The field of biomarker medicine is an exponentially growing field of significant scientific, clinical, and financial interest in the United States and globally. This interest has resulted in the creation of hundreds of annual national and international biomarker meetings and millions of dollars in financial support of this research discipline. The clinical impact of accurate and predictive disease biomarkers is now widely appreciated. For example, the success of personalized cancer therapies is dependent on the ability to select both appropriate patients and to monitor their response to therapy longitudinally. The significant financial impact is impressive as well, with the global biomarkers market predicted to grow from $29.3 billion in 2013 to $53.6 billion in 2018 with a 12.8% compound annual growth rate (MarketWatch, 2014, http://www.reportlinker.com/p0488647/Biomarkers-Technologies-and-Global-Markets.html).
Within the context of this review, we examine state-of-the-art approaches to the discovery and validation of cancer biomarkers, with a specific emphasis on protein and protein-associated ones. We review sample selection strategies, currently utilized proteomic approaches for both discovery and validation requirements, and data analysis standards. Finally, we provide examples of these elements of biomarker discovery and validation from our own biomarker research.
By definition, a biomarker is a biological molecule detected in body fluids and tissues that represents normal and/or abnormal biological processes, conditions, and diseases. Nucleic acids, metabolites, proteins, lipids, steroids, and carbohydrates are among the biological molecules that have been identified as potential disease diagnostics and prognostics.1–3 Biomarkers may be used to develop diagnostic tests for the early detection of cancer as well as provide diagnostic and prognostic information. They may guide risk assessment and be valuable for monitoring high-risk populations. Biomarkers may also be useful for patient stratification for targeted therapy and to monitor response to therapy. Specimen type selection is critically important and, unfortunately to date, has predominantly required invasive procedures for sample acquisition. These specimens include blood (serum, plasma), tissues and tissue extracts, cerebrospinal fluid (CSF), nipple aspirates, bronchial lavage, and others.
The importance of biomarker discovery study design and the validation of potential disease biomarkers cannot be overestimated. A number of important guidelines and recommendations, in recent years, have been established to guide researchers with respect to both reporting biomarker research results and advancing their translation to the clinic. For example, REporting recommendations for tumor MARKer prognostic studies (REMARK) have been created to specifically limit the problems that have arisen as a function of poor study design, general methodical differences, inappropriately small study design, and misleading statistical analyses among other issues.4,5 These guidelines have now been adopted by most peer-reviewed journals as a prerequisite for manuscript submission. The guidelines complement those of EDRN (Early Detection Research Network) of the National Cancer Institute, National Institutes of Health, which has recommended a 5-phase approach to the successful clinical translation of biomarker discoveries. These phases are the preclinical exploratory phase, the validation phase, the retrospective longitudinal phase, the prospective screening study phase, and finally the cancer control phase. The reader is referred to the network’s guidelines for complete details of this and other aspects of biomarker discovery and validation (http://edrn.nci.nih.gov/FOA-guidelines).
Our laboratory has had a longstanding interest in the discovery and validation of noninvasive biomarkers for cancer and other diseases. Our goal has been to leverage the discovery and validation of biomarkers for a variety of diseases to provide useful clinical information with respect to disease development, progression, response to therapy, and even risk potential. We have created and maintained one of the largest institutional review board–approved urine repositories available; have established best practice protocols for the collection, storing, and shipping of human and rodent urine samples; and have been fortunate to be joined by a multidisciplinary team of committed scientists and clinicians who have made these accomplishments possible.
Our group made a strategic decision some years ago to focus on urine as our biosample of choice. The reasons for our decision are obvious. Urine tests are inherently noninvasive, thereby increasing the frequency with which they could be conducted while increasing patient compliance.1–3 The levels of unrelated, “contaminating” proteins are lower in urine than those of other biofluids, making discovery protocols more efficient; the profile of the human urinary proteome is known to change with disease status; urine can be easily adapted to high-throughput diagnostic and prognostic techniques; and sample acquisition is simple and economical.
We have utilized two discovery approaches in our studies of the urinary proteome. The first approach, the candidate biomarker approach, is one in which the biology of the disease or condition of interest drives discovery. It is, inherently, a hypothesis-driven, biased approach to identifying novel biomarkers. The second approach utilized by our laboratory is a global proteomics strategy in which we have utilized a variety of mass spectrometry (MS) approaches to identify all proteins that are differentially present, in a significant manner, in urine samples from patients when compared with age- and sex-matched control subjects. This approach is inherently hypothesis-free and unbiased. A small sampling of representative noninvasive, urinary cancer biomarkers that we have discovered and validated using these approaches is reviewed in the following sections.
BIOLOGICAL SAMPLES FOR CANCER BIOMARKER DISCOVERY
A wide variety of biological samples may be used for cancer biomarker discovery via proteomic analysis. We provide a brief discussion of types of samples used and the advantages and disadvantages of each of these. Ultimately, the choice of sample type for cancer biomarker discovery and/or validation will depend on the clinical question being addressed, the availability of clinical samples, and the discovery techniques used.
Serum and plasma are the most commonly used sources of clinical samples for biomarker detection and validation because of availability and clinical convention. Other biofluids including urine, CSF (central nervous system tumors), saliva, ascitis fluid, gastric fluid (gastrointestinal tumors), bronchoalveolar lavage or pleural fluid (lung tumors), and tumor interstitial fluid are rapidly gaining popularity as sources of potential cancer biomarkers. A majority of the biofluids listed above are perhaps more suited for initial biomarker discovery but may not be practical for ultimate clinical use because of difficulty and expense of obtaining patient samples. In contrast, urine is easily available and can be noninvasively collected and readily archived for processing. The urinary proteome is a rich source of disease biomarkers that may be queried using a wide variety of high-throughput analytical approaches.
Biofluids are complex mixtures of proteins derived from the tumor as well as many different normal tissues; therefore, there may be a lack of specificity as to the source of differentially expressed proteins. Highly abundant proteins in biofluids (such as albumin or immunoglobulin G) may mask other low-abundance proteins that are likely to constitute potential cancer biomarkers. This may be overcome by initial sample preprocessing to deplete the sample of the abundant proteins before proteomic analysis. Finally, care must be taken to establish and follow guidelines for biofluid collection, storage, and processing to ensure maintenance of sample integrity and reproducibility of results. In the current era of personalized medicine and/or precision medicine, based on the molecular characteristics of the tumor, it is becoming apparent that continuous disease monitoring is essential for therapeutic success. Therefore, in the future, it is highly likely that liquid biopsy-based biomarker detection approaches will prove to be essential in clinical settings for cancer patient management.
Comparison of protein expression profiles between tumor tissue and normal adjacent tissues can provide important prognostic information for a variety of diseases including cancer. There are several advantages to using tissue for cancer biomarker discovery including the fact that the identified biomarkers would clearly be of tumor origin. Not all putative tumor markers are secreted from tumor cells; therefore, the local concentration may be much higher in the tumor tissue compared with other biofluids, making it easier to detect candidate markers. Techniques such as laser capture microdissection of the tumor epithelial and/or stromal components followed by proteomic analysis can often identify the exact source of the protein biomarker within the tumor microenvironment. However, proteomic studies with tumor tissues have several drawbacks as well. Being an inherently invasive approach, the availability of tumor and/or healthy tissue materials may be limited. Furthermore, tumor tissue is a heterogeneous mixture of malignant cells as well as connective tissue, adipose tissue, and inflammatory cells, presenting technical challenges for sample handling and a lack of tumor specificity for the potential biomarker identified.
In contrast to tissue, tumor cell lines may represent homogenous cell populations that can be cultured easily, can be available in unlimited amounts, and may be easily adapted to a variety of biomarker discovery studies. It is relatively easy to prepare subcellular fractions for study including plasma membrane, cytosolic or nuclear fractions, secretomes (conditioned medium), or exosomes from cultured cells for use in proteomic studies. Tumor cells may also be manipulated in vitro to simulate clinical conditions such as knockdown or overexpression of genes of interest, chemotherapy resistance, and response to therapeutic regimens prior to proteomic analysis. Nevertheless, the use of cell lines also has several limitations including the lack of availability of appropriate tumor cell lines and/or normal epithelial cells to serve as controls, changes in gene or protein expression profile that may occur in 2-dimensional culture conditions, and the lack of interaction of tumor cells with stromal or immune cells that characterize the tumor microenvironment in vivo.
