WO2005036180A1

WO2005036180A1 - Analysis methods using biomarkers concentrated with biomarkers attractant molecules

Info

Publication number: WO2005036180A1
Application number: PCT/US2004/033305
Authority: WO
Inventors: Arpita I. Mehta; Lance A. Liotta; Emmanuel F. Petricoin, Iii
Original assignee: The Government Of The United States Of America As Represented By The Secretary Of Department Of Health And Human Services
Priority date: 2003-10-08
Filing date: 2004-10-08
Publication date: 2005-04-21

Abstract

A method for discovering biomarkers by examining the molecules that adhere to and are concentrated by larger biomarker attractant molecules is disclosed. Also disclosed is a method for diagnosing biological states based on biomarkers bound to one or more biomarker attractant molecules obtained from one or more biological fluids. In a particular embodiment, molecules bound to high abundance proteins with a molecular weight above the kidney filtration cutoff (about 45 kDa) are collected from serum and compared between groups of subjects to detect biomarkers. In another embodiment, artificial materials such as nanoparticles are introduced into the bloodstream of different groups of subjects and allowed to harvest smaller molecules. The harvested molecules may then be compared between the groups to detect new biomarkers. Once discovered, biomarkers that bind to one or more particular biomarker attractant molecules may be used to provide a diagnosis.

Description

ANALYSIS METHODS USING BIOMARKERS CONCENTRATED WITH BIOMARKERS ATTRACTANT MOLECULES

Cross-reference to Related Applications This application claims the benefit of U.S. Provisional Patent Application No. 60/509,782, filed October 8, 2003, which application is incorporated by reference herein.

Field The invention relates to analysis of molecules present in biological fluids. More specifically, the invention relates to the use of biomarker attractant molecules to concentrate low abundance biomarkers such as peptides, proteins and protein fragments for analysis.

Background Biological fluids such as blood and lymph are the repositories of a vast number of molecules that are excreted or otherwise shed by cells. The molecules that are present in biological fluids reflect the physiological and pathological states of the cells that are contacted by the fluids or which were contacted by the fluids from which they are derived (for example, lymphatic fluid enters the blood through the thoracic duct). Therefore, a major goal of clinical diagnostics is to correlate the particular molecules (biomarkers) present in biological fluids with particular disease states. Although the biomarker approach to diagnostics holds great promise, the number of biomarkers that are reaching routine clinical use remains low. The low molecular mass range (for example, between about 0.3 kDa and 12 kDa) of the plasma proteome remains largely uncharacterized, principally because no adequate method for discovering and utilizing biomarkers in this mass range currently exists. For example, the lower limit of effective resolution achieved by conventional 2-D gel electrophoresis (2D-GE) is about 10 kDa. While mass spectrometry (MS) exhibits its optimal performance in the low molecular mass range, it is nonetheless inadequate for comprehensive analysis of plasma proteins because the dynamic range of concentrations of proteins in plasma is much greater than the dynamic range of the method. Traditionally, analysis of complex protein mixtures by MS has involved a combination of chromatographic separation and enzymatic fragmentation prior to analysis by MS-MS. Enzymatic treatment may, however, destroy or mask the information content of the sample by cleaving disease biomarkers and by creating large quantities of interfering enzymatic fragments from high abundance proteins. To avoid this problem, recent applications of mass spectrometry for biomarker analysis have turned to the native undigested serum proteome as a launch point for biomarker discovery (see, for example, Petricoin et al., "Use of Proteomic Patterns in Serum and Identification of Ovarian Cancer," Lancet, 359: 572-577, 2002). In this approach, chemical fractionation based on protein affinity for derivatized MS targets (surface-enhanced laser desorption/ionization or SELDI targets) is used to reduce sample complexity (Anderson and Anderson, "The Human Plasma Proteome," Molecular and Cellular Proteomics, 1: 845-867, 2002). However, because many proteins of similar mass are detected, the SELDI approach to analysis of whole samples of proteins does not directly provide a sequence-based identification of particular protein peaks. Furthermore, peaks due to high abundance proteins tend to mask underlying peaks due to low abundance molecules. Therefore, under the assumptions that low molecular mass biomarkers contain important diagnostic information and that protein biomarkers useful in disease detection are of very low abundance, the search for low molecular mass biomarkers typically begins with a separation step to remove the abundant high molecular mass proteins such as albumin. For example, Kantor ( antor, "Comprehensive Phenotyping and Biological Marker Discovery, Disease Markers, 18: 91-97, 2002) states that "[b]iological samples, like unfractionated serum are very complex and dominated by a few components present at very high concentration that can interfere with the detection of all other molecules by MS." Some biomarkers that were first identified as diagnostic for particular diseases have later been found to exist in chemically-stable complexes with other proteins. For example, prostate- specific antigen (PSA), first identified in prostatic tissue, was later found to be present in significant part as a complex with α antichymotrypsin (See, for example, Wang et al., Invest Urol., 17: 159-63, 1979 and Lilja et al, Clin. Chem., 37: 1618-1625, 1991). The existence of these two forms of PSA has both complicated diagnostic assays for prostate cancer, and been exploited in such assays to provide better differentiation between benign prostatic hypertrophy and prostate cancer (see, for example, Peter et al., Clinical Chemistry, 46: 474-482, 2000). Similarly, PSP 94 (prostate secretory protein of 94 amino acids) was isolated and later found to form chemically stable complexes with serum proteins (See, for example, Dube et al, J. Androl., 8: 182-189, 1987; Wu et al., J. Cell Biochem., 76: 71-83, 1999; and Guo et al., "Serum Bound Forms of PSP94 in Prostate Cancer Patients," J. Cell Biochem., 76: 71-83, 1999). Again, the existence of multiple forms of PSP94 has complicated the development of assays for this biomarker (See, for example, U.S. Patent No. 6,107,103). In contrast to such later discoveries that known biomarkers exist as complexes in serum, it has been not reported that complexes of molecules adhered to each other in biological fluids can be separated from biological fluids and probed for the presence of potential biomarkers.

Summary The assumption that high molecular weight, high abundance proteins interfere with detection of biomarkers has now been found to diminish the chances of finding important low abundance, low molecular mass disease biomarkers. Surprisingly, it has been discovered by the inventors that much of the important diagnostic information present in serum is in the form of low molecular weight biomarkers (such as less than about 12 kDa) bound to high abundance, high molecular weight proteins such as albumin. For example, low molecular weight biomarkers that are introduced into the blood stream at surprisingly low rates, and that would otherwise be rapidly cleared from serum by filtration through the kidneys, are concentrated by binding to high abundance, high molecular weight proteins, making their detection possible. Molecules and substances which bind and concentrate low abundance, low molecular weight biomarkers are referred to herein as "biomarker attractant molecules." A method is disclosed for discovering or identifying potential biomarkers by separating biomarker attractant molecules from biological fluids and detecting the molecules that are adhered to the biomarker attractant molecules, for example, by separating or isolating the biomarker from the biomarker attractant molecules and analyzing those molecules. This method is contrary to known methods of biomarker discovery in that high abundance molecules present in the biological fluid are not separated out of the biological fluid to improve analysis. Rather, the adherence of biomarkers to such high abundance molecules is exploited to increase detection sensitivity. In one embodiment, a naturally occurring biomarker attractant molecule is isolated from a biological fluid obtained from different populations of subjects exhibiting different biological states (such as the presence and absence of a type of cancer). Molecules differentially bound to the biomarker attractant molecule between the populations are then detected and identified as biomarkers that discriminate between the biological states. In another embodiment, a biomarker attractant molecule is added to the biological fluid of the subjects and allowed to concentrate low molecular weight molecules for some period of time (such as more than a day). After the period of time, the biomarker attractant molecule is isolated from the biological fluid of the subjects and analyzed for bound biomarkers that discriminate between the biological states. Identification of low molecular mass peptides bound to naturally occurring high abundance biomarker attractant molecules, for example, by sequencing the peptides, has also unexpectedly revealed that many such peptides are fragments of known cellular proteins. In many instances, the cellular proteins that are the source of the fragments are themselves too large to pass through cell membranes (walls) and into biological fluids. Thus, the fragments provide a window into specific disease-related enzymatic cleavage patterns and/or post-translational modifications of the larger proteins that are occurring within cells. In addition, the cellular proteins that are the source of the fragments have, in some instances, already been correlated with a particular biological state (such as a disease state or pathological state) in a particular location (such as a particular organ, sub-anatomic structure, cell or organelle). Furthermore, a diagnostic method is disclosed in which biomarker attractant molecules are separated from a biological fluid and biomarkers adhered to the biomarker attractant molecule are detected to determine the biological state of a subject. In some embodiments, proteins, fragments thereof, and combinations of such proteins and protein fragment that have been identified as biomarkers are used to provide a diagnosis for a subject. In other instances, molecules adhered to biomarker attractant molecules that are separated from a biological fluid are analyzed for the presence of fragments of particular cellular proteins that are known to correlate with particular biological states, such as biological states in particular anatomic or physiologic locations. Such fragments that might otherwise be cleared from the biological fluid and that would otherwise never reach detectable levels are surprisingly concentrated by adherence to biomarker attractant molecules and therefore can be detected. For example, in one particular embodiment, biomarker attractant molecule bound fragments of proteins associated with hypoxia and/or apoptosis in the heart muscle are detected and used to diagnose myocardial ischemia or a myocardial infarction (heart attack). Concentration of such fragments by biomarker attractant molecules facilitates their detection. Once identified, particular diagnostic fragment/biomarker attractant molecule complexes or biomarker/biomarker attractant molecule complexes may then be more sensitively detected through further amplification using, for example, enzyme-linked antibodies that recognize the fragment/biomarker attractant complex. In one embodiment, a capture surface that specifically binds one or more biomarker attractant molecules is used as a bait surface to separate the biomarker attractant molecule(s) from a biological fluid. Antibodies directed to one or more of the diagnostic fragments, biomarkers, or complexes thereof with the biomarker attractant molecule, are then applied to the captured complexes. Binding of the antibodies to the adhered complexes can be detected using any known antibody detection method including colorimetric methods, chemiluminescent methods, fluorescence methods, evanescent wave detection methods, electrochemical methods, magnetic methods or electrical methods. In another embodiment, the capture surface is used as a target for laser desoφtion and molecules bound to the separated biomarker attractant molecules are analyzed by mass spectrometry.

Brief Description of the Drawings FIG. 1 is a schematic diagram of a process by which proteins shed by cells are clipped into fragments small enough to pass into the blood stream. FIG. 2 is a schematic diagram of a model of biomarker production and distribution between cellular and vascular compartments. FIG. 3 is a series of graphs showing the results of a theoretical model of biomarker amplification/concentration according to the disclosed method. FIG. 4 is a diagram illustrating a separation protocol used to generate the fractions (A-F) that were analyzed to generate the corresponding mass spectra of FIG. 5. FIG. 5 is a diagram illustrating the mass spectra obtained for each of the fractions illustrated in FIG. 4. The letters A-F denote the spectra obtained for fractions A-F as illustrated in FIG. 4. FIG. 6 is a series of SELDI-TOF mass spectra obtained for serum and fractions thereof that were derived from an ovarian cancer sera study set using the protocol shown in FIG. 4. FIG. 7 is a series of mass spectra in which the spectral region between 5000-7500 m/z has been magnified, and comparing albumin associated ions normalized for sample concentration and amplitude. FIG. 8 is series of SELDI-TOF mass spectra showing the dependence of ion amplitude on quantity of albumin captured (0.025-0.175 mg) from the serum. FIG. 9 is a schematic diagram of an exemplary process by which biomarker attractant molecules (denoted carrier with diagnostic cargo in the figure) may be isolated from biological fluids by binding the biomarker attractant molecule to capture agents inside of a flow cell, and the biomarkers bound to the biomarker attractant molecule are separated therefrom. FIG. 10 is a schematic diagram of a process by which the biomarkers bound to the biomarker attractant molecule immobilized by the process of FIG. 8 are eluted directly into a mass spectrometer for analysis, and the mass spectral data is used to detect previously sequenced protein fragments that are indicative of a subject's physiological and/or pathological profile. FIG. 11 is a schematic diagram showing a process whereby a fluorescent labeled capture agent (in this case an antibody) immobilized to a silica substrate (A) is used to capture a biomarker attractant molecule (in this case a carrier protein) with biomarkers attached (B). Light (in this case UV light) is used to covalently attach the biomarker attractant molecule to the substrate through the immobilized antibody (C) and a laser pulse is used to desorb biomarkers directly from the biomarker attractant molecule, for example, after an energy absorbing molecule such as cinnapinic acid is used to coat the substrate with its bound molecules. FIG. 12 is a schematic diagram showing a method of diagnosis utilizing the information obtained from analysis of biomarkers that bind to a biomarker attractant molecule.

Detailed Description of Several Disclosed Embodiments In order to facilitate an understanding of the embodiments presented, the following explanations are provided. I. Abbreviations SELDI - surface-enhanced laser desoφtion/ionization MALDI - matric -enhanced laser desoφtion/ionization MS - mass spectrometry TOF - time-of-flight LMW- low molecular weight HMW-high molecular weight LMM- low molecular mass HDVIM - high molecular mass BAM- biomarker attractant molecule ESI-MS - electrospray ionization mass spectrometry m/z -mass-to-charge ratio

