Abstract
In spite of the progress made in treatment and early diagnosis, breast cancer remains a major public health issue worldwide. Although modern image-based screening modalities have significantly improved early diagnosis, around 15% to 20% of breast cancers still go undetected. In underdeveloped countries, lack of resources and cost concerns prevent implementing mammography for routine screening. Noninvasive, low-cost, blood-based markers for early breast cancer diagnosis would be an invaluable alternative that would complement mammography screening. Tumor-specific autoantibodies are excellent biosensors that could be exploited to monitor disease-specific changes years before disease onset. Although clinically informative autoantibody markers for early breast cancer screening have yet to emerge, progress has been made in the development of tools to discover and validate promising autoantibody signatures. This review focuses on the current progress toward the development of autoantibody-based early screening markers for breast cancer.
See all articles in this CEBP Focus section, “NCI Early Detection Research Network: Making Cancer Detection Possible.”
Introduction
Breast cancer is the leading cancer type among women with over 2 million new cases expected annually worldwide. In 2020, it is estimated over 279 000 new cases of breast cancer will be diagnosed in the United States and over 42,000 may succumb to the disease (1). Based on recent reports, the death rate for breast cancer has dropped by 40% between 1989 and 2017 (1, 2). Advancement in treatment and early detection has contributed to this decrease in mortality rate (3). This emphasizes the importance of early detection and screening for timely intervention and better therapeutic outcomes. For instance, the 5-year relative survival rate for 44% of patients with breast cancer approaches 100% if diagnosed at stage 1, but decreases to 26% with stage IV (3). In the United States, mammography and physical exams are widely used screening methods for breast cancer (4). For an average-risk woman, screening mammography has the benefit of reducing breast cancer mortality by 40% and thus improving survival (5–8). Although modern screening digital mammography has improved the sensitivity of breast cancer detection (86.9% vs. 78.7% predigital era), it does not detect all breast cancers (9). Cancers in women with high breast density are often obscured by dense breast tissues (10). Some breast carcinomas tend to grow along the normal breast architecture, making them difficult to detect with mammography (8). False-positive results are one of the most common issues encountered in mammography, especially among young women and women with dense breasts, which leads to follow-up studies including biopsies (11, 12). In the global health setting, low and middle-income countries have a lower frequency of mammography as a population-based screening tool due to affordability, inadequate resources, lack of medical education, and various other logistical limitations. Therefore, there is an intense effort in the search for simple, rapid, and cost-effective blood-based biomarkers for early detection of breast cancers which can be used in parallel with mammography. Many circulating biomarkers, including proteins, autoantibodies (AAb), circulating tumor cells, microRNAs, circulating tumor DNA, and exosomes, have been investigated as promising tools to fill this clinical niche (13–17). This review will mainly focus on the development and progress made on tumor-specific AAbs for diagnosis and early detection of breast cancer.
Autoantibodies as Potential Biomarkers
Cancers can induce an immunologic response resulting in the production of AAbs directed against self-antigens. Tumor-associated antigens can have abnormal structures, altered protein expression levels, or changes in posttranslational modifications (glycosylation, acetylation, methylation, phosphorylation, etc.) that are no longer recognized as “self” by the immune system, thus triggering the production of AAbs (18–20). These AAbs can be exploited as sensors to monitor disease-related proteomic changes to develop useful diagnostic assays. AAbs possess many attractive features as a diagnostic marker for early detection. First, compared with other serum proteins, AAbs are highly stable and less prone to proteolysis, making sample processing much easier (21). Second, AAbs may show persistent response over time because they are known to circulate for extended periods as opposed to tumor antigens. Tumor antigens suffer from low concentrations and brief circulation time due to degradation and rapid clearance (21, 22). Third, AAbs are detectable in archived samples and have well-characterized secondary reagents for easy identification, facilitating the development of cost-effective screening tools easily adaptable in a clinical setting. Finally, tumor AAbs are produced early in the tumorigenesis process and have been detected several years before the development of clinical symptoms (23–25). To be clinically useful as an early diagnostic marker, the AAbs should allow clear discrimination against the healthy and disease state preferably at the early stages of cancer (22). Moreover, the screening AAbs should be able to distinguish breast cancer patients with high accuracy, sensitivity, and specificity, and therefore quantitative parameters should be established to clearly discriminate positive and negative tests (26). A better way to determine if selected AAbs will make a good early screening marker is to select a series of cutoff values for the assay and determine the specificity and sensitivity. This can be plotted in a receiver operating characteristic (ROC) curve to assess the diagnostic parameters including the area under the ROC curve (AUC) to establish a cutoff value for positivity for early screening (26). The ideal AUC would be 1.0, and the ideal specificity and sensitivity values would each be 100%. But such numbers are virtually never achievable in real circumstances. In most cases, there is a tradeoff between specificity and sensitivity, wherein finding conditions to increase one often diminishes the other. This often depends on the intended application. If the test is to be used as a screening assay, where the greatest cost of error is the missed detection of disease, then optimizing sensitivity is paramount. This is especially true if another test, such as an imaging study, can provide specificity in a second round. In the case of AAbs, it is sometimes possible to combine multiple AAbs, each having good specificity, to improve sensitivity while still maintaining high specificity. Also, it is highly recommended that all AAb studies follow the five-phase schema and the Prospective Sample Collection Retrospective Evaluation (ProBE) guidelines to vigorously evaluate new diagnostic and screening markers before implementing them as clinical tools (27, 28).
Methods to Identify and Validate Autoantibodies in Breast Cancer
During the early years, AAb discovery studies were performed on small-scale, targeting just a few tumor antigens. Many studies lacked a systemic approach and failed to validate markers beyond the discovery stage. To develop a successful AAb-based screening tool for breast cancer, it is essential to have technologies capable of screening AAbs for thousands of antigens at the initial phase of malignancy. Because many single AAbs show poor performance as screening tools, it is critical to have high-throughput assays at the discovery phase to find complex panels of AAbs with suitable features to develop useful tools for early diagnosis (29, 30). High-throughput approaches can process a large number of patient sera rapidly and identify many tumor-associated AAbs at the discovery phase, which is important in the process of developing reliable assays. In recent years, new technologies were developed, facilitating the discovery of novel AAb markers in high-throughput (31–34).