PROTEOMIC TECHNIQUES FOR CANCER BIOMARKER DISCOVERY
It is now widely appreciated that both the human genome and proteome can provide valuable insight into the disease process. The human proteome is more complex than the genome as a consequence of posttranslational modifications, proteolytic processing, alternative splicing, and RNA editing. Proteomics is the large-scale study of proteins that can be not only used to elucidate the complex molecular mechanisms of disease but also applied to the discovery of disease biomarkers. Comparative proteomics of samples from normal healthy individuals and cancer patients often serves as the basis for discovery of candidate cancer biomarkers. The following techniques are commonly used in proteomic-based biomarker discovery workflow (Fig. 1).
Protein identification in complex protein mixtures often requires fractionation. Gel-based approaches including 1-dimensional (1D) and 2-dimensional (2D) gel electrophoresis are the most commonly utilized methods to fractionate and separate proteins in a gel-based matrix. These proteins are then excised from the gel, digested with a protease (typically trypsin), and the resulting peptide fragments analyzed using MS.6
One-dimensional gel electrophoresis. One-dimensional gel electrophoresis is a common separation method used to analyze macromolecules, including proteins. Proteins are solubilized in sodium dodecyl sulfate (SDS) and then separated according to their mass: charge ratio (nondenaturing polyacrylamide gel electrophoresis [PAGE]) or primarily by mass (denaturing SDS-PAGE) in a polyacrylamide gel.7 This method is generally used to resolve proteins between 10 and 300 kd.8 One-dimensional gel electrophoresis has been used in combination with other proteomic and mass spectroscopy methods to successfully identify biomarkers. For example, we have utilized 1D gel electrophoresis in combination with Q-sepharose anion exchange and gelatin-sepharose affinity chromatography followed by unbiased protein identification using MALDI-TOF (matrix-assisted laser desorption/ionization time-of-flight) MS, to identify a disintegrin and metalloprotease 12 [ADAM12], which was elevated in urine of breast cancer patients. In addition, elevated levels of ADAM12 correlated with breast cancer progression, making it a potential noninvasive biomarker for breast cancer.9 We utilized a similar approach in a different study, where 1D gel electrophoresis in combination with column chromatography, zymography, and tandem MS (MS/MS) was used to identify MMP-9 (matrix metalloproteinase 9)/TIMP-1 (tissue inhibitor of metalloproteinase 1) complex, MMP-9 dimer, and ADAMTS-7 (a disintegrin and metalloproteinase with thrombospondin motifs 7) to be elevated in the urine of prostate and bladder cancer patients. These biomarkers provide noninvasive diagnostic and prognostic value in the detection and clinical monitoring of disease progression and therapeutic efficacy in patients with these cancers.10 Many other biomarkers from other groups have been identified using 1D gel electrophoresis in combination with other proteomic methods in various biological fluids, including serum, pleural effusion, plasma, tissue interstitial fluid, ascites, and urine in numerous epithelial cancers.1,2,11
Two-dimensional gel electrophoresis. Two-dimensional gel electrophoresis has become an important primary tool in proteomics research, as it allows proteins to be separated in two dimensions. The first dimension is represented by isoelectric focusing in which solubilized, denatured proteins are separated in polyacrylamide gel strips by their isoelectric points. In the second step, the polyacrylamide gel strip is subsequently analyzed via SDS-PAGE to permit the separation of the focused proteins by molecular weight. Spots generated by proteins are identified by comparing databases of gel maps and by other proteomic methods.12 However, 2D possesses several disadvantages including gel-to-gel variation and low sensitivity to very small (<10 kd) and very large (>150 kd) proteins. Proteins with very basic pI values and with high salt content are also seldom detected.7,12 To overcome this obstacle, 2D differential in-gel electrophoresis was introduced. This method allows separation of two samples on a single gel by first labeling the proteins with two different fluorescent dyes.6 Following electrophoresis, the 2D pattern on the gel is visualized using a fluorescence scanner that scans the sequential excitation wavelengths of the different fluorescent dyes.7 Two-dimensional gel electrophoresis has been successfully used to identify numerous biomarkers from a variety of biofluids, including interstitial fluids, saliva, urine, serum, peripheral blood, gastric fluid, ascites, CSF, pleural effusion, and bronchoalveolar lavage fluid in variety of cancers.1,2,11
Mass spectrometry is a commonly used discovery tool that plays an essential role in biomarker discovery. Mass spectrometry methods provide high sensitivity and specificity to the detection of proteins in various biological fluids and have been utilized in the discovery of numerous protein cancer biomarkers in various cancers,1,2,11 thus representing a cornerstone of protein-based cancer biomarker discovery.
Mass spectrometry detects analyte ions in the gas phase, and the detection is specific to the mass-to-charge ratio (m/z) of the ions.13 One advantage of MS over antibody-based assays is that it enables identification of protein isomers and posttranslational modifications of proteins in addition to the quantification of detected proteins. The two most commonly used ionization techniques are electron spray ionization (ESI) and MALDI. It is advantageous to utilize these “soft” ionization methods during the ionization process when converting large biomolecules into molecular ions, because they mitigate any fragmentation of the analyte.13
In general, two global profiling MS-based strategies are used for protein characterization. In the first strategy, the “top-down” approach, a full-length protein is ionized (mainly by ESI or MALDI), without enzymatic digestion and analyzed by MS. The measured mass corresponds to a specific protein and includes posttranslational modifications. In the second strategy, the “shotgun” proteomics or “bottom-up” approach, proteins are enzymatically digested into smaller peptides (usually with trypsin) and subsequently analyzed by MS. The peptides are identified by peptide mass fingerprinting (usually MALDI-TOF or ESI-TOF) or by MS/MS.13,14
Current MS-based strategies for protein quantification can be divided into two main groups: strategies based on labeling a specific amino acid residue and so-called label-free proteomics.7 From an MS analysis point of view, stable isotope label–based quantitative analysis can be categorized as “isotopic” or “isobaric.”15 Isotopic methods such as SILAC (stable isotope labeling by amino acids in cell culture), ICAT (isotope-coded affinity tag), and ICPL (isotope-coded protein label) quantify peptides at the MS level based on ion intensities of light and heavy isotopes of a peptide. On the other hand, isobaric methods including tandem mass tags and iTRAQ (isobaric tags for relative and absolute quantitation) quantify peptides at the MS/MS level based on comparison of measured reporter ion intensities.16 Label-free LC-MS (liquid chromatography MS) methods are also attractive for high-throughput quantitative proteomics and, compared with labeling-based methods; the sample processing is straightforward and can be scaled to a large number of samples. Label-free quantification is often achieved using peak intensity or spectral counting.16,17
In contrast to global profiling, targeted quantitative proteomics is a candidate-based technique that enables the specific detection of selected analytes in a complex system. This approach is based on the concept of stable isotope dilution and uses isotopically labeled synthetic peptides that mimic endogenous targets and can be used as internal standards to achieve absolute quantification.15 Recently, selected reaction monitoring MS (SRM-MS), which can be used for targeted quantification and posttranslational modifications of proteins, has become increasingly popular in proteomics research.18 This targeted MS approach has emerged to bridge the gap between biomarker discovery and clinical validation.19
A very important part of MS-based strategies is to accurately and quantitatively analyze data to identify specific peptides or proteins. New software, algorithms, and databases have been developed to facilitate data analysis relatively quickly. The most common computational methods for peptide and protein identification include search engines and protein sequence databases, de novo sequencing, and the spectral library searches.20 As a result, multiple computer programs were developed and are now being used worldwide, including Mascot,21 X!Tandem,22,23 Sequest,23 MyriMatch,24 SpectraST,25 OMSSA,26 Andromeda,27 and many others.