//. Terms In order to facilitate an understanding of the embodiments presented, the following explanations are provided. The singular terms "a," "an," and "the" include plural referents unless context clearly indicates otherwise. Similarly, the word "or" is intended to include "and" unless the context clearly indicates otherwise. The term "comprises" means "includes." Hence, "comprising A or B" means including A or B, or A and B, unless the context clearly indicates otherwise. It is further to be understood that all molecular weight or molecular mass values given for compounds are approximate, and are provided for description. Although methods and materials similar or equivalent to those described herein can be used in the practice or testing of this disclosure, suitable methods and materials are described below. In addition, the materials, methods, and examples are illustrative only and not intended to be limiting. The term "adhered to a biomarker attractant molecule" refers to physical or chemical association of another molecule (such as a known or potential biomarker) with the biomarker attractant molecule. The association can be either direct or indirect, in that molecules can be adhered in one or more layers to a surface (external or internal) of the biomarker attractant molecule. For example, molecules exhibiting hydrophilic and hydrophobic domains (an amphipathic molecule) on opposite sides of the molecule can adhere to a hydrophobic domain on a biomarker attractant molecule, thereby presenting the hydrophilic domain of the adhered molecule to the solution. A hydrophilic molecule can then adhere to the hydrophilic side of the amphipathic molecule, and thus become indirectly adhered to the biomarker attractant molecule. In some embodiments, adhered molecules are molecules that do not form covalent bonds with the biomarker attractant molecule. In other embodiments, the adhered molecule is bound non-specifically (that is, not to the substantial exclusion of other molecules) to one or more domains on the biomarker attractant molecule. In further embodiments, the molecule is adhered to the biomarker attractant molecule with an association constant (K_a) of less than about 10⁸, for example, a K_a of less than about IO⁷, such as less than about IO⁶ or about 10^s. Adhered molecules also can be distinguished according to the strength of their association based upon the identity, strength (concentration), and or type of denaturant required to dissociate them from the biomarker attractant molecule. The term "biological fluid" refers to a liquid into which biomarkers are released, or a liquid derived from the liquid into which biomarkers are initially released. Such derivation may occur either in vivo or in vitro. In some instances, the biological fluid is a circulating fluid such as blood or lymph, or a fraction thereof, such as serum or plasma. In other cases, the biological fluid remains substantially in a particular locus, for example, synovial fluid, cerebrospinal fluid or interstitial fluid. In still further cases, the biological fluid is an excreted fluid, for example, urine, breast milk, saliva, sweat, tears, mucous, nipple aspirants, semen, vaginal fluid, pre-ejaculate and the like. A biological fluid also refers to a liquid in which cells are cultured in vitro such as a growth medium, or a liquid in which a cell sample is homogenized, such as a buffer. The term "biomarker" refers to one or more molecules (or signals due to such molecules in an analytical method such as mass spectrometry) that are differentially released into a biological fluid by any means (including secretion or by leakage through the cell membrane) in one or more subjects in each of two or more populations that exhibit different biological states. Examples of biomarkers include proteins, lipids, lipoproteins, glycoproteins, nucleic acids (such as circulating nucleic acids, see, for example, Krishna et al., Jpn. J. Clin. Oncol., 34: 307-311, 2004; Ngan et al., Ann. NY Acad. Sci.,, 1022: 263-270, 2004; Carstensen et si., Ann. NY Acad. Sci., 1022: 210-210, 2004; Watanagara et al., Ann. NY Acad. Sci., LU22: 90-99, 2004; Herman, "Circulating Methylated DNA," Ann. NY Acad. Sci., 1022: 33-39, 2004; and Schwarzenbach, Ann. NY Acad. Sci, 1022: 25-32, 2004), carbohydrates, lipopolysaccharides, small molecule metabolites, and fragments thereof. Typically, the presence and/or the concentration of a biomarker (or biomarkers, or pattern or patterns of biomarkers) in a biological fluid is discriminatory between physiological and pathological states of the cells from which they are released. The term "biomarker atfractant molecule," or "BAM," refers to a molecule, or other substance to which biomarkers in a biological fluid adhere. In particular examples, biomarkers adhere to a BAM with a low binding affinity (for example, a binding affinity of less than 10^"3, 10^"4, 10°, IO^"6, IO^"7 or IO^"8 L/mol-min). An antibody may be a BAM to the extent that it binds biomarkers, other than through the specific antigen antibody interaction that results from the immune response that stimulated its production. For example, biomarker binding to an antibody BAM may occur outside of the complementarity defining region (CDR), or outside of the variable region altogether, for example by binding to the Fc portion of the antibody. However, in certain embodiments of the disclosed methods, the BAM is not an antibody. Although a particular BAM may selectively bind a class of biomarkers, the binding affinities of the biomarkers in a particular class do not differ as significantly as the binding affinities of an antigen to a particular antibody compared to other non- recognized molecules. The less specific nature of biomarker binding may be illustrated in certain examples of the BAM in which more than one biomarker binds to the BAM, for example, at least 2, at least 5, at least 10, at least 20, or even 50 or more biomarkers bind to the BAM. Typically, BAMs have a half-life of existence in a particular biological fluid (for example in the body) that is longer than the half-life of biomarkers that become adhered to the BAMs and thereby concentrate the biomarker in the biological fluid. For example, BAMs can have a half-life of greater than about 1 day, such as greater than 2, 5, 10, 20 or 50 days. In particular examples, the BAM has as size and/or shape such that it is not substantially filtered from the blood stream by the kidneys. In other particular examples, the BAM has a molecular weight of greater than 25 kDa, for example, greater than 30, 50, 75, 100, 150, 200 or 300 kDa. In yet other particular examples, the BAM molecule has a molecular weight falling within a particular range, for example between 30 and 50 kDa, between 50 and 75 kDa, between 75 and 100 kDa, between 100 and 150 kDa, between 150 and 200 kDa, between 200 and 300 kDa, or any other range between 30 kDa and 300 kDa. Biomarkers may adsorb to the surface or be absorbed into the interior of the BAM, or both.

Examples of BAMs include proteins (including natural and engineered proteins such as chimeric proteins proteins with modified amino acid composition, proteins modified posfranslationally, nucleic acids, carbohydrate decorated molecules, and organic polymers), dendrimers and particles (such as microparticles and nanoparticles, including silica, metal, ceramic and carbohydrate microparticles and nanoparticles ), and cellular microparticles (see, for example, Diamant et al, Cellular microparticles: new players in the field of vascular disease?, EurJ Clin Invest. 34: 392-401, 2004). BAMs may be produced or derivatized to provide ionic groups (such as carboxylate, protonated arnine, quaternary ammonium, and sulfate groups), hydrogen-bond acceptors or hydrogen-bond donors, electron donors or electron acceptors, polar groups (such as amino, hydroxyl, ester, sulfhydryl and nitrile groups), hydrophobic groups (such as alkyl, alkenyl and alkynyl groups or groups with specific partition coefficients), peptides, proteins, nucleic acids, carbohydrates, lipids or any combination thereof, on their surfaces or in their interiors. Where the BAM is a protein, such as a naturally occurring protein, it may also be referred to as a "carrier protein" to reflect its role in collecting and concentrating LMM biomarkers from biological fluids. Examples of carrier proteins include albumin, iron binding proteins (such as transferrin), fibrinogen, alpha-2-macroglobulin, immunoglobulins (such as IgA, IgE and IgG), complement, haptoglobulin, lipoproteins, prealbumin, alpha- 1 -acid glycoprotein, fibronectin, and ceruloplasmin, and fragments, combinations and chemical derivatives thereof. The term "biological state" refers to any characterizing feature of an organism or a cell. The feature can be a biological trait such as a genotypic trait or a phenotypic trait. The feature can be a physiological or disease trait, such as the presence, stage or absence of a particular disease, including infectious disease. The feature also can be a condition (environmental, social, psychological, time- dependent, etc.) to which the organism or cell has been exposed or not exposed, including exposure/non-exposure to a drug or exposure/non-exposure to a toxin. Examples of biological states include a pathologic diagnosis, a toxicity state, efficacy of a drug, prognosis of a disease, stage of a disease, condition of an organ, presence of a pathogen (such as a virus), or toxicity of one or more drugs. Particular examples of disease states include presence, absence or stage of cancers, such as carcinomas, melanomas, lymphomas, sarcomas, blastomas, leukemias, myelomas, and neural tumors. Particular examples of cancers further include prostate cancer, breast cancer, pancreatic cancer, ovarian cancer, testicular cancer and lung cancer. The term "capture agent" refers to an agent that preferentially binds or selects a particular molecule, substance or material from a sample to the substantial exclusion of others present in the sample. Examples of capture agents include antibodies, apatamers, and reactive groups that selectively react with a particular functional group present on the molecule, substance or material to be captured and not on others. Where the molecule, substance or material to be captured is magnetic, the capture agent may be a magnet. Where the molecule, substance or material to be captured has a molecular weight that is substantially different from others in a sample, the capture agent may be a molecular sieve or size exclusion material that either retains or passes the molecule, substance or material to be captured while passing or retaining others. Where the molecule, substance or material to be captured comprises a sugar, the capture agent may be lectin that binds the sugar. The term "denaturant" refers to an agent that disrupts the adherence of at least a portion of biomarkers adhered to a biomarker attractant molecule. Examples of denaturants include high (>1M) concentration salt solutions, solutions comprising a chaotropic agent such as urea, guanidinium chloride, and acetonitrile, solutions having low pH (for example, pH<5) or high pH (for example, pH>9), a high temperature solution (for example a solution having a temperature greater than 30 degrees celcius), an electrochemical perturbation, or application of electromagnetic radiation (for example, a laser pulse). A denaturant, may, in some embodiments, be a series of solutions with a continuous or discontinuous gradient of increasing or decreasing salt concentration, increasing or decreasing concentration of a chaotropic agent, increasing or decreasing pH, increasing or decreasing temperature, or any combination thereof. In this instance, binding of different biomarkers may be sequentially disrupted as a function of the gradient. In some particular embodiments, a denaturant is other than a reducing agent such as β-mercaptoethanol or an amino alcohol or diamine compound (such as ethanolamine) that has at least two nucleophilic groups, at least one of which is an amine group. Conversely, the term "non-denaturant" refers to an agent that does not disrupt binding of a biomarker to a biomarker attractant molecule. Examples of non-denaturants include low concentration solutions of salts such as buffers and substantially pure water, such as distilled or reverse-osmosis purified water. The term "non-denaturing conditions" refers to conditions do not significantly disrupt adherence of a biomarker to a biomarker attractant molecule. For example, such conditions include those under which biomarkers adhered to a biomarker attractant molecule with an association constant (K_a) of greater than IO⁴, such as greater than IO⁵, 10⁶, IO⁷ or 10^s are remain substantially adhered to the biomarker attractant molecule. In other embodiments, biomarkers having a rate of association with the biomarker attractant molecule of greater than 10^-6 L/mol min, such as greater than 10^"5, IO^"4 or 10^"3 L/mol min remain adhered to the biomarker atfractant molecule. Biomarker attractant molecules can be separated from a biological fluid under non-denaturing conditions to help reduce loss of biomarkers adhered to the biomarker attractant molecule during the separation. The term "domain" refers to a portion of a molecule that exhibits a particular chemical or physical property. A domain can be as small as a particular amino acid residue on a peptide or as large as the entire surface of a biomarker attractant molecule, although more typically a domain will represent some portion of the surface of biomarker attractant molecule. Within a domain, a chemical or physical property can be substantially homogeneous, for example, substantially hydrophobic or substantially hydrophilic. Domains also can exhibit a set of chemical or physical properties that lead to adherence of particular classes of molecules, for example, phosphorylated or ubiquitinated protein fragments. The term "eluting" refers to the act of disrupting intermolecular bonds (such as hydrogen bonds, dipole-dipole bonds, hydrophobic interactions and the like) between two molecules, substance, materials and the like, for example, a biomarker and a biomarker attractant molecule, and separating the two, for example, into separate containers or physical locations such as wells on a microtiter plate. The term "identifiable molecule" refers to a molecule for which an analytical method is available for identifying its presence in a sample, such as substantially (for example, to a greater than 98% or 99% certainty) or unambiguously identifying its presence in a sample. Identifiable molecules can, for example, be identified according to their mass (such by a single or multiple peaks in a mass spectrum or an electrophoretogram), according to their immunochemical properties (such as by direct or sandwich antibody detection), or according to their sequence (such as using a cDNA array or RT- PCR). The term "at least an identifiable fragment of a protein" refers to fragment of the protein that, if detected, identifies (such as substantially or unambiguously) the protein from which the fragment was derived. An identifiable fragment of a protein can include, for example, one or more proteolytic peptide sequences (such as tryptic peptide sequences), the masses or sequences of which can be used to identify the protein with substantial certainty (for example, with >98% or >99% probability), for example unambiguously (such as accurate mass tag). An identifiable fragment of a protein can include an epitopic sequence of amino acids that is recognized by one or more antibodies that also recognize the whole protein from which the fragment is derived. Methods for using antibodies to immunochemically detect molecules are thoroughly discussed in Harlow and Lane, Using Antibodies: A Laboratory Manual, CSHL, New York, 1999 (incoφorated by reference herein), and in Paul, Fundamental Immunology, 3rd ed., 243-247 (Raven Press, 1993) and references cited therein. The term "protein array analysis" refers to any technique in which samples comprising proteins or proteins fragments are placed on a substrate in particular locations and probed in a location specific manner, for example, with fluorescent antibodies to the proteins. Protein array analysis is discussed in detail in Liotta et al, Method and Devices for the isolation and analysis of cellular protein content," U.S. Patent Application No. 09/913,667, which is incoφorated by reference herein.

