1. Introduction
Inflammatory breast cancer is a very aggressive type of breast cancer with a poor prognosis. In the United States, it accounts for 2–6% of all patients with breast cancer [
1,
2,
3]. The principal clinical symptoms of inflammatory breast cancer (IBC) are breast erythema, edema, peau d’orange, and dermal lymphatic invasion [
4]. Despite its name, IBC is not associated with a profuse inflammatory response; the characteristic redness and swelling of the breast are due to obstruction of lymphatic channels in the dermis by tumor cells [
5,
6]. Although IBC is a rare clinical subtype of locally advanced breast cancer, it is responsible for approximately 10% of breast cancer-associated deaths annually in the US, which translates into 4000 deaths per year [
2,
6]. IBC is either stage III or IV at the time of diagnosis and the majority of patients have lymph node metastases, with one third of the patients having metastases in distant organs such as the brain, bone, other visceral organs, and soft tissue [
6]. The median overall survival (OS) for patients with stage III IBC is 4.75 years, compared to 13.40 years in those with non-IBC. In stage IV disease, the median OS is 2.27 years in IBC patients versus 3.40 years in non-IBC patients [
7,
8]. Although IBC, like non-IBC breast cancers, is a heterogeneous disease and can occur as any of the four molecular subtypes, it is most commonly either human epidermal growth factor receptor 2 (HER2) overexpressive or triple negative (TN) [
9,
10]. TN breast cancer, which is defined by the absence of estrogen and progesterone receptors, and a lack of HER2 overexpression, has a poorer prognosis than other subtypes [
11].
The principal objective of precision oncology is to improve the diagnosis and treatment of cancer patients, and its focus is increasingly turning to liquid biopsies such as cell-free DNA from blood. Due to tumor heterogeneity, the analysis of an individual tissue biopsy may not accurately reflect the genomic profile of a patient’s cancer and this could introduce bias to the selection and efficacy of personalized therapies. Instead, as DNA is released from multiple tumor regions into the bloodstream, cfDNA may reflect the aggregate genetic composition of intra-tumor heterogeneity [
12,
13] as well as metastases [
14,
15,
16,
17]. Furthermore, blood samples can be collected at multiple time points, facilitating longitudinal disease monitoring [
18]. As IBC progresses rapidly, the use of blood cfDNA could be important in following disease progression and selection of new treatments. In the present work, we studied genetic variants in IBC patients using tissue and paired blood cfDNA samples to evaluate the concordance of observed mutation profiles. Genetic variants of clinical relevance were studied in TN and non-TN IBC.
3. Discussion
Genetic studies using cfDNA from blood in IBC patients may allow the profiling of genetic changes over time, enabling the use of more efficient therapies in this rapidly progressing disease. cfDNA from blood has been evaluated to determine whether it can be used as an alternative to tissue biopsies in several types of cancers. Our studies showed a high concordance between genetic variants detected in tumor tissue and blood cfDNA from IBC patients with advanced disease, in particular for those variants with AF > 5% in tissue. For variants with AF ≥ 10% in malignant tissue or cells, 93.7% were also detected in matching blood cfDNA, and for variants with 5% ≤ AF < 10% in malignant tissue or cells, 73.3% were also detected in paired cfDNA from blood. However, for variants with AF < 5% in tissue or cells, only 36.8% were also detected in cfDNA. Furthermore, 85.6% of the variants detected in blood cfDNA were also detected in paired malignant tissue/cells. Previous studies in other types of cancer have shown a concordance of 50–88% when comparing tumor DNA and blood cfDNA using NGS [
23,
24]. Low concordance (10.8–15.1%) was found in an NGS study that included 45 breast cancer patients, 34 of whom had IBC [
25]. The low concordance could be explained by the fact that two different platforms, with different sequencing techniques, were used to study tissue and paired cfDNA [
25]. In that study, the FoundationOne (Foundation Medicine) test was used for formalin-fixed, paraffin-embedded specimen, and cfDNA was tested using the Guardant360 (Guardant Health) platform; concordance was particularly low for copy number variants (CNVs) [
25]. The numbers of genes tested in the FoundationOne and Guardant 360 panels were 315 and 70, respectively. Kuderer et al. found also low concordance between tumor and paired blood cfDNA using these two platforms [
26], however, in both studies, the concordance increased after restricting comparisons to variants found in the cfDNA at AF greater than 1% [
25,
26]. In our work, we used the same platform to study tissue and blood cfDNA samples, and only variants with AF ≥ 1.5% in cfDNA were considered. We studied single nucleotide variants (SNVs) and small indels (insertions and deletions), but did not include CNVs in the analysis. Although combining liquid biopsy with NGS technology provides a noninvasive method to analyze numerous cancer-related genes in a single assay, detecting low AF of SNVs through NGS still presents significant challenges due to the high rates of false positives when tumor DNA is in low concentration as in the case of cfDNA.