Protein microarrays, analogous to gene microarrays, can be used to simultaneously analyze many different proteins in a sample. There are two main types of protein arrays: forward-phase protein arrays or reverse-phase protein arrays. Forward-phase protein arrays, also known as analyte capture array, can use a variety of capture molecules (antibodies, proteins, nucleotides, or aptamers) that can specifically bind the protein of interest. For example, a set of antibodies can be immobilized onto a solid surface or membrane, which can then be queried with tumor lysates, clinical biofluids or tumor cell–conditioned media, in order to identify protein expression or protein phosphorylation status in the samples. In reverse-phase protein arrays, tumor sections, tumor lysates, or other clinical samples are immobilized onto a solid surface or membrane, and specific antibodies are used to detect an epitope or a particular protein modification or structure.28,29 Protein microarrays may be scaled up or automated for high-throughput analysis to provide comparative data on a large number of clinical samples in parallel. We refer the reader to several recent reports on the utility of protein arrays in the proteomic profiling of a variety of cancers.30–33
Activity-Based Proteomic Profiling
Aberrant expression of proteases within the tumor microenvironment promotes tumor growth, metastasis, and angiogenesis.34 Measurement of protease expression and/or function in tumor biopsies and/or body fluids may be used for cancer diagnosis and prognosis as well as to monitor therapeutic responses in patients. In particular, expression and activity of the gelatinases, MMP-2 and MMP-9, are significantly up-regulated in tumor tissue and in the circulation of patients with a variety of cancers. Detection of biological activity (via gelatin zymography) of MMP-2, MMP-9, or MMP-9/NGAL (neutrophil gelatinase-associated lipocalin) complex in urine from cancer patients is an independent predictor of a variety of organ-confined cancers including breast, prostate, bladder, and brain tumors.10,35–38 Similarly, elevated activity and expression of MMP-2 and MMP-9 in serum are significantly associated with node metastasis in breast cancer.39 Serum and saliva MMP-2 activity might also serve as markers for detecting lung cancer.40 In addition to zymography, FRET (fluorescence resonance energy transfer)–based substrates may also be used for the detection of proteolytic activity in clinical specimens. For instance, Flsub21 and Flsub13 (fluorescent-labeled peptide substrates for MMP/ADAM proteases) cleavage activities were detected in the urine of patients with invasive and metastatic breast cancer at significantly higher frequencies compared with urine from control subjects.41 Functional protease profiling, an alternate approach for monitoring tumor-associated protease activity, involves the MS-based detection of fragments of proteolytically degraded proteins in biofluids.42,43 For example, reporter peptides with a known cleavage site for cysteine-endopeptidase were added to serum samples from colorectal cancer patients or control subjects, and subsequently, proteolytic fragments of the reporter peptides were quantified via LC/MS.44 Using this approach, the median concentration of reporter peptide fragments was found to be significantly higher in serum from colorectal cancer patients compared with control subjects.44
In Silico Approaches
The wide availability of data derived from high-throughput omic technologies such as proteomics, MS, microarray, and next-generation sequencing has provided novel avenues for biomarker discovery. Pathway and network-based analyses may be applied to omics data to understand underlying biological functions, protein-protein interactions, and processes perturbed in cancer such as cell signaling and metabolic pathways. These methods can be useful in the discovery of novel biomarkers that may have been otherwise overlooked using other approaches. For instance, in silico modeling of gene expression data from inflammatory breast cancer (IBC) and non-IBC datasets was used to identify novel master regulators that may drive IBC.45 This systems biology approach identified the transcription factor NFAT5 (nuclear factor of activated T cells 5) as a putative biomarker for IBC. Inflammatory breast cancer tumor tissues displayed a higher frequency of NFAT5-positive expression and nuclear accumulation compared with non-IBC tumors via immunohistochemistry (IHC) analysis.45 Text or literature mining has also been suggested as useful approaches in disease biomarker discovery.46,47 Several publicly available software tools such as Ingenuity Pathway Analysis (Qiagen, Valencia, CA), Kyoto Encyclopedia of Genes and Genomes, STRING, and Pathway Studio may be used to integrate text-mining results or other data resources for the identification of candidate biomarkers.48–51 Such in silico approaches are transforming biomarker discovery from identification of individual biomarkers to the global analyses of perturbed pathways and protein networks in cancer.
CANDIDATE BIOMARKER VALIDATION
Rigorous validation studies are necessary in order to translate novel biomarkers into clinical use for improved detection and treatment of cancer. Some of the commonly used approaches for biomarker validation include antibody-based methods (enzyme-linked immunosorbent assay [ELISA], immunoblot, bead-based immunoassays, and protein arrays), IHC and quantitative MS for protein and peptide analysis, and activity-based assays.
Immunohistochemistry permits the visualization of proteins in tissues or cells using an antigen-antibody–based method. Immunohistochemistry, combined with bright-field or fluorescent microscopy, enables the processing of both formalin-fixed paraffin-embedded and frozen tissues and is an important tool for biomarker validation in clinical patient cohorts. One widely used, US Food and Drug Administration–approved IHC application is the test to detect estrogen receptor (ER), progesterone receptor, and HER-2 (human epidermal growth factor receptor 2)/neu expression in breast tumor tissue. The presence of these biomarkers is used to identify breast cancer subtypes and to stratify patients for targeted therapies such as tamoxifen (for estrogen receptor–positive tumors) and trastuzumab (for HER-2–positive tumors). For both biomarker discovery and validation efforts, however, it is important to keep in mind those factors that may influence IHC analysis such as tissue collection, handling and storage, and choice of antibody and detection system. It is also important to adopt recommended guidelines for the interpretation of immunostaining results and IHC scoring.
Enzyme-linked immunosorbent assay is currently the most commonly used biomarker validation approach. Enzyme-linked immunosorbent assay can serve as a highly sensitive and quantitative method for biomarker detection in a variety of sample types including clinical biofluids.52 In general, single-plex ELISAs use a sandwich format, comprising the following sequential steps: a capture antibody that is immobilized, sample addition and binding, detection antibody, and detection via a colorimetric enzyme substrate. For the simultaneous detection of multiple biomarkers in the same sample, multiplex ELISAs may use chemiluminescent/fluorescent reporter systems that allow the analysis of many analytes. Common multiplex ELISA formats include the Luminex Cytometric Bead Array and Bio-PlexPro, in which high-affinity capture ligands are immobilized on fluorescently activated plastic microbeads. After mixing the beads with the clinical sample, detection antibodies labeled with a reporter dye enable high-resolution analysis of specific fluorescent signal via flow cytometry methods.53 One drawback of the immunoassay approach for biomarker validation is that high-quality, commercially manufactured ELISAs for the biomarker of interest may not be available. One alternative is the development of custom ELISA assays; however, for the development of in-house ELISAs assays, it is important to thoroughly characterize the antibodies used (polyclonal or monoclonal), determine the antibody-binding kinetics, and optimize the detection method.
Quantitative Mass Spectrometry
Targeted protein analysis allows simultaneous quantization and identity confirmation of multiple biomarkers within the same LC-MS/MS experiment. Selected reaction monitoring (SRM or multiple selected reaction monitoring [MRM] for multiplex) is a mass spectrometric approach that can monitor one or more specific ion transition/s at high sensitivity.54 Absolute quantization of each protein can be achieved by using isotope-labeled peptides as internal standards. Multiple selected reaction assays may be used to monitor up to 100 proteins simultaneously in human serum, plasma, and urine samples.55–57 To approach the picogram-per-milliliter sensitivity ranges typically possible for ELISAs, MRM assays can be combined with a sample preparation step such as immunoaffinity chromatography.58,59 In the future, MRM assays may prove to be more cost-effective and time saving than ELISA methods for biomarker validation.