Phage display–based methods
Phage display-based microarray approaches have emerged as a powerful method to identify AAb profiles in cancers (33, 35, 36). Phage display evolved as a high-throughput modification of the SEREX (Serological Screening of cDNA Expression Library) method, which resulted in the identification of over 2,000 tumor-associated antigens (37–39). In SEREX, total RNA is isolated from tumor cells or tissues to construct a cDNA library. The cDNA library is inserted into the phage vectors, and proteins are expressed in E.coli. The primary discovery is performed by transferring recombinant proteins into nitrocellulose membranes and probing with patient samples. The current phage display strategies avoid the immunoblotting step and instead subject the library to several cycles of affinity selection to enrich phages for specific clones. The enriched clones can then be eluted and propagated and lysates can be printed onto glass slides to develop a phage–protein microarray (33, 40). The array can be used to incubate serum samples from patients to discover novel AAbs (33). The method does not require large volumes of serum and allows screening thousands of antigens. However, frameshifts, truncated protein expression, lack of mammalian posttranslational modifications, bias toward high abundant transcripts, and labor-intensive procedures are drawbacks of this method. Several groups have implemented this technology for AAb discovery in numerous cancers, including breast cancer (33, 35, 36, 41–43). Zhong and colleagues used a phage display strategy to report an AAb panel (ASB-9, SERAC1, and RELT) for early detection of breast cancer (AUC = 0.86; ref. 44). In this study, the authors utilized a breast cancer cDNA T7 phage library to screen 87 breast cancer patients and 87 normal serum samples and showed that the combined panel has a sensitivity of 80% at 100% specificity in predicting breast cancer. However, further validation studies are needed to evaluate its potential for early screening.
Serological proteome analysis
Serological proteome analysis (SERPA) is a technology that combines two-dimensional electrophoresis, Western blotting, and mass spectrometry to identify tumor-associated antigens and autoantibodies (34, 45, 46). In this approach, the proteome from tumor tissues or cancer cell lines is first separated with 2D electrophoresis. Separated proteins are transferred onto a membrane and probed with sera from healthy individuals and cancer patients. The protein spots that specifically react with cancer patient sera are located by superimposing the silver-stained 2D gels with the Western blot. Proteins of interest are extracted from the gel and analyzed by mass spectrometry. SERPA utilizes in vivo–derived tumor-associated antigens to identify AAb profiles. It avoids time-consuming construction of cDNA libraries and identifies tumor-specific posttranslational modifications and various isoforms. However, many proteins are too low in abundance, and only a fraction of proteins are detectable when extracted from cells or tumors. It is also challenging to detect membrane-associated antigen. In general, this method is biased toward abundant proteins. SERPA can only recognize responses against linear epitopes and could be labor intensive to profile large cohorts of serum samples. SERPA has been used as an AAb discovery platform in several cancer types including breast cancer (30, 47–49). Desmetz and colleagues used SERPA to develop an AAb panel with five candidates to discriminate early-stage breast cancer and healthy controls (AUC = 0.73; ref. 30). The authors used both a discovery and validation cohort to develop this panel with 55.2% sensitivity at 87.9% specificity by combining three markers (PPIA, PRDX2, and FKB52) they discovered by SERPA with two other previously reported markers (HSP60 and MUC1) in the literature. However, it needs further validation in larger cohorts of retrospective and prospective studies before clinical development.
Protein microarrays
Protein microarray is an alternate approach to discover AAbs in high throughput. Protein microarrays enable the screening of a large number of antigens with low sample consumption. Several types of protein microarrays can be used for this purpose (50).
Purified recombinant protein arrays
Proteins can be expressed in heterologous systems (insect cells, E. coli, etc.), purified, and then printed on a surface (51, 52). They allow proteome scale screening (∼19,000 human proteins with various isoforms) for AAbs with low sample consumption, albeit requiring proteins to be immobilized on the array substrate. Both known and predicted cancer-related antigens can be immobilized on a single microscopic slide to generate a comprehensive screening array to be assayed with serum and control samples. When protein arrays display consistent levels of protein at each spot, they avoid some of the biases of cDNA libraries. However, purified protein microarrays are costly, labor intensive, and need significant quality control measures to ensure proper functionality and maintain stability. Based on the type of protein expression system used, some proteins immobilized on these arrays may lack posttranslational modifications. E. coli–based protein expression systems are capable of producing large quantities of antigens in a cost-effective manner but fail to incorporate posttranslational modifications. This can be overcome by expressing antigens in human cells.
Native protein arrays
Some AAb discovery studies utilize native protein arrays where human cell or tissue lysates with naturally expressed proteins are captured on a surface or fractionated with separation methods (53, 54). Once probed with patient sera, targets can be identified by mass spectrometry. This method closely mimics the in vivo environment by printing posttranslationally modified antigens. However, it is difficult to control the proper orientation of the proteins during immobilization, and may sterically block protein surfaces. Ladd and colleagues used this method to discover glycolysis and spliceosome AAb signatures from patients recently diagnosed with breast cancer (53). In this study, native protein arrays generated from fractionated MMTV-neu and MCF7 cell lysates were used for discovering immunogenic pathways and autoantibody signatures in breast cancer plasmas. This is an interesting study where they used prediagnostic plasmas from 48 women with ER+/PR+ breast cancer and 65 healthy controls and discovered significant enrichment of proteins in the glycolysis and spliceosome biological pathways. The ROC analysis on the glycolysis gene set (9 proteins) and spliceosome gene set (14 proteins) signatures gave AUCs of 0.68 and 0.73, respectively. They also reported an AUC of 0.77 with 35% sensitivity at 95% specificity for combined signatures. However, this study was limited due to a small sample cohort and requires more validation studies. In a follow-up study, native protein arrays were also used to report autoimmune response signatures associated with the development of triple-negative breast cancer (TNBC; ref. 55). Katayama and colleagues used a high-density protein array developed from MDA-MB-231 cell lysates to probe serum samples collected before clinical diagnosis of TNBC along with samples collected at the time of diagnosis from participants in the Women's Health Initiative cohort (n = 13 for cases and controls). The proteins that exhibited immunoreactivity in prediagnostic TNBC samples represented major nodes of TP53 and PI3K genes that were commonly mutated in TNBC. The study also reported AAb signatures for cytokeratin proteins associated with a mesenchymal/basal phenotype in prediagnostic human TNBC samples.
