. These include, with alteration frequencies of about 30%,
PIK3CA and
TP53, and, with amplification frequencies around 15%,
ERBB2,
FGFR1, and
CCND. Nearly all primary BCs consist of a dominant subclone (>50% of tumor cells), commonly comprised of
TP53 and
PIK3CA mutations
[1]. In basal-like tumors,
TP53 loss of function mutation occurs at frequencies as high as 80%, rendering it a potential target of interest. Successful therapy against
HER2 amplifications has established a precedent for identifying gene targets in cancer subtypes, such as basal-like cancers, where chemotherapy is the only established medical treatment
[2]. Sequencing the exons of 173 genes in 2433 primary breast tumors likewise identified
PIK3CA and
TP53 mutations in 40.1% and 35.4% of samples, respectively. Just five other genes—
MUC16,
AHNAK2,
SYNE1,
KMT2C, and
GATA3—were found in at least 10% of the samples. Notably, several different cancers harbor
MUC16,
AHNAK2, and
SYNE1 mutations, although their significance is still unknown
[3]. Additionally, the mapping of all mutations in 31 tumor types to variant interpretations merged from three genetic knowledge bases showed BC to have the greatest fraction of patients who might benefit from existing investigational targeted therapies, due to frequent mutations of
AKT1,
ERBB2, and
PIK3CA [4]. Important to note, however, is that many driver mutations have low allele fractions, and even uncommon driver mutations may be clinically relevant therapeutic targets
[1].
1.2. Other Genes of Interest
Multiple copy number alterations have also been identified in HER2+ BC, including loss of
PTEN and
INPP4B. PI3K pathway inhibitors may prove potential therapeutic targets in BCs harboring these deletions
[5]. Identified mostly in basal-like/triple-negative BC, INPP4B is a novel tumor suppressor and inhibitor of PI3K/Akt signaling and cell survival in ER+ BC and, in one study, was linked to loss of
PTEN in BC subtypes with poor prognosis
[6]. Moreover, due to high mutational frequency,
PIK3CA activated kinase or signaling pathway inhibitors may be helpful in the treatment of luminal/ER+ cancer
[7].
In a genomic analysis of 2000 breast tumors, Curtis et al. identified heterozygous and homozygous deletions in
PPP2R2A (8p21, region 11),
MTAP (9p21, region 15), and
MAP2K4 (17p11, region 33), which have also been observed in numerous other cancers.
PPP2R2A is a B-regulatory subunit of the PP2A mitotic exit holoenzyme complex, and its absence has also been detected in clear cell ovarian, endometrioid, and colorectal cancers
[8]. PP2A phosphatase is a serine/threonine phosphatase tumor suppressor controlling mitosis, cytokinesis, and other aspects of the cell cycle. While previous studies have reported breast tumor growth and poor outcomes associated with PP2A inhibitory protein overexpression and
PP2A inactivation, Watt et al. demonstrated that loss of specific PP2A regulatory subunits is of likely functional significance in breast tumorigenesis
[9]. Thus, activation of
PP2A and modulation of the enzymes involved in PP2A’s suppression and inactivation could serve as potential targets for the prevention of aberrant cell cycle progression and chromosome division.
MTAP is often co-deleted with
CDKN2A and
CDKN2B, and the enzyme MTAP plays a role in adenine and methionine salvage from endogenous MTA. Consequently, cells deficient in MTAP are more susceptible to de novo purine synthesis inhibitors and methionine starvation
[10]. More recent studies have identified MAT2A as a viable target in
MTAP-deleted cancers, and clinical development has begun for the MAT2A inhibitor, AG-270
[11][12]. Stephens et al.
[13] identified somatic
MAP3K1 mutations in 6% of mostly ER+ BC tumors. MAP3K1 is involved in cell migration, mitosis, and apoptosis via caspase activation
[13][14]. Additionally,
MAP3K1 and the more common
PIK3CA co-occur in approximately 11% of
PIK3CA mutant tumors and, thus, may serve as a biomarker in PI3K pathway inhibitor trials
[15].