URINARY BIOMARKERS FOR HUMAN CANCERS
A longstanding interest of our group has been to identify noninvasive diagnostic and prognostic biomarkers for cancers. As discussed in greater detail in the following sections, we have focused on discovery and validation studies not only of some of the most common human cancers (e.g., breast and prostate cancer) but also of those human cancers that, although they may affect a relatively smaller number of people compared with the more common epithelial cancers, are equally devastating. For example, ovarian and pancreatic cancers are often detected at late, advanced stages because of lack of symptoms and limitations of detection methods at an early stage. Patients with these cancers would have a much better prognosis, better quality of life after treatment, and a greater potential to be cured were there biomarkers that could accurately detect the cancer at an early resectable and/or treatable stage. In addition to our interest in discovering biomarkers that can be utilized in early detection, we have also focused our work on the discovery of biomarkers that could be used to predict cancer recurrence, monitor cancer progression, evaluate therapeutic efficacy, and predict cancer risk.
As discussed previously in this article, our laboratory has made a strategic decision of identifying noninvasive cancer biomarkers in patient urine samples. For the discovery of novel noninvasive cancer biomarkers, our laboratory has utilized both the hypothesis-driven candidate approach as well as the global proteomics approach. Our earlier candidate biomarker work focused on MMPs, which are a family of enzymes actively involved in tumorigenesis, tumor progression, metastasis, and angiogenesis. Their functions include remodeling of extracellular matrix; modulation of cell-cell adhesion and cell-matrix adhesion, which are critical for cancer cell invasion and metastasis; and for the release of certain growth factors that are otherwise tethered to cell surface or matrix and therefore less effective.1,3,34 An initial study from our group demonstrated that the activities of gelatinase-type MMPs (MMP-2 and MMP-9) and their complexes in urine samples, analyzed by studying their activity via substrate gel electrophoresis (zymography), were significantly elevated in patients with a variety of cancers, including breast, prostate, bladder, and renal cancer compared with age- and sex-matched control subjects, and were independent predictors of organ-confined cancer.35 These early studies were eventually validated in large cohorts of human urine samples and confirmed by others as well. Since then, we have expanded that initial study by identifying additional MMP-associated proteins in urine samples and assessing whether these proteins, either alone or when multiplexed, could accurately predict the presence, status, and stage of different human cancers. We have also utilized methods in addition to the activity-based assays to evaluate these noninvasive biomarkers, including immunoblot, ELISA, IHC, and other effective assays. We have also utilized global proteomics approaches to discover novel biomarkers by analyzing the entire proteome of urinary proteins from cancer patients and have not only discovered promising biomarker candidates but have also gained insights into the pathways and mechanisms that are operative at different stages of cancer progression. Our findings using both the biologically driven, candidate marker discovery approach and the global proteomic approach will be discussed in the following sections. We, and others, have comprehensively reviewed other markers for human cancers elsewhere. 1–3,11,60,61
Breast cancer is the most common cancer among women (other than skin cancer) in the United States (American Cancer Society, http://www.cancer.org). At the early stages of breast cancer, cancer cells localize within the breast tissue and the tumors are identified as in situ (e.g., ductal carcinoma in situ). At more advanced stages, cancer cells invade into the neighboring tissues and lymph nodes and, at the most malignant stage, metastasize to distant organs, such as bone, lung, and brain.
In our initial studies, we found that MMP-2, MMP-9, and associated high-molecular-weight (HMW) species that also possess gelatinase activity were independent predictors of cancer status, both organ confined and metastatic, in a variety of cancers including breast cancer.35 One of the HMW species that migrated at ∼125 kd was later identified by our group as being the complex of MMP-9 and NGAL (also referred to as lipocalin 2),62 similar to the complex purified from human neutrophils.63 In an early pilot study, the MMP-9/NGAL complex was demonstrated by our group to be present in ∼86.4% urine samples from breast cancer patients but in none of the control subjects tested.36 We have also shown that NGAL monomer levels are elevated in urine samples from metastatic breast cancer patients and may be a potential urinary biomarker for metastatic breast cancer.64 Over time, we have confirmed these findings in large sample cohorts of urine from breast cancer patients at different stages compared with age- and sex-matched control subjects.
Subsequently, we utilized a classic proteomic approach, combining chromatography, zymography, and MS to identify other unknown HMW species. ADAM12 was identified as one of these urinary species.9 At the time, immunoblot analyses were used to validate this identification given the absence of monospecific ELISA for this protein. ADAM12 was detected both more frequently and at higher levels in the urine samples from breast cancer patients than in samples from age-matched control subjects in this early study. Most interestingly, ADAM12 levels were significantly correlated with breast cancer progression with the lowest level found in urine samples from patients with precursor lesions atypical ductal hyperplasia/lobular carcinoma in situ (LCIS), with increasing levels detected in urine from women with ductal carcinoma in situ and locally invasive cancer samples and the highest levels detected in the urine of metastatic samples.9 Subsequently, we have also identified the other HMW gelatinase species of ∼140, 190, and greater than 220 kd as the complex of MMP-9/TIMP-1, MMP-9 dimer, and ADAMTS-7, respectively.10
Identifying patients at high risk of developing breast cancer in order to take advantage of potential risk reduction measures is critically important in breast cancer prevention.65 Small precursor lesions that can lead to malignant disease are often below the limit of current detection methods, including mammography, magnetic resonance imaging, and palpation.66,67 Atypical ductal hyperplasia and LCIS are such precursor marker lesions indicative of increased breast cancer risk. The GAIL model is a widely used tool in breast cancer risk assessment. This model calculates a woman’s risk of developing breast cancer by combining several risk factors for this disease, including age, race, family history, and number of biopsies.68,69
To identify biomarkers that can noninvasively predict breast cancer risk and identify high-risk populations, we analyzed a panel of urinary MMP candidates including MMP-2, MMP-9, MMP-9 dimer, MMP-9/NGAL complex, and ADAM12 in 148 samples, including those from women with atypical ductal hyperplasia, LCIS, and age-matched control subjects.38 Both MMP-9 and ADAM12 levels were significantly increased in samples with the risk marker lesions with ADAM12 distinguishing atypical hyperplasia or LCIS samples from control subjects with excellent accuracy (receiver operating characteristic [ROC] area under curve [AUC] of 0.914 and 0.950, respectively). Most importantly, we developed a predictive index that combined ADAM12 urinary levels with GAIL 5-year risk scores, which yielded excellent risk assessment capability with index scores greater than 2.8 predicting atypical hyperplasia at a sensitivity of 0.976 and a specificity of 0.977. Taken together, we concluded that these novel urinary biomarkers can be used clinically in conjunction with the GAIL score to even more accurately identify women at high risk of breast cancer.
We have also conducted a global proteomics search for the urinary biomarkers that can differentiate patients with metastatic breast cancer from patients with LCIS lesions and from control subjects. We have utilized quantitative time-of-flight LC-MS/MS to identify proteins in urine samples from the 3 groups and identified ∼117 proteins to be uniquely present in the metastatic breast cancer samples, ∼29 unique proteins in the LCIS group, and ∼25 unique proteins in the control subject group (unpublished data). Interestingly, bioinformatics analyses have recently revealed distinct patterns associated with each of the 3 groups. For example, urinary proteins in the metastatic group were enriched in pathways associated with cell proliferation, angiogenesis, extracellular matrix remodeling, invasion, and metastasis, whereas proteins associated with host immunity and defense were significantly reduced in both of the disease groups compared with the control subject group. This study is ongoing in our laboratory.
Prostate cancer is the most frequently diagnosed cancer (after skin cancer) among men (American Cancer Society, http://www.cancer.org). Patients with prostate cancer often share similar symptoms with patients with other prostate diseases, such as benign prostatic hyperplasia (BPH).70–74 Currently, age, digital rectal examination, and PSA (prostate-specific antigen) levels are used to evaluate prostate cancer risk before biopsy; however, these criteria cannot distinguish between BPH and prostate cancer75 and may lead to unnecessary biopsies and severe adverse effects.76 For example, PSA levels are frequently increased in benign diseases such as BPH,77 whereas 6.6% of patients with extremely low PSA levels (<0.5 ng/mL) actually had prostate cancer.78 Therefore, although both the treatment and prognosis of BPH and prostate cancer are drastically different, current detection methods cannot reliably distinguish between these two diseases, representing a significant clinical challenge.
