Biomarker panel detection of HDL-related protein

Abstract

To provide a method, a system, and a composition for detecting one or several types of an HDL-related protein (for example, ApoC3; ApoC3 and ApoA1; ApoC3 and SAA1 / 2; or proteins of biomarker panels 1-30) from a subject suffering from a cardiovascular disease (CVD) or another HDL-related disease or from a sample derived from a subject suspected of having the disease.SOLUTION: In a certain embodiment, the method, system and composition are used to determine an estimated risk of CVD (or another disease) in a subject and / or an approximate cholesterol efflux capacity (CEC) of a sample.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to U.S. Provisional Application No. 62 / 365,175, filed July 21, 2016, which is incorporated by reference in its entirety.

[0002] FIELD OF THEINVENTION Provided herein are methods, systems, and compositions for detecting one or more HDL-associated proteins (e.g., ApoC3; ApoC3 and ApoA1; ApoC3 and SAA1 / 2; or proteins of biomarker panels 1-30) from a sample from a subject suffering from or suspected of suffering from cardiovascular disease (CVD) or other HDL-associated disease. In certain embodiments, such methods, systems, and compositions are used to determine an estimated risk of CVD (or other disease) in a subject, and / or an estimated cholesterol efflux capacity (CEC) of the sample. In certain embodiments, the systems and compositions are comprised of a sample from a subject suffering from or suspected of suffering from CVD, and an HDL-associated binding agent or mass spectrometry standard. [Background technology]

[0003] Despite recent advances in both our understanding of the pathophysiology of cardiovascular disease and imaging techniques for atherosclerotic plaques, accurately determining risk in stable cardiac patients remains challenging. Clinically unidentified high-risk patients who are not undergoing aggressive risk factor modification and who experience significant adverse cardiac events are a significant problem. Similarly, more accurate identification of low-risk subjects is needed to refocus finite healthcare resources on those most likely to benefit. Current clinical risk assessment tools relate mostly to algorithms developed from epidemiologically based studies of untreated primary prevention populations, with limited application in high-risk and medicated cardiac outpatient settings. Although interest is high, efforts to incorporate more comprehensive array-based phenotyping techniques (e.g., genomic, proteomic, metabolomic, and expression arrays) to improve cardiac risk stratification are still in their infancy, and no efficient and robust platform has yet been developed that can keep up with the high speeds required for clinical practice. Summary of the Invention

[0004] Provided herein are methods, systems, and compositions for detecting one or more HDL-associated proteins (e.g., ApoC3; ApoC3 and ApoA1; ApoC3 and SAA1 / 2; or proteins of biomarker panels 1-30) from a sample from a subject suffering from or suspected of suffering from cardiovascular disease (CVD) or other HDL-associated disease. In certain embodiments, such methods, systems, and compositions are used to determine an estimated risk of CVD (or other disease) in a subject, and / or an estimated cholesterol efflux capacity (CEC) of the sample. In certain embodiments, the systems and compositions are comprised of a sample from a subject suffering from or suspected of suffering from CVD, and an HDL-associated binding agent or mass spectrometry standard.

[0005] In some embodiments, the method provided herein comprises detecting the level of at least one HDL-associated protein in a sample from a subject, wherein the at least one HDL-associated protein comprises ApoC3, and the subject is suffering from or suspected of suffering from cardiovascular disease. In certain embodiments, the sample is a purified high density lipoprotein sample. In other embodiments, the sample is selected from the group consisting of a serum sample, a plasma sample, and a blood sample. In additional embodiments, the method further comprises normalizing the level of the at least one HDL-associated protein to the estimated level of total HDL particles, or ApoA1, or HDL cholesterol to generate a normalized value of the at least one HDL protein.

[0006] In certain embodiments, determining the approximate level of total HDL particles, or ApoA1, or HDL cholesterol in sample comprises determining the level of an internal standard added to sample, where the internal standard comprises labeled HDL protein.In some embodiments, detecting the level of at least one HDL protein comprises adding a reagent to sample, where the reagent digests HDL protein in sample (e.g., Lyc-C, pepsin, trypsin, etc.).In other embodiments, HDL-associated protein is detected without prior digestion (e.g., detect intact HDL-associated protein).

[0007] In other embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay for detecting ApoC3 or ApoC3 fragments on at least a portion of the sample. In certain embodiments, detecting the level of at least one HDL-associated protein further comprises adding a labeled or unlabeled ApoC3 protein or protein fragment standard to the sample and detecting the ApoC3 standard. In further embodiments, the ApoC3 standard comprises or consists of the amino acid sequence shown in SEQ ID NO: 11. In certain embodiments, the assay is a mass spectrometry assay or an immunoassay.