Additional examples of protein arrays are described in U.S. Patent No. 6,630,358. Use of antibodies in such protein arrays as is discussed in these references is also generally applicable to detection/identification of molecules. "Mass spectrometry" refers to any technique in which molecules are analyzed by converting them to gas phase ions and measuring the mass-to-charge ratio of the ions. Examples of mass spectrometric techniques include matrix-assisted laser desoφtion ionization time-of-flight (MALDI- TOF), surface-enhanced laser desoφtion/ionization time-of-flight (SELDI-TOF), electrospray ionization mass spectrometry (ESI-MS), and Fourier-transform ion-cyclotron resonance (FT-ICR), and related techniques. Such techniques can further be coupled to a method of separation such as HPLC or capillary electrophoresis. The term "population" refers to one or more subjects, but is more typically two or more subjects, for example, many more, such as more than 10, 100, 500 or 1000. The term "separating" refers to dividing a substance (such as a biomarker attractant molecule) from other substances (such as other molecules in a biological fluid) in order to obtain a purified form of the substance, such as an isolated form of the substance. In general, "purifying" is a relative term that does not require absolute purity and, for example, refers to separating a substance from other substances to provide a preparation wherein the separated substance represents at least 50% of the preparation. The term "isolating" refers to substantially separating a substance from other substances to provide an even more purified preparation, for example, a preparation in which the substance represents at least 90%, 95% or 98% or more of the preparation. The term "size exclusion" refers to any technique that separates molecules based upon their size, including liquid chromatographic techniques such as column and high-pressure liquid chromatography (HPLC) and electrophoretic techniques such as denaturing polyacrylamide electrophoresis. The term "subject" refers to an animal, and more particularly to a mammal (for example, a human or a veterinary animal such as a dog, a cat, a pig, a horse, a sheep, or a cow). ///. Overview Blood contains thousands of molecules shed from tissues into the blood stream. Similarly, molecules are shed into other biological fluids such as lymph fluid, interstitial fluid, synovial fluid, cerebrospinal fluid, saliva, seminal fluid and the like. In addition, molecules initially shed into one fluid may become part of another biological fluid. For example, lymph fluid is ultimately introduced into blood and blood is filtered by the kidneys to provide urine. Many of the molecules carried by these fluids are fragments or enzymatically modified forms of normal functioning proteins emanating from all levels of cellular and extra-cellular compartments. Every cell in the body can be viewed as leaving a record of its physiologic state in its waste products, or the products of its interactions with neighboring cells (FIG. 1). As shown in FIG. 1, such waste products may, for example, enter the bloodstream by passing through the vascular wall. To enter the bloodstream, however, large proteins are often be degraded into fragments small enough to go through or between the endothelial cells lining the blood vessels. The lower molecular weight (LMW, such as less than about 12 kDa) proteome has not previously been systematically studied because this region of the proteome has been viewed as a non- interesting dumping ground of metabolic and proteolytic byproducts. Moreover, traditional protein- based discovery tools such as 2-D gel electrophoresis cannot discriminate in the LMW range. Finally, and very importantly, previous methods for preparing serum for biomarker discovery efforts have destroyed, contaminated, or discarded, the LMW biomarker information archive by enzymatic treatment (e.g. trypsin digestion) or by removing the high abundance serum proteins, such as albumin. The inventors have found that virtually all of the LMW species are bound or complexed with the high molecular weight proteins, which act to accumulate and thereby amplify low abundance biomarkers. Biomarkers enter the circulation and adhere to long-half-life proteins such as albumin to provide an amplified and integrated history of biomarker production. Thus, the carrier proteins are a previously unappreciated storehouse of diagnostic information, which is exploited in certain of the disclosed methods. Accordingly, a method is disclosed for identifying a biomarker in a biological fluid that includes separating a biomarker attractant molecule from samples of the biological fluid and detecting a biomarker that is differentially adhered to the biomarker atfractant molecules between different populations of subjects that exhibit two or more different biological states. The detected biomarker that is adhered to the biomarker attractant molecule in the biological fluid discriminates between the different populations of subjects, and therefore can be used to discriminate between the biological states of subjects. For example, detection of the biomarker adhered to the biomarker attractant molecule in another subject can indicate that the subject exhibits a particular biological state such as a disease state. In some embodiments, the biomarker attractant molecule is separated from the biological fluid using a capture agent that specifically or αon-specifically binds the biomarker attractant molecule (such as an antibody anchored to a substrate or a substrate that preferentially attracts the biomarker attractant molecule to its surface), and thereby removes at least a portion of the biomarker atfractant molecule from the fluid. It is also possible to form a covalent bond between the biomarker atfractant molecule and the capture agent to more firmly anchor the biomarker atfractant molecule to the capture agent. In other embodiments, the biomarker atfractant molecule can be separated from the biological fluid by size exclusion chromatography, for example, by cutoff at a particular molecular weight or by fractionation in to ranges of molecular weight. However, in general, any method that can concentrate the biomarker attractant molecule away from the biological fluid can be used to provide isolated biomarker atfractant molecules for analysis of their adhered diagnostic cargo. Typically, the biomarker attractant molecule that is separated from the biological fluid will have a half-life in the biological fluid that is longer than the half-life of biomarker that it amplifies and concentrates. For example, the biomarker attractant molecule can have a half-life in the biological fluid of greater than 1 day, for example, 3 days or more, such as 5 days, 10 days, 20 days, 30 days, 50 days, 100 days or more. In some instance, biomarker atfractant molecules will have a molecular weight greater than about 25 kDa, for example, greater than about 40 kDa, such as greater than about 60 kDa or greater than about 80 kDa. In particular embodiments, the biomarker attractant molecule is selected from the group consisting of albumin, transferrin, fϊbrinogen, alpha-2- macroglobulin, an immunoglobulin, complement, haptoglobulin, a lipoprotein, prealbumin, alpha 1 acid glycoprotein, fibronectin, and ceruloplasmin, and fragments, chemical modifications and combinations thereof. Such biomarker attractant molecules can be obtained directly from a biological fluid, or can be Biomarker attractant molecules can be native to a subject's body or not native to a mammalian body (such as a human body) and introduced into the biological fluid of the subjects in the different populations to harvest diagnostic information from the biological fluid. Alternatively, native biomarker attractant molecules (such as blood proteins) can be extracted from a subject, stripped of bound molecules, and reintroduced to harvest a new constellation of bound molecules. Examples of non-native biomarker attractant molecules include nanoparticles formed, for example, from silica, a ceramic, a magnetic metal oxide, or a biodegradable material. The biomarker atfractant molecule also can include a dendrimer. Like other examples of biomarker attractant molecules, nanoparticles and dendrimers can be chemically modified to provide surfaces having particular chemical/physical properties. For example, nanoparticles can be modified to include at least one region (or domain) on their surface that is substantially hydrophobic and at least one region (or domain) on their surface that is substantially hydrophilic such that hydrophobic and hydrophilic biomarkers become adhered to the regions of the surface, respectively. Another way to modify the surface properties of a non-native biomarker attractant molecule is to conjugate it with a biological molecule, such as a protein or a protein fragment. Conjugated biomolecules can be chosen to specifically or non-specifically adhere particular classes of biomarkers. Since serum proteins (such as albumin, transferrin, fibrinogen, alpha 2 macroglobulin, immunoglobulins, complement, haptoglobulin, a lipoprotein, prealbumin, εlpha 1 acid glycoprotein, fibronectin, and ceruloplasmin, or fragments, chemical modifications and combinations thereof) often have domains on their surface exhibiting diverse chemical and physical properties that are capable of binding many different types of biomarker molecules, conjugation of such proteins to, for example, a nanoparticle or a dendrimer can provide a molecular "mop" that can be especially useful when the goal is to identify biomarkers in a biological fluid. Biomarker attractant molecules also can be produced to include properties or groups of atoms that facilitate their separation from a biological fluid, such as magnetism or an affinity tag such as a poly-histidine affinity tag or a sugar recognized by a lectin. Biomarkers can be harvested from any biological fluid through adherence to biomarker attractant molecules. Non-limiting examples of biological fluids include blood, plasma, serum, lymph fluid, synovial fluid, cerebrospinal fluid, breast milk, nipple aspirants, sweat, tears, saliva, mucous, pre-ejaculate, semen, vaginal fluid, a cell culture medium, or interstitial fluid, or a fraction thereof, a combination thereof, or a fluid derived therefrom. While it is possible to detect biomarkers by directly probing a separated biomarker attractant molecule for the presence of adhered molecules (for example, with panels of antibodies to known proteins), in some embodiments, it is advantageous to separate at least a portion of the molecules adhered to the biomarker attractant molecule from the biomarker attractant molecule prior to facilitate their detection. Molecules can be separated from a biomarker attractant molecule by dissociating them from the biomarker attractant molecule. This can be accomplished, for example, by contacting the biomarker attractant molecule with a denaturant to elute bound molecules from the biomarker attractant molecule. In this embodiment, it may be advantageous to immobilize the biomarker attractant molecule with a capture agent. Elution of biomarkers can be partial or complete, and can include a gradient elution with increasingly stronger denaturants and fractionation of the molecules eluted from the biomarker atfractant molecule during the gradient elution. In other embodiments, molecules that are eluted from a biomarker atfractant molecule are further separated from the biomarker atfractant molecule by size exclusion chromatography. An alternative method of dissociating molecules from biomarker attractant molecules is to directly dissociate them using using a laser. Molecules adhered to biomarker attractant molecules that are discriminatory between biological states and hence diagnostic for particular biological states can be identified according to the disclosed methods. There is virtually no limit to the biological states that can be discriminated using the methods. Examples of biological states include absence, presence and/or stage of a disease such as a cancer, exposure to a drug and non-exposure to the drug, exposure to a toxin and non- exposure to the toxin, and before, during and/or after a change in physiological, hormonal or developmental state (such as menopause) Molecules adhered to biomarker attractant molecules can be detected by any method, but some methods include mass specfrometric detection (which can be coupled to pattern recognition, especially for discovery of complex biomarkers comprising multiple molecule or identification of such complex biomarkers). Particular methods of mass specfrometric detection include MALDI- TOF, SELDI-TOF, FT-ICR, and ESI-MS. Other methods for detecting molecules adhered to biomarker atfractant molecules include protein array analysis, which also can be coupled to pattern recognition. Other methods include immunochemical analysis of molecules adhered to the biomarker attractant molecule, either with or without dissociation of the molecules from the biomarker attractant molecule. The diverse methods of detection that can be used in the disclosed methods make it possible to identify and utilize biomarkers of diverse types. For example, in some embodiments, the biomarker comprises a protein, a lipid, a carbohydrate or a nucleic acid, or a fragment or combination thereof. In some embodiments, the biomarker includes two or more different molecules adhered to the biomarker attractant molecule, for example 5 or more, such as more than 10, 20 or 50. In particular embodiments, the biomarker includes two or more different identifiable proteins or protein fragments. In some embodiments the biomarker adheres to the biomarker attractant molecule non- covalently, for example, with a binding constant of less than 10^s M^"1. In other embodiments, the biomarker attractant molecule adheres and thereby concentrates the biomarker because the concentration of the biomarker attractant molecule in the biological fluid is at least 10 times greater (for example, at least 100 times greater, such as at least 10³, IO⁴, IO⁵ or IO⁶ times greater) than the concentration of the biomarker in the biological fluid. In still other embodiments, the biomarker atfractant molecule further concentrates the biomarker because the biomarker attractant molecule has a half-life of greater than 1 day in the biological fluid. In some embodiments, the biomarker attractant molecule circulates in the biological fluid and collects the biomarker as it is released into the biological fluid, and wherein the biomarker adheres specifically or non-specifically to one or more domains on the surface of the biomarker attractant molecule, the one or more domains on the surface of the biomarker further adhering additional molecules found in the biological fluid. In other embodiments, the biological fluid comprises serum or plasma and the biomarker, which would otherwise be filtered out of the serum or plasma at a rate equal to or greater than a rate of its introduction in the serum or plasma, is retained and concentrated over time in the serum or plasma because the biomarker atfractant molecule to which it adheres is large enough and in sufficient excess concentration relative to the biomarker to prevent the biomarker from being filtered from the serum or plasma by normal kidney/glomerular function. In still other embodiments, the biomarker adhered to the biomarker attractant molecule has a molecular weight of less than about 12 kDa (for example, less than about 10 kDa, such as less than about 8 kDa, 6 kDa or 5 kDa.). In other embodiments, the biomarker attractant molecule includes two or more different biomarker attractant molecules. For example, a combination of two or more biomarker atfractant molecules such as native carrier proteins can be used to provide more discrimination between biological states. In particular embodiments, the two or more different biomarker attractant molecules include all molecules naturally present in the biological fluid having a molecular weight of greater than about 25 kDa (for example, greater than about 30 kDa, such as greater than 40, kDa, 50 kDa, 70kDa or 90 kDa). And, the biomarker itself can include two or more different molecules differentially adhered between the two or more different biomarker atfractant molecules. Biomarkers identified (or discovered) by the disclosed method are contemplated. Also disclosed is a method of detecting a biological state of a subject. The method includes separating a biomarker attractant molecule from a biological fluid obtained from the subject, wherein a biomarker of the biological state adheres to the biomarker attractant molecule. Detection of the biomarker adhered to the biomarker attractant molecule indicates that the subject exhibits the biological state. All of the permutations with regard to the separation of the biomarker attractant molecule from the biological fluid, detection of the biomarker, the biomarker itself, the biological fluids and the biomarker attractant molecules that were discussed above in regard to method of identifying a biomarker can be used in the method of detecting a biological state. However, in particular embodiments, a denaturant used to dissociate a biomarker from a biomarker attractant molecule does not include a reducing agent or a nucleophilic compound comprising at least two nucleophilic groups, wherein at least one of the at least two nucleophilic groups in the nucleophilic compound is an amine group. In other embodiments, the biological state detected includes a pathological condition. For example, the pathological condition can include a disease or condition of a particular organ and the biomarker is a protein or a fragment thereof that is known to be produced in the particular organ (such as the heart) as a result of the disease or condition that becomes concentrated by the biomarker attractant molecule. In others, the pathological condition is the presence of a tumor, for example, of a particular type or a toxin exposure. Particular embodiments, wherein a particular pathological condition is detected using a particular biomarker attractant molecule by detecting at least an identifiable fragment of each of two or more particular proteins (previously identified by the disclosed discovery method to be discriminatory) adhered to the particular biomarker attractant molecule are disclosed in Examples 4-7. The following examples are provided to illustrate certain particular features and/or embodiments, but these examples should not be construed to limit the invention to the particular features or embodiments described. Example 1 This example demonstrates the method and provides experimental results that illustrate how low molecular mass (LMM) biomarkers reaching the blood may be significantly concentrated over time by binding (even with very low affinities such IO^"4 L/mol min or lower, for example, IO^"5 L/mol-min, or lower) to larger blood proteins having substantial half-lives (such as greater than 1 day, for example greater than 2, 5, 10, 20 or 50 days). Theoretical analysis of the binding kinetics, and experimental studies of human serum following molecular mass fractionation, demonstrates that the majority of low molecular mass biomarkers are bound to carrier proteins. Moreover, the pattern of LMM biomarkers bound specifically to albumin is distinct from those bound to non-albumin carriers. Several suφrising insights emerge from these results: a) accumulation of LMM biomarkers on circulating carrier proteins greatly amplifies the total serum/plasma concentration of the biomarkers, making them more easily measured, b) the total serum/plasma biomarker concentration is largely determined by the carrier protein clearance rate, not the unbound biomarker clearance rate itself, and c) examination of the LMM species bound to a specific carrier protein contains important diagnostic information, as does examination of LMM species bound to differing carrier proteins. Carrier proteins and their biomarker content provide an unexpectedly superior means to detect biomarkers and sets of biomarkers for diagnostic puφoses. The following analytical and experimental approach was used to demonstrate the concept of biomarker amplification through the use of carrier protein binding. Amplification refers to an increase in the total plasma biomarker concentration due to the accumulation of biomarker molecules over time as they remain bound to carrier proteins. The proportion of low molecular mass species detectable by SELDI (surface enhanced laser desoφtion and ionization) that are associated with the higher molecular mass serum proteome was determined. Human serum was fractionated into high molecular mass and low molecular mass native fractions. Each fraction was assayed by SELDI to assess whether the preponderance of low molecular mass ions is found in the low or the high molecular mass fraction. The subpopulation of molecular species bound to albumin compared to the total carrier protein fraction was also determined. Further, it was determined whether SELDI-TOF identified biomarkers correlating with presence of ovarian cancer were associated with high molecular mass carrier proteins. The results demonstrate the suφrising usefulness of carrier proteins as an affinity capture means for disease relevant biomarkers. A mathematical model of biomarker amplification by carrier proteins was developed in which the kinetics of biomarker production, binding to carrier protein(s), and clearance, was modeled as a deterministic compartmental model with first order kinetics. A schematic diagram of the compartmental model is shown in FIG. 2. Here, the tissue compartment generating the biomarker is composed of target cell (representing, for example, a tumor cell) and host cell (representing, for example, a normal cell) compartments, and the net rate of biomarker shed into the vascular compartment from the tissue compartment is represented by k^. The biomarker in the vascular compartment exists either in the bound or free (unbound) state. The carrier protein and the biomarker leave the vascular compartment at different rates, represented by k^_b and k_oU . The binding state of the biomarker determines its rate of clearance from the vascular compartment. Additional abbreviations used in the equations of the model that follows are shown below in Table 1. Table 1: Theoretical Model Abbreviations

Numerical integration of the equations that follow was performed using Runge-Kutta approximation and simulation by MATLAB software (The MathWorks Inc; Natick, MA). In this model, biomarkers are considered to be molecules shed into the circulatory compartment from perfused tissue (See FIG. 1). The type and relative abundance of single or multiple biomarkers can reflect the diseased or physiologic state of the tissue source. Biomarkers arise either directly from the cellular target itself, or from an interaction between the target cell and its surrounding host cells. Exchange of cytokines, enzymes, ligands, and metabolic products at the interface between the target cell and the host can potentially generate a population of modified molecules. For example, in the case of tumor biomarkers, enzymes elaborated by the tumor cells can modify substrates donated by the host to ultimately generate cleavage products that enter the venous drainage of the tissue. The exchange of materials between compartments exists in equilibrium. In this model, the assumption that the biomarker is introduced into the vascular compartment at a low constant rate kj-_>b is used. The biomarker enters the vascular compartment in a form that is not bound to a serum protein carrier, and becomes evenly distributed into the vascular compartment volume. Circulating biomarkers cycle between the bound and free states, depending upon their affinity to the carrier. This is expressed by the associative and dissociative rate constants, k_o„ and k^ respectively. For a biomarker emanating from the tissue, it is assumed that the biomarker becomes evenly distributed within the circulation compartment. At any point in time the total concentration of the biomarker is dependent upon the biomarker production rate, the biomarker clearance /excretion rate, the binding of the biomarker to a circulating carrier protein, and the clearance / excretion rate of the carrier protein (see FIG. 2). When the system is represented by a single compartment model, V is the volume of distribution or total blood volume, and C_B is the total concentration of the biomarker b consisting of the sum of the bound and free forms of the biomarker. The rate of biomarker production is k_{in b}, and the rate of its elimination, degradation, or excretion is k_oUt. Assuming first order kinetics, the total change in mass of the compartment by the biomarker can be represented through the principles of mass balance as: dM_B(f) = k_{in h} - k_n,„ M_B(t) (1) dt

where M_B is the mass of the total biomarker within the compartment. An individual biomarker molecule can exist in the free form or it can become bound to a circulating carrier protein, "r." As a result, the system may be represented as

koff in which case the total mass of biomarker as a function of time is

M_B(t) = M_br(t) + M_b(t)

where M_B is the mass of the total biomarker, M_b is the mass of the free form biomarker, M_r is the mass of the carrier protein, M _r is the mass of the biomarker bound to the carrier protein, k„_n is the rate of biomarker associating with the carrier protein, and k^ is the rate that the bound biomarker dissociates from the carrier protein. Thus, dM_b(t} = ._b - k_o„,_,b M_b(t) + k_off M_br(t) - k_o„ (M (t) - M .(t)) M_b(t) (2) dt V

where (M_r(t) - M_br(t))/V represents the quantity of carrier proteins available to bind the free form of the biomarker. Biomarker concentrations are represented as follows: Total C_B(t), Free C_b(t),& Bound to Carrier C_br(t): By the definition the concenfration, M_b(t) = C_b(t) V and its change with respect to time is,

and the change in free biomarker concentration can be expressed as the following non-linear equation, C_b(t) = f k^ + k_off C_br(t) 1 - f k_oUt,_b + k_on (C_r(t) - C_br(t)) 1 C_b(t) (3) dt^" l V J I J

where C_b is the free form biomarker concentration, C_r is the concentration of the carrier protein and