To meet the urgent need for biomarkers that can differentiate BPH from prostate cancer, we compared the global urinary proteome of men with BPH to that of men with localized and, in the majority of the cases, early prostate cancer using quantitative iTRAQ LC-MS/MS analysis and identified, with high confidence, 25 proteins whose levels were significantly different between the two groups of patients.79 Three proteins, β2 M (β2-microglobulin), PGA3 (pepsinogen 3), and MUC3 (mucin 3), were further validated using immunoblot analyses in urine samples from BPH patients and samples from prostate cancer patients. These proteins, either alone or when multiplexed, showed significant sensitivity and specificity in differentiating between patients with BPH and those with early prostate cancer.
Our group has also conducted studies aimed at noninvasively monitoring therapeutic efficacy in prostate cancer. In a proof-of-principle study, we focused on longitudinal urine samples from cancer patients both at diagnosis and at different points during and after treatment. Analysis of these longitudinal samples would inform us whether any of the urinary biomarkers correlate with cancer recurrence, predict patient survival, and help the evaluation of therapeutic efficacy. In this study, we analyzed MMP activity and levels of the proangiogenic growth factor, vascular endothelial growth factor (VEGF), in 242 urine samples from 65 cancer patients (including 42 with prostate cancer) at first evaluation, during radiation therapy, and at follow-up, as well as 16 samples from control subjects.80 In keeping with our findings in breast cancer showing the correlation between MMPs and cancer status and progression,9,35,36,64 MMP activity, as well as VEGF levels, was significantly higher in cancer patients than in control subjects and was significantly higher in patients with metastatic cancer than in patients with local disease. Importantly, trends in MMPs and VEGF during and after radiotherapy significantly correlated with disease presence after therapy and with patient 1-year progression-free survival, suggesting that longitudinal changes in these urinary markers may also predict patient survival after therapy.
During the course of studying gelatinase-type MMPs in urine samples, we found that samples from different cancers had distinct MMP fingerprints. We analyzed urine samples from patients with organ-confined prostate cancer and samples with organ-confined bladder cancer, as well as samples from control subjects and found that MMP-9 dimer and monomer were multivariable predictors for discriminating prostate cancer from bladder cancer.10 The frequency of MMP-9 dimer in samples of bladder cancer was 1.5-fold higher than those of prostate cancer, whereas MMP-9 was 2-fold more frequent in samples of prostate cancer. These findings suggest that, in addition to predicting cancer presence, the tumor-specific urinary MMP fingerprints may also facilitate the diagnosis of tumor type.
Central nervous system tumors, including brain tumors and spinal cord tumors, are the second most diagnosed cancer type in children (American Cancer Society, http://www.cancer.org). Although brain tumors are not as common in adults, the number of patients with primary and metastatic brain tumors is steadily increasing.81 The complete lack of accurate markers to detect the disease is one of the biggest obstacles in treating brain tumors.
We collected urine samples from both pediatric patients and adult patients with a diagnosis of primary glial tumors (glioblastoma, anaplastic astrocytoma, fibrillary astrocytoma, and pilocytic astrocytoma) and other primary central nervous system tumors.37,82 We found in this training set of samples that levels of MMP-2, MMP-9, MMP-9/NGAL complex, and VEGF, as analyzed by ELISA, were significantly increased in urine samples from brain tumor patients compared with age- and sex-matched control subjects. Among these markers, MMP-2 and VEGF, as revealed by multiple stepwise logistic regression analysis, significantly differentiated brain tumor samples from control samples with excellent accuracy (AUC for MMP-2, 0.894; AUC for VEGF, 0.965), and the multiplexing of these two markers performed even better in predicting the tumor presence than either marker alone. We detected the same elevated MMP species in CSF and brain tumor tissues, demonstrating that the source of these marker proteins was the brain tumor tissue. In the same study, we also analyzed a series of longitudinal urine samples collected from patients at the time of diagnosis and at different time points after surgery. In all patients, increased urinary MMP activity at the time of diagnosis was followed by clearance of the MMP activity either immediately after surgery or one year after, precisely reflecting the tumor presence as documented by imaging. This study demonstrates that MMPs, their associated proteins, and VEGF are novel noninvasive urinary markers that can predict the presence and recurrence of brain tumors.
Although pancreatic cancer is accountable for only 3% of all cancers in the United States (American Cancer Society, http://www.cancer.org), patients with a diagnosis of this disease have extremely dismal prognoses. Approximately 90% of pancreatic cancer patients receive a diagnosis of exocrine pancreatic ductal adenocarcinoma (PDAC).83 The 5-year survival rate of patients with PDAC is only ∼5%, and the median survival time is less than 6 months.84 Another subcategory of pancreatic cancer is pancreatic neuroendocrine tumors (pNETs). Although this cancer is represented at a much lower frequency (∼1%–2%)85 and is associated with a better prognosis, the disease is incurable once the tumor progresses to the unresectable metastatic stage.86 Diagnosis of PDAC or pNETs at an early stage is difficult because of the lack of specific symptoms and effective detection methods. CA-19-9 and chromogranin A are 2 markers most widely used for PDAC and pNETs, respectively. However, neither marker is satisfactory in terms of sensitivity and specificity: their levels are often increased in diseases other than pancreatic cancer and sometimes not elevated in patients who do have pancreatic cancer.87–92 There is therefore an unmet and urgent need for accurate and reliable markers to detect PDAC and pNETs.
In a recent proof-of-principle study, we analyzed urinary MMP-2 and TIMP-1 levels in PDAC samples, pNET samples, and age- and sex-matched control subject samples.93 Using multivariable logistic regression analysis, urinary MMP-2 was found to be an independent predictor that distinguished either PDAC or pNET samples from control subjects. Urinary TIMP-1 not only differentiated PDAC samples from control subjects but also differentiated PDAC samples from pNET samples. Analysis of MMP-2 and TIMP-1 in tissues using IHC confirmed that levels of both proteins were significantly higher in PDAC tissues than the normal pancreas. Taken together, in this study, we identified noninvasive biomarkers that may not only facilitate the diagnosis of pancreatic malignancies but also distinguish the subgroups of this disease.
Gastric cancer is the second leading cause of all cancer deaths in the world.94 Early detection of gastric cancer is critically important both for the survival of the patients and the quality of life for patients with this disease after treatment. The 5-year survival rate for patients at the earliest stage of gastric cancer (T1a) is greater than 90%; however, the rate drops drastically to 15% for patients at stage IV, the most advanced stage with metastasis.95 The standard treatment for early-stage gastric cancer is endoscopic resection, which can cause few complications and gives patients a much better life quality compared with partial or total gastrectomy applied to patients at more advanced stages.
To discover noninvasive biomarkers for the detection of early gastric cancer, we conducted a pilot study that followed the REMARK4 and STROBE guidelines96 with urine samples from a patient cohort enriched with stage IA cancer (62.9%) as well as age- and sex-matched control subjects. Importantly, most of these stage IA patients (81.8%) could be treated by endoscopy.97 We found that urinary levels of MMP-9/NGAL complex and ADAM12 as analyzed with ELISA were significantly elevated in the cancer group compared with the control group, and we also demonstrated with multivariate analysis that both markers were significant independent biomarkers for gastric cancer. Furthermore, in this study, multiplexing of MMP-9/NGAL and ADAM12, as revealed by multivariate logistic regression analysis, significantly distinguished gastric cancer samples from control subject samples with an AUC of 0.825 in a ROC analysis. We confirmed with IHC that levels of MMP-9, NGAL, and ADAM12 were all significantly up-regulated in cancer tissues compared with adjacent normal tissues, consistent with the source of these biomarkers being the cancer tissue.
