Abstract
Background/Aim: Pregnancy-associated breast cancer (PABC) is one of the most frequently diagnosed pregnancy-related malignancies, characterized by a notably rising incidence. Data remain scarce in the literature regarding the molecular nature and pathophysiology of PABC. Proteomic analyses are known to reflect cellular functions more accurately when compared to genomic or transcriptomic studies.
Materials and Methods: In the present study, two-dimensional gel electrophoresis and matrix-assisted laser desorption ionization time-of-flight mass spectrometry were employed to identify differentially expressed serum proteins among five patients with PABC, five matched (according to age, histological type and stage) non-pregnant patients diagnosed with BC, and five healthy pregnant controls.
Results: A panel of 53 differentially expressed (>1.5-fold) proteins with diverse biological roles and various functional interactions was identified among the three groups examined in our study. Of the 53 differentially expressed proteins, 23 proteins were identified in the PABC group, 8 proteins in the non-PABC group, and 22 proteins in healthy pregnant controls. Many of the proteins differentially expressed in patients with PABC were involved in biological processes known to be deregulated in carcinogenesis, such as metabolism (e.g., apolipoprotein-E, and apolipoprotein-A1) and immune system regulation (e.g., complement factor B, and defensin-5).
Conclusion: Differential proteomic expression was detected in PABC-derived serum samples, implying distinct PABC molecular features that require further investigation. The identification of PABC proteomic signatures provides significant insight into PABC pathophysiology and offers novel targets for early diagnosis and optimal treatment.
Introduction
Pregnancy-associated breast cancer (PABC) defined as breast cancer (BC) diagnosed anytime during gestation, lactation or within 1 year after delivery, represents one of the most frequently diagnosed pregnancy-related malignancies (1, 2). Vast discussion exists in the literature regarding PABC definition as multiple researchers insist that BC diagnosed during pregnancy should be differentiated from that diagnosed in the postpartum period. Several differences have been observed between these two entities in terms of underlying biological mechanisms and prognostic outcome (3). Nevertheless, in the present study, we are staying in accordance with the most commonly used PABC definition. Thus, all patients enrolled were diagnosed with PABC anytime during pregnancy, lactation or within 1 year after childbirth.
PABC remains relatively rare and represents approximately 4% of early-onset BC, i.e., BC diagnosed in pre-menopausal women of age 18-45 years (4). However, PABC incidence has been notably rising during the past decade and is expected to further increase in the upcoming years as more women tend to delay childbearing to a later age (5). Advanced maternal age is considered to be a major risk factor for PABC development as older first-time mothers are more likely to be diagnosed with the disease than younger women (6). Additionally, the use of non-invasive prenatal testing of fetal aneuploidy in obstetric monitoring of women in developed countries has incidentally identified pre-symptomatic maternal tumors with chromosomal imbalances, thereby increasing pregnancy-related malignancy rates (7, 8).
The molecular nature and pathophysiology of PABC remain largely unknown. Solid data on the genomic profile of PABC are solely based on retrospective studies and meta-analyses, and little progress has been made in exploring its biological features despite major advances in genomic technologies. According to a recently published systematic review by members of our research team, multiple differentially expressed genes have been detected in PABC-derived tissues, including the tumor suppressor gene TP53, the cell-cycle regulator phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha (PIK3CA), members of the mucin gene family related to glycosylation, and the breast cancer gene 1 (BRCA1) (9). Even though enhancing our insights into the genetic basis of cancer is certainly required to deeply understand PABC pathophysiology, it is currently thought that the PABC proteome represents a more dynamic state of the disease (10). Of note, the term ‘oncoproteogenomics’ has recently been coined to highlight any alteration observed in the mutational landscape being eventually translated into an aberrantly expressed proteome that defines a cancer phenotype (11, 12). We presently know that major post-genomic biological mechanisms such as alternative gene splicing and post-translational modifications (e.g., glycosylation, phosphorylation, acetylation, and ubiquitination), define the ultimate active protein product, thus directly affecting several cellular procedures (13, 14). Therefore, proteomics represents a field that bridges genomics and cellular functions, and proteomic studies are considered to reflect cellular mechanisms more accurately when compared to genomic or transcriptomic analyses.