C^ is the concentration of the biomarker bound to the carrier protein. The change in mass of the biomarker bound to the carrier can be expressed as dM_bjϋ) = k_on MM - MJM M_b(t) - k_oUt,_br M_br(t) - k_off M_br(t) (4) dt V

while the change in total concentration of the biomarker bound to the carrier protein is represented by the non-linear equation dC_brO) = ko„ C_r(t) C_b(t) - (ko_ff + k_out,_br + k_on C_b(t)) C_br(t) (5) dt

where C_b is the free biomarker concentration, C_r is the concentration of the carrier protein and C_br is the concentration of the biomarker bound to the carrier protein. The total concentration of biomarker can be expressed as the sum of the bound and free forms of the biomarker: thus,

C_B(t) = C_br(t) + C_b(t) (6)

and its change can be expressed as the sum of equations (3) and (5), or

dC»(t) = dCκ,(t) + dCκ(t) dt dt dt dCβ(t) = kn_jj, _ ko^_b Cb(t) _ k_ou_t,_br C_br(t) (7) dt V

Then C_B(t) = k^ t kout br f C_br(τ) dτ koutb f C_b(τ) dτ (8) V J o J o

For the earner protem concentration, C_r(t) within the same single compartment model with volume V, where the concentration of the earner protein r is defined as Cr, and where the rate of earner protein production is k_{ιn r}, and the rate of its elimination or removal is k_{oUt r} Then, the change in the earner concentration can be expressed as the linear equation dC_r(t) = k^ - l C t) dt V

Using the LaPlace Transform, the concenfration of the earner protem, at tune t becomes

C_r(t) — kui_j _ k_{]! E}_ I - kout r I (9) * ^out r ^» ^ oouutt rr I I ^

The following equation represents amplification of biomarker concentration in the presence versus the absence of the earner protein, under the assumption that the biomarker is continuously produced or shed at the tissue source, and leads to the conclusion that biomarker molecules can accumulate over time in a earner-bound form At steady state, the total concentration of a biomarker measured m a blood sample can therefore become elevated due to its association with the earner protem T he level of amplification (A) of the biomarker concentration at steady state, due to the presence of the carrier protem can be defined as the following ratio

A ^— ,IΠ the presence of earner proteiniϋ \ *V) β in the absence of carrier proteinl.*-)

When considermg the special case solution for excess earner protein compared to biomarker, which is reasonable because plasma biomarkers reflecting a physiologic or disease state of perfused tissues are expected to exist at concentrations many orders of magnitude below the concentration of large earner proteins such as albumin and immunoglobulins Smce the bound biomarker concentration is much greater than the free biomarker concentration (C_br(t) » C_b(t)), equation 8 simplifies to

C_B(t) = kπj, t _ o _hr P C_br(τ) dτ V J o In other words, the majority of the biomarker molecules will exist in the bound state, C_br(t), and since C_b(t) is negligible and equation (7) becomes dCB(t} = km,!, kout.br Cbr(t) dt V

Therefore, C_B(t) ≡ C_br(t), and

C_B(t) = _k_1! b_ _ ^■Kjgjj " ^out.br * V ^out,br * ^out,br A

At t ->°o or steady state,

because the biomarker bound to the earner protem acquires the clearance rate of the earner protem, _out,br ⁼ _out,_r, and the total steady state concentration of biomarker in the plasma becomes a simple function of the biomarker production rate and the clearance, excretion rate of the earner protein

and equation 1 can now be viewed as follows dNJ_fl(t) * k_ιn,_b - MB(t) ^{^} (14) dt

To illustrate the results predicted by these equations, some assumptions about rate constants are made With respect to the biomarker production rate, the biomarker production rate from the tissue is likely to be less than an actively secreted hormone. Consequently, production rates ranging three orders of magnitude below the typical rate of hormone secretion were examined. The association and dissociation rates for binding of biomarkers to carrier proteins are assumed to span the known range extendmg from peptides and small molecules binding to human serum albumin up to and mcludmg the affinity constants for hormone binding protems and the bmdmg of PSA (prostate specific antigen) to serum bmding proteins. The clearance rates for example serum earner proteins listed in Table 2 are known. The clearance and excretion rate for free biomarkers was chosen to span the known range for small molecules. The influence on total biomarker concentration over time of the caπier protein/biomarker complex clearance rate is illustrated in FIG. 3 A, which is a graphical representation of the numerical solutions to Equation 8 for a series of total biomarker clearance rates (half-lives). The total biomarker concentration is shown as a function of time after commencement of biomarker production. The production rate of the biomarker is assumed to be continuous at 1.0 femtomole/day. The biomarker mass is assumed to be 10 kDa, and possess a half-life of one hour. Even for the low level of biomarker production of one femtomole per day, steady state total biomarker concentrations are reached within 10 to 30 days. The steady state concentration of the biomarker is markedly influenced by the clearance rate of the carrier protein. For the example of FIG. 3 A, if the lower limit of sensitivity of a biomarker assay is one femtomole per liter, and the production rate is one femtomole per day, then the biomarker must have a half-life of greater than several days in order to accumulate a measurable level in the blood. Thus, without amplification due to adherence to the carrier protein, small biomarkers that are quickly filtered from blood would be undetectable. Influence of biomarker attractant molecule association is dramatic. Since a low molecular weight biomarker will be rapidly cleared by the kidney, its binding to high molecular weight carrier protein becomes the major mechanism for increasing the half-life of the biomarker in the circulation compartment. The amplification of biomarker molecules that are bound to the carrier protein at steady state is shown in FIG. 3B as a function of both the biomarker/biomarker attractant concentration and the carrier protein half-life. Numerical integration of equation 8 was conducted for a series of different ratios of biomarker / carrier protein as a function of the carrier protein half-life, assuming a very low affinity (IO^"4) between the biomarker and the carrier protein, a biomarker of mass 10 kDa with a half-life of one hour produced at a rate of 1.0 fmol/day. Under these assumptions, FIG 3B demonstrates that even for short half-life carrier proteins and relatively low affinity (k_on=10^"4) between the biomarker and the carrier protein, high concentration ratios of the carrier protein to the biomarker can lead to dramatic amplification of the biomarker. This result shows that an excess in concentration of carrier proteins compared to the biomarker, and a carrier with a relatively long half-life compared to the biomarker, can both contribute to amplification of the biomarker. FIG. 3C shows the effect that both biomarker affinity for a carrier protein and carrier protein concentration have on the percentage of biomarkers that are bound rather than free in a biological fluid. Numerical integration of equation 8 also shows that assuming a low affinity of 10^"4 between the biomarker and the carrier protein, a biomarker of mass 10 kDa with a half-life of one hour produced at a rate of 1.0 fmol/day, half-life of the carrier protein equal to albumin (18 days), and a vast excess of the albumin, virtually all biomarker will exist in the bound form. That is, even if a putative biomarker possesses even a relatively weak affinity for a high abundance carrier protein such as albumin (half life 18 days), the majority of the total circulating biomarker will still exist bound to albumin and be concentrated (amplified) in the biological fluid. Detection of such biomarkers, thus advantageously focuses on detecting them adhered to the biomarker attractant molecules, rather than free in solution as in prior methods. Table 2 below demonstrate the dramatic amplification of the biomarker concentration as a result of carrier protein binding. As shown in FIG. 3C and Table 2, the amplification A (Equation 10) can be several orders of magnitude. Canier protein amplification thus becomes a major factor determining whether a low abundance biomarker can reach a threshold of concentration that is above the lower limits of detection. A corollary of this conclusion is embodied in the generalized solution to equation 13, and is displayed in the results of Table 2: if the concentration of the biomarker and the half-life of the carrier protein are known, then an estimate can be provided for the minimum rate of production of the biomarker emanating from the tissue.

Table 2

Biomarker Concentration (fmol/L) ϊiomarker Absence of Any

Productio Carrier nRate Protein Albumin Ig IgG IgA/M IgD IgE (fmol/da t'Λ: 1/8 t'Λ: 18 t'Λ: 21 t'Λ: 23 t'Λ: 5 t'Λ: 2.8 t'Λ: 2 y) days days days days days days days 0.01 0.0005 0.07 0.08 0.09 0.02 0.01 0.01 0.10 0.005 0.69 0.80 0.88 0.19 0.11 0.08 1 0.05 6.88 8.02 8.79 1.91 1.07 0.76 10 0.5 68.8 80.2 87.9 19.1 10.7 7.6 100 5 688 802 879 191 107 76 1000 48 6877 8023 8788 1910 1070 764 10000 478 68773 80235 87876 19103 10698 7641 100000 4776 687725 802346 878760 191035 106980 76414 1000000 47759 6877254 8023463 8787602 1910348 1069795 764139 ^♦Theoretical prediction based on equation 10 of measured total biomarker concentration for selected high abundance serum carrier proteins, as a function of biomarker production rate, and carrier protein half-life. The assumption is that due to the vast excess of carrier protein, and even for a weak association constant, the majority of the biomarker species exists in a state bound to the selected carrier protein.

With regard to mass spectrometry serum proteomics, the most significant result of the theoretical analysis is that the vast majority of low molecular mass serum plasma biomarkers will exist bound to large carrier proteins. This conclusion was tested experimentally using SELDI analysis of native serum fractions of high and low molecular mass. Montage Albumin Deplete Kit, Centriplus centrifugal filters with average MWCO

(molecular weight cutoff) of 30 kDa, and ZipTip Cι₈ reversed-phase desalting columns (6 μm bed) were purchased from Millipore Coφoration (Bedford, MA); human serum acquired and processed in house; sequencing grade modified trypsin purchased from Promega (Madison, WI); dithiothreitol

(DTT) and iodoacetamide, (97%) were purchased from Aldrich Chemical Co (Milwaukee, WI); glacial acetic acid and formic acid (88%) were purchased from Maillinckrodt Baker Inc (Phillipsburg,

NJ); trifluoroacetic acid (TFA, sequanal grade) purchased from Pierce (Rockford, IL); acetonitrile (ACN) spectromettic grade purchased from Merck (Darmstadt, GE); ammonium bicarbonate

(NH₄HCO₃) certified purchased from Fisher Scientific (Fair Lawn, NJ); urea, enzyme grade, purchased from BRL Life Tech (Gaithersburg, MD). Serum samples were derived from the ovarian cancer clinical study set of the National

Ovarian Cancer Early Detection Program, Northwestern University. The full characteristics of this study set has been described previously and bioinformatic serum proteomic pattern analysis was conducted as described therein (see, Petricon et al., Lancet, 359: 572-577, 2002, the entirety of which is incoφorated by reference herein). Surface Enhanced Laser Desoφtion and Ionization (SELDI) was conducted using a PBSII

(Ciphergen Systems) and human serum was collected and anonymized as previously reported in Bonk and Humeny, Neuroscientist, 7:6-12, 2001 and Weinberger et al., J. Chrom. B. Analyt. Technol

Biomed. Life Sci., 782: 307-316. Analysis was conducted on a WCX2 (weak cationic exchange) chip.

The serum was fractionated into molecular mass classes under native conditions. Mass fractionation was carried out as follows. Thirty microliters of unfractionated human native serum was introduced into a Sephadex G-25^® or a Sephadex G-50 molecular sieve spin column according to the manufacture's instructions. The column was centrifuged at 3000 x g for three minutes, and approximately 30 microliters of eluate containing the high molecular mass fraction was collected. The eluate was treated with 50% acetonitrile (w/w in water) to dissociate bound molecules for 30 minutes and was transferred to the inlet of a molecular filtration microcolumn. (Microcon

YM-30 Millipore Centrifugal Filter Device) and the column was centrifuged at 1000 x g. The eluate containing the low molecular weight fraction was collected. All fractions at each stage were sampled and one microliter was analyzed by SELDI. Segregation of albumin and its low molecular mass binding constituents was conducted using the Montage Albumin Deplete Kit. 100 μL of human serum was diluted one to one with

Equilibration Buffer provided with the kit for a final volume of 200 μL and vortex sample. The column was rehydrated twice with 400 μL of Equilibration Buffer and centrifuge through the column insert for 2 minutes at 2,000 φrα 200 μL of diluted serum was introduced into the rehydrated albumin column and centrifuged for 2 minutes at 2,000 φm. The eluate from the column contains the serum without albumin. The bound fraction contains the albumin and the low molecular weight species bound to the albumin. 400 μL EAM solution were then added, composed of 50% acetonitrile and 0.1% TFA, to the column to strip the column and dissociate albumin from its bound species. After 30 minutes, EAM solution was centrifuged through the column at 2,000 φm for 3 minutes. The eluate contained the dissociated albumin and low molecular weight species that binds to albumin. Analysis of the proteins bound to the column using ion trap mass spectrometry was performed in line with an LCQ Classic MS (ThermoFinnigan, San Jose, CA) with a modified nanospray source. Dynamic exclusion of the three most abundant peptide hits from a full MS scan were selected for MS/MS analysis by collision induced dissociation with normalized collision energy of 35% and an activation time of 30 ms. Ion spray voltage was 2.00 kV with a capillary voltage of 26.20 V and a capillary temperature of 160 °C. Results for MS/MS scans were searched and compared with theoretical spectra in the Sequest Browser database specified for human proteins. WCX2 protein arrays for SELDI-TOF analysis were processed in a bioprocessor (Ciphergen Biosystems, Inc). 100 μl of 10 mM HCL was applied to the protein arrays in the bioprocessor and allowed to incubate for 5 minutes. The HCL wash was aspirated and discarded and 100 μl of H₂0 was applied and allowed to incubate for one minute. The H₂0 was aspirated and discarded, then reapplied for another minute. 100 μl of 10 mM ammonium acetate with 0.1% TritonX was applied to the surface and allowed to incubate for 5 minutes. The ammonium acetate was aspirated and discarded. A second application of ammonium acetate was applied and allowed to incubate for 5 minutes. The chip surfaces were then dried using a vacuum to remove any excess amount of liquid. Five μl of raw sera, or molecular mass fraction, or eluate was then applied to each chip surface and allowed to incubate for 55 minutes. Each protein chip was washed six times with 150 μl of PBS and H₂0 and then vacuum dried. Cross contamination was eliminated between spots by using a bioprocessor gasket. The gasket was removed and 1.0 μl of a saturated solution of the energy absorbing molecule cinnamic acid (25% saturation) in 50% (v/v) acetonitrile, 0.5% trifluoroacetic acid was applied to each spot on the protein array twice allowing the solution to dry between applications. A schematic of the fractionation scheme is shown in FIG. 4, and the mass spectrometry results following fractionation are shown in FIG. 5. The results demonstrate that most of the detectable ions by MS are derived from molecular species, putative biomarkers, that are bound to large carrier proteins. In fact, removal of the high molecular mass proteins under native conditions (FIG. 5B), a common method used for biomarker discovery, removes a significant proportion of the ions generated by SELDI-TOF. As shown, the preponderance of mass spectrometry detected low molecular weight ions are retained within the high molecular weight protein carrier fractions (FIGS. 5C-F), and do not exist in the free phase (FIG 5B). FIG. 5A is the spectrum obtained for a control of the SELDI matrix alone. Spectra generated from whole, native, unfractionated serum applied directly to the WCX2 chip surface. FIG 5B is the spectrum after removal of all species greater than 30,000 MW with a Microcon^® YM-30 Molecular Filter from the unfractionated serum. FIG. 5C is a direct analysis of high molecular weight native species by MS after fractionation through a Sephadex G-25 Molecular Sieve. The MS profile displays the low molecular weight species ionized and desorbed away from the high molecular weight carriers. FIG. 5D shows the spectrum after the high molecular weight fraction eluted from the G-25 sieve was treated with a 50% acetonitrile solution to dissociate potential low molecular weight species bound to their high molecular weight carriers, and the dissociated eluate was then passed through the YM-30 Filter. Only the low molecular weight species previously bound to the G-25 sieved high molecular weight fraction should pass through the filter. FIG 5E is a spectrum of the ion species associated with HMW earner species depleted of albumin and albumin binding partners. FIG 5F is a spectrum showing the ion species associated specifically with albumin only. A comparison of the spectrum of FIG. 5B to that of FIG. 5C shows that the majority of ions generated from unfractionated serum are derived from species associated with larger carrier proteins. FIG. 5D displays the spectrum of ion species previously bound, and then dissociated and separated from the higher molecular mass fraction. The intensity and number of ion species is augmented compared to FIG. 5A and FIG 5C. FIGS. 5E and 5F compare the proportion of SELDI-TOF generated ions associated with albumin compared to all the other serum carrier proteins. FIG. 5E displays the ions associated with the non-albumin earner proteins, and FIG. 5F displays the ions generated from species bound only to albumin. A significant proportion of the ions in the spectra appear to be derived from species associated with albumin. Microcapillary LC MS/MS was used to verify that the albumin bound fraction acquired through stripping the Montage Albumin Deplete Column was entirely albumin and its bound low molecular mass species. Since there was no indication of other high molecular mass proteins bound to the albumin specific column, the low molecular mass species detected were specific to albumin. Furthermore, a number of low molecular mass species that were dissociated from the albumin carrier were positively identified. Additional results characterizing the entire repertoire of low molecular mass species bound to serum carrier proteins by LC MS/MS were obtained. It was then shown that the ions generated from the species bound to albumin contained disease biomarker information. SELDI-TOF ion patterns, generated on the WCX2 chips, correlating with ovarian cancer were identified by methods described previously. The sample set included 91 controls and 162 ovarian cancers. The entire process of preparing the chips was done using a robotic instrument. The following selection of ion mass/charge (m/z) values generated a pattern that was