Ovarian cancer is the most deadly cancer of the female reproductive system and the fifth leading cause of cancer deaths among women (American Cancer Society, http://www.cancer.org). Approximately 70% of all ovarian cancer patients receive a diagnosis of advanced FIGO (International Federation of Gynecology and Obstetrics) stages III and IV disease because of lack of symptoms at early stages.98 The 5-year survival rate drops drastically from 90% in patients with local cancer (stage I) to only 17% in patients with metastatic disease (stage IV). These dismal statistics highlight the significant clinical need for accurate biomarkers of ovarian cancer.
CA-125 (cancer antigen 125) is currently the only widely accepted biomarker for monitoring ovarian cancer recurrence and evaluating ovarian tumor response to therapies.99 However, about half of stage I and II patients and 20% of late-stage patients do not have elevated CA-125 levels. Their CA-125 levels are within or less than the reference of 35 U/mL and are therefore clinically uninstructive.100–102 This cohort of ovarian cancer patients that has CA-125 levels within the reference completely lack useful diagnostic tests to inform them of their disease status and allow for timely treatments, which can potentially lead to improved survival.
To begin to help overcome these challenges, we analyzed the levels of MMP-2, MMP-9, and NGAL using ELISA in urine samples from ovarian cancer patients, including patients with normal CA-125 levels (<35 U/mL), and age-matched control subjects.103 All three markers could significantly distinguish between ovarian cancer patients with normal CA-125 and the control subjects as determined by ROC-AUC analysis. Multivariate logistic regression analysis revealed that multiplexing MMP-2 and MMP-9 and age provided a highly significant diagnostic accuracy with an AUC of 0.820 in patients with normal CA-125 and with an AUC of 0.881 in all ovarian cancer samples tested, completely independent of CA-125 levels. This study therefore identified noninvasive biomarkers that can aid in the diagnosis of ovarian cancer in that patient cohort in which CA-125 levels are not informative and no other biomarker is available, as well as for all women with ovarian cancer.
As noted earlier, the full impact of current clinical advances and the practical success of the emerging field of precision medicine are dependent on the discovery and validation of sensitive and accurate biomarkers that can enable appropriate and rigorous sample type and patient selection, reliable longitudinal monitoring of therapeutic efficacy, and even early detection and risk assessment. As discoveries such as ours and those of others are validated in even larger cohort studies, their potential to provide useful diagnostic and prognostic information to clinicians and to their patients can be realized in a variety of settings including at point of care, in clinical laboratories, and eventually by the patients themselves. The noninvasive nature of our own biomarker work uniquely enables the valuable opportunity to test frequently and economically and also provides the opportunity to bring cancer diagnostics and prognostics to those parts of the world where access to reliable cancer testing is extremely limited or even nonexistent.
This work was supported by NIH R01CA185530, The Ellison Foundation and The Breast Cancer Research Foundation. Due to space constraints, we have, in some cases, been relegated to citing review articles as opposed to the original papers.
1. Roy R, Yang J, Moses MA. Matrix metalloproteinases as novel biomarkers and potential therapeutic targets in human cancer. J Clin Oncol
. 2009; 27: 5287–5297.
2. Roy RCC, Yang J, Moses MA. Biomarkers in cancer biology. In: Vidaya V, Bonventre J, eds. Biomarkers of Toxicity: A New Era in Medicine
. New York: John Wiley and Sons; 2010: 355–380.
3. Jia D, Roy R, Moses MA. MMPs in biology and medicine. In: Sagi I, Gaffney J, eds. Matrix Metalloproteinase Biology
. New York: John Wiley and Sons, Inc; 2015. In press.
4. McShane LM, Altman DG, Sauerbrei W, et al. REporting recommendations for tumour MARKer prognostic studies (REMARK). Eur J Cancer
. 2005; 41: 1690–1696.
5. McShane LM, Altman DG, Sauerbrei W, et al. Reporting recommendations for tumor marker prognostic studies. J Clin Oncol
. 2005; 23: 9067–9072.
6. Filip S, Zoidakis J, Vlahou A, et al. Advances in urinary proteome analysis and applications in systems biology. Bioanalysis
. 2014; 6: 2549–2569.
7. Alvarez-Chaver P, Otero-Estevez O, Paez de la Cadena M, et al. Proteomics for discovery of candidate colorectal cancer biomarkers
. World J Gastroenterol
. 2014; 20: 3804–3824.
8. Graves PR, Haystead TA. Molecular biologist’s guide to proteomics. Microbiol Mol Biol Rev
. 2002; 66: 39–63; table of contents.
9. Roy R, Wewer UM, Zurakowski D, et al. ADAM 12 cleaves extracellular matrix proteins and correlates with cancer status and stage. J Biol Chem
. 2004; 279: 51323–51330.
10. Roy R, Louis G, Loughlin KR, et al. Tumor-specific urinary matrix metalloproteinase fingerprinting: identification of high molecular weight urinary matrix metalloproteinase species. Clin Cancer Res
. 2008; 14: 6610–6617.
11. Coticchia CM, Yang J, Moses MA. Ovarian cancer biomarkers
: current options and future promise. J Natl Compr Cancer Netw
. 2008; 6: 795–802.
12. Pastor MD, Nogal A, Molina-Pinelo S, et al. Proteomic biomarkers in lung cancer. Clin Transl Oncol
. 2013; 15: 671–682.
13. Scherl A. Clinical protein mass spectrometry. Methods
. 2015; 81: 3–14.
14. Clauser KR, Baker P, Burlingame AL. Role of accurate mass measurement (+/− 10 ppm) in protein identification strategies employing MS or MS/MS and database searching. Anal Chem
. 1999; 71: 2871–2882.
15. Pan S, Brentnall TA, Kelly K, et al. Tissue proteomics in pancreatic cancer study: discovery, emerging technologies, and challenges. Proteomics
. 2013; 13: 710–721.
16. Ansari D, Aronsson L, Sasor A, et al. The role of quantitative mass spectrometry in the discovery of pancreatic cancer biomarkers
for translational science. J Transl Med
. 2014; 12: 87.
17. Sandin M, Chawade A, Levander F. Is label-free LC-MS/MS ready for biomarker discovery? Proteomics Clin Appl
. 2015; 9: 289–294.
18. Elschenbroich S, Kislinger T. Targeted proteomics by selected reaction monitoring mass spectrometry: applications to systems biology and biomarker discovery. Mol Biosyst
. 2011; 7: 292–303.
19. Chambers AG, Percy AJ, Simon R, et al. MRM for the verification of cancer biomarker proteins: recent applications to human plasma and serum. Expert Rev Proteomics
. 2014; 11: 137–148.
20. Perez-Riverol Y, Wang R, Hermjakob H, et al. Open source libraries and frameworks for mass spectrometry based proteomics: a developer’s perspective. Biochim Biophys Acta
. 1844; 2014 (1 Pt A): 63–76.
21. Perkins DN, Pappin DJ, Creasy DM, et al. Probability-based protein identification by searching sequence databases using mass spectrometry data. Electrophoresis
. 1999; 20: 3551–3567.
22. Craig R, Beavis RC. TANDEM: matching proteins with tandem mass spectra. Bioinformatics
. 2004; 20: 1466–1467.
23. Eng JK, McCormack AL, Yates JR. An approach to correlate tandem mass spectral data of peptides with amino acid sequences in a protein database. J Am Soc Mass Spectrom
. 1994; 5: 976–989.
24. Tabb DL, Fernando CG, Chambers MC. MyriMatch: highly accurate tandem mass spectral peptide identification by multivariate hypergeometric analysis. J Proteome Res
. 2007; 6: 654–661.
25. Lam H, Aebersold R. Building and searching tandem mass (MS/MS) spectral libraries for peptide identification in proteomics. Methods
. 2011; 54: 424–431.