Programmable protein arrays
On the other hand, programmable protein microarrays like Nucleic Acid Protein Programmable Arrays (NAPPA) print cDNAs encoding the target genes on the matrix instead of purified proteins (31, 32). The slides can then be incubated with a cell-free protein expression system to transcribe and translate the genes to produce proteins within a few hours and can be captured on to the surface via the aid of fusion tags. This method avoids protein purification and proteins can be expressed just in time, just before probing the array with patient sera, minimizing protein degradation, and maximizing the likelihood of natural folding. Cell-free systems with chaperone proteins assist in producing well-folded functional proteins and proteome scale discovery can be performed for discovering AAbs (29, 56). The proteins are displayed at the same level so the likelihood of measuring AAb against all types of antigen targets is high. It also facilitates the incorporation of posttranslational modifications and allows to monitor AAb responses against both modified and unmodified targets. Anderson and colleagues utilized NAPPA arrays to develop an AAb panel (28 antigens) for early detection of breast cancer with 80.8% sensitivity and 61.6% specificity (AUC = 0.756; ref. 56). This was the first study that used a programmable protein microarray platform for the detection of novel AAb markers with nearly 5,000 proteins displayed on the array. This was also the first serum biomarker panel developed for the discrimination of invasive breast cancer from benign breast disease. This study (discussed in detail under AAb panels) aided in the development of the first CLIA-certified blood-based assay (Videssa Breast) for breast cancer detection (57–59). In 2015, Wang and colleagues used NAPPA arrays to build a 13-AAb panel for the detection of basal-like breast cancers with 33% sensitivity and 98% specificity (AUC = 0.68; ref. 29). The programmable array was constructed with 10,000 antigens, approximately 50% of the human genome. This study also identified AAb markers reported in other studies (TP53, NY-ESO-1).
Glycan arrays
Glycan arrays are high-throughput devices capable of detecting autoantibodies against aberrant glycans (60, 61). Around 1% of human genes undergo glycosylation, a posttranslational modification where carbohydrates are linked to proteins via glycosidic bonds with the aid of enzymes (glycotransferases; ref. 62). When the activity of these enzymes is compromised, it results in the synthesis of aberrant glycans responsible for many diseases including cancer. These unusual glycan structures can trigger an immune response to produce anti-glycan antibodies long before disease onset (63). Some groups have fabricated high-throughput devices where glycan structures are immobilized on glass surfaces to screen for anti-glycan antibodies in patient samples (60, 63, 64). Blixt and colleagues used a glycan array to look for anti-glycan antibodies against Mucin 1 (MUC1) glycopeptides and found cancer-associated glycoforms of MUC1 at higher levels in early-stage breast cancer patients (64).
Validation assays
The AAbs discovered in the discovery phase need to be validated with clinically acceptable assay platforms to determine performance measures. Traditionally, singleplex-enzyme linked immunosorbent assays (ELISA) are the most commonly used platform for validation. Various formats of ELISAs exist, and for AAb validation studies, different groups use different antigenic sources, variable attachment chemistries, and detection methods, which causes difficulty when comparing results. Most ELISA assays use purified recombinant proteins from heterologous expression systems (E.coli, insects) as the antigenic source, which is either adsorbed or captured on to the ELISA plate (30, 44). Antigens expressed from different systems may lack or may show variations in posttranslation modifications. Changes may also occur to the conformational structure of the antigen, affecting the reproducibility of the assay. Some labs use antigen-specific antibodies to capture the antigen of interest from human tumor cells and thereby bypass the need for producing purified antigens (65, 66). RAPID ELISA (rapid antigenic protein in situ display) is another assay adapted from the NAPPA technology that processes a large number of clinical samples against a limited number of antigens during AAb validation (31). Here, cDNAs for the gene of interest can be readily added and proteins can be produced just in time for testing. Antibodies or ligands against a fusion tag can be used to capture the protein on to the ELISA plate obviating protein purification. Bead-based methods such as Luminex-xMAP technology are increasingly popular for AAb validation studies due to multiplexing capabilities (67–69). In these bead-based assays, fluorophore-labeled beads can be coated with antigen-specific capture antibodies to immobilize the antigens on to the surface. It allows simultaneous analysis of serum antibodies against 100 different antigens saving sample consumption, cost, and time, although absorption of antibodies in human sera on beads leads to nonspecific background.
Autoantibodies to Individual Tumor Antigens in Breast Cancer
Tumor-specific AAbs, which have the potential to be used for early screening, have been reported in the sera of breast cancer patients (summarized in Tables 1 and 2). However, only very few have been studied in detail to understand the diagnostic utility in early cancer detection.
List of individual autoantibodies discussed in this review.
Name . | Patient cohort . | . | % Positive . | Sensitivity (%) . | Specificity (%) . | AUC . | Methods . | Reference . | Year . |
---|---|---|---|---|---|---|---|---|---|
MUC1 | 24 BC | 8.3 | ELISA | Kotera et al. | 1994 | ||||
TP53 | 176 BC, 76HC | 25.5 | ELISA | Green et al. | 1994 | ||||