Of recent, but lesser-known, interest is the long non-coding RNA (lncRNA) homeobox
MNX1 (7q36.3), which encodes for the transcription factor motor neuron and pancreas homeobox 1 (MNX1). lncRNAs are highly sensitive regulators of cancer growth, and MNX1′s expression has been found to be upregulated in not only BC, but also in all other cancers, although its exact contribution to tumorigenesis is uncertain
[16][17]. Tian et al. found elevated expression of MNX1 in patients with larger tumor size, more extensive lymph node involvement, and poorer prognosis, thus supporting
MNX1′s role as a cancer promoter
[17]. Li et al. showed that MNX1 antisense RNA 1 (MNX1-AS1) interacted with and upregulated STAT3′s phosphorylation via enhanced STAT3 and p-JAK interaction five times more in triple negative breast cancer (TNBC) versus normal breast and non-TNBC tissue. An additional in vivo experiment involving nude mice demonstrated that the silencing of MNX1-AS1 reduces tumor growth and lung metastasis, further supporting MNX1-AS1 as a novel therapeutic target in TNBC
[18].
Table 1 provides a non-comprehensive overview of genes discussed in these contents.
Table 1. Several of the genes linked to primary and metastatic breast cancer discussed in these contents. Those presented below are not intended to reflect a comprehensive list, but rather the ones highlighted in these contents and prevalent throughout literature. References for this table include those provided in these contents. Bold indicates genes were commonly involved in more than one site (ERBB2, AKT1), noting that this is a very limited overview.
2. Genomics of Metastasis: Most Common Sites
2.1. Bone Metastasis
Bone is the most common location for BC metastases. Of the main subtypes of BC, luminal hormone positive (HR+) tumors and even more specifically, ER+ tumors, have the strongest propensity for bone metastasis (
Figure 1)
[19]. A study analyzing gene expression in 69 bone metastasis samples and 39 non-bone metastasis samples revealed 69 differentially expressed genes. Five genes demonstrated significantly higher expression in breast cancer bone metastasis (BCBM):
TFF1,
TFF3,
AGR2,
NAT1, and
CR1P1 [20]. The highest-ranking gene expressed,
TFF1, produces a cysteine rich protein that is normally expressed in the gastric mucosa. The overexpression of
TFF1 in ER+ BC was later supported; however, its functional role in BCBM metastasis, if any, is still unknown
[21].
Figure 1. Heterogeneity of BCSCs contributes to the metastatic organotropism of breast cancer.
Additionally, a limited retrospective study on archival tissue of 41 metastatic BCs found increased expression of the chemokine receptors CXCR4 and CCR7
[22]. CXCR4 was exclusively observed in BCBM, whereas CCR7 was also observed in other metastases, albeit expressed in a larger percentage of the bone metastasis cases. CXCR4 was noted to work in close relation to CXCL12, acting as a signal for the chemotaxis and migration of BCBM
[23]. There is also evidence suggesting that an upregulation of SNAI1 is associated with BCBM
[24]. SNAI1 is a zinc finger transcriptional repressor of CDH1, which encodes E-cadherin. Downregulation of E-cadherin is necessary for the dissemination and invasion of cancer cells, loss of epithelial differentiation, and acquisition of a mesenchymal phenotype, which might augment BC metastasis into the bone
[25].
A multigenetic study in mice, utilizing the parental BC cell line MDA-MB-231, was conducted to analyze BCBM’s genetic associations. Subpopulations of the parental cells were organized based upon their preference of metastasis location. Compared to other sites, the subgroups with first metastasis to the bone showed a substantial increase in IL11, CTGF, CXCR4, and MMP-1 expression. When IL11 was expressed alone, it did not show highly metastatic effects, but when in combination with osteopontin, it was highly metastatic to bone. Osteopontin, however, was found in many highly metastatic populations and was not specific to bone. Similarly, with the remaining genes, the individual expression was not overtly metastatic, but instead, the summation of three or four of the genes resulting in a highly bone specified metastatic tumor was found. Although this work was completed in a mouse model, it aids in demonstrating the functional pro-metastatic effect of these specific genes
[23].