100% predictive in the training and blinded testing- 2760.6685, 19643.409, 465.56916, 6631.7043, 14051.976, 435.46452, and 3497.5508. Randomly selected representative serum samples from this study set were analyzed by MALDI-TOF comparing the spectra generated by unfractionated sera to the spectra generated only from the species bound to albumin. As demonstrated in FIG. 6, the spectrum generated from the species bound to albumin is complex and exhibits a number of differences between the cancer and the unaffected ("normal") cases shown in the example. A comparison of the peak intensities between the unfractionated serum (FIGS. 6C and 6D) and the albumin-bound fraction (FIGS. 6A 6B) shows that a significant set of putative disease biomarkers may be associated with albumin. FIG. 8 displays the ion spectra for an example serum sample in which the ion species bound only to albumin are compared for different amounts of albumin captured. FIGS. 7A-D show the example of ion 6631.7043, a member of the ion pattern 100% correlated with ovarian cancer in this clinical study set. Matched for dilution and amplitude, the predicate ion is highly associated with albumin, and the ionization intensity is augmented in the albumin bound fraction. Finally the complexity and amplitude of the ion species of the SELDI-TOF spectra is directly related to the amount of albumin captured as shown in FIG. 8. The ion species increase in amplitude and complexity when they are derived from larger quantities of bound albumin. This supports the concept that the biomarkers are quantitatively associated with the albumin The theoretical and experimental data support the concept that the vast majority of small mass ions detected by mass spectrometry of native human serum exist in association with circulating earner proteins of higher molecular weight. This conclusion is important for biomarker physiology and biomarker measurement technology. One result is that the concentration of a biomarker measured in serum or plasma is directly related to the clearance rate or half life of the carrier protein, not the biomarker clearance rate itself. As shown above in equation 13, the concentration of the biomarker is a function of the ratio between the biomarker production rate from the tissue and the clearance rate of the carrier protein. This means that carrier protein binding amplifies the total biomarker concentration levels measured in serum or plasma. Amplification occurs because the carrier protein acts as a reservoir to accumulate the biomarker over time, as the tissue is continuously producing the biomarker. Thus, a biomarker produced by a small volume of tissue such as the ovary, prostate, or breast at a low concentration (for example, one femtomole per day) can accumulate to a concentration of one picomole in the serum because it binds with a carrier protein with much larger half-life. In this example, the existence of the carrier protein can raise the concentration of the biomarker to a range detectable by conventional assay technology. Without the carrier protein, the free biomarker would be rapidly cleared by the kidney and would therefore reside at a steady state concentration many fold below the detection limits of assay technology. Another conclusion is that the results extend beyond current mass spectrometry detection technology. Small biomarkers are commonly not the province of two-site sandwich immunoassays, because it is difficult to develop two antibody-binding sites on the same small molecule. In contrast, if one half of the immunoassay sandwich is the carrier protein and the other half was the small biomarker, a sandwich immunoassay can be achieved. The biomarker clearance rate becomes the carrier protein clearance rate because the carrier protein, even if it has low affinity for the biomarker, is in vast excess. Therefore, another implication is that if the clearance rate of a given carrier protein is known, and the serum/plasma concentration of the biomarker bound to that carrier protein is known; an estimate a lower limit on the continuous production rate by the tissue (Equation 13) may be made. For example, if the concentration of a generic, bound biomarker α is 3.8 ng/mL (42.22 fmol/mL, where the biomarker α with its carrier protein have a combined molecular weight of 90 kDa) and the carrier protein half-life is 2.43 days, then the production rate of α from the tissue is at least 45,000 femtomoles per day. If α is produced by a one cubic centimeter tumor composed of IO⁹ cells, then each cell would produce approximately 16,000 molecules per day. This approximation is consistent with previous experimental findings. As shown in the experimental data presented in FIGS. 5-7, the theoretical predictions of FIG. 3 have been demonstrated. Thus, in analyzing unfractionated native sera, the majority of ions generated by SELDI-TOF analysis are found to be associated with carrier proteins, rather than free in solution phase (compare FIG. 5B to FIGS. 5C-F). Moreover, as shown in FIGS. 6 and 7, ion species altered in disease study sets may be those specifically captured on a single carrier protein. In the example, the carrier protein is albumin. In the past, extensive effort has been placed to separate and throw away the high abundance, large carrier proteins in the native plasma so that the remaining low abundance, disease-related markers can be discovered. The present results demonstrate that the search for biomarkers must be directed at those bound to the carrier proteins or other BAMs. Removal of high abundance serum/plasma proteins prior to proteomic analysis should be conducted after, not before, dissociation from binding partners, since the results indicate that the low molecular mass proteome, existing within the range detectable by MALDI-TOF exists predominately in the bound phase. Thus, technologies that focus on efficient capture of the carrier proteins and specific elution of the low molecular weight biomarkers will likely yield the greatest amount of diagnostic information. Although the bound biomarkers may exist in concentrations ten to 500 times greater compared to their free counteφarts, and since the carrier proteins exist in vast excess compared to the biomarkers, it is unlikely that the carrier proteins will become saturated with bound biomarkers.

Moreover, based on its unique affinity topology, each carrier protein may have its own constellation of bound biomarkers. Indeed, the distribution of biomarkers among specific plasma/ serum carrier proteins may contains important diagnostic information.

Example 2 - Identification of Proteins Having Fragments Bound to Natural Circulating BAMS Hundreds of peptides and clipped proteins bound to carrier proteins such as albumin have been sequenced by MS/MS. Following trypsin digestion of the canier bound fragments, identification of the proteins/peptides was performed in line with an LCQ Classic MS (ThermoFinnigan, San Jose, CA) with a modified nanospray source. Dynamic exclusion of the three most abundant peptide hits from a full MS scan were selected for MS/MS analysis by collision induced dissociation with normalized collision energy of 35% and an activation time of 30 ms. Ion spray voltage was 2.00 kV with a capillary voltage of 26.20 V and a capillary temperature of 160 °C. Results for MS/MS scans were searched and compared with theoretical spectra in the Sequest Browser database specified for human proteins. The sequencing results for trypsin fragments of carrier bound peptides/proteins reveal molecules derived from all parts of cells, ranging from low concentration transcription factors from the nucleus, to membrane bound receptors, to secreted growth factors. Table 3 below illustrates the general types of proteins found. Table 3

Example 3 - Discovery of ovarian cancer biomarkers bound to human serum albumin. The Clementine Data Mining Bioinformatics Toolset (SPSS, Inc., Chicago, IL) was used to discover patterns of ions representing biomarkers bound to albumin that discriminate between serum samples obtained from subjects with and without ovarian cancer. Prior to introducing the dataset to the Clementine system, a feature selection was performed by computing and then comparing the aggregate mean values of the total ion count (sum of amps) on a bucket by bucket (bin by bin) basis, and selecting the 100 buckets (bins) with the largest difference in mean values between normal and cancer samples. The Clementine system was then trained using un-normalized total ion count per bin values. Using a decision tree analysis, a set of ion species bound to albumin were found to discriminate between 127 non cancer and 115 ovarian cancer cases with 100% specificity and 100% sensitivity. A Qstar mass spectrometer (described in Example 1) was used to analyze albumin bound proteins (and protein fragments) in the mass to charge regions of 1000-7760 and 7762-11,000. These ranges excluded the matrix effect area and the region of Transferrin. The ion amplitude and density of ion peaks was ten fold greater using the albumin eluted fraction, compared to direct analysis of whole unfractionated sera, making detection of the discriminatory patterns of ions suφrisingly reliable. Although several groupings of ions were found to be discriminatory, the key ions in one highly discriminatory group had mass to charge ratios as follows: 3884, 1033, 2366, 4033, 4070, 3157, 7953, 6678, 7837, 2375, in order of importance. Within the molecular mass regions identified by the decision tree analysis, fragments of a number of previously unknown serum biomarkers were also identified as being bound to albumin in the ovarian cancer patients. The molecular masses of the fragments were validated using Western Blotting with polyclonal antibodies directed against the candidate proteins. The new ovarian cancer biomarker proteins, identified as being bound to albumin are listed in Table 4 below. These proteins are considered to be low abundance cellular proteins, and were not previously expected or reported to be found in serum. Table 4

In conclusion, the mass specfrometric analysis of molecules bound to a BAM (in this case a serum carrier protein, specifically albumin) circulating in the blood of the women test subjects who either had biopsy proven ovarian cancer, or were disease free for 5 years, in combination with pattern recognition, revealed a large number set of previously unknown biomarkers associated with ovarian cancer. In this case the BAM is natural albumin. Sequencing and identification of the carrier protein associated molecules revealed that they are primarily derived from low abundance cellular and tissue sources, and represent all major classes of cellular compartments, including membrane, cytosol, nucleus, and extracellular matrix.

Example 4 - Detection of Stage Specific Ovarian Cancer Biomarkers and Biomarkers Indicating Lack of Ovarian Cancer in a Subject. In this example, albumin bound molecules were analyzed to detect protein and protein fragment biomarkers that discriminate between different populations of subjects having no evidence of ovarian cancer, Stage I ovarian cancer or Stage III ovarian cancer. Biomarkers adhered to albumin were analyzed and identified by mass spectrometry. Serum Samples Serum was collected prior to physical evaluation, diagnosis and treatment and was stored at -80 centigrade. The study set used consisted of 98 unaffected "high risk" patient samples, and 146 cancer patients. Serum samples were obtained from the National Ovarian Cancer Early Detection Program (NOCEDP) and gynecologic oncology clinic at Northwestern University (Chicago, Illinois). Specimens from women enrolled in the NOCEDP who had no evidence of any cancer for 5 years were evaluated as healthy (no cancer) women. Similarly, only preoperative specimens were used from women who were surgically staged and found to have epithelial ovarian carcinoma. Two hundred and forty eight samples were prepared using a Biomek 2000 robotic liquid handler (Beckman Coulter, Inc., Palo Alto, California). All analyses were performed using ProteinChip weak cation exchange interaction chips (WCX2, Ciphergen Biosystems Inc., Fremont, California). A control reference sample was randomly applied to one spot on each protein array as a quality control for overall process integrity, sample preparation and mass spectrometer function. The control sample, SRM 1951 A, which is comprised of pooled normal human sera, was provided by the National Institute of Standards and Technology (Gaithersburg, Maryland). Application of Quality Control Analysis The 248 mass spectra acquired by the QqTOF MS were analyzed with a wide variety of statistical tools to evaluate the spectral quality [i.e. record count and mean amplitude] and statistical variances greater than the population norm. Mean spectral amplitude and the file record count (i.e. the total number of data points within a mass spectrum) were selected as global parameters for statistical analysis. Using these parameters, the mass spectra from serum albumin elutions of unaffected individuals and cancer patients were tracked over several days of analysis and over different lots of ProteinChips. The total record count of each mass spectrum was compared to that recorded for the reference sample. QqTOF MS Analysis ProteinChip arrays were analyzed using a hybrid quadrupole time-of-flight mass spectrometer (QSTAR pulsar i, Applied Biosystems Inc., Framingham, Massachusetts) fitted with a ProteinChip array interface (Ciphergen Biosystems Inc.). Samples were ionized with a 337 nm pulsed nitrogen laser (ThermoLaser Sciences model VSL-337-ND-S, Waltham, Massachusetts) operating at 30 Hz. Approximately 20 mTorr of nitrogen gas was used for collisional ion cooling. Each spectrum represents 100 multi-channel averaged scans (1.667 min acquisition/spectrum). The mass spectrometer was externally calibrated using a mixture of known peptides. Elution and sequencing of albumin bound proteins Chemicals: acetonitrile (ACN) (HPLC grade), ammonium bicarbonate (NH₄HC0₃), iodoacetamide (97%), methanol (99+%) and dithiothreitol (DTT) were purchased from Sigma- Aldrich Co. (St. Louis, MO). Trifluoroacetic acid (TFA) (sequanal grade) purchased from Pierce (Rockford, II). Formic acid (88%), acetic acid (glacial) purchased from Mallinckrodt Baker (Phillipsburg, NJ). H₂0 was doubly distilled in house with Kontes High Purity Water System. Porcine sequencing grade modified trypsin was purchased from Promega (Madison, WI). Coomassie Brilliant Blue R 250 was purchased from Fluka Chemical (Switzerland). Sequencing of albumin associated proteins was conducted on groups of 30 stage 1 cancers,