Proteomic technologies including mass-spectrometry (MS), may identify variations in circulating proteins and detect clinically meaningful biomarkers that may potentially be integrated into personalized patient care. Identification of blood-circulating biomarkers (e.g., serum proteins), namely ‘liquid biopsy’, offers an accurate and less invasive alternative to histopathological assessment and can further classify biomarkers as diagnostic, prognostic or predictive (15, 16).
To the best of our knowledge, there is a significant lack of data available in the literature regarding proteomic signatures of PABC. The aim of our study, focusing on the distinct molecular nature of PABC and on high-throughput proteomic techniques, was to further elucidate the proteomic landscape and to identify novel serum biomarkers in PABC that may potentially serve as targets for early diagnosis and optimal treatment.
Materials and Methods
The present study was conducted at Alexandra General Hospital of Athens (affiliated with the National and Kapodistrian University of Athens), Greece and at the Proteomics Research Unit of the Biomedical Research Foundation of the Academy of Athens, Greece. All participants were required to sign an informed consent form according to the principles of the Declaration of Helsinki, to have completed their 18th year of age, and to have attended the Department of Clinical Therapeutics or the Breast Unit of the First Department of Obstetrics and Gynecology, National and Kapodistrian University of Athens, Greece. The study protocol was approved by the Institutional Review Board of Alexandra General Hospital of Athens (protocol number: 508/14-07-2020, date of approval: 14-07-2020).
Patients. Based on the abovementioned inclusion criteria, blood serum samples from five patients diagnosed with PABC, five matched non-pregnant patients diagnosed with BC (non-PABC), and five healthy pregnant controls were retrieved from the Department of Clinical Therapeutics/Breast Unit of the First Department of Obstetrics and Gynecology at Alexandra General Hospital of Athens. Of note, BC was diagnosed based on a combination of standard clinical, radiological and histopathological criteria and all cases of in-vitro fertilization were excluded from the present study. All patients with PABC enrolled in our study were matched in a 1:1 ratio with non-pregnant patients diagnosed with BC according to several confounding factors including age (±5 years), histopathological subtype, clinical and pathological stage, and comorbidities. Lastly, the control samples collected from healthy pregnant women were acquired during scheduled obstetric monitoring visits; these women had no history of malignancy or other chronic diseases. Patients with PABC were matched in a 1:1 ratio with controls based on age, race and gestational age at blood withdrawal.
Proteomic analysis. At initial diagnosis, a blood sample from each participant was collected and immediately centrifuged in order to isolate the serum supernatant. Subsequently, serum was stored at −80°C. Using Protein 200 plus kits (Agilent Technologies Inc., Waldbronn, Germany), serum protein concentration was estimated via a Bioanalyzer system (Agilent Technologies Inc.). Two-dimensional gel electrophoresis (2-DGE) was performed as previously described by Vaiopoulou et al. (17-19). Gel images were scanned in a GS-800 Calibrated Densitometer (Bio-Rad, Hercules, CA, USA) and images were saved in digital format for additional analysis.
Image analysis. Protein spots from all gels analyzed were identified, aligned, matched, and quantified using PD-Quest version 8.0 image processing software (Bio-Rad) according to the manufacturer’s protocol. In order to verify matching accuracy, additional manual inspection of the spots was performed. Spot volume was the foundation of protein expression quantification. Having subtracted the background values, normalization of each individual spot was carried out according to the total quantity of the valid spots in each gel. The optical density of each protein from each group was determined separately and calculated as the sum of the volume percentage of all spots from all gels containing the same protein. Based on the optical density alterations observed in the three groups of our study, protein spots or entire gel regions were selected for MS identification. A minimum of 1.5-fold alteration in the expression level between groups was used as a selection criterion.