TP53 | 182 BC, 76 HC | 26 | ELISA | Mudenda et al. | 1994 | ||||
MUC1 | 40 BBD, 140 BCP, 96 HC | 26 | 88 | ELISA | Von Mensdorff-Pouilly et al. | 1996 | |||
HER2 | 107 BC, 200 HC | 11 | ELISA, Western blotting | Disis et al. | 1997 | ||||
SURVIVIN | 46 BC, 10 HC | 23.9 | ELISA | Yagihashi et al. | 2005 | ||||
LIVIN | 32.6 | ||||||||
CDKN2A (p16) | 41 BC and 82 HC | 7 | ELISA, Western blotting | Looi et al. | 2006 | ||||
c-MYC | 9 | ||||||||
TP53 | 5 | ||||||||
c-MYC | 97 PBC, 40 DCIS, 94 HC | PBC vs. HC | 13 | 97 | ELISA | Chapman et al. | 2007 | ||
DCIS vs. HC | 8 | 97 | |||||||
TP53 | PBC vs. HC | 24 | 96 | ||||||
DCIS vs. HC | 15 | 96 | |||||||
NY-ESO-1 | PBC vs. HC | 26 | 94 | ||||||
DCIS vs. HC | 8 | 94 | |||||||
BRCA1 | PBC vs. HC | 8 | 91 | ||||||
DCIS vs. HC | 3 | 91 | |||||||
BRCA2 | PBC vs. HC | 34 | 92 | ||||||
DCIS vs. HC | 23 | 92 | |||||||
HER2 | PBC vs. HC | 18 | 94 | ||||||
DCIS vs. HC | 13 | 94 | |||||||
MUC1 | PBC vs. HC | 20 | 98 | ||||||
DCIS vs. HC | 23 | 98 | |||||||
AHSG | 81 BC and 73 HC | 79.1 | 2DE, immunoblot, mass spectrometry | Yi et al. | 2009 | ||||
HSP60 | Discovery | SERPA, ELISA | Desmetz et al. | 2008 | |||||
20 BC, 20 other cancers, 10 AID, 20 HC | |||||||||
Validation | BC vs. HC | 31.8 | 95.7 | 0.637 | |||||
49 DCIS, 58 T1N0, 93 HC | T1N0 vs. HC | 31 | 95.7 | 0.634 | |||||
DCIS vs. HC | 32.7 | 95.7 | 0.642 | ||||||
Globo H | 58 BC, 47 HC | Glycan arrays | Wang et al. | 2008 | |||||
MUC1 (comb) | 395 BC, 108 BBD, 99 HC | BC vs. HC | 10.6 | 95.9 | 0.72 | Glycopeptide arrays | Blixt et al. | 2011 | |
BC vs. BBD | 10.6 | 97.2 | 0.77 | ||||||
SERPINA1 (Alpha 1-antitrypsin) | 25 BC, 20 HC | 96 | 2D-gel, mass spectroscopy | Lopez-Arias et al. | 2012 | ||||
GAL3 | Discovery | HC vs. PBC | 47 | 88 | 0.67 | SERPA, ELISA | Lacombe et al. | 2013 | |
10 PBC, 10 DCIS, 20 BBL, 20 AID, and 20 HC | HC vs. DCIS | 24 | 97 | 0.56 | |||||
PAK2 | Validation | HC vs. PBC | 31 | 94 | 0.59 | ||||
59 PBC, 55 DCIS, and 68 HC | HC vs. DCIS | 27 | 91 | 0.52 | |||||
PHB2 | HC vs. PBC | 32 | 94 | 0.62 | |||||
HC vs. DCIS | 18 | 97 | 0.50 | ||||||
RACK1 | HC vs. PBC | 31 | 97 | 0.61 | |||||
HC vs. DCIS | 29 | 94 | 0.57 | ||||||
RUVBL1 | HC vs. PBC | 31 | 93 | 0.59 | |||||
HC vs. DCIS | 18 | 94 | 0.50 | ||||||
HNRNPF | 155 BC, 40 other cancers, 155 HC | HC vs. BC | 84.2 | 60.8 | 0.72 | SEREX, ELISA | Dong et al. | 2013 | |
FTH1 | HC vs. BC | 81.2 | 56.1 | 0.68 | |||||
BRCA1, TP53 | 13 prediagnostic (TNBC), 13 HC | Microarray, mass spectrometry | Katayama et al. | 2015 | |||||
Cytokeratin | |||||||||
Network proteins | |||||||||
Plasminogen | 29 BC, 43 HC | 69 | 2D-gel, mass spectrometry, ELISA | Goufman et al. | 2015 |
Name . | Patient cohort . | . | % Positive . | Sensitivity (%) . | Specificity (%) . | AUC . | Methods . | Reference . | Year . |
---|---|---|---|---|---|---|---|---|---|
MUC1 | 24 BC | 8.3 | ELISA | Kotera et al. | 1994 | ||||
TP53 | 176 BC, 76HC | 25.5 | ELISA | Green et al. | 1994 | ||||
TP53 | 182 BC, 76 HC | 26 | ELISA | Mudenda et al. | 1994 | ||||
MUC1 | 40 BBD, 140 BCP, 96 HC | 26 | 88 | ELISA | Von Mensdorff-Pouilly et al. | 1996 | |||
HER2 | 107 BC, 200 HC | 11 | ELISA, Western blotting | Disis et al. | 1997 | ||||
SURVIVIN | 46 BC, 10 HC | 23.9 | ELISA | Yagihashi et al. | 2005 | ||||
LIVIN | 32.6 | ||||||||
CDKN2A (p16) | 41 BC and 82 HC | 7 | ELISA, Western blotting | Looi et al. | 2006 | ||||
c-MYC | 9 | ||||||||
TP53 | 5 | ||||||||
c-MYC | 97 PBC, 40 DCIS, 94 HC | PBC vs. HC | 13 | 97 | ELISA | Chapman et al. | 2007 | ||
DCIS vs. HC | 8 | 97 | |||||||
TP53 | PBC vs. HC | 24 | 96 | ||||||
DCIS vs. HC | 15 | 96 | |||||||
NY-ESO-1 | PBC vs. HC | 26 | 94 | ||||||
DCIS vs. HC | 8 | 94 | |||||||
BRCA1 | PBC vs. HC | 8 | 91 | ||||||
DCIS vs. HC | 3 | 91 | |||||||
BRCA2 | PBC vs. HC | 34 | 92 | ||||||
DCIS vs. HC | 23 | 92 | |||||||
HER2 | PBC vs. HC | 18 | 94 | ||||||
DCIS vs. HC | 13 | 94 | |||||||
MUC1 | PBC vs. HC | 20 | 98 | ||||||
DCIS vs. HC | 23 | 98 | |||||||
AHSG | 81 BC and 73 HC | 79.1 | 2DE, immunoblot, mass spectrometry | Yi et al. | 2009 | ||||
HSP60 | Discovery | SERPA, ELISA | Desmetz et al. | 2008 | |||||
20 BC, 20 other cancers, 10 AID, 20 HC | |||||||||
Validation | BC vs. HC | 31.8 | 95.7 | 0.637 | |||||
49 DCIS, 58 T1N0, 93 HC | T1N0 vs. HC | 31 | 95.7 | 0.634 | |||||
DCIS vs. HC | 32.7 | 95.7 | 0.642 | ||||||
Globo H | 58 BC, 47 HC | Glycan arrays | Wang et al. | 2008 | |||||
MUC1 (comb) | 395 BC, 108 BBD, 99 HC | BC vs. HC | 10.6 | 95.9 | 0.72 | Glycopeptide arrays | Blixt et al. | 2011 | |
BC vs. BBD | 10.6 | 97.2 | 0.77 | ||||||
SERPINA1 (Alpha 1-antitrypsin) | 25 BC, 20 HC | 96 | 2D-gel, mass spectroscopy | Lopez-Arias et al. | 2012 | ||||
GAL3 | Discovery | HC vs. PBC | 47 | 88 | 0.67 | SERPA, ELISA | Lacombe et al. | 2013 | |
10 PBC, 10 DCIS, 20 BBL, 20 AID, and 20 HC | HC vs. DCIS | 24 | 97 | 0.56 | |||||
PAK2 | Validation | HC vs. PBC | 31 | 94 | 0.59 | ||||
59 PBC, 55 DCIS, and 68 HC | HC vs. DCIS | 27 | 91 | 0.52 | |||||
PHB2 | HC vs. PBC | 32 | 94 | 0.62 | |||||
HC vs. DCIS | 18 | 97 | 0.50 | ||||||
RACK1 | HC vs. PBC | 31 | 97 | 0.61 | |||||
HC vs. DCIS | 29 | 94 | 0.57 | ||||||
RUVBL1 | HC vs. PBC | 31 | 93 | 0.59 | |||||
HC vs. DCIS | 18 | 94 | 0.50 | ||||||
HNRNPF | 155 BC, 40 other cancers, 155 HC | HC vs. BC | 84.2 | 60.8 | 0.72 | SEREX, ELISA | Dong et al. | 2013 | |
FTH1 | HC vs. BC | 81.2 | 56.1 | 0.68 | |||||
BRCA1, TP53 | 13 prediagnostic (TNBC), 13 HC | Microarray, mass spectrometry | Katayama et al. | 2015 | |||||
Cytokeratin | |||||||||
Network proteins | |||||||||
Plasminogen | 29 BC, 43 HC | 69 | 2D-gel, mass spectrometry, ELISA | Goufman et al. | 2015 |
Abbreviations: AUC, area under curve; BBD, benign breast disease; BC, breast cancer; BCP, breast cancer pretreatment; BLBC, basal-like breast cancer; DCIS, ductal carcinoma in situ; HC, healthy control; PBC, primary breast cancer; TNBC, triple-negative breast cancer.