30 stage 3 cancers, and 30 high high risk ovarian human serum, obtained as described above. Vivaspin 500 centrifugal membrane filters were purchased from VivaScience. ZipTip g (6 μm bed) desalting columns and Montage Albumin Depletion Columns were purchased from Millipore. Elution buffer was replaced with a 70% ACN/ 30% dd H₂0/ 0.2% TFA mixture. Pre cast gels, sample, and running buffers were purchased from Invifrogen Co. Fused silica is from Polymicro Technologies (Phoenix, AZ). Albumin and bound peptide purification: 40 μL of human stage specific (pooled) ovarian cancer serum (~5 mg protein) was diluted to 200 μL with Equilibration Buffer (Millipore) and run through a (Montage) albumin specific affinity column twice. The bound protein was washed thoroughly, and eluted from the column by equilibrating with 70% ACN/ 30% H₂0/ 0.2% TFA for 30 minutes, followed by a slow spin-through of the elution mixture. The eluate was lyophilized to <10 μL in a HetoVac rotofor (CT 110) and reconstituted in a 95% H₂OZ 5% ACN / 0.1 % formic acid buffer (Buffer A). Samples were sometimes desalted with a ZipTip cleanup or with Vivaspin 500 centrifugal membranes and always reconstituted in a 1 : 1 mixture of water and SDS sample buffer (20 μL total volume). ID gel separation and digestion: 20 μL of sample in SDS sample buffer (40 μL of original serum) was boiled for 5 minutes at 95 °C and run on a ID pre-cast gel (16% Tricene or 4-20% Tris-Glycine) to separate albumin from the proteins/peptides/fragments of interest. The gel was stained with Coomassie Blue for one hour and destained overnight in 30% methanol/ 10 % acetic acid solution. The entire lane containing stage specific serum proteins was excised from the gel and finely sliced into very small molecular weight regions (~ 40 slices/ lane). Gel bands were reduced and alkylated with 10 mM DTT and 55 mM iodoacetamide, incubated at 4 °C for 1 hour in trypsin (20 ng/ uL) and allowed to digest overnight at 37°C in 25 mM NH₄HCO₃. The following moming, proteins were extracted from the gel with a 70% ACN/ 5% formic acid solution. μLC/MS/MS analysis: Samples were lyophilized to near dryness and reconstituted in 6.5 μL of Buffer A for mass spec analysis. Microcapillary reverse phase LC/MS/MS analysis was performed with Dionex's LC Packings liquid chromatography system coupled online to a ThermoFinnigan LCQ Classic ion trap mass spectrometer (San Jose, CA) with a nanospray source. Reverse phase separations were performed with an in-house, slurry packed capillary column. The C₁₈ silica-bonded column is 75 μm i.d., 360 μm o.d., 10 cm long fused silica packed with 5 μm beads with 300 Angstrom pores (Vydac, Hesperia, CA). A μ-precolumn PepMap C|₈ cartridge (Dionex) acts as a desalting column. Sample is injected in μL pick-up mode and washed with Buffer A for five minutes prior to a linear gradient elution with Buffer B (95% ACN/ 5% H₂0/ 0.1% formic acid) up to 85% over 95 minutes at a flow rate of 200 nL/ minute. Full MS scans are followed by three MS/MS scans of the most abundant peptide ions (in a data dependant mode) and collision induced dissociation is performed at a collision energy of 38%. Data analysis: Data analysis was performed by searching MS/MS spectra against the European Bioinformatics Institute of the non-redundant proteome set of Swiss-Prot, TrEMBL and Ensembl entries through the Sequest Bioworks Browser (ThermoFinnigan). Peptides were considered legitimate hits after filtering the correlation scores (refer below) and manual inspection of the MS/MS data. The criteria used to filter data were at least as stringent as most reported in the literature.

Charge -^corr DeltaCNSp Ions Rsp

+1 >1.9 >0.1 >500 >50% <5

+2 >2.5 >0.1 >500 >50% <5

+3 >3.5 >0.1 >500 >50% <5

Accepted peptide hits were required to have an X_corr ranking = 1 relative to all other peptides in the database. The albumin extraction, gel electrophoresis, protein digestion/extraction, and LC/MS/MS analysis was repeated in more than five distinct trials - each time yielding diminishing returns of new identifications for low abundance peptide hits. Repetitive sequencing of peptides in multiple trials further validated the experimental procedure - both within a stage and between ovarian cancer stages. All of the proteins identified adhered to albumin, and their presence or absence in one or more of the no-cancer high risk, stage I cancer and stage III cancer populations are shown in Table 5 below. Biomarkers useful for providing diagnoses of ovarian cancer can be chosen from Table 5 as follows: a) Biomarkers that indicate a subject exhibits no cancer can be chosen to include at least an identifiable fragment of each of two or more proteins in Table 5 (such as 3 or more, or 5 or more) that are were identified only in the no cancer population. b) Biomarkers that indicate a subject exhibits ovarian cancer regardless of its stage can be chosen to include at least an identifiable fragment of each of two or more proteins in Table 5 (such as 3 or more, or 5 or more) that were identified only in either or both of Stage I and Stage III populations, and can further include fragments of proteins in Table 5. c) Biomarkers that indicate a subject exhibits Stage I ovarian cancer can be chosen to include at least an identifiable fragment of each of two or more proteins in Table 5 (such as 3 or more, or 5 or more) that were identified only in the Stage I population. d) Biomarkers that indicate a subject exhibits Stage III ovarian cancer can be chosen to include at least an identifiable fragment of each of two or more proteins in Table 5 (such as 3 or more, or 5 or more) that were identified only in the Stage III population. Table 5

Example 5 -Prostate Cancer Biomarkers This example demonstrates that the disclosed method can detect serum biomarkers for prostate cancer. The sera of subjects in each of a prostate cancer population and a no cancer population (hospital control) were pooled and analyzed as in Example 4, except that potential biomarkers adhered to all serum proteins of MW > 25 kDa were separated from the pooled sera samples, and potential protein and protein fragment biomarkers were eluted therefrom and identified. The results of this analysis are shown in Table 6. Useful biomarkers for prostate cancer can be selected to include at least an identifiable fragment of each of two or more proteins in Table 6 (such as 3 or more, or 5 or more) that were found only in the prostate cancer population. Useful biomarkers for the absence of prostate cancer can be selected to include at least an identifiable fragment of each of two or more proteins in Table 6 (such as 3 or more, or 5 or more) that were found only in the no prostate cancer population. Table 6

Example 6 - Breast Cancer This example demonstrates that the disclosed method can detect serum biomarkers for breast cancer. The sera of subjects in each of a breast cancer population and a no breast cancer population (hospital control) were pooled and analyzed as in Example 4. The results of this analysis are shown in Table 7. Useful biomarkers for breast cancer can be selected to include at least an identifiable fragment of each of two or more proteins in Table 7 (such as 3 or more, or 5 or more) that were found only in the breast cancer population. Useful biomarkers for the absence of breast cancer can be selected to include at least an identifiable fragment of each of two or more proteins in Table 7 (such as 3 or more, or 5 or more) that were found only in the no breast cancer population.

Table 7

Example 7 - Lung Cancer Biomarkers This example demonstrates that the disclosed method can detect serum biomarkers for lung cancer, and biomarkers that differentiate the histopathological type of the lung cancer. The sera of subjects in each of an adenocarcinoma population, a squamous cell carcinoma population and a no cancer population (hospital control) were pooled and analyzed as in Example 4. The results of this analysis are shown in Table 8. Useful biomarkers for lung cancer (of either type) can be selected to include at least an identifiable fragment of each of two or more proteins in Table 8 (such as 3 or more, or 5 or more) that were found only in one or both of the adenocarcinoma population and the squamous cell carcinoma population. Useful biomarkers for adenocarcinoma can be selected to include at least an identifiable fragment of each of two or more proteins in Table 8 (such as 3 or more, or 5 or more) found only in the adenocarcinoma population. Useful biomarkers for squamous cell carcinoma can be selected to include at least an identifiable fragment of each of two or more proteins in Table 8 (such as 3 or more, or 5 or more) found only in the squamous cell carcinoma population. Useful biomarkers for the absence of lung cancer can be selected to include at least an identifiable fragment of each of two or more proteins in Table 8 (such as 3 or more, or 5 or more) found only in the no cancer population. Table 8

Example 8 - High Throughput Chamber for Collection of Biomarkers from BAMs Collection of BAMs and subsequent elution of their bound biomarker cargo may be accomplished in a high throughput flow chamber. With reference to FIG. 9, the sample body fluid containing the carrier molecule and its associated diagnostic molecule analytes are passed into a column or passed along a surface which contains immobilized antibodies or other ligand directed against a BAM, such as the illustrated carrier protein. In one embodiment, immobilized antibodies that also comprise a means to couple (for example a means to covalently couple) the captured BAMs are used. Following capture of the BAM (and possibly covalent coupling of the BAM to the capture agent) and its analyte cargo, the analytes can be dissociated off the earner molecule using a denaturation or suitable elution buffer. Because the carrier protein itself is covalently attached to the solid phase through the antibody/ligand it will not be eluted. Thus, only the analytes previously associated with the carrier molecule will be eluted and can be analyzed for the diagnostic information they contain based on their biochemical or physiologic characteristics. An example means for coupling the antibody ligand to the captured BAM is the use of light activated coupling described by Holden and Cramer (JACS, 125: 8074-8075, 2003). This method couples fluorescently labeled molecules to binding partners using a photo-bleaching reaction that initiates a reactive singlet oxygen to covalently couple with highly localized electron rich sites on the binding partner. Using this method to capture albumin-associated analytes a serum sample is first passed through a column containing immobilized fluorescein (or Alexa dye) labeled antibodies to albumin. The albumin is captured on the antibodies. A photo-bleaching illumination is then provided to couple the antibodies to the albumin through the fluorescein moiety. This is then followed by an elution step which elutes the albumin associated analytes but leaves the albumin itself bound to the column. The whole process can be in-line and continuous, with sample introduction followed by pulses of light energy to couple the carrier molecule and followed by subsequent elutions of the analytes. The capture itself can take place in a filter format, a microwell chamber, etched channels, packed nozzels, in-solution on particles which are subsequently filtered out or captured by another means or in a microfluidic device. Further separation and processing can be done before or after this capture and elution step. The output can even go directly into a mass spectrometer, or into a chemical and enzymatic step (for example, a trypsin or cyanogen bromide cleavage) and then be analyzed by MS as illustrated in FIGS. 10. FIG. 11 illustrates an embodiment in which capture and covalent coupling of a BAM to an immobilized antibody is followed directly by laser desoφtion into a mass spectrometer. In this embodiment, an antibody labeled with fluorescein is immobilized on a silica surface through a linker molecule (FIG. 11 A). A biological fluid comprising a BAM that is bound by the particular antibody (such as the illustrated carrier protein) is passed through the device and the BAM with its diagnostic cargo is captured by the immobilized antibody (FIG. 1 IB). UV (or visible) light is then used to excite the fluorescein moiety, which then catalyzes covalent coupling between antibody and the BAM (FIG. 11C). Once the BAM is immobilized, a laser pulse is used to desorb the biomarkers, such as peptides, from the BAM and into a mass spectrometer. Any method known in the art may be used to immobilize a capture agent (such as an antibody or protein A) to a surface in the devices of FIGS. 9-11, and the device itself may be constructed from a wide variety of materials, including, for example, a metal, silicon or a plastic. Examples of methods for attaching capture agents to particular substrate materials include the photoactivated polystyrene method of Bora et al. (J. Immunol. Methods, 268: 171-7, 2002), the method for immobilizing proteins to modified silicon surfaces described by Yakovleva et al. (Anal. Chem., 74: 2994-3004, 2002), the use of microporous membranes made of poly(2- hydroxyethylmethacrylate) carrying protein A as described by Denizli and Arica (J. Biomater. Sci. Polm. Ed., 11: 367-382, 2000), the method for immobilizing antibodies to a metal coated with polyimide described by Wessa et al. (Biosens Bioelectron. 14: 93-98, 1999), or the method for coupling immunoglobulin to a graft polymer of poly(vinyl)alcohol-poly(acrylic acid) described by Disley et al., (Biosens. Bioelectron, 13: 383-396, 1998). Methods for attaching antibodies to a metal surface are described in Storri et al., Biosens. Bioelectron., 13: 347-357, 1998. A method of immobilizing proteins such as antibodies using heterobifunctional crosslinkers is described by Shriver-Lake et al., Biosens. Bioelectron., 12: 1101-1106, 1997. Photochemical linkage of antibodies to silicon chips is described by Sundarababu et al, Photochem. Photobiol, 61: 540-546, 1995. A variety of methods for immobilization of protein monolayers are reviewed in Williams and Blanch, Biosens. Bioelectron., 9: 159-167, 1994. Chelating peptides may also be used to immobilize antiobodies on a solid support (see, for example, Loεtscher et al., J. Chromatogr., 595: 113-119, 1992). Antibodies may also be immobilized on ε polystyrene copolymer surface, such as onto styrene copolymer beads, which could be used in a column format for the disclosed device where a sample is passed through a column of such beads to remove BAMs, from which biomarkers may later be eluted (see, for example, Bale-Oemck et al , Ann Biol Clin, 48 651-654, 1990) A combmation of thiol-terminated silanes and heterobifunctional crosslinkers may be used to immobilize antiobodies on silica surfaces (see, for example, Bhatia et al , Anal Biochem , 178 408-413, 1989) Additional methods exist and are contemplated If the BAM is a magnetic particle, a magnet may be mcluded m the device of FIGS 9-11, and used to capture the BAM, and retam it as the biomarkers are eluted therefrom If the BAM is a dendnmer, either antibodies directed to the type of dendnmer, or a chemical reaction between the surface functional groups of the dendnmer and the surface of the device, may be used to retain the BAM while the biomarkers are eluted therefrom Alternatively, the dendnmer BAM could be denvatized, for example, with biotm and captured by streptavidin covalently bound to the surface of the device The device itself may be constructed using nano fabrication methods as is described, for example, in Marnan and Tennant, J Vac Sci Technol A, 21 S207-S215, 2003 (incoφorated by reference herein) to provide a microfluidic device