Protein identification by matrix-assisted laser desorption ionization tandem time-of-flight (MALDI-TOF)-MS. For MALDI-TOF-MS analysis, all protein spots of interest were manually annotated using Melanie 4.02 software (GeneBio, Geneva, Switzerland) and excised from the 2D gels with a Proteiner SPII instrument (Bruker Daltonics, Bremen, Germany). Subsequently, gel pieces were placed in 96-well microtiter plates, destained with 180 μl of 30% acetonitrile in 50 mM ammonium bicarbonate and dried in a speed vacuum concentrator (MaxiDry Plus; Heto, Allered, Denmark). Using 0.01 μg/μl trypsin (Roche Diagnostics, Basel, Switzerland) for 16 h at room temperature, in-gel digestion was then performed. Afterwards, 10 μl of 50% acetonitrile containing 0.1% trifluoroacetic acid were added to each dried gel piece and tryptic-digested peptides were extracted. Tryptic peptide mixtures (1 μl) were applied on an anchor chip MALDI plate mixed with 1 μl of matrix solution, consisting of 0.08% α-cyano-4-hydroxycinnamic acid (Sigma-Aldrich, St. Louis, MO, USA) and internal standard peptides des-Arg-bradykinin (904.4681 Da; Sigma-Aldrich) and adrenocorticotropic hormone fragment 18-39 (2,465.1989 Da; Sigma-Aldrich) in 50% distilled water, 50% acetonitrile and 0.1% trifluoroacetic acid. Peptide mixtures were analyzed in a MALDI-TOF-MS as described by Vaiopoulou et al. (17). Using the SWISS-PROT (www.expassy.ch/sprot/) and the TREmBL databases (www.ebi.ac.uk./swissprot/), peptide masses were compared with the theoretical peptide masses of all available proteins from Homo sapiens. For protein identification, stringent criteria were applied with a maximum accepted mass error of 25 ppm and a minimum error of four matching peptides. For affirmative protein identification, a probability score with p<0.05 was defined as a criterion. Additionally, monoisotopic masses were used, and one missed trypsin-cleavage site was calculated for proteolytic products. Search parameters also included potential residue mass modification for carbamido-methylation and oxidation. Of note, any redundancy of proteins recognized in the database under different names and accession numbers was eliminated. When more than one protein was identified under one spot, the single protein member with the highest protein score was selected from the multi-protein family.
Verification methods. Two proteins were further measured in the same serum samples using automated methods, as previously described (20). Specifically, these were apolipoprotein-A1 (APOA1) and apolipoprotein-E (APOE) and were measured with an automated immuno-turbidimetric assay on a Atellica CH Analyzer (Siemens Healthineers, Tarrytown, NY, USA).
Pathway analysis. Using the STRING database v.12 (Search Tool for the Retrieval of interacting Genes/Proteins; http://string-db.org), we further analyzed all functional interactions among the differentially expressed proteins. The simplified version of the network produced was adopted. Lastly, aiming to reveal their molecular function, the signaling pathways and biological processes associated with each identified protein were determined using the PANTHER database (http://panther.appliedbiosystems.com).
Data analysis. In order to ensure a high level of confidence in the present study, we implemented an experimental design that involved using duplicate 2D gels per sample and separate preparations for each replicate sample per experiment. Comparisons were made among samples (PABC, non-PABC, controls). Mean densitometric values of all spots corresponding to specific proteins from each group were first checked for normal distribution and unequal variances. Data with normally distributed densitometric values were exported to Microsoft Excel 2024 software (Microsoft Corp., Redmond, WA, USA) and compared with the two-pair t-test assuming unequal variance. Means of spot intensities for proteins with non-normally distributed values were compared for statistical significance with the Mann-Whitney non-parametric test (GraphPad Instant 3 software; GraphPad Software Inc., La Jolla, CA, USA). Statistical significance (alpha-level) was defined as a value of p<0.05. To control the false-discovery rate, individual alpha-levels for each spot were adjusted following the false-discovery rate-correction procedure.
Results
Patients. All demographic characteristics of the participants of the present study are presented by group in detail in Table I. In the PABC subgroup the age at diagnosis ranged from 32 to 45 years, with a mean age of 37.2 years, whereas in the non-PABC subgroup, the age at diagnosis ranged from 37 to 47 years, with a mean age of 40 years. Additionally, the gestational age at diagnosis in the PABC subgroup demonstrated variation; only one patient was diagnosed with the disease during the second gestational trimester, two patients during the third trimester, and two patients in the postpartum period. One African American patient with PABC was accordingly matched with one non-PABC patient and one control of the same race. Lastly, three out of five of the patients with PABC reported a negative family cancer history, whereas two patients with PABC examined had a positive family history of prostate, ovarian or colorectal cancer among first-degree relatives.