List of autoantibody panels discussed in this review.
Name . | Sample cohort . | . | Sensitivity (%) . | Specificity (%) . | AUC . | Methods . | Reference . | Year . |
---|---|---|---|---|---|---|---|---|
CDKN2A (p16), c-MYC, TP53 | 41 BC, 82 HC | 43.9 | 97.6 | ELISA, Western blotting | Looi et al. | 2006 | ||
TP53, c-MYC, NY-ESO-1, BRCA2, HER2, MUC1 | 97 PBC, 40 DCIS, 94 HC | HC vs. PBC | 64 | 85 | ELISA | Chapman et al. | 2007 | |
HC vs. DCIS | 45 | 85 | ||||||
ASB-9, SERAC1, RELT | 87 BC, 87 HC | 77 | 82.8 | Phage display, ELISA | Zhong et al. | 2008 | ||
PPIA, PRDX, FKBP52, MUC1, HSP60 | Discovery | HC vs. Cancer | 60.5 | 77.2 | 0.74 | SERPA, ELISA | Desmetz et al. | 2009 |
20 BC, 10 other cancers, 10 AID, 20 HC | HC vs. PBC | 55.2 | 87.9 | 0.73 | ||||
Validation | HC vs. DCIS | 72.2 | 72.6 | 0.80 | ||||
82 DCIS, 60 PBS, and 93 HC | ||||||||
ATP6AP1, PDCD6IP, DBT, CSNK1E, FRS3, RAC3, HOXD1, SF3A1, CTBP1, C15ORF48, MYOZ2, EIF3E, BAT4, ATF3, BMX, RAB5A, UBAP1, SOX2, GPR157, BDNF, ZMYM6, SLC33A1, TRIM32, ALG10, TFCP2, SERPINH1, SELL, ZNF510 | Phase I-discovery | NAPPA protein array, ELISA | Anderson et al. | 2011 | ||||
53 BC, 53 HC | ||||||||
Phase II-training | BC vs. BBD | 9–40 | 91 | |||||
51 BC, 39 BBD | ||||||||
Phase III-validation | BC vs. HC | 80.8 | 61.6 | 0.756 | ||||
51 BC, 38 HC | ||||||||
HER2, TP53, CEA, CCNB1 (cyclin B1) | Initial triage set | ELISA | Lu H et al. | 2012 | ||||
98 BC time of treatment, 98 HC | ||||||||
HER2, TP53, CCNB1 (cyclin B1) | Primary validation | 0.73 | ||||||
20 BC time of diagnosis, 20 HC | ||||||||
Secondary validation | 0.60 | |||||||
33 BC before diagnosis, 45 HC | ||||||||
GAL3, PAK2, PHB2, RACK1, RUVBL1 | Discovery | 2D-gel-mass spectrometry, ELISA | Lacombe et al. | 2013 | ||||
10 PBC, 10 DCIS, 20 BBL, 20 AID, and 20 HC | ||||||||
Validation | HC vs. PBS | 66 | 84 | 0.81 | ||||
59 PBC, 55 DCIS, and 68 HC | HC vs. DCIS | 82 | 74 | 0.85 | ||||
IMP1, P62, KOC1, TP53, c-MYCc, SURVIVIN, CDKN2A (p16), CCNB1 (cyclin B1) | 41 BC, 82 HC | HC vs. BC | 61 | 86.6 | Mini array, ELISA | Ye et al. | 2013 | |
CCND1 (cyclin D1), CDK2 | ||||||||
Glycolysis signature (9 proteins) | Discovery | 0.68 | Native microarrays, ELISA | Ladd et al. | 2013 | |||
Spliceosome signature (14 proteins) | 48 BC (prediagnosed), 65 HC | 0.73 | ||||||
Glycolysis signature + Spliceosome signature | Validation | 35 | 95 | 0.77 | ||||
61 BC (newly diagnosed), 61 C | ||||||||
118 BC (prediagnosed), 120 HC | ||||||||
CCNB1, FKBP52, GAL3, PAK2, PRDX2, PPIA, TP53, MUC1 | Discovery and validation | BC vs. HC | 90 | 42 | ELISA | Lacombe et al. | 2014 | |
87 DCIS | PBC vs. HC | 90 | 51 | |||||
153 PBS | DCIS vs. HC | 90 | 32 | |||||
156 HC | ||||||||
CTAG1B, CTAG2, TP53, RNF216, PPHLN1, PIP4K2C, ZBTB16, TAS2R8, WBP2NL, DOK2, PSRC1, MN1, TRIM21 | Discovery | NAPPA Protein array, ELISA | Wang et al. | 2015 | ||||
45 BLBC, 45 HC | ||||||||
Validation | 33 | 98 | 0.68 | |||||
145 BLBC, 145 HC | ||||||||
LGALS3, PHB2, MUC1, GK2, and (CA15–3) | Discovery | SEREX, ELISA | Zuo et al. | 2016 | ||||
10 BC, 5 HC | ||||||||
Validation | 87 | 76 | 0.872 | |||||
100 BC, 50 HC | ||||||||
CDKN2A (p16), c-MYC, TP53, ANXA | 102 BC, 146 HC | 33.3 | 90 | 0.725 | ELISA | Liu et al. | 2017 |
Name . | Sample cohort . | . | Sensitivity (%) . | Specificity (%) . | AUC . | Methods . | Reference . | Year . |