[0008] In certain embodiments, the at least one HDL-associated protein further comprises serum amyloid A1 or 2 (SAA1 and / or SAA2, referred to herein as "SAA1 / 2"). In further embodiments, detecting the level of at least one HDL protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, and ii) SAA1 / 2 or an SAA1 / 2 fragment. In additional embodiments, detecting the level of at least one HDL-associated protein further comprises adding a labeled or unlabeled SAA1 / 2 protein or protein fragment standard to the sample and detecting the SAA1 / 2 standard. In certain embodiments, the SAA1 / 2 standard comprises or consists of the amino acid sequence set forth in SEQ ID NO:35.

[0009] In further embodiments, the at least one HDL-associated protein further comprises apolipoprotein 1 (ApoA1). In some embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, and ii) ApoA1 or an ApoA1 fragment. In other embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, ii) SAA1 / 2 or an SAA1 / 2 fragment, and iii) ApoA1 or an ApoA1 fragment. In certain embodiments, detecting the level of at least one HDL-associated protein further comprises adding a labeled or unlabeled ApoA1 protein or protein fragment standard to the sample and detecting the ApoA1 standard. In other embodiments, the ApoA1 standard comprises or consists of the amino acid sequence set forth in SEQ ID NO: 1 or 2.

[0010] In further embodiments, the at least one HDL-associated protein further comprises apolipoprotein C1 (ApoC1). In some embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, and ii) ApoC1 or an ApoC1 fragment. In further embodiments, the at least one HDL-associated protein further comprises apolipoprotein C1 (ApoC2). In some embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, and ii) ApoC1 or an ApoC2 fragment. In further embodiments, the at least one HDL-associated protein further comprises apolipoprotein C4 (ApoC4). In some embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, and ii) ApoC1 or an ApoC4 fragment. In certain embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, ii) ApoA1 or an ApoA1 fragment, and iii) ApoC1 or an ApoC1 fragment. In certain embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, ii) ApoA1 or an ApoA1 fragment, and iii) ApoC2 or an ApoC2 fragment.

[0011] In some embodiments, the at least one HDL-associated protein further comprises apolipoprotein L1 (ApoL1) and phospholipid transfer protein (PLTP). In other embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or ApoC3 fragments, ii) SAA1 / 2 or SAA1 / 2 fragments, iii) ApoL1 or ApoL1 fragments, and iv) PLTP or PLTP fragments. In other embodiments, detecting the level of at least one HDL-associated protein further comprises adding a labeled or unlabeled ApoL1 protein or protein fragment standard and / or a labeled or unlabeled PLTP protein or protein fragment standard to the sample and detecting the ApoL1 standard and / or the PLTP standard. In additional embodiments, the ApoL1 standard comprises or consists of the amino acid sequence set forth in SEQ ID NO: 18 or 19, and the PLTP standard comprises or consists of SEQ ID NO: 31 or 32.

[0012] In other embodiments, the at least one HDL-associated protein further comprises apolipoprotein E (ApoE). In some embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or ApoC3 fragment, ii) SAA1 / 2 or SAA1 / 2 fragment, iii) ApoL1 or ApoL1 fragment, iv) PLTP or PLTP fragment, and v) ApoE or ApoE fragment. In other embodiments, detecting the level of at least one HDL-associated protein further comprises adding a labeled or unlabeled ApoE protein or protein fragment standard to the sample and detecting the ApoE standard. In further embodiments, the ApoE standard comprises or consists of the amino acid sequence set forth in SEQ ID NO: 15 or 16.

[0013] In certain embodiments, the at least one HDL-associated protein further comprises apolipoprotein A1 (ApoA1) and apolipoprotein D (ApoD). In further embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or ApoC3 fragment, ii) SAA1 / 2 or SAA1 / 2 fragment, iii) ApoL1 or ApoL1 fragment, iv) PLTP or PLTP fragment, v) ApoE or ApoE fragment, vi) ApoA1 or ApoA1 fragment, and vii) ApoD or ApoD fragment. In some embodiments, detecting the level of at least one HDL-associated protein further comprises adding a labeled or unlabeled ApoD protein or protein fragment standard to the sample and detecting the ApoD standard. In certain embodiments, the ApoD standard comprises or consists of the amino acid sequence set forth in SEQ ID NO: 13 or 14. In other embodiments, detecting the level of at least one HDL-associated protein further comprises adding a labeled or unlabeled ApoA1 protein or protein fragment standard to the sample and detecting the ApoA1 standard.