Demographic characteristics of pregnancy-associated breast cancer (PABC), non-PABC and healthy pregnant control groups.
Histopathological characteristics. The main histo-pathological features of the eligible PABC and the non-PABC patients are summarized in Table II. All patients with BC, both pregnancy-associated and non-pregnancy-associated, were diagnosed with invasive ductal carcinoma. They were matched among the two subgroups based on pathological characteristics, according to the abovementioned inclusion criteria. Most of the PABC tumors were of high grade and stage III or IV disease was most frequently diagnosed (21, 22). All PABC tumors demonstrated positive hormone receptor status and high Ki-67 levels >20% were largely detected. Of note, only one patient with PABC carried a germline mutation of unknown significance [namely of BRCA1-interacting DNA helicase 1 (BRIP1) p.Arg762His – VUS]. Two participants were diagnosed with primary metastatic PABC; one patient had liver and bone metastases at the time of diagnosis, whereas the other one had solely liver metastasis. Lastly, two patients with non-metastatic PABC received neoadjuvant chemotherapy regimens. Of note, a 37-year-old woman of African American race diagnosed with inflammatory PABC is presented in Figure 1, highlighting the impressive clinical features of the disease.
Histopathological characteristics of pregnancy-associated breast cancer (PABC) and non-PABC groups.
A 37-year-old African American woman diagnosed with inflammatory pregnancy-associated breast cancer, highlighting the impressive clinical features of the disease.
Proteomic analysis. In order to identify serum proteins differentially expressed among the participants of our study, each serum sample was separated via 2-DGE. Figure 2A, Figure 2B, and Figure 2C demonstrate the results of the 2-DGE analysis of the serum samples from the PABC, the non-PABC, and the control groups, respectively. Image analysis was performed and revealed 57 protein spots in total corresponding to 58 proteins, since more than one protein was identified in the same spot, and vice versa. After omitting high-abundance proteins such as albumin and haptoglobin, 53 proteins were marked as being differentially expressed among the three subgroups of our study and are presented in Table III. This table provides their identities (names, abbreviations), theoretical pI, molecular weight, MASCOT score and protein coverage. Of the 53 differentially expressed proteins, 23 were identified in the PABC group, 8 in the non-PABC group, and 22 in healthy pregnant controls, as shown in detail in Table IV.
Proteins differentially expressed in serum samples from pregnancy-associated breast cancer (PABC) (A), non-PABC (B) and healthy pregnant control (C) participants. Serum samples were analyzed by two-dimensional gel electrophoresis and further analyzed by mass spectrometry, as described in the Materials and Methods. In the gel images, each spot represents a protein. Comparison of protein profiles revealed proteins differentially expressed among the three subgroups of our study. See Table III for protein abbreviations.
The total of the differentially expressed proteins in the serum of pregnancy-associated breast cancer (PABC) and non-PABC and healthy pregnant control groups.
Proteins differentially expressed in the serum from pregnancy-associated breast cancer (PACB) and non-PABC and healthy pregnant control (CTRL) groups.
Verification methods. Serum levels of APOA1 and APOE, after verification methods were applied, are demonstrated in Figure 3 and were shown to be de-regulated in cases with malignancy compared with healthy controls.
Quantification of Apolipoprotein A1 (APOA1) (A) and apolipoprotein E (APOE) (B) in serum from pregnancy-associated breast cancer (PABC), non-PABC and healthy pregnant control participants. Data are the mean±standard deviation from each group.
Pathway analysis. Aiming to determine the pathways involving the differentially expressed proteins that were identified as previously described, we interrogated the STRING database (v.12). As presented in Figure 4, several interactions among the proteins identified in the PABC, non-PABC, and the control groups, were recognized. A non-directed graph with the functional interactions in the dataset presented as predicted by STRING is given for each group of our study. Each node represents a protein and each line between proteins indicates both the predicted functional and physical protein association. Of note, line thickness indicates the strength of data support as well as neighboring positions. The protein–protein interplay detected may represent one protein being regulated by another, a genetic or physical interaction, their involvement in the same biological process, or the regulation of both proteins by a third molecule, among others.