Example 9 - Artificial BAMs In addition to using natural circulating protems to collect molecules of diagnostic importance, it is possible to use engineered molecules and matenals to collect biomarkers from biological fluids Advances in nanotechnology and microfabncation provide a means to mass- produce "nanoharvesting" BAMs designed to specifically capture and amplify different classes of LMW biomarkers by altermg, for example, their surface chemistry For example, BAMs tailored to monitor specific diseases developing in target tissues may be produced using existing targeting technologies Such BAMs agents may be instilled into the blood of a patient at one pomt m time, and then collected at a later time point with their bound diagnostic cargo In addition, nanoharvesting BAMs, filtered and arrayed m a high throughput platform may be directly quened with mass spectroscopy to obtain an individual global health profile rendered at an affordable cost, all from a drop of blood Examples of suitable artificial BAMs include chimeπc protems and proteins altered to enhance separation of the BAM from a biological fluid following exposure to the fluid For example, native protem sequences having an added polyhistidine tag may be used to enhance separation of the altered protein from the biological fluid using a metal-affinity column Metal affinity columns are descnbed, for example, in U S Patent No 5,962,641 Further examples of artificial BAMs include silica-coated nanoparticles having, for example, a magnetic core that facilitates separation of the BAM from a biological fluid Functional groups, such as carboxyl and amine groups, may be denvatized to the surface of the silica shell by any known mεtaod Such functional groups may also compnse proteins, enzymes, biotin, streptavidin, nucleic acids, and chemosensors such as fluorescent probes, that function to assist separation of the BAM from a biological fluid, permit detection of the BAM, or alter the types of LMM biomarkers that accumulate on the BAM when exposed to the biological fluid. Silica-coated nanoparticles, their preparation, and derivatization are described in detail in U.S. Patent No. 6,548,264. Another class of artificial BAM useful for the disclosed methods are the ultrafine lightly coated supeφaramagnetic particles described in U.S. Patent No. 6,207,134. These particles comprise a metal oxide core, such as an iron oxide core and a shell comprising a polyelectrolyte such as a structural polysaccharide or a synthetic polymer such as a polyaminoacid. Micoφarticles and nanoparticles comprising cross-linked monosaccharides and oligosaccharides may also be utilized as BAMs. Examples of such particles are described in U.S. Patent No. 6, 197,757. Such particles are biodegradable and may, for example, be separated from biological fluids with lectins specific for the sugar moieties found on their surface. In one embodiment, the particles comprise cyclodextrins that may be chosen to be of a size that captures biomarkers of particular mass ranges. The surface of such particles may be derivatized with, for example, amino or carboxylate groups according to known methods. Biodegradable, injectable nanoparticles that are not rapidly cleared from blood, and may be modified to target specific cells or organs, and to change their lipophilicity and hydrophilicity, are described in U.S. Patent No. 5,543,159. A wide variety of microspheres and nanospheres that may be used directly or derivatized for use as BAMs are available from Bangs Laboratories, Inc. (Fishers, IN). Examples of such microspheres include polystyrene, latex, silica, and polymethylmethacrylate particles. Covalent coupling of various biomolecules (such as Protein- A) and derivatization of the surface of such spheres are discussed in TechNotes 201 and 205 published by Bangs Laboratories (TechNote 201, October 25, 2002/TechNote 205, March 30, 2002). Another class of BAMs that may be used in the disclosed methods and apparatus are dendrimers. Examples of dendrimers, their synthesis, and surface derivatization are discussed in U.S. Patent Nos. 4,587,329 and 5,338,532. Dendrimers may also be derivatized with antibodies to provide targeted access to particular types of tissue, and collection of biomarkers therefrom. Both polyamidoamine and polyalkylenimine dendrimers may be used. Propylenimine dendrimers with primary amino surface groups, polyamidoamine dendrimers with primary amino surface groups, polyamidoamine dendrimers with carboxylate surface groups and polyamidoamine dendrimers with hydroxyl surface groups are available from Aldrich (Milwaukee, WI). Over 200 modifications of the surface groups of dendrimers are described in Dvornic et al, Poly. Prepr., 40:408, 1999. Example 10 - Pattern Recognition Methods Any known pattern recognition method may be used to identify BAM-associated molecules that discriminate between biological states, and to provide a probable diagnosis of a subject's biological state based on BAM-associated molecules. Statistical methods of pattern recognition may be used and are reviewed in Jain et al., IEEE Transactions on Pattern Analysis and Machine Intelligence, 22: 4-37, 2000, which is incoφorated by reference herein. Artificial intelligence methods may also be used (see, for example, Zupan and Gasteiger, Neural Networks for Chemists, VCH, 1993). In general, pattern recognition relates to the discovery of hidden patterns that characterize subsets of data within a larger data set, for example, data patterns that distinguish between objects of different classifications. Pattern recognition methods may be divided into three major components: clustering or classification, association rules, and sequential analysis. In classification or clustering, a set of data is analyzed to generate a set of grouping rules that may, for example, be used to classify diseases. An association rule is a rule that implies certain relationships between objects in a database. For example, sets of symptoms often occurring together with certain kinds of diseases may be identified. In sequential analysis, patterns that occur in sequence are discovered. For example, patterns that correlate with progression of disease may be detected. Pattern recognition involves a training (learning) mode and a classification (testing) mode. During the training mode, data for a set of objects (such as mass spectra obtained for a number of samples) is used to establish patterns of data that characterize various subsets of objects present amongst all objects (such as patterns of mass spectral data that are characteristic of those samples from subjects with cancer). Class (subset) labels may be assigned to the objects before patterns are detected, or after objects are first grouped into subsets based on their similar patterns of data. In the former case, the pattern discovery process is known as supervised learning, and in the latter, unsupervised learning. In both supervised and unsupervised learning, the goal is to extract pattems of data that may be used to classify future data. In the classification mode, data for objects of known (validation) or unknown (testing) class is compared to the patterns detected during the training mode. For example, a diagnostic pattern of mass spectral data may be validated by using the pattern to classify the spectra of samples known to be from subjects with or without cancer. If the diagnostic pattern is "robust," it will provide the correct classification of the known samples. Often this process is repeated for each of the known samples used during the training mode in what is called a leave-one-out (LOO) cross-validation. Once the pattern is deteπnined to be effective for classification, testing is performed to assign a class to an object of unknown class. Both the training mode and the classification mode may involve preprocessing of raw data to separate the desired pattern(s) from background and/or noise. Preprocessing may also include normalization of the data, or any other process that ultimately leads to a compact representation of the pattern(s) within the data. For example, mass spectral data may be normalized such that the highest intensity signal is given an intensity value of 100 and all other signals are given an intensity value equal to their intensity as a percentage of that seen for the highest intensity signal. Alternatively, an even simpler way of preprocessing mass spectral data is to convert the data to a series of ones and zeroes for each mass value (or mass/charge ratio), where a one means that a signal (regardless of intensity) is present at a particular mass value and a zero represents the absence of a signal at a particular mass value. Feature extraction selection is performed during training to identify a subset of features within the data (such as mass spectral signals at particular mass values) that may be used to classify objects into their respective classes without having to consider all the available features. More specifically, feature selection refers to using an algorithm to select the most discriminatory subset of features within the input feature set, and feature extraction refers to applying an algorithm that creates new features based on transformations or combinations of the original feature set. Feature selection may be performed after feature extraction to select the most discriminatory set of extracted features. Feature extraction/selection may involve an exhaustive search for the optimal subset of features (i.e. all combinations are tried), or when the number of features is so large that an exhaustive search is computationally infeasible, a heuristic search is used. In either case, feature extraction/selection leads to greater computational efficiency in learning. In the training mode, learning involves applying a classifier to partition the feature space into regions characteristic of each class of object within the training set. In other words, combinations of features that are characteristic of objects of a particular class are identified. Since objects may be represented by points in multidimensional feature space with coordinates conesponding to the values of their features, objects of the same class will tend to appear close to each other (cluster) in the feature space. The classifier tries to draw boundaries between the portions of the feature space where objects of the same class tend to appear together. In the classification mode, an object is represented by a point in the feature space and assigned a class according to the class of objects that are nearest to it in the feature space. A feedback process may also be employed to refine the preprocessing and feature/extraction steps in order to provide data subsets that supply pattems that best separate objects into their known classes. Pattern recognition methods that involve iterative identification of feature subsets are sometimes called "wrapper" methods. Feature subsets that provide improved diagnostic patterns are retained and used to generate new feature subsets, which are then tested for improved diagnostic ability. This process is repeated until an optimal subset of features is identified. If, however, the feedback process is not employed, the technique is known as a "filter" method, and feature selection is performed independently of the learning algorithm. Examples of classifier methods that may be used for pattern recognition during learning include decision tree methods, statistical methods, neural networks and evolutionary algorithms. These same methods may also be used to extract and select appropriate subsets of features. Combinations of these methods are also possible. For example, one type of algorithm may be used for feature extraction/selection and a second algorithm used as the classifier, in either a filter or wrapper method. Decision tree methods classify objects using a series of tests based on data features. For example, in the classification and regression tree (CART) method, the algorithm considers all the possible tests (e.g. presence, absence or range of values of each feature) that can split the objects of the data set into two subsets. The test that provides the "best" separation of the objects into known classes (or subsets of similar pattern) is then selected. The remaining tests are then explored to find the test that best separates each of these subsets into two smaller subsets with increasing homogeneity of class membership, and the process is repeated until each subset hopefully contains objects of only one class. The resulting scheme may be represented by a tree-like structure (dendrogram) with branches corresponding to each test that was used to partition the data into separate classes. Unknown objects are classified by applying the series of feature tests used to partition the training data, and assigned the class membership of the subset into which it falls. In statistical pattern recognition, each object in a data set is represented as a vector in multidimensional space where the space is defined by the features or combinations of the features used to describe the objects. Statistical decision concepts are then used to establish boundaries that partition the feature space into regions characteristic of different classes of objects. Examples of statistical pattern recognition methods include principal component analysis (PCA) and discriminant analysis. Artificial neural networks mimic the pattern-finding capacity of the human brain and may be viewed as a parallel computing system consisting of a large number of simple processors with many interconnections. Neural networks relate particular input patterns to a classification through these interconnections. A principal difference between neural networks and other approaches to pattern recognition are that neural networks have the ability to learn complex non-linear relationships between data and the classes of objects they represent. Neural networks also have the characteristic of adapting themselves to the data. Examples of neural networks include the Self-Organizing Map (SOM) and multilayer feed-forward neural networks. Evolutionary algorithms, such as genetic algorithms, are optimization techniques that mimic the processes of natural selection, and in the case of genetic algorithms, genetic mutation and recombination. In genetic algorithms, individuals represent candidate solutions to the optimization problem being solved. For example, in the feature subset selection problem, each individual would represent a feature subset. In each "generation," the most "fit" individuals (feature subsets that are best for classifying objects) are retained and undergo mutation and crossover (of the features within the subsets) to yield the next generation. Successive generations undergo mutation and crossover until a satisfactory solution is found. "Cluster analysis" is a broad term used to describe methods that group, or cluster, objects into subsets based on their data patterns, often in an unsupervised manner. Since each object may be represented as a point in multidimensional feature space with coordinates corresponding to its values for the features, a cluster analysis will group objects based on their similar positions in the feature space. Generally, clustering techniques employ some measure of similarity to decide cluster membership of objects. Common similarity measures include distances between objects in the feature space (e.g. Euclidian or Mahalanobis distances) and affinity of the objects in the feature space (e.g. density or neighborhood). For example, clusters of objects in the feature space may be selected in a manner that minimizes the distances between objects in the same cluster and maximizes the distances between objects in different clusters. The -means algorithm and the Self-Organizing Map are examples of cluster analyses. Particular examples of pattern recognition techniques that may be used in the disclosed methods including the methods described in U.S. Patent Application Publication No. 20030004402, which is incoφorated by reference herein, and the methods described in PCT Publication WO 02/42733, which is also incoφorated by reference herein. Additional methods for pattern recognition are described in U.S. Patent Publication Nos. 20020193950 and 20030134304. Examples of statistical pattern recognition software that is commercially available include

SPSS (SPSS Inc., Chicago, IL), JMP (SAS Inc., Cary NC), Stata (Stata Inc., College Station, TX) and Cluster. Artificial neural network software is available from, among other sources, Neurodimension, Inc., Gainesville, FL (Neurosolutions) and The Mathworks, Inc., Natick, MA (MATLAB Neural Network Toolbox).

Example 11 - Diagnostic Method Based on Identity of BAM-associated Biomarkers This example describes a particular embodiment of a method for diagnosing the biological state of a subject. In this method the molecules bound to one or more biomarker attractant molecules are analyzed for the presence of previously identified biomarkers. With reference to FIG. 12, biomarkers shed, for example, by a tumor cell pass into the blood stream where they encounter and bind to a circulating earner protein such as albumin. Over time, such biomarkers accumulate on the carrier protein along with other molecules. After isolating the carrier protein from the blood (typically from serum), laser desoφtion of the bound molecules may be used to introduce them into a mass spectrometer which provides an ion spectrum readout of the molecules bound to the carrier protein. Based on the masses observed in the mass spectrum, it is possible to make a direct identification of the ions comprising the mass spectral pattern by reference to a database of previously identified biomarkers. In some cases, ions appearing at particular masses are fragments of known proteins that are correlated with a biological state of a particular anatomic of physiologic location. Thus, identification of particular ions in a mass spectrum of the molecules bound to a particular BAM is correlated with the presence of a particular biological state. For example, identification of a particular fragment from a particular protein (such as a protein known to be associated with growth or metastasis of a tumor or a particular type of tumor) may be used to make a specific diagnosis. Similarly, BAM associated fragments of proteins associated with hypoxia or apoptosis of the heart muscle may be detected to provide a diagnosis of myocardial infarction (heart attack). It should be understood that the foregoing relates only to particular embodiments and that numerous modifications or alterations may be made without departing from the true scope and spirit of the invention as defined by the following claims.

Claims

We Claim: 1. A method for identifying a biomarker in a biological fluid, comprising: separating a biomarker attractant molecule from samples of the biological fluid obtained from one or more subjects in each of two or more different populations of subjects, wherein the different populations are populations exhibiting two or more different biological states; detecting a biomarker that is differentially adhered to the biomarker atfractant molecule between the different populations and that discriminates between the two or more biological states of the different populations. 2. The method of claim 1, wherein separating the biomarker attractant molecule from the biological fluid comprises separating the biomarker attractant molecule from the fluid using a capture agent that specifically or non-specifically binds the biomarker atfractant molecule.

3. The method of claim 2, further comprising forming a covalent bond between the biomarker atfractant molecule and the capture agent.

4. The method of claim 1, wherein separating the biomarker atfractant molecule comprises separating the biomarker attractant molecule from the biological fluid by size exclusion chromatography.

5. The method of claim 1, wherein the biomarker attractant molecule has a half-life in the biological fluid that is longer than the half-life of biomarker and amplifies concentration of the biomarker in the biological fluid. 6. The method of claim 1, wherein the biomarker attractant molecule has a half-life in the biological fluid of greater than 1 day.

7. The method of claim 1, wherein the biomarker attractant molecule has a molecular weight greater than about 25 kDa.

8. The method of claim 6, wherein the biomarker attractant molecule has a molecular weight of greater than about 40 kDa.

9. The method of claim 1, wherein the biomarker attractant molecule is selected from the group consisting of albumin, transferrin, fibrinogen, alpha-2-macroglobulin, an immunoglobulin, complement, haptoglobulin, a lipoprotein, prealbumin, alpha 1 acid glycoprotein, fibronectin, and ceruloplasmin, and fragments, chemical modifications and combinations thereof.

10. The method of claim 1, wherein the biomarker atfractant molecule is not native to a mammalian body and is introduced into the biological fluid of the subjects in the different populations. 11. The method of claim 10, wherein the biomarker attractant molecule is not native to a human body.

12. The method of claim 10, wherein the biomarker atfractant molecule comprises a nanoparticle.

13. The method of claim 12, wherein the nanoparticle comprises silica, a ceramic, a magnetic metal oxide, or a biodegradable material.

14. The method of claim 10, wherein the biomarker attractant molecule comprises a dendrimer.

15. The method of claim 12, wherein the nanoparticle has at least one region on its surface that is substantially hydrophobic and at least one region on its surface that is substantially hydrophilic.

16. The method of claim 12, wherein the biomarker attractant molecule comprises a protein conjugated to the nanoparticle.

17. The method of claim 16, wherein the protein is selected from the group consisting of albumin, transferrin, fibrinogen, alpha 2 macroglobulin, an immunoglobulin, complement, haptoglobulin, a lipoprotein, prealbumin, alpha 1 acid glycoprotein, fibronectin, and ceruloplasmin, and fragments, chemical modifications and combinations thereof.

18. The method of claim 1, wherein the biological fluid comprises blood, plasma, serum, lymph fluid, synovial fluid, cerebrospinal fluid, breast milk, nipple aspirants, sweat, tears, saliva, mucous, pre-ejaculate, semen, vaginal fluid, a cell culture medium, or interstitial fluid, or a fraction thereof, a combination thereof, or a fluid derived therefrom.

19. The method of claim 1 , wherein detecting the biomarker further comprises dissociating at least a portion of molecules adhered to the biomarker attractant molecule from the separated biomarker attractant molecule.

20. The method of claim 19, wherein dissociating comprises eluting by contacting the biomarker atfractant molecule with a denaturant to elute bound molecules from the biomarker attractant molecule. 21. The method of claim 20, wherein contacting the biomarker adhered to the biomarker attractant molecule with a denaturant comprises a gradient elution with increasingly stronger denaturants and fractionation of the molecules eluted from the biomarker attractant molecule. 22. The method of claim 19, further comprising separating the dissociated molecules from the biomarker attractant molecule by size exclusion chromatography.

23. The method of claim 19, wherein dissociating comprises using a laser to dissociate molecules adhered to the separated biomarker atfractant molecule.

24. The method of claim 1, wherein the two or more biological states of the different populations comprise a presence of a disease and an absence of the disease.

25. The method of claim 24, wherein the disease is a cancer.

26. The method of claim 1, wherein the two or more biological states of the different populations comprise two or more different stages of a disease.