26. Geer LY, Markey SP, Kowalak JA, et al. Open mass spectrometry search algorithm. J Proteome Res
. 2004; 3: 958–964.
27. Cox J, Neuhauser N, Michalski A, et al. Andromeda: a peptide search engine integrated into the MaxQuant environment. J Proteome Res
. 2011; 10: 1794–1805.
28. Lin JC. Protein microarrays for cancer diagnostics
and therapy. Medical principles and practice. Int J Kuwait Univ
. 2010; 19: 247–254.
29. Hu B, Niu X, Cheng L, et al. Discovering cancer biomarkers
from clinical samples by protein microarrays. Proteomics Clin Appl
. 2015; 9: 98–110.
30. Kobel M, Madore J, Ramus SJ, et al. Evidence for a time-dependent association between FOLR1 expression and survival from ovarian carcinoma: implications for clinical testing. An Ovarian Tumour Tissue Analysis consortium study. Br J Cancer
. 2014; 111: 2297–2307.
31. Sonntag J, Schluter K, Bernhardt S, et al. Subtyping of breast cancer using reverse phase protein arrays. Expert Rev Proteomics
. 2014; 11: 757–770.
32. Jakobsen KR, Paulsen BS, Baek R, et al. Exosomal proteins as potential diagnostic markers in advanced non–small cell lung carcinoma. J Extracell Vesicles
. 2015; 4: 26659.
33. Qin R, Smyrk TC, Reed NR, et al. Combining clinicopathological predictors and molecular biomarkers in the oncogenic K-RAS/Ki67/HIF-1alpha pathway to predict survival in resectable pancreatic cancer. Br J Cancer
. 2015; 112: 514–522.
34. Roy R, Zhang B, Moses MA. Making the cut: protease-mediated regulation of angiogenesis. Exp Cell Res
. 2006; 312: 608–622.
35. Moses MA, Wiederschain D, Loughlin KR, et al. Increased incidence of matrix metalloproteinases in urine of cancer patients. Cancer Res
. 1998; 58: 1395–1399.
36. Fernandez CA, Yan L, Louis G, et al. The matrix metalloproteinase-9/neutrophil gelatinase-associated lipocalin complex plays a role in breast tumor growth and is present in the urine of breast cancer patients. Clin Cancer Res
. 2005; 11: 5390–5395.
37. Smith ER, Zurakowski D, Saad A, et al. Urinary biomarkers predict brain tumor presence and response to therapy. Clin Cancer Res
. 2008; 14: 2378–2386.
38. Pories SE, Zurakowski D, Roy R, et al. Urinary metalloproteinases: noninvasive biomarkers
for breast cancer risk assessment. Cancer Epidemiol Biomarkers Prev
. 2008; 17: 1034–1042.
39. Heo DS, Choi H, Yeom MY, et al. Serum levels of matrix metalloproteinase-9 predict lymph node metastasis in breast cancer patients. Oncol Rep
. 2014; 31: 1567–1572.
40. Ali-Labib R, Louka ML, Galal IH, et al. Evaluation of matrix metalloproteinase-2 in lung cancer. Proteomics Clin Appl
. 2014; 8: 251–257.
41. Roy R, Zurakowski D, Pories S, et al. Potential of fluorescent metalloproteinase substrates for cancer detection. Clin Biochem
. 2011; 44: 1434–1439.
42. Peccerella T, Lukan N, Hofheinz R, et al. Endoprotease profiling with double-tagged peptide substrates: a new diagnostic approach in oncology. Clin Chem
. 2010; 56: 272–280.
43. Findeisen P, Neumaier M. Functional protease profiling for diagnosis of malignant disease. Proteomics Clin Appl
. 2012; 6: 60–78.
44. Findeisen P, Costina V, Yepes D, et al. Functional protease profiling with reporter peptides in serum specimens of colorectal cancer patients: demonstration of its routine diagnostic applicability. J Exp Clin Cancer Res
. 2012; 31: 56.
45. Remo A, Simeone I, Pancione M, et al. Systems biology analysis reveals NFAT5 as a novel biomarker and master regulator of inflammatory breast cancer. J Transl Med
. 2015; 13: 138.
46. Elkin PL, Tuttle MS, Trusko BE, et al. BioProspecting: novel marker discovery obtained by mining the bibleome. BMC Bioinform
. 2009; 10: S9.
47. Wang J, Zuo Y, Man YG, et al. Pathway and network approaches for identification of cancer signature markers from omics data. J Cancer Educ
. 2015; 6: 54–65.
48. Guo X, Hao Y, Kamilijiang M, et al. Potential predictive plasma biomarkers for cervical cancer by 2D-DIGE proteomics and Ingenuity Pathway Analysis. Tumour Biol
. 2015; 36: 1711–1720.
49. Poisson LM, Munkarah A, Madi H, et al. A metabolomic approach to identifying platinum resistance in ovarian cancer. J Ovarian Res
. 2015; 8: 13.
50. Tenga MJ, Lazar IM. Proteomic study reveals a functional network of cancer markers in the G1-stage of the breast cancer cell cycle. BMC Cancer
. 2014; 14: 710.
51. Rogers A, Smith MJ, Doolan P, et al. Invasive markers identified by gene expression profiling in pancreatic cancer. Pancreatology
. 2012; 12: 130–140.
52. Solier C, Langen H. Antibody-based proteomics and biomarker research—current status and limitations. Proteomics
. 2014; 14: 774–783.
53. Tighe PJ, Ryder RR, Todd I, et al. ELISA in the multiplex era: potentials and pitfalls. Proteomics Clin Appl
. 2015; 9: 406–422.
54. Anderson L, Hunter CL. Quantitative mass spectrometric multiple reaction monitoring assays for major plasma proteins. Mol Cell Proteomics
. 2006; 5: 573–588.
55. Percy AJ, Yang J, Hardie DB, et al. Precise quantitation of 136 urinary proteins by LC/MRM-MS using stable isotope labeled peptides as internal standards for biomarker discovery and/or verification studies. Methods
. 2015; 81: 24–33.
56. Abbatiello SE, Schilling B, Mani DR, et al. Large-scale inter-laboratory study to develop, analytically validate and apply highly multiplexed, quantitative peptide assays to measure cancer-relevant proteins in plasma [published online ahead of print February 18, 2015]. Mol Cell Proteomics
57. Li XJ, Lee LW, Hayward C, et al. An integrated quantification method to increase the precision, robustness, and resolution of protein measurement in human plasma samples. Clin Proteomics
. 2015; 12: 3.
58. Whiteaker JR, Zhao L, Anderson L, et al. An automated and multiplexed method for high throughput peptide immunoaffinity enrichment and multiple reaction monitoring mass spectrometry-based quantification of protein biomarkers. Mol Cell Proteomics
. 2010; 9: 184–196.
59. Whiteaker JR, Zhao L, Lin C, et al. Sequential multiplexed analyte quantification using peptide immunoaffinity enrichment coupled to mass spectrometry. Mol Cell Proteomics
. 2012; 11: M111 015347.
60. Prensner JR, Rubin MA, Wei JT, et al. Beyond PSA: the next generation of prostate cancer biomarkers
. Sci Transl Med
. 2012; 4: 127rv3.
61. Bauca JM, Martinez-Morillo E, Diamandis EP. Peptidomics of urine and other biofluids for cancer diagnostics
. Clin Chem
. 2014; 60: 1052–1061.
62. Yan L, Borregaard N, Kjeldsen L, et al. The high molecular weight urinary matrix metalloproteinase (MMP) activity is a complex of gelatinase B/MMP-9 and neutrophil gelatinase-associated lipocalin (NGAL). Modulation of MMP-9 activity by NGAL. J Biol Chem
. 2001; 276: 37258–37265.
63. Kjeldsen L, Johnsen AH, Sengelov H, et al. Isolation and primary structure of NGAL, a novel protein associated with human neutrophil gelatinase. J Biol Chem
. 1993; 268: 10425–10432.