STRING network of proteins differentially expressed in serum samples from pregnancy-associated breast cancer (PABC) (A), non-PABC (B) and healthy pregnant control (C) participants. Each node represents a protein. Lines represent functional interactions between proteins. Heavier network lines demonstrate stronger protein relations as well as neighboring positions. See Table III for protein abbreviations.
For the interpretation of the identified protein interplays, Gene Ontology categorization was performed using the PANTHER classification system, which sorts the proteins into respective groups based on biological processes and their involvement in different signaling pathways. In Figure 5, the biological processes of the differentially expressed proteins identified in the groups of our study are presented in detail. It should be highlighted that proteins expressed in patients with PABC were mainly related to cellular processes (20.8%), localization (14.6%), biological regulation (12.5%), metabolic processes (8.3%), response to stimulus (8.3%), and immune system processes (6.3%), processes that are known to be disrupted in carcinogenesis. Regarding the proteins identified in non-PABC, it is of great interest that molecules assigned were solely involved in cellular processes (50%), metabolic processes (25%), and localization (16.7%). Lastly, multiple differentially expressed proteins were detected in controls, mainly participating in cellular processes (30.6%), metabolic processes (13.9%), biological regulation (13.9%), and response to stimulus (11.1%). Additionally, in Figure 6, all protein classes of the molecules identified in each group of our study based on the PANTHER classification system are demonstrated.
PANTHER classification of biological processes of the differentially expressed proteins identified in serum samples from pregnancy-associated breast cancer (PABC), non-PABC and healthy pregnant control groups of our study. The number of the differentially expressed proteins that are involved in each biological process is presented.
PANTHER classification of the differentially expressed proteins identified in serum samples from pregnancy-associated breast cancer (PABC) (A), non-PABC (B) and healthy pregnant control (C) participants.
Discussion
In the present study, we investigated proteomic signatures in PABC using high-throughput proteomic techniques, including 2-DGE and MALDI-TOF-MS, while aiming to detect the differential PABC proteome and the distinct molecular features of the disease. Our analysis identified a panel of 53 differentially expressed proteins with diverse biological roles and various functional interactions among PABC patients, non-PABC patients and healthy pregnant controls. Of the 53 differentially expressed proteins, 23 proteins were identified in the PABC group, 8 proteins in the non-PABC group and 22 proteins in healthy pregnant controls as shown in detail in Table IV and Table V.
Proteins overexpressed or underexpressed in the serum of patients with pregnancy-associated breast cancer (PABC) compared to non-PABC and healthy pregnant control groups.
Studies investigating PABC proteomic profile are lacking in the literature while most of the published proteomic analyses have been applied to non-PABC or male BC (15, 23-26). MS-based proteomic technologies have further enabled researchers to detect protein variations and validate proteomic biomarkers that more accurately reflect carcinogenesis and dissemination mechanism, especially when performed on cancerous tissue samples (27-29). Thus, in the present study, we attempted to apply high-throughput proteomic techniques, including MALDI-TOF-MS, on PABC-derived serum samples to fill the gap regarding the PABC proteome that exists in the literature.
Differential expression of several clinically meaningful serum proteins was identified in patients with PABC enrolled in our study. APOE and APOA1, widely known for their involvement in lipid transport and metabolism, were both overexpressed in serum samples from patients with PABC as was verified by biochemical assays, and a significant functional protein interplay was reported among the molecules as demonstrated in Figure 4A. Recently published studies have also highlighted the influence apolipoproteins may have on tumorigenesis (e.g., nasopharyngeal carcinoma, non-squamous non-small-cell lung cancer, colorectal cancer etc.) through inflammation and oxidative stress pathways, and their potential application as biomarkers for diagnosis and prognosis (30-32). For instance, according to Ren et al., reduced serum APOA1 expression has been described as an independent predictor of metastasis or unfavorable prognosis in case of ovarian cancer, whereas increased APOA1 levels may indicate recurrence of small cell lung carcinoma (30, 33, 34). Xu et al. also detected elevated APOE levels in patients diagnosed with BC and demonstrated that increased serum APOE expression may serve as an independent prognostic variable for inferior outcomes (35). Of note, aberrant apolipoprotein expression has also been strongly associated with drug resistance (e.g., aromatase inhibitors) in various cancer types (30, 36). Thus, apolipoproteins are currently considered to be promising non-invasive diagnostic or therapeutic targets in cancer research, but further investigation is certainly required to draw firm conclusions and examine their potential clinical application in PABC diagnosis and management.