---|---|---|---|---|---|---|---|---|
CDKN2A (p16), c-MYC, TP53 | 41 BC, 82 HC | 43.9 | 97.6 | ELISA, Western blotting | Looi et al. | 2006 | ||
TP53, c-MYC, NY-ESO-1, BRCA2, HER2, MUC1 | 97 PBC, 40 DCIS, 94 HC | HC vs. PBC | 64 | 85 | ELISA | Chapman et al. | 2007 | |
HC vs. DCIS | 45 | 85 | ||||||
ASB-9, SERAC1, RELT | 87 BC, 87 HC | 77 | 82.8 | Phage display, ELISA | Zhong et al. | 2008 | ||
PPIA, PRDX, FKBP52, MUC1, HSP60 | Discovery | HC vs. Cancer | 60.5 | 77.2 | 0.74 | SERPA, ELISA | Desmetz et al. | 2009 |
20 BC, 10 other cancers, 10 AID, 20 HC | HC vs. PBC | 55.2 | 87.9 | 0.73 | ||||
Validation | HC vs. DCIS | 72.2 | 72.6 | 0.80 | ||||
82 DCIS, 60 PBS, and 93 HC | ||||||||
ATP6AP1, PDCD6IP, DBT, CSNK1E, FRS3, RAC3, HOXD1, SF3A1, CTBP1, C15ORF48, MYOZ2, EIF3E, BAT4, ATF3, BMX, RAB5A, UBAP1, SOX2, GPR157, BDNF, ZMYM6, SLC33A1, TRIM32, ALG10, TFCP2, SERPINH1, SELL, ZNF510 | Phase I-discovery | NAPPA protein array, ELISA | Anderson et al. | 2011 | ||||
53 BC, 53 HC | ||||||||
Phase II-training | BC vs. BBD | 9–40 | 91 | |||||
51 BC, 39 BBD | ||||||||
Phase III-validation | BC vs. HC | 80.8 | 61.6 | 0.756 | ||||
51 BC, 38 HC | ||||||||
HER2, TP53, CEA, CCNB1 (cyclin B1) | Initial triage set | ELISA | Lu H et al. | 2012 | ||||
98 BC time of treatment, 98 HC | ||||||||
HER2, TP53, CCNB1 (cyclin B1) | Primary validation | 0.73 | ||||||
20 BC time of diagnosis, 20 HC | ||||||||
Secondary validation | 0.60 | |||||||
33 BC before diagnosis, 45 HC | ||||||||
GAL3, PAK2, PHB2, RACK1, RUVBL1 | Discovery | 2D-gel-mass spectrometry, ELISA | Lacombe et al. | 2013 | ||||
10 PBC, 10 DCIS, 20 BBL, 20 AID, and 20 HC | ||||||||
Validation | HC vs. PBS | 66 | 84 | 0.81 | ||||
59 PBC, 55 DCIS, and 68 HC | HC vs. DCIS | 82 | 74 | 0.85 | ||||
IMP1, P62, KOC1, TP53, c-MYCc, SURVIVIN, CDKN2A (p16), CCNB1 (cyclin B1) | 41 BC, 82 HC | HC vs. BC | 61 | 86.6 | Mini array, ELISA | Ye et al. | 2013 | |
CCND1 (cyclin D1), CDK2 | ||||||||
Glycolysis signature (9 proteins) | Discovery | 0.68 | Native microarrays, ELISA | Ladd et al. | 2013 | |||
Spliceosome signature (14 proteins) | 48 BC (prediagnosed), 65 HC | 0.73 | ||||||
Glycolysis signature + Spliceosome signature | Validation | 35 | 95 | 0.77 | ||||
61 BC (newly diagnosed), 61 C | ||||||||
118 BC (prediagnosed), 120 HC | ||||||||
CCNB1, FKBP52, GAL3, PAK2, PRDX2, PPIA, TP53, MUC1 | Discovery and validation | BC vs. HC | 90 | 42 | ELISA | Lacombe et al. | 2014 | |
87 DCIS | PBC vs. HC | 90 | 51 | |||||
153 PBS | DCIS vs. HC | 90 | 32 | |||||
156 HC | ||||||||
CTAG1B, CTAG2, TP53, RNF216, PPHLN1, PIP4K2C, ZBTB16, TAS2R8, WBP2NL, DOK2, PSRC1, MN1, TRIM21 | Discovery | NAPPA Protein array, ELISA | Wang et al. | 2015 | ||||
45 BLBC, 45 HC | ||||||||
Validation | 33 | 98 | 0.68 | |||||
145 BLBC, 145 HC | ||||||||
LGALS3, PHB2, MUC1, GK2, and (CA15–3) | Discovery | SEREX, ELISA | Zuo et al. | 2016 | ||||
10 BC, 5 HC | ||||||||
Validation | 87 | 76 | 0.872 | |||||
100 BC, 50 HC | ||||||||
CDKN2A (p16), c-MYC, TP53, ANXA | 102 BC, 146 HC | 33.3 | 90 | 0.725 | ELISA | Liu et al. | 2017 |
Abbreviations: AUC, area under curve; BBD, benign breast disease; BC, breast cancer; BCP, breast cancer pretreatment; BLBC, basal-like breast cancer; DCIS, ductal carcinoma in situ; HC, healthy control; PBC, primary breast cancer.