64. Yang J, Bielenberg DR, Rodig SJ, et al. Lipocalin 2 promotes breast cancer progression. Proc Natl Acad Sci U S A
. 2009; 106: 3913–3918.
65. Hollingsworth AB, Singletary SE, Morrow M, et al. Current comprehensive assessment and management of women at increased risk for breast cancer. Am J Surg
. 2004; 187: 349–362.
66. Kronqvist P, Kuopio T, Nykanen M, et al. Predicting aggressive outcome in T1N0M0 breast cancer. Br J Cancer
. 2004; 91: 277–281.
67. Verschraegen C, Vinh-Hung V, Cserni G, et al. Modeling the effect of tumor size in early breast cancer. Ann Surg
. 2005; 241: 309–318.
68. Gail MH, Brinton LA, Byar DP, et al. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst
. 1989; 81: 1879–1886.
69. Costantino JP, Gail MH, Pee D, et al. Validation studies for models projecting the risk of invasive and total breast cancer incidence. J Natl Cancer Inst
. 1999; 91: 1541–1548.
70. Abrams P. New words for old: lower urinary tract symptoms for “prostatism”. BMJ
. 1994; 308: 929–930.
71. Bosch JL, Hop WC, Kirkels WJ, et al. Natural history of benign prostatic hyperplasia: appropriate case definition and estimation of its prevalence in the community. Urology
. 1995; 46(3 suppl A): 34–40.
72. Guess HA. Epidemiology and natural history of benign prostatic hyperplasia. Urol Clin North Am
. 1995; 22: 247–261.
73. Eckhardt MD, van Venrooij GE, van Melick HH, et al. Prevalence and bothersomeness of lower urinary tract symptoms in benign prostatic hyperplasia and their impact on well-being. J Urol
. 2001; 166: 563–568.
74. Siegel R, DeSantis C, Virgo K, et al. Cancer treatment and survivorship statistics, 2012. CA Cancer J Clin
. 2012; 62: 220–241.
75. Yang CC, Sun SS, Lin CY, et al. Differentiation of prostate cancer and benign prostatic hyperplasia: the clinical value of 201Tl SPECT—a pilot study. Ann Nucl Med
. 2003; 17: 521–524.
76. Basch E, Oliver TK, Vickers A, et al. Screening for prostate cancer with prostate-specific antigen testing: American Society of Clinical Oncology Provisional Clinical Opinion. J Clin Oncol
. 2012; 30: 3020–3025.
77. Mazzucchelli R, Colanzi P, Pomante R, et al. Prostate tissue and serum markers. Adv Clin Path
. 2000; 4: 111–120.
78. Thompson IM, Pauler DK, Goodman PJ, et al. Prevalence of prostate cancer among men with a prostate-specific antigen level < or =4.0 ng per milliliter. N Engl J Med
. 2004; 350: 2239–2246.
79. Jedinak A, Curatolo A, Zurakowski D, et al. Novel non-invasive biomarkers that distinguish between benign prostate hyperplasia and prostate cancer. BMC Cancer
. 2015; 15: 259.
80. Chan LW, Moses MA, Goley E, et al. Urinary VEGF and MMP levels as predictive markers of 1-year progression-free survival in cancer patients treated with radiation therapy: a longitudinal study of protein kinetics throughout tumor progression and therapy. J Clin Oncol
. 2004; 22: 499–506.
81. Barker FG 2nd, Curry WT Jr, Carter BS. Surgery for primary supratentorial brain tumors in the United States, 1988 to 2000: the effect of provider caseload and centralization of care. Neuro Oncol
. 2005; 7: 49–63.
82. Smith ER, Manfredi M, Scott RM, et al. A recurrent craniopharyngioma illustrates the potential usefulness of urinary matrix metalloproteinases as noninvasive biomarkers
: case report. Neurosurgery
. 2007; 60: E1148–E1149; discussion E9.
83. Kaur S, Baine MJ, Jain M, et al. Early diagnosis of pancreatic cancer: challenges and new developments. Biomark Med
. 2012; 6: 597–612.
84. Maitra A, Hruban RH. Pancreatic cancer. Annu Rev Pathol
. 2008; 3: 157–188.
85. Plockinger U, Rindi G, Arnold R, et al. Guidelines for the diagnosis and treatment of neuroendocrine gastrointestinal tumours. A consensus statement on behalf of the European Neuroendocrine Tumour Society (ENETS). Neuroendocrinology
. 2004; 80: 394–424.
86. Yao JC, Hassan M, Phan A, et al. One hundred years after “carcinoid”: epidemiology of and prognostic factors for neuroendocrine tumors in 35,825 cases in the United States. J Clin Oncol
. 2008; 26: 3063–3072.
87. Yeo TP, Hruban RH, Leach SD, et al. Pancreatic cancer. Curr Probl Cancer
. 2002; 26: 176–275.
88. Ozkan H, Kaya M, Cengiz A. Comparison of tumor marker CA 242 with CA 19-9 and carcinoembryonic antigen (CEA) in pancreatic cancer. Hepatogastroenterology
. 2003; 50: 1669–1674.
89. Goggins M. Molecular markers of early pancreatic cancer. J Clin Oncol
. 2005; 23: 4524–4531.
90. Ballehaninna UK, Chamberlain RS. The clinical utility of serum CA 19-9 in the diagnosis, prognosis and management of pancreatic adenocarcinoma: an evidence based appraisal. J Gastrointest Oncol
. 2012; 3: 105–119.
91. Modlin IM, Gustafsson BI, Moss SF, et al. Chromogranin A—biological function and clinical utility in neuro endocrine tumor disease. Ann Surg Oncol
. 2010; 17: 2427–2443.
92. Zatelli MC, Torta M, Leon A, et al. Chromogranin A as a marker of neuroendocrine neoplasia: an Italian multicenter study. Endocr Relat Cancer
. 2007; 14: 473–482.
93. Roy R, Zurakowski D, Wischhusen J, et al. Urinary TIMP-1 and MMP-2 levels detect the presence of pancreatic malignancies. Br J Cancer
. 2014; 111: 1772–1779.
94. Ferlay J, Shin HR, Bray F, et al. Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008. Int J Cancer
. 2010; 127: 2893–2917.
95. Nashimoto A, Akazawa K, Isobe Y, et al. Gastric cancer treated in 2002 in Japan: 2009 annual report of the JGCA nationwide registry. Gastric Cancer
. 2013; 16: 1–27.
96. Vandenbroucke JP, von Elm E, Altman DG, et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration. Epidemiology
. 2007; 18: 805–835.
97. Shimura T, Dagher A, Sachdev M, et al. Urinary ADAM12 and MMP-9/NGAL complex detect the presence of gastric cancer. Cancer Prev Res (Phila)
. 2015; 8: 240–248.
98. Liu J, Matulonis U. Rational use of cytotoxic chemotherapy for recurrent ovarian cancer. J Natl Compr Cancer Netw
. 2006; 4: 947–953.
99. Riedinger JM, Bonnetain F, Basuyau JP, et al. Change in CA 125 levels after the first cycle of induction chemotherapy is an independent predictor of epithelial ovarian tumour outcome. Ann Oncol
. 2007; 18: 881–885.
100. Jacobs I, Bast RC Jr. The CA 125 tumour-associated antigen: a review of the literature. Hum Reprod
. 1989; 4: 1–12.
101. Niloff JM, Knapp RC, Schaetzl E, et al. CA125 antigen levels in obstetric and gynecologic patients. Obstet Gynecol
. 1984; 64: 703–707.
102. Paramasivam S, Tripcony L, Crandon A, et al. Prognostic importance of preoperative CA-125 in International Federation of Gynecology and Obstetrics stage I epithelial ovarian cancer: an Australian multicenter study. J Clin Oncol
. 2005; 23: 5938–5942.
103. Coticchia CM, Curatolo AS, Zurakowski D, et al. Urinary MMP-2 and MMP-9 predict the presence of ovarian cancer in women with normal CA125 levels. Gynecol Oncol
. 2011; 123: 295–300.