Furthermore, aberrant expression of proteins related to the immune system [e.g. complement factor B (CFAB), immunoglobulin heavy variable 2-26 (HV226), defensin-5 (DEF5)] were identified in patients with PABC and healthy pregnant controls examined in the present study. The maternal-fetal interface established during pregnancy has been extensively analyzed in multiple species and the immunological challenge posed by placentation leads to immunological barriers concurrently sustaining maternal tolerance and immune response to microbial infection (37-39). However, several researchers have postulated that the co-existence of immunosuppression, immuno-tolerance and enriched inflammatory responses due to pregnancy-induced mammary gland remodeling also allows for tumor growth and dissemination via escape mechanisms (40). Azim et al. evaluated the presence of tumor infiltrating lymphocytes (TILs) in PABC-derived tissues and reported significantly lower prevalence of TILs in tumors diagnosed during pregnancy, suggesting an altered immune status (41). On the other hand, Sajjadi et al. also analyzed the immune tumor microenvironment in PABC and identified high levels of cytotoxic CD8+ TILs in hormone receptor-positive–human epidermal growth factor receptor 2-negative PABC tissues, implying a potential tumor-suppressor mechanism (42). Notably, another recently published study by Sajjadi et al. postulated that the immune status may also be dynamic based on gestational age, allowing for aggressive tumor behavior and inferior clinical outcome in late pregnancy (43). Hence, exploring PABC immunological features represents a real challenge that may enable us to deeply understand the underlying molecular mechanisms in PABC and may also offer major clinical implications in therapeutic options.
Moreover, aberrantly expressed proteins in serum samples from patients with PABC analyzed in the present study were related to localization [e.g., islet cell autoantigen 1-like protein (ICA1L), intraflagellar transport protein 81 homolog (IFT81), centrosomal protein 83 (CEP83), claudin 20 (CLD20)] and were identified as transporters, membrane trafficking proteins, cytoskeleton or cell junction-related proteins. Increasing evidence in the literature suggests that claudin proteins, controlling intracellular signaling and paracellular permeability, play key role in various cancer types when over- or underexpressed (44). CLD20, which was overexpressed in the PABC group of our analysis, has been strongly associated with aggressive phenotype in BC as reviewed in (45). Lastly, molecules involved in cell-cycle regulation [e.g., Checkpoint protein HUS1B (HUS1B)] were also detected in serum from patients with PABC, potentially implying impaired cell-cycle arrest in response to DNA damage as demonstrated in other cancer types such as epithelial ovarian tumors (46).
Concerning the limitations of our study, it should be highlighted that even though the number of patients included was small due to the rarity of PABC, we managed to obtain high-throughput results that the literature is currently lacking. Larger cohort of patients is necessary to ensure safe conclusions to be drawn on the distinct proteomic profile of PABC.
Conclusion
The findings of our analysis provide valuable evidence regarding the differential serum proteomic landscape in PABC compared to non-PABC and pregnancy-related protein-expression patterns. Integrating protein expression discrepancies into individualized patient care represents the ultimate goal of personalized medicine and our study aimed to further enhance this effort by characterizing the proteomic profile of PABC.
Footnotes
Authors’ Contributions
FZ conceptualized the project and GTT the methodology. AMK and EZ analyzed the data, generated the tables and figures, and prepared the original draft under the supervision of FZ, GTT, SM, CD, GX, DM, CS and MAD. SCP, AMP, KA, and CZ performed the experimental procedures. All Authors provided critical feedback, contributed to the manuscript, and approved the final version according to criteria established by the International Committee of Medical Journal Editors.
Conflicts of Interest
The Authors declare that they have no potential conflicts of interest.
Funding
This study was funded by the Hellenic Society of Medical Oncologists (HeSMO). The sponsors of the study were not involved in the study design, data collection, data analysis, data interpretation, or writing of the manuscript.
- Received March 23, 2025.
- Revision received April 16, 2025.
- Accepted May 1, 2025.
- Copyright © 2025 The Author(s). Published by the International Institute of Anticancer Research.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.