27. The method of claim 1, wherein the two or more biological states comprise exposure to a drug and non-exposure to the drug, exposure to a toxin and non-exposure to the toxin, or before and after a change in physiological, hormonal or developmental state of the subjects in the different populations.

28. The method of claim 1, wherein detecting comprises mass specfrometric detection of molecules adhered to the biomarker attractant molecule.

29. The method of claim 28, wherein detecting further comprises pattern recognition of mass spectrometric patterns that differentiate molecules adhered to the biomarker attractant molecule between the different populations.

30. The method of claim 28, wherein mass specfrometric detection comprises MALDI-

TOF, SELDI-TOF, FT-ICR, or ESI-MS.

31. The method of claim 1 , wherein detecting comprises protein array analysis of molecules adhered to the biomarker atfractant molecule.

32. The method of claim 30, further comprising analyzing a pattern obtained by protein aπay analysis using pattern recognition to detect a pattern of molecules adhered to the biomarker atfractant molecule that differentiates the two or more populations.

33. The method of claim 1, wherein detecting comprises immunochemical analysis of molecules adhered to the biomarker atfractant molecule.

34. The method of claim 1, wherein the biomarker comprises a protein, a lipid, a carbohydrate or a nucleic acid, or a fragment or combination thereof.

35. The method of claim 1, wherein the biomarker comprises two or more different molecules adhered to the biomarker attractant molecule.

36. The method of claim 35, wherein the biomarker comprises two or more different proteins or protein fragments. 37. The method of claim 10, wherein the biomarker attractant molecule comprises an affinity tag that can be used to separate the biomarker atfractant molecule from the biological fluid after it is introduced into the biological fluid.

38. The method of claim 1, wherein the biomarker adheres to the biomarker attractant molecule non-covalently.

39. The method of claim 1, wherein the biomarker adheres to the biomarker attractant molecule with a binding constant of less than 10^s M^"'. 40. The method of claim 1, wherein the biomarker attractant molecule adheres and thereby concentrates the biomarker because the concentration of the biomarker atfractant molecule in the biological fluid is at least 10 times greater than the concenfration of the biomarker in the biological fluid. 41. The method of claim 40, wherein the biomarker attractant molecule further concentrates the biomarker because the biomarker attractant molecule has a half-life of greater than 1 day in the biological fluid.

42. The method of claim 1, wherein the biomarker atfractant molecule circulates in the biological fluid and collects the biomarker as it is released into the biological fluid, and wherein the biomarker adheres specifically or non-specifically to one or more domains on the surface of the biomarker attractant molecule, the one or more domains on the surface of the biomarker further adhering additional molecules found in the biological fluid.

43. The method of claim 1, wherein the biological fluid comprises serum or plasma and the biomarker, which would otherwise be filtered out of the serum or plasma at a rate equal to or greater than a rate of its introduction in the serum or plasma, is retained and concentrated over time in the serum or plasma because the biomarker atfractant molecule to which it adheres is large enough and in sufficient excess concentration relative to the biomarker to prevent the biomarker from being filtered from the serum or plasma by normal kidney/glomerular function.

44. The method of claim 1, wherein the biomarker has a molecular weight of less than about 12 kDa.

45. The method of claim 1, wherein the biomarker attractant molecule comprises two or more different biomarker atfractant molecules. 46. The method of claim 45, wherein the two or more different biomarker attractant molecules comprise all molecules naturally present in the biological fluid having a molecular weight of greater than about 25 kDa.

47. The method of claim 46, wherein the two or more different biomarker attractant molecules comprise all molecules naturally present in the biological fluid having a molecular weight of greater than about 40 kDa.

48. The method of claim 45, wherein the biomarker comprises two or more different molecules differentially adhered between the two or more different biomarker atfractant molecules.

49. A biomarker identified by the method of claim 1.

50. A method of detecting a biological state of a subject, comprising: separating a biomarker attractant molecule from a biological fluid obtained from the subject, wherein a biomarker of the biological state adheres to the biomarker atfractant molecule; and detecting the biomarker adhered to the biomarker atfractant molecule, wherein the presence of the biomarker adhered to the biomarker attractant molecule indicates that the subject exhibits the biological state.

51. The method of claim 50, wherein separating the biomarker attractant molecule from the biological fluid comprises separating the biomarker attractant molecule from the fluid using a capture agent that specifically or non-specifically binds the biomarker atfractant molecule. 52. The method of claim 51 , further comprising forming a covalent bond between the biomarker atfractant molecule and the capture agent.

53. The method of claim 50, wherein separating the biomarker attractant molecule comprises separating the biomarker atfractant molecule from the biological fluid by size exclusion chromatography.

54. The method of claim 50, wherein the biomarker comprises two or more different molecules adhered to the biomarker attractant molecule. 55. The method of claim 54, wherein the two or more different molecules adhered to the biomarker attractant molecule comprise two or more proteins, or protein fragments.

56. The method of claim 50, wherein the biomarker attractant molecule comprises two or more different biomarker attractant molecules.

57. The method of claim 56, wherein the biomarker comprises two or more different molecules adhered to at least one of the two or more different biomarker atfractant molecules.

58. The method of claim 56, wherein the two or more different biomarker attractant molecules comprise all molecules naturally present in the biological fluid having a molecular weight of greater than about 25 kDa.

59. The method of claim 58, wherein the two or more different biomarker atfractant molecules comprise all molecules naturally present in the biological fluid having a molecular weight of greater than about 40 kDa.

60. The method of claim 50, wherein the biomarker is non-covalently adhered to the biomarker atfractant molecule. 61. The method of claim 50, wherein detecting the biomarker adhered to the biomarker attractant molecule further comprises dissociating the biomarker from the separated biomarker attractant molecule.

62. The method of claim 50, wherein detecting the biomarker further comprises dissociating at least a portion of molecules adhered to the biomarker attractant molecule from the separated biomarker attractant molecule. 63. The method of claim 62, wherein dissociating comprises eluting by contacting the biomarker attractant molecule with a denaturant to elute bound molecules from the biomarker atfractant molecule.

64. The method of claim 63, wherein contacting the biomarker adhered to the biomarker attractant molecule with a denaturant comprises a gradient elution with increasingly stronger denaturants and fractionation of the molecules eluted from the biomarker attractant molecule, wherein the biomarker elutes in a particular fraction.

65. The method of claim 63, wherein the denaturant does not comprise a reducing agent or a nucleophilic compound comprising at least two nucleophilic groups, wherein at least one of the at least two nucleophilic groups in the nucleophilic compound is an amine group.

66. The method of claim 62, further comprising separating the dissociated molecules from the biomarker atfractant molecule by size exclusion chromatography.

67. The method of claim 62, wherein dissociating comprises using a laser to dissociate molecules adhered to the separated biomarker attractant molecule.

68. The method of claim 50, wherein the biological state comprises the presence or absence of a disease.

69. The method of claim 68, wherein the disease is a cancer.

70. The method of claim 50, wherein the biological state is a stage of a disease.

71. The method of claim 50, wherein the biological state is exposure to a drug or non- exposure to the drug.

72. The method of claim 50, wherein detecting comprises mass specfrometric detection of molecules adhered to the biomarker attractant molecule.

73. The method of claim 72, wherein the biomarker is a mass specfrometric pattern and detecting further comprises comparing a mass specfrometric pattem of molecules adhered to the biomarker attractant molecule to a known mass specfrometric pattern of molecules adhered to the biomarker atfractant molecule m other subjects known to exhibit the biological state, wherein the presence of the know pattern indicates the subject exhibits the biological state

74 The method of claim 72, further compnsmg analyzing a mass specfromefric pattern obtamed by mass specfromefric detection to detect a pattern of molecules adhered to the biomarker atfractant molecule that indicates the subject exhibits the biological state

75 The method of claim 72, wherem mass specfromefric detection compnses MALDI- TOF, SELDI-TOF, FT-ICR, or ESI-MS

76 The method of claim 50, wherein detecting comprises protein aπay analysis of molecules adhered to the biomarker attractant molecule

77 The method of claim 76, further comprising analyzing a protem array pattern to detect a pattern of molecules adhered to the biomarker atfractant molecule that indicates the subject exhibits the biological state

78 The method of claim 50, wherem detecting comprises lmmunochemical analysis of molecules adhered to the biomarker attractant molecule

79 The method of claim 50, wherem the biomarker compnses a protem, a lipid, a carbohydrate or a nucleic acid, or a fragment or combination thereof

80 The method of claim 50, wherein the biomarker compnses a molecule havmg a molecular weight of less than about 12 kDa

81 The method of claim 50, wherein the biomarker adheres to the biomarker atfractant molecule with an association constant of less 10^s M ' 82 The method of claim 50, wherein the biomarker attractant molecule has a half-life in the biological fluid that is longer than the half-life of biomarker and amplifies concenfration of the biomarker in the biological fluid

83 The method of claim 50, wherem the biomarker attractant molecule has a half-life in the biological fluid of greater than 1 day

84 The method of claim 50, wherein the biomarker attractant molecule has a molecular weight greater than about 25 kDa

85. The method of claim 84, wherein the biomarker attractant molecule has a molecular weight of greater than about 40 kDa.

86. The method of claim 50, wherein the biomarker atfractant molecule is selected from the group consisting of albumin, ttansfeπin, fibrinogen, alpha 2 macroglobulin, an immunoglobulin, complement, haptoglobulin, a lipoprotein, prealbumin, alpha 1 acid glycoprotein, fibronectin, and ceruloplasmin, and fragments, chemical modifications and combinations thereof.

87. The method of claim 50, wherein the biomarker atfractant molecule is not native to a mammalian body and is introduced into the biological fluid.

88. The method of claim 87, wherein the biomarker attractant molecule is not native to a human body. 89. The method of claim 87, wherein the biomarker attractant molecule comprises a nanoparticle.

90. The method of claim 89 wherein the nanoparticle comprises silica, a ceramic, a magnetic metal oxide, or a biodegradable material.

91. The method of claim 87, wherein the biomarker attractant molecule comprises a dendrimer.

92. The method of claim 89, wherein the nanoparticle has at least one region on its surface that is substantially hydrophobic and at least one region on its surface that is substantially hydrophilic.

93. The method of claim 89, wherein the biomarker atfractant molecule comprises a protein conjugated to the nanoparticle.

94. The method of claim 93, wherein the protein is selected from the group consisting of albumin, transferrin, fibrinogen, alpha 2 macroglobulin, an immunoglobulin, complement, haptoglobulin, a lipoprotein, prealbumin, alpha 1 acid glycoprotein, fibronectin, and ceruloplasmin, and fragments, chemical modifications and combinations thereof.

95. The method of claim 87, wherein the biomarker attractant molecule comprises an affinity tag that can be used to separate the biomarker atfractant molecule from the biological fluid after it is introduced.

96. The method of claim 50, wherein the biomarker atfractant molecule adheres and thereby concentrates the biomarker because the concentration of the biomarker attractant molecule in the biological fluid is at least 10 times greater than the concenfration of the biomarker in the biological fluid.

97. The method of claim 96, wherein the biomarker atfractant molecule further concentrates the biomarker because the biomarker atfractant molecule has a half-life of greater than 1 day in the biological fluid. 98. The method of claim 50, wherein the biomarker attractant molecule circulates in the biological fluid and collects the biomarker as it is released into the biological fluid, and wherein the biomarker adheres specifically or non-specifically to one or more domains on the surface of the biomarker attractant molecule, the one or more domains on the surface of the biomarker further adhering additional molecules found in the biological fluid.

99. The method of claim 50, wherein the biological fluid comprises serum or plasma and the biomarker, which would otherwise be filtered out of the serum or plasma at a rate equal to or greater than a rate of its introduction in the serum or plasma, is retained and concentrated over time in the serum or plasma because the biomarker attractant molecule to which it adheres is large enough and in sufficient excess concenfration relative to the biomarker to prevent the biomarker from being filtered from the serum or plasma by normal kidney/glomerular function.

100. The method of claim 50, wherein biological state comprises a pathological condition.

101. The method of claim 100, wherein the pathological condition comprises a disease or condition of a particular organ and the biomarker is a protein or a fragment thereof that is known to be produced in the particular organ as a result of the disease or condition. 102. The method of claim 100, wherein the particular organ is the heart.

103. The method of claim 100, wherein the pathological condition comprises a tumor.

104. The method of claim 103, wherein the tumor comprises a tumor of a particular histological type.

105. The method of claim 103, wherein the tumor comprises an ovarian tumor.

106. The method of claim 100, wherein the pathological condition is a toxin exposure, and the toxin exposure comprises a drug.

107. The method of claim 50, wherein the biological fluid comprises blood, plasma, serum, lymph fluid, synovial fluid, cerebrospinal fluid, breast milk, nipple aspirants, sweat, tears, saliva, mucous, pre-ejaculate, semen, vaginal fluid, a cell culture medium, or interstitial fluid, or a fraction thereof, a combination thereof, or a fluid derived therefrom.

108. The method of claim 100, wherein the pathological condition is absence of ovarian cancer, the biomarker attractant molecule is albumin, and detecting at least an identifiable fragment of each of two or more proteins in Table 5 that are were identified only in no cancer population indicates the subject exhibits an absence of ovarian cancer.

109. The method of claim 100, wherein the pathological condition is ovarian cancer, the biomarker attractant molecule is albumin, and detecting at least an identifiable fragment of each of two or more proteins in Table 5 that were identified only in either or both Stage I and Stage III populations indicates the subject exhibits ovarian cancer.

110. The method of claim 100, wherein the pathological condition is Stage I ovarian cancer, the biomarker attractant molecule is albumin, and detecting at least an identifiable fragment of each of two or more proteins in Table 5 that were identified only in Stage I population indicates the subject exhibits Stage I ovarian cancer.

111. The method of claim 100, wherein the pathological condition is Stage III ovarian cancer, the biomarker attractant molecule is albumin, and detecting at least an identifiable fragment of each of two or more proteins in Table 5 that were identified only in Stage III population indicates the subject exhibits Stage I ovarian cancer.

112. The method of claim 100, wherein the pathological condition is prostate cancer, the biomarker attractant molecule is all serum proteins with a MW >25 kDa, and detecting at least an identifiable fragment of each of two or more proteins in Table 6 that were found only in prostate cancer population indicates the subject exhibits prostate cancer.

113. The method of claim 100, wherein the pathological condition is the absence of prostate cancer, the biomarker atfractant molecule is all serum proteins with a MW >25 kDa, and detecting at least an identifiable fragment of each of two or more proteins in Table 6 that were found only in no prostate cancer population indicates the subject exhibits the absence of prostate cancer.

114. The method of claim 100, wherein the pathological condition is lung cancer, the biomarker atfractant molecule is albumin, and detecting at least an identifiable fragment of each of two or more proteins in Table 8 that were found only in one or both of adenocarcinoma population and squamous cell carcinoma population indicates the subject exhibits lung cancer.

115. The method of claim 100, wherein the pathological condition is lung adenocarcinoma, the biomarker attractant molecule is albumin, and detecting at least an identifiable fragment of each of two or more proteins in Table 8 that were found only in adenocarcinoma population indicates the subject exhibits lung adenocarcinoma.

116. The method of claim 100, wherein the pathological condition is squamous cell lung carcinoma, the biomarker attractant molecule is albumin, and detecting at least an identifiable fragment of each of two or more proteins in Table 8 that were found only in squamous cell carcinoma population indicates the subject exhibits squamous cell lung carcinoma.

117. The method of claim 100, wherein the pathological condition is absence of lung cancer, the biomarker attractant molecule is albumin, and detecting at least an identifiable fragment of each of two or more proteins in Table 8 that were found only in no cancer population indicates the subject exhibits an absence of lung cancer.