Notably, a study of 389 primary BC tumors found no significant somatic mutation associations in BCBM for their cohort of samples. It should be stated, however, that the study analyzed only 46–50 cancer related genes in each tumor sample. Although no genes were found to be specifically associated with bone metastasis, increased expression of
TP53,
PIK3CA, and
AKT, which have been previously implicated in many other BC metastases, was observed
[26].
2.2. Lung Metastasis
As BC lung metastasis (BCLuM) is typically symptomatic only once the lungs are overtaken with secondary tumors, identification of its cellular and molecular processes is critical for treatment development
[27]. The basal-like subtype of BC has a propensity for BCLuM
[28], and another study of 2933 BC patients found that 75.8% of BCs that gave first rise to lung metastasis expressed either HER2 or EGFR
[24]. Minn et al. identified a set of genes that not only promote, but are also clinically correlated with BCLuM. They include
MMP1,
MMP2,
CXCL1,
PTGS2,
ID1,
VCAM1,
EREG,
SPARC, and
IL13RA2. Many of these genes encode for metastasis-promoting extracellular products such as growth and survival factors (the HER/ErbB receptor ligand epiregulin), chemokines (CXCL1), cell adhesion receptors (ROBO1), and extracellular proteases (MMP1). Others, such as ID1 and COX2, encode for transcriptional regulators and intracellular enzymes, respectively
[29].
Overexpressed in ER+ BC cells,
AGR2 may also promote BCLuM via the (de)regulation of tumor cell adhesion and spread
[27]. AGR2′s potential role in tumorigenesis can be attributed to its likely function in protein folding and endoplasmic reticulum-assisted degradation of proteins
[30]. Thus,
AGR2 may facilitate tumor resistance against proteotoxic stress, preventing tumor cell death. With therapies targeting inhibition of AGR2, tumor cells are more susceptible to proteotoxic stress, which helps prevent tumor metastasis and slow its spread
[31].
Additionally, KLF5 is a transcription factor that is expressed in high-grade ER− tumors, such as basal-like BC
[32]. It promotes overall BC cell development, survival, spread, and overall tumor growth
[33] and has been found to promote BC cell proliferation through its target genes
FGF-BP,
mPGES1, p27, and the tumor necrosis factor-α induced gene,
TNFAIP2 [34][35][36][37]. Specifically, TNFAIP2 interacts with the two small GTPases Rac1 and Cdc42, increasing motility via their activities, to alter actin cytoskeleton and cell morphology. Therefore, the proliferation, migration, and invasion of triple negative BC can be stimulated by KLF5-upregulation of TNFAIP2
[34]. KLF5′s function is also stabilized by the deubiquitinase, BAP1. In fact, BAP1 helps promote BCLuM via KLF5 stabilization
[33].
Only a small fraction of cancer cells from the primary tumor successfully metastasizes. What allows these cancer cells to proliferate at a particular site is dependent on that organ’s microenvironment
[38]. In an in vivo mice study, Salvador et al. demonstrated that
Loxl2 facilitates the premetastatic niche, thereby promoting dedifferentiation and tumor invasion. In contrast, it was demonstrated that
Loxl2 deletion in the primary breast tumor led to a marked decrease in lung metastatic burden, whereas its overexpression produced the opposite effect
[39].