[0014] In other embodiments, the at least one HDL-associated protein further comprises apolipoprotein M (ApoM) and phospholipid transfer protein (PLTP). In some embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, ii) SAA1 / 2 or an SAA1 / 2 fragment, iii) ApoM or an ApoM fragment, and iv) PLTP or a PLTP fragment. In further embodiments, detecting the level of at least one HDL-associated protein further comprises adding a labeled or unlabeled ApoM protein or protein fragment standard and / or a labeled or unlabeled PLTP protein or protein fragment standard to the sample and detecting the ApoM and / or PLTP standard. In additional embodiments, the ApoM standard comprises or consists of the amino acid sequence set forth in SEQ ID NO: 20, and the PLTP standard comprises or consists of the amino acid sequence set forth in SEQ ID NO: 31 or 32.

[0015] In additional embodiments, the at least one HDL-associated protein further comprises apolipoprotein C1 (ApoC1). In further embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or ApoC3 fragment, ii) SAA1 / 2 or SAA1 / 2 fragment, and iii) ApoC1 or ApoC1 fragment. In some embodiments, detecting the level of at least one HDL-associated protein further comprises adding a labeled or unlabeled ApoC1 protein or protein fragment standard to the sample and detecting the ApoC1 standard. In certain embodiments, the ApoC1 standard comprises or consists of the amino acid sequence shown in SEQ ID NO: 7 or 8.

[0016] In additional embodiments, the at least one HDL-associated protein further comprises apolipoprotein D (ApoD). In further embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, ii) SAA1 / 2 or an SAA1 / 2 fragment, and iii) ApoD or an ApoD fragment. In other embodiments, detecting the level of at least one HDL-associated protein further comprises adding a labeled or unlabeled ApoD protein or protein fragment standard to the sample and detecting the ApoD standard. In some embodiments, the ApoD standard comprises or consists of the amino acid sequence set forth in SEQ ID NO: 13 or 14.

[0017] In some embodiments, the at least one HDL-associated protein further comprises or consists of an additional protein selected from the group consisting of at least CLU, ApoE, CETP, PON1, ApoC1, ApoA2, ApoC2, ApoM, PLTP, and ApoL1. In other embodiments, detecting the level of the at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, ii) SAA1 / 2 or an SAA1 / 2 fragment, and iii) at least one additional protein or fragment thereof. In certain embodiments, detecting the level of the at least one HDL-associated protein further comprises adding a labeled or unlabeled protein or protein fragment standard corresponding to the at least one additional protein and detecting the corresponding standard. In further embodiments, the corresponding standard comprises or consists of the amino acid sequence set forth in SEQ ID NOs: 23, 24, 15, 16, 21, 22, 33, 34, 7, 8, 3, 4, 9, 10, 20, 31, 32, 18, and 19.

[0018] In certain embodiments, the at least one HDL-associated protein further comprises ApoC1 and ApoC2. In other embodiments, detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or ApoC3 fragments, ii) ApoC1 or ApoC1 fragments, and iii) ApoC2 or ApoC2 fragments. In additional embodiments, detecting the level of at least one HDL-associated protein further comprises adding a labeled or unlabeled ApoC1 protein or protein fragment standard to the sample, and / or adding a labeled or unlabeled ApoC2 protein or protein fragment standard to the sample, and detecting the ApoC1 and / or ApoC2 standard. In further embodiments, the ApoC1 standard comprises or consists of the amino acid sequence shown in SEQ ID NO: 7 or 8, and the ApoC2 standard comprises or consists of the amino acid sequence shown in SEQ ID NO: 9 or 10. In a further embodiment, the at least one HDL-associated protein further comprises apolipoprotein C1 (ApoC1) and at least one additional protein selected from the group consisting of ApoM, ApoA1, ApoC2, ApoC4, CLU, SAA4, ApoL1, HP, C3, and PLTP. In a further embodiment, detecting the level of the at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, ii) SAA1 / 2 or an SAA1 / 2 fragment, iii) ApoC1 or an ApoC1 fragment, and iv) at least one additional protein or fragment thereof. In a further embodiment, detecting the level of the at least one HDL-associated protein further comprises adding a labeled or unlabeled ApoC1 protein or protein fragment standard and / or a labeled or unlabeled additional protein or protein fragment standard, and detecting the ApoC1 standard and / or the additional protein standard.In other embodiments, the ApoC1 standard comprises or consists of the amino acid sequence set forth in SEQ ID NO: 7 or 8, and the additional protein standard comprises or consists of the amino acid sequence set forth in SEQ ID NO: 23, 24, 9, 10, 20, 31, 32, 18, 19, 1, 2, 12, 36, 37, 27, 28, 25, or 26.