p53 AAbs
p53 is one of the most studied tumor-associated antigens in cancer (70). In healthy cells, wild-type p53 is predominantly present in the nuclei in low concentrations. It plays a key role as a mediator in cell-cycle arrest and apoptosis and is crucial for suppressing uncontrolled cell growth (71). p53 is often mutated in many solid tumors and various reports have shown these mutations can occur during the early stages of cancer development (72). Mutant p53 proteins are more stable with a half-life of several hours compared with wild-type p53, which lasts only a couple of minutes. This causes mutant p53 to accumulate in the nucleus and escape into the cytoplasm, eventually inducing the immune system to generate autoantibodies (73, 74). The earliest report providing evidence for p53 AAbs goes back to 1979 when DeLeo and colleagues reported that the humoral response of mice to some chemically induced tumor cells was directed against the protein p53 (75). A few years later, Crawford and colleagues demonstrated that around 9% of stage 1 and 2 breast cancer patients had p53 autoantibodies in their sera (76). Since then numerous studies have reported the presence of p53 AAbs in various cancers, including breast cancer (77–80). Around 10% to 15% of early-stage breast cancers have detected p53 AAbs (81–84). Several studies have reported a positive correlation between the presence of p53 AAbs and p53 missense mutations and/or accumulation (70, 74). However, as a standalone marker p53, AAbs have low sensitivity for screening (85). Although there is a significant correlation between the anti-p53 AAb levels among healthy controls and cancer patients, the AAb is not specific enough to distinguish one cancer from another. In addition, only around 20% to 40% of patients harboring p53 mutations develop AAbs (70). Patients with similar mutations in similar cancer types could be either positive or negative for p53 AAbs, indicating the influence of other factors in antibody response (70, 86, 87). Therefore, p53 AAbs alone will not be sufficient for early disease screening of breast cancer. Moreover, the association of p53 AAbs has shown conflicting results for different tumor stages of breast cancer. Several studies have reported that the p53 AAb levels do not correlate with the disease stage (88, 89). Others have shown a higher frequency of AAbs in late-stage breast cancers (83, 84). Although p53 alone is less useful as an early screening marker, its discovery has aided in developing AAb panels with better diagnostic characteristics (25, 78).
MUC1 AAbs
MUC1 is a single-pass type 1 transmembrane protein with a heavily glycosylated extracellular domain (90, 91). MUC1 is normally expressed in the cell surface of secretory epithelia including the mammary gland, respiratory, urinary, gastrointestinal, and reproductive tract (92). Mucins are a family of glycoproteins with a high molecular weight with extracellular domains extending up to 200 to 500 nm from the cell surface (90, 91). In healthy tissues, MUC1 protects the epithelia and acts as a barrier against pathogen colonization (90). MUC1 overexpression is observed in more than 90% of breast cancers and frequently appears in other cancers, including pancreatic, ovarian, colon, and lung cancers (92, 93). The MUC1 protein present in tumor cells shows aberrant glycosylation patterns and changes in cellular distribution (94, 95). High expression levels of MUC1 with altered glycan patterns can induce an immune response, which leads to the production of glycoprotein-specific AAbs. In the early 1990s, many groups reported the presence of humoral immune response to MUC1 in patients with benign and malignant breast tumors (96–98). Since then, a number of studies have implicated the usefulness of anti-MUC1 antibodies for early detection of breast cancer (64, 78, 99). By using glycopeptide microarrays with 60mer MUC1 glycopeptides, Blixt and colleagues reported significantly higher levels of MUC1 AAbs in early-stage breast cancer patients (n = 365) than in women with benign breast disease (n = 108) or healthy controls (n = 99; ref. 64). The data reported a sensitivity of 10.6% for MUC1 glycan combinations with 95% specificity. However, a large-scale follow-up study performed by the same group with both discovery (breast cancer patients n = 240, controls n = 273) and validation samples (breast cancer patients n = 431, controls n = 431) showed no difference between the cases and controls (100). This study emphasized the importance of performing independent validation on diagnostic markers. Another study conducted with a population of women with BRCA1 and BRCA2 mutations (n = 127) reported lower levels of MUC1 AAbs among the mutation carriers than the healthy controls (101).
HER2/neu AAbs
HER2/neu belongs to the family of epidermal growth factor receptors and plays an important role in cell proliferation (102). Around 20% of newly diagnosed breast cancers have amplification or overexpression of HER2 and show more aggressive disease with worse prognosis (103, 104). In 1997, Disis and colleagues reported antibody titers of > 1:100 of HER2/neu antibodies in 11% of breast cancer patients demonstrating a correlation with HER2/neu protein overexpression in the primary tumor (105, 106). In a study reported with patients newly diagnosed with primary invasive breast cancer (PBC) and ductal carcinoma in situ (DCIS), AAbs for HER2 reported a sensitivity of 18% for PBC and 13% for DCIS with 94% specificity (78). In a more recent study, Lu and colleagues used an initial triage set of breast cancer samples collected at the time of treatment (n = 98) with matched controls (n = 98) to measure the AAb response against eight known tumor-associated antigens, which also included HER2 (25). HER2 demonstrated a significant increase in AAb response in cancer patients. When subjected to primary validation (20 breast cancer samples collected at the time of diagnosis along with matched controls) followed with secondary validation (breast cancer samples collected before the time of diagnosis n = 33 with matched controls n = 45), they observed a significantly high serum antibody response for HER2 (AUC = 0.63, P = 0.026) in prediagnostic sera. About 15% of the prediagnostic breast cancer patients were positive for HER2 AAbs. This study was one of the first studies that reported the occurrence of serum AAbs in prediagnostic sera from patients with breast cancer using samples collected using the ProBE guidelines. However, the study was conducted with a small patient cohort and needs to be validated in a larger sample size.
Other markers
Numerous other tumor-associated AAbs (Tables 1 and 2) have been reported as potential markers for early diagnosis of breast cancer. More investigation and validation are needed to evaluate the true clinical potential of these markers (55, 107–115).
Autoantibody Panels for Early Detection of Breast Cancers
Most of the individual autoantibodies identified to date suffer from low clinical sensitivity; hence, they cannot be used for early disease screening. Only a fraction of patients respond to tumor antigens, and no single serum marker exists that can be used for early breast cancer screening. Therefore, to increase the sensitivity for early diagnosis, several groups have developed tailor-made autoantibody panels (Table 2; refs. 29, 30, 56, 85, 116–118). When Chapman and colleagues used seven tumor antigens (p53, c-MYC, HER2, NY-ESO-1, BRCA1, BRCA2, and MUC1) to investigate AAbs in PBC and DCIS patients, sensitivities for individual AAbs varied between 8% and 34% (PBC) and 3% and 23% (DCIS) compared with healthy controls for 91% to 98% specificity (78). However, when used as a panel, 64% of patients with PBC and 45% from DCIS showed elevation of at least one of the six autoantibodies at 85% specificity (78). In another study serum, AAbs detected against a combined panel of five tumor antigens (FKBP52, PPIA, PRDX2, HSP60, and MUC1) accurately discriminated between early-stage breast cancer (AUC = 0.73; 55.2% sensitivity, 87.9% specificity) and carcinoma in situ (AUC = 0.80; 72% sensitivity, 72.6% specificity) from healthy individuals (30). These initial AAb panels demonstrated a promising trend for early breast cancer screening. However, these studies are in phase I/II of the ProBE guidelines and need large-scale retrospective and prospective studies to understand their usefulness as early diagnostic markers.