2.3. Liver Metastasis
Breast cancer liver metastasis (BCLM) is a complex process, involving specific gene mutations with few known therapeutic options, which often leads to poor outcomes. Liver metastasis is reported in 15% of newly diagnosed BC patients and is a complex multistep process, involving signaling pathways
[40]. The MAPK, NFκB, and VEGF signaling pathways are mechanisms highlighted in regulating BCLM. According to one study
[41], the prognosis following liver metastasis has the second-worst outcome after breast cancer brain metastasis (BCBrM), and about half of patients with metastatic BC eventually develop liver metastasis. BCLM has low expression of immune genes in comparison to other organ sites. BCLM has a very poor prognosis and has a survival time of four to eight months if left untreated, even though new therapies in the last decade have resulted in a yearly 1–2% decrease in mortality rates
[40]. The many signaling pathways and genetic mutations involved in BCLM contribute to the poor prognosis associated with the disease.
A study conducted by Chen et al.
[40] identified MAPK, NFκB, and VEGF as the three most critical molecular pathways in the regulation of BCLM. These three pathways may be responsible for the distinct pathology seen in liver metastasis, due to the high observed gene count in BCLM genomes. The MAPK cascade is the most important mitogenic pathway to human cancer pathogenesis. Activation of MAPK can lead to downstream changes in both protein expression and activity, because of its role in growth factor mediation. In contrast, the NFκB pathway plays a role in apoptosis, allowing cancer cells to initiate cell death in healthy cells. Finally, the VEGF pathway enables cancer cells to induce new blood vessel formation and growth, which allows the tumor to expand and travel throughout the body, leading to metastasis.
Mutations in driver genes
ESR1,
AKT1,
ERBB2,
FGFR4, and the MS APOBEC cytidine deaminases negatively alter cellular mechanisms. Defective DNA mismatch repair further contributes to the progression of liver metastasis.
ESR1 has been discovered as the most mutually exclusive mutant gene pair in liver metastasis
[42].
ESR1 is an estrogen receptor protein coding gene that is a biological indicator of tumor status in BC. Most of these mutations occur in BC cells during metastasis to the liver.
PPFIA1, which has a key oncogenic role in BC, was discovered to also contribute to metastatic relapse, due to its upregulation in BCLM
[43].
PPFIA1 is a protein coding gene that is important in regulating cell migration and invasion. Expression of
PPFIA1 is significantly higher in liver metastatic breast tumors compared to primary tumors.
Rhodes et al.
[44] determined that activation of LKB1 signaling represents the possibility of developing new therapeutics, particularly for patients exhibiting basal-like BC or triple-negative breast disease with low endogenous LKB1 expression. The lack of frequent LKB1 mutations in sporadic BC further supports the possibility for successful therapeutic intervention of normal LKB1 signaling. LKB1 is a kinase that acts as a blockade upstream to many oncogenic pathways, and LKB1 expression could regulate the invasive and metastatic properties of the basal-like BC subtype. Therefore, targeting LKB1 expression by increasing the kinase mechanisms in patients with BCLM could be a possible treatment target, as it can suppress invasion and metastasis of certain BC types.
2.4. Brain Metastasis
The brain is the third most common site for BC metastasis, behind bone and liver
[45]. Epidemiologic studies have found that 10–16% of BC patients have brain metastases, while large autopsy studies have suggested frequencies as high as 18–30%
[46]. Brain metastasis is typically diagnosed within 2–3 years of initial BC diagnosis, and it usually develops after metastasis to the bone, lung, and/or liver
[47]. The median survival for patients with BCBrM is 13 months, and fewer than 2% of patients with BCBrM survive past 2 years
[46].
Each of the PAM50-based BC subtypes exhibits a unique tropism for different metastasis locations, as exhibited by a study
[48] conducted in 2010 on the metastatic behavior of these subtypes. The frequency of BCBrM from each PAM50 subtype examined in this study were: Luminal A = 7.6% (ER+ and/or PR+, Ki-67 < 14%); Luminal B = 10.8% (ER+ and/or PR+, and Ki-67 ≥ 14%); Luminal/HER2 = 15.45% (HER2+, ER+ and/or PR+); HER2 enriched = 28.76% (HER2+, ER−, and PR−); Basal-like = 25.23% (HER2−, ER−, and PR−, EGFR+ and/or CK5/6+); Triple Negative non-basal = 22.06% (HER2−, ER−, PR−, EGFR−, and CK5/6−). The more common HER2+ and Triple Negative are illustrated in
Figure 2 [49].