[0019] In further embodiments, the methods provided herein include detecting in a sample from a subject a level of at least one of the HDL-associated proteins, each of biomarker panels 1-30. Biomarker panels 1-26 are set forth in Table 2, and biomarker panels 27-30 are set forth in Table 54. In certain embodiments, the subject is a human (e.g., a male or female human). In certain embodiments, the human has been diagnosed with cardiovascular disease.

[0020] In some embodiments, the systems and compositions provided herein comprise: a) a sample from a subject suffering from or suspected of suffering from cardiovascular disease; and b) a first component, the first component comprising i) an apolipoprotein C3 (ApoC3) binding agent, and / or ii) an ApoC3 mass spectrometry standard.

[0021] In certain embodiments, the system and composition further comprises c) a second component comprising a composition comprising a detectable labeled HDL protein that can serve as a calibrator for normalizing the signal in the sample to the estimated level of total HDL particles, ApoA1, or HDL cholesterol present in the sample. In other embodiments, the detectable labeled HDL protein comprises a labeled (e.g., isotopically labeled) ApoA1 protein or a fragment thereof. In certain embodiments, the sample is selected from the group consisting of a serum sample, a plasma sample, a blood sample, and a purified high density lipoprotein (HDL) sample. In certain embodiments, the ApoC3 binding agent comprises an anti-ApoC3 antibody or a binding portion thereof, or an anti-ApoC3 nucleic acid or protein aptamer or a binding portion thereof. In certain embodiments, the ApoC3 mass spectrometry standard comprises an isotopically labeled or unlabeled ApoC3 protein or protein fragment. In a further embodiment, the protein fragment comprises or consists of the amino acid sequence shown in SEQ ID NO:11.

[0022] In other embodiments, the system or composition further comprises c) a second component comprising i) a serum amyloid A1 / 2 (SAA1 / 2) binding agent, and / or ii) an SAA1 / 2 mass spectrometry standard. In certain embodiments, the SAA1 / 2 binding agent comprises an anti-SAA1 / 2 antibody or binding portion thereof, or an anti-SAA1 / 2 nucleic acid or protein aptamer or binding portion thereof. In some embodiments, the SAA1 / 2 mass spectrometry standard comprises a labeled (e.g., isotopically labeled) or unlabeled SAA1 / 2 protein or protein fragment. In further embodiments, the protein fragment comprises or consists of the amino acid sequence set forth in SEQ ID NO:35.

[0023] In additional embodiments, the system or composition further comprises d) a third component comprising i) an apolipoprotein A1 (ApoA1) binding agent, and / or ii) an ApoA1 mass spectrometry standard. In further embodiments, the ApoA1 binding agent comprises an anti-ApoA1 antibody or binding portion thereof, or an anti-ApoA1 nucleic acid or protein aptamer or binding portion thereof. In further embodiments, the ApoA1 mass spectrometry standard comprises an isotopically labeled or unlabeled ApoA1 protein or protein fragment. In certain embodiments, the protein fragment comprises or consists of the amino acid sequence set forth in SEQ ID NO: 1 or 2.

[0024] In additional embodiments, the system or composition further comprises d) a third component comprising i) an apolipoprotein L1 (ApoL1) binding agent, and / or ii) an ApoL1 mass spectrometry standard, and e) a fourth component comprising i) a phospholipid transfer protein (PLTP) binding agent, and / or ii) a PLTP mass spectrometry standard. In further embodiments, the ApoL1 binding agent comprises an anti-ApoA1 antibody or binding portion thereof, or an anti-ApoL1 nucleic acid or protein aptamer or binding portion thereof, and the PLTP binding agent comprises an anti-PLTP antibody or binding portion thereof, or an anti-PLTP nucleic acid or protein aptamer or binding portion thereof. In certain embodiments, the ApoL1 mass spectrometry standard comprises an isotopically labeled or unlabeled ApoL1 protein or protein fragment, and / or the PLTP mass spectrometry standard comprises an isotopically labeled or unlabeled PLTP protein or protein fragment. In further embodiments, the ApoL1 protein fragment comprises or consists of the amino acid sequence set forth in SEQ ID NO: 18 or 19, and / or the PLTP protein fragment comprises or consists of the amino acid sequence set forth in SEQ ID NO: 31 or 32.