In 2011, Anderson and colleagues used a three-phase screening approach to detect AAbs for early-stage breast cancers (IBC). In the first stage, sera from IBC (n = 53) and healthy control (n = 53) were screened against 4,988 antigens via a high-density protein array, NAPPA (56). After eliminating uninformative antigens, 761 antigens with high responses were screened in the second stage using an independent set of IBC sera (n = 51) and sera from women with benign breast disease (n = 39). One hundred nineteen antigens were selected from the second phase (sensitivities from 9%–40% at 91% specificity) to conduct the phase III validation study. In the third phase, with an independent serum cohort (n = 51 cases/38 controls, also benign disease), 28 of these antigens were confirmed with an ELISA assay under blinded conditions. The 28-AAb panel had a sensitivity of 80.8% and a specificity of 61.6% (AUC = 0.756) for early detection of breast cancer. This discovery and validation study (phase I/II), later led to the development of Videssa Breast, a blood-based combinatorial proteomic biomarker assay (57). In 2017, two prospective clinical trials were conducted to evaluate the potential of Videssa Breast, which combined 10 AAbs with 8 serum protein biomarkers to detect breast cancer in women under the age of 50 years (59). The validation cohort reported a sensitivity and specificity of 66.7% and 81.5%, for Videssa Breast, demonstrating its potential to effectively detect breast cancer and indicating it would be useful to combine this assay with image-based screening modalities. In other studies, Videssa Breast has further demonstrated that breast density does not affect the ability of the assay to detect breast cancer, and it may provide clinicians extra information that potentially would aid in reducing false positives in breast cancer imaging (58, 119).
In a subsequent study, Wang and colleagues reported plasma autoantibodies associated with basal-like breast cancer, a rare aggressive subtype less likely to be detected via mammography (29). This study used basal-like breast cancer patients (n = 45) and controls (n = 45) from the Polish Breast Cancer study to screen 10,000 antigens on high-density NAPPA protein arrays in the discovery phase. From the initial screen, 748 promising AAbs were identified and subjected to further validation using a cohort of basal-like breast cancer (n = 145) and age-matched controls (n = 145). This study reported a 13-AAb panel to distinguish basal-like breast cancer from controls with 33% sensitivity and 98% specificity. This study (phase I/II) was focused mainly on basal-like breast cancer subtype. They used a large number of basal-like breast cancer patient samples and age-matched controls with detailed data on tumor characteristics, demographics, and treatment information. However, it is unclear from this study how early these markers are present and require further validation in prospective cohorts.
Lacombe and colleagues reported a panel of five AAbs (GAL3, PAK2, PHB2, RACK1, and RUVBL1) as a diagnostic tool for screening early-stage and preinvasive breast cancer (116). To discover new AAbs, they used 2D gel analysis and mass spectrometry on a discovery cohort (n = 80) and identified 67 interesting targets that elicited a humoral response. Five of the targets were selected and validated with an independent sample set (n = 182) with ELISA. As a panel, the five markers discriminated early-stage cancer from healthy controls (AUC = 0.81; 95% CI) and reported a sensitivity of 66% and a specificity of 84% for early-stage breast cancer patients. Here, they used two independent serum cohorts to identify and validate the five protein panel (phase I/II). However, a systematic, prospective trial is needed to further investigate the clinical effectiveness of this panel for early diagnosis. A follow-up study reported a multiparametric serum marker panel for screening early-stage breast cancer that could be used along with mammography to improve early diagnosis (85). Here, the authors explored the AAb response against 13 antigens (HSP60, FKBP52, PRDX2, PPIA, MUC1, GAL3, PAK2, p53, CCNB1, PHB2, RACK1, RUVBL1, and HER2) already identified in the literature in a large prospective cohort of 240 patients with node-negative early-stage disease or DCIS with 156 healthy controls. Single AAbs demonstrated a weak performance in discriminating breast cancer from healthy controls (AUC ranging from 0.52–0.65). When used as an AAb panel the discrimination power was improved (AUC = 0.82; 95% CI) between the breast cancer and the healthy cohort. The screening test showed 90% sensitivity with 42% specificity. For a different subtype, the panel discriminated node-negative early disease with 51% and DCIS with 32% specificities at 90% sensitivity. Patients younger than 50 years with node-negative early-stage breast cancers showed 59% specificity with 90% fixed sensitivity. However, large-scale trials are needed to further evaluate its potential for early screening and clinical management.
Conclusions and Future Directions
Multiple breast cancer-specific AAbs have been identified for early diagnosis of breast cancer. Although individual AAbs have shown poor performance for population-based screening, autoantibody panels have shown encouraging results. Modern screening digital mammography has a sensitivity of 86.9% for breast cancer screening (9). None of the AAb panels reported so far for breast cancer qualify as standalone screening assays but could be useful in combination with routine mammography screening (58). Moving these promising AAb candidates into clinical use necessitates a rigorous systematic approach. Proper study design, statistical models, and extensive analytical and clinical validation with well-defined quantitative parameters are necessary attributes to develop a useful AAb-based diagnostic screening tool for breast cancer. Future studies should follow the recommended five-phase schema and the ProBE guidelines suggested for all biomarker studies before any clinical evaluation (27, 28). Many AAbs reported for breast cancer early screening have not gone beyond the discovery phase and most lack blinded validation studies. Most of the studies discussed here belong to phase I, preclinical exploratory, or phase II, clinical assay, and validation phases in biomarker development (29, 64, 78, 116). Only a few studies have conducted retrospective longitudinal studies (phase III) where they have attempted to detect preclinical disease (25). It is vital to have newly diagnosed patient sera and samples collected before diagnosis when validating AAb panels to determine the true potential of the markers for early detection. The NCI Early Detection Research Network (NCI-EDRN) has supported early screening studies by developing a multicenter breast cancer reference set of plasma and sera, a precious resource for validation studies. Despite many challenges, AAbs have great potential for early screening of breast cancers and would be useful when used in conjunction with mammography.
Disclosure of Potential Conflicts of Interest
K.S. Anderson reports personal fees from ProvistaDx and nonfinancial support from FlexBioTech outside the submitted work, as well as a patent for breast cancer biomarkers pending, issued, licensed, and with royalties paid from Provista Dx. J. LaBaer reports grants from NIH NCI EDRN during the conduct of the study, as well as a patent for US 9857374 issued and some ownership of Ordinatrix, a small start-up company that produces protein microarrays and provides protein microarray screening services. No potential conflicts of interest were disclosed by the other author.