From the genetic variants detected in blood, 14.3% of them were not present in paired tissue samples. Metastatic lesions have a genomic fingerprint that may evolve and become discordant from the primary tumor [
27]. Most of the patients in this study had stage IV disease and most of them developed distal metastases to the lung, bone, liver, and/or abdomen that could explain the presence of variants in blood cfDNA coming from these metastatic sites. cfDNA may be derived from a primary tumor, metastatic lesions, or circulating tumor cells (CTCs) [
28]; both apoptosis and necrosis, alongside active secretion, play important roles in the cfDNA presence in liquid biopsies [
29,
30].
In addition to cfDNA found in plasma and serum, cfDNA in urine has shown promise as a biomarker for certain cancers. For example, in patients with non-muscle-invasive bladder cancer, high levels of cfDNA were found in urine samples in patients with progressive disease, including samples from patients where levels of cfDNA were low in plasma [
31]. Moreover, in a genomic analysis of urine cfDNA in patients with urothelial bladder cancer, there was a high rate of concordance between mutations found in urine cfDNA and tumor tissue [
32]. Saliva cfDNA was used to study variants in patients with oral cancer [
33]. In patients with brain tumors, ctDNA in blood is rarely found, presumably due to the blood–brain barrier [
34]. Cerebrospinal fluid (CSF) ctDNA was identified in primary and metastatic brain tumors [
17]. Three patients from our cohort study developed brain metastases, one of whom also had leptomeningeal disease. Genomic profiling of CSF might guide clinical decisions in IBC patients who develop brain or leptomeningeal metastasis [
35].
In the present work, pathogenic or likely pathogenic variants were most frequently detected in
TP53 (47.3%),
PMS2 (26.3%),
MRE11 (26.3%),
BRCA1 (10.5%),
RB1 (10.5%),
AR (10.5%) and
PTEN (10.5%); others in
PMS1,
KMT2C,
BRCA2,
PALB2,
MUTYH,
MEN1,
MSH2,
CHEK2,
NCOR1,
PIK3CA,
ESR1 and
MAP2K4 were detected in 5.3% of patients. Most of these variants correspond to proteins involved in DNA repair (PMS2, MRE11, BRCA1, BRCA2, PALB2, PMS1, MUTYH, CHEK2, MSH2) and control of the cell cycle (TP53, RB1, CHEK2). Deficient DNA repair and control of cell cycle would contribute to disease progression. In a recent study of 101 untreated primary IBC tumors aggregated from four public datasets, Bertucci et al. showed that the genomic profile of IBC is different from non-IBC breast cancer [
36]. Genes involved in DNA repair were found more frequently altered in IBC than in non-IBC breast cancer [
36].