Figure 2. Heterogeneity of BCSCs contributes to the metastatic organotropism of breast cancer.
Several genes and deeper genomic similarities have been identified in BCs with subsequent brain metastases. For instance, a 2004 study
[50] found that cells from BCs selective for brain metastasis produced higher levels of the angiogenic factors VEGF-A and IL-8 in vitro compared to the parent line of BC cells. A 2014 study
[46] found that BCs with brain metastasis exhibited large chromosomal gains in 1q, 5p, 8q, 11q, and 20q, and chromosomal deletions in 8p, 17p, 21p, and Xq.
Several genes have exhibited overexpression, such as
ATAD2,
DERL1, and
NEK2A.
ATAD2 is believed to be a transcription coactivator of
ESR1, enabling the expression of many estradiol target genes, and may also be required for histone hyperacetylation
[51]. Moreover, ATAD2 is a known cofactor for the oncogene
MYC, which is associated with poor BC outcomes
[52].
DERL1 encodes a member of the derlin family of proteins, which are responsible for the endoplasmic reticulum (ER)-associated translocation of misfolded proteins for proteasomal degradation. BC cells have shown increased DERL1 expression during ER-stress (which is associated with solid tumor progression), while knockout of
DERL1 leads to decreased cancer cell development
[53].
NEK2A encodes a serine/threonine kinase involved in mitotic regulation, which has been found to contribute to the growth potential of ductal carcinoma in situ and invasive ductal carcinoma. Additionally, NEK2A expression is correlated to a higher histological grade and lymph node metastasis
[54].
The
ATM,
CRYAB, and
HSPB2 genes are commonly deleted and/or under-expressed in BCBrM. The
CRYAB and
HSPB2 genes are both members of the multi-gene small heat shock protein family located on 11q23, and their specific biological roles are unclear. However, 11q23 is believed to be a tumor suppressor for many solid tumors, including breast, cervical, ovarian, gastric, bladder carcinoma, and melanoma
[55]. The
ATM gene encodes the ATM serine/threonine kinase, which has a major role in DNA repair, and the absence of ATM expression leads to genomic instability and cancer predisposition
[56].
Certain cellular pathways were also found to be enriched in BC cells with brain metastases. The “IL-8 signaling” pathway demonstrated enrichment, mainly as a result of hypermethylation and downregulation of many genes, including
ANGPT1,
KDR,
ITGAM,
PIK3CG, and
TEK.
BRAF and
BCL2, in contrast, were hypomethylated and overexpressed in this pathway. Similarly, the cellular pathways of “hepatic fibrosis/hepatic stellate cell activation signaling” and “thyroid hormone metabolism signaling” were also frequently enriched. Genes involved in cell cycle progression and the G2/M transition pathway have also been found to be enriched, such as
AURKA,
AURKB, and
FOXM [46].
DNA methylation has also shown to play a role in BCBrM. Hypermethylation and downregulation of the
PENK,
EDN3,
RELN, and
ITGAM genes appear to cause defects in cell migration and adhesion in BCBrM, while hypomethylation and subsequent upregulation of the
KRT8 gene increased cell adhesion and permeability in cancer cells. Another study
[57] from 2004 found the
HIN1 and
RARB genes to be hypermethylated in brain metastases compared to primary BCs. These data suggest that demethylating agents may have therapeutic significance for BCBrM patients.