TP53 was found to be the most frequently altered gene in IBC and its rate of mutation in IBC was found to be significantly higher than in non-IBC patients [
36,
37,
38]. Matsuda et al. found that
TP53 was altered in 75% of IBC (18/24 patients) and in 28.2% (106/376 patients) of non-IBC patients [
38]. Liang et al. found alterations in
TP53 in 43% (67/156) of the IBC patients (61/197) and 31% of the non-IBC breast cancer patients [
37]. The likely pathogenic variant androgen receptor (AR) c.170T > A (p. L57Q) detected in two IBC patients from our study represents a missense mutation in the amino-terminal domain of the AR with partial loss of function of the protein. The AR, like the estrogen receptor (ER) and the progesterone receptor (PR), is a member of the steroid hormone receptor family. There is a significant association between AR and ER expression in breast carcinoma [
39]. In general, AR-positive status is significantly associated with better clinical outcomes than AR-negative tumors; however, in some studies, the significant prognostic relevance of AR was observed in ER-positive tumors, but not in ER-negative tumors or triple negative tumors [
39]. In another study, AR negativity was associated with a greater frequency of recurrence and distant metastasis in triple negative tumors [
40]. IBC patients who were found to have AR-negative/ER-negative tumors had the worst survival outcomes compared to patients who had tumors that exhibited other AR/ER combinations [
41].
Importantly, although our studies showed high concordance between variants detected in tissue and paired cfDNA from blood, some pathogenic variants detected at high AF in tissue were not detected in cfDNA. These variants were found in MEN1, PIK3CA and ESR1, suggesting that the information detected from blood cfDNA could, at most, be complementary to the variants detected in tissue.
Although the number of samples studied in the present work was low, it must be taken into consideration that IBC is a rare disease that accounts for only 2–5% of all patients with breast cancer. Many inflammatory breast cancer genetic studies face challenges of a paucity of samples given the rarity of the disease. Five NGS-based studies have been published regarding IBC using tissue samples in which the number of genes tested varied between 50 to 255 [
37,
38,
42,
43,
44]. In these studies, targeted NGS [
37,
38,
43,
44] and whole-exome sequencing [
42] were used.
The fact that IBC patients tend to be younger than other breast cancer patients (52 years in IBC vs. 57 in non-IBC) has suggested a genetic component in the etiology of IBC [
45]. In a recent study of 368 IBC patients, it was found that 14.4% carried pathogenic germline variants [
46].
BRCA1 and
BRCA2 germline pathogenic variants were found in 7.3% of the IBC patients, 6.3% had a mutation in other cancer genes (
PALB2,
CHEK2,
ATM and
BARD1), and 1.6% had a germline pathogenic variant in other genes not related with breast cancer [
46]. In this study, putative pathogenic variants with AF~50% in both tissue and paired cfDNA were detected in
BRCA,
BRCA2 and
TP53 in three patients; those
BRCA1 and
BRCA2 variants were confirmed to be of germline origin since those patients had clinical genetic tests performed. Pathogenic germline variants in
BRCA1 and
BRCA2 genes are highly penetrant, conferring a risk that is more than four times that of the non-mutated population [
47]. The patient who carried the
TP53 mutation at high AF had also Ehlers–Danlos syndrome and a first degree relative with prostate cancer. A putative germline variant of uncertain significance found commonly in patients from this study was
BARD1 V507M, which was carried by 7 of the 19 patients. BARD1 (BRCA1-associated ring domain) encodes a protein which interacts with the N-terminal region of BRCA1. BARD1 is vital in the rapid relocation of BRCA1 to DNA damage sites [
48] and has been associated with increased risk of breast cancer in postmenopausal Japanese women [
49]. Although this alteration has not been associated with familial or sporadic breast cancer in other populations [
50], a statistically significant association of this variant with high-risk neuroblastoma has been demonstrated [
51,
52]. One limitation of our study design is that it was retrospective and control samples for germline profiling such as leukocytes, fibroblasts, or other normal tissue samples were not available. Internal validation in future studies will provide a more accurate estimate of the expected germline mutation prevalence in IBC, and somatic mutations that could arise from clonal hematopoiesis that could confound cfDNA analysis [
53,
54].
Our results suggest that the information regarding genetic variants in blood cfDNA from IBC patients is complementary to the variants detected using malignant tissue samples. Further studies are ongoing to improve the sensitivity of these assays, such as deeper sequencing using a different panel to increase the sensitivity of the assays in cfDNA from blood. Prospective studies are necessary in order to distinguish germline and somatic variants in IBC.