3. Genomics of Breast Cancer Metastasis: Selected Rare Sites
3.1. Orbital Metastasis
Orbital metastases (OM) are an infrequent occurrence, comprising anywhere from 1 to 13% of malignant orbital tumors
[58][59][60]. Although the breast is the most common primary site of OM, small studies provide limited insight as to its true incidence, with reports ranging anywhere from 29 to 53%
[61][62]. One systematic meta-analysis
[63] of 72 orbital metastatic tumors found that infiltrating lobular breast cancer (ILBC) was five-times more likely to be implicated in OM than invasive ductal BC, although it represents just 10–15% of mammary carcinomas overall. ILBC has been associated with loss of E-cadherin, P-cadherin, HER2, EGFR, and p53 expression, and estrogen promotes its growth
[64]. In a study of eight OMs from BC tumors, seven exhibited MGBN, GATA3, and BCL2 positivity, as well as ERBB2 negativity. A concordant
PIK3CA activating oncogene mutation was found in both the primary tumor and OM of one patient, with findings consistent with ILBC
[63]. BC ocular tropism may be linked to the expression of estrogen receptors in normal conjunctiva, tear glands, and tarsal conjunctiva
[65][66]. Additionally, periorbital fat produces the steroid hormones required for lacrimation, potentially explaining hormone-sensitive tumors’ affinity for the orbit
[66]. This suggestion is supported in another study
[67], where 16/20 orbital metastases were confirmed as ER+.
3.2. Gynecologic Metastasis
Gynecologic metastases are uncommon across all cancer types. Due to proximity, colorectal cancer most commonly metastasizes to this area, followed by BC, indicating BC’s spread is likely targeted. As a possible consequence of hormone signaling, patients with BC gynecologic metastases present at a younger age (46–54 years) with ER+ and HER2− ILBC
[68]. Kutasovic et al. found recurrent gene amplifications at
FGFR1 and
CCND1 that coincided with therapy resistance in ER+ BC metastasized to gynecologic sites. Additionally, many of the metastatic tumors harbored these mutations solely in the metastasis, indicating these mutations were either present in an unsampled portion of the primary tumor or later acquired as a means for dissemination and/or survival
[68]. FGFR1 signaling through the MAPK and PI3K pathways has been specifically implicated in BC growth, survival, and metastasis. Moreover, 7.5–17% of all BCs and 16–27% of luminal-B type tumors possess a
FGFR1 mutation.
FGFR1 amplifications often demonstrate therapy resistance
[69]. However, several studies have shown promise in the utility and practicality of targeting
FGFR1 [70].
3.3. Pancreatic Metastasis
Comprising just 2–5% of pancreatic tumors overall, metastases to the pancreas is a rare event
[71]. Of these, the breast is the source of only 3–5% of these metastases
[72]. Owing to this, genetic analyses of BC metastases to the pancreas are limited. One study
[72] investigating biomarkers for BC to pancreas metastasis identified a
ERBB2 I767M mutation within the pancreatic tumor consistent with BC origin. This gene mutation’s functional significance is undetermined. Although one study
[73] found no distinction in cell proliferative ability or viability between wild-type and
I767M HER2, two other studies
[74][75] found evidence for
I767M as a gain of function mutation, leading to enhanced HER2 kinase activity and AKT phosphorylation.
4. Paired Primary and Metastatic Breast Cancer
Although there has been extensive investigation of the genes involved in both primary and metastatic BC, the exact events leading to dissemination of disease have not been completely elucidated. However, progress has been made in understanding the links between certain signaling pathways and the trigger for metastasis. Paul et al. studied this link by executing whole-exome and shallow whole-genome sequencing. The authors found specific genes that were not involved in primary BC but found in certain metastases. Seven of the genes included
MYLK,
PEAK1,
ESR1,
PALB2,
XIRP2,
EVC2, and
SLC2A4RG. In addition, regions including
STK11,
CDKN2A/B loss and
PTK6,
PAQR8 gain were also involved. The specific pathways implicated in metastases involved mTOR, CDK/RB, WNT, HKMT, cAMP/PKA, and focal adhesion. Importantly, these pathways can potentially have implications in the treatment of metastatic BC by the development of said pathway inhibitors
[76]. PARP inhibitors, for instance (as previously discussed), are currently being used in clinical trials, to treat metastatic BC.