[0025] In some embodiments, the system or composition further comprises a fifth component comprising i) an apolipoprotein E (ApoE) binding agent, and / or ii) an ApoE mass spectrometry standard. In other embodiments, the ApoE binding agent comprises an anti-ApoE antibody or binding portion thereof, or an anti-ApoE nucleic acid or protein aptamer or binding portion thereof. In additional embodiments, the ApoE mass spectrometry standard comprises an isotopically labeled or unlabeled ApoE protein or protein fragment. In other embodiments, the ApoE protein fragment comprises or consists of the amino acid sequence set forth in SEQ ID NO: 15 or 16.

[0026] In certain embodiments, the system and composition further comprises a sixth component comprising i) an apolipoprotein A1 (ApoA1) binding agent, and / or ii) an ApoA1 mass spectrometry standard. In other embodiments, the system further comprises d) a third component comprising i) an apolipoprotein M (ApoM) binding agent, and / or ii) an ApoM mass spectrometry standard, and e) a fourth component comprising i) a phospholipid transfer protein (PLTP) binding agent, and / or ii) a PLTP mass spectrometry standard. In certain embodiments, the ApoM binding agent comprises an anti-ApoM antibody or binding portion thereof, or an anti-ApoM nucleic acid or protein aptamer or binding portion thereof. In certain embodiments, the ApoM mass spectrometry standard comprises an isotopically labeled or unlabeled ApoM protein or protein fragment. In other embodiments, the ApoM protein fragment comprises or consists of the amino acid sequence set forth in SEQ ID NO:20.

[0027] In additional embodiments, the systems and compositions further comprise d) a third component comprising i) an apolipoprotein C1 (ApoC1) binding agent, and / or ii) an ApoC1 mass spectrometry standard. In additional embodiments, the ApoC1 binding agent comprises an anti-ApoC1 antibody or binding portion thereof, or an anti-ApoC1 nucleic acid or protein aptamer or binding portion thereof. In other embodiments, the ApoC1 mass spectrometry standard comprises a labeled (e.g., isotopically labeled) or unlabeled ApoC1 protein or protein fragment. In certain embodiments, the ApoC1 protein fragment comprises or consists of the amino acid sequence set forth in SEQ ID NO: 7 or 8.

[0028] In other embodiments, the systems and compositions further comprise d) a third component comprising i) an apolipoprotein D (ApoD) binding agent, and / or ii) an ApoD mass spectrometry standard. In some embodiments, the ApoD binding agent comprises an anti-ApoD antibody or binding portion thereof, or an anti-ApoD nucleic acid or protein aptamer or binding portion thereof. In further embodiments, the ApoD mass spectrometry standard comprises a labeled (e.g., isotopically labeled) or unlabeled ApoD protein or protein fragment. In further embodiments, the ApoD protein fragment comprises or consists of the amino acid sequence set forth in SEQ ID NO: 13 or 14.

[0029] In certain embodiments, the systems and compositions further comprise d) a fourth component comprising i) an additional protein binder, and / or ii) an additional protein mass spectrometry standard, where the additional protein is selected from the group consisting of CLU, ApoE, CETP, PON1, ApoC1, ApoA2, ApoC2, ApoM, PLTP, and ApoL1. In further embodiments, the additional protein binder comprises an anti-additional protein antibody or binding portion thereof, or an anti-additional protein nucleic acid or protein aptamer or binding portion thereof. In other embodiments, the additional protein mass spectrometry standard comprises an isotopically labeled or unlabeled additional protein or protein fragment. In other embodiments, the additional protein fragment comprises or consists of the amino acid sequence set forth in SEQ ID NOs: 23, 24, 15, 16, 21, 22, 33, 34, 7, 8, 3, 4, 9, 10, 20, 31, 32, 18, and 19.

[0030] In some embodiments, the systems and compositions further comprise c) a second component comprising i) an ApoC1 binding agent and / or ii) an ApoC1 mass spectrometry standard, and d) a third component comprising i) an ApoC2 binding agent and / or ii) an ApoC2 mass spectrometry standard. In further embodiments, the ApoC1 binding agent comprises an anti-ApoC1 antibody or binding portion thereof, or an anti-ApoC1 nucleic acid or protein aptamer or binding portion thereof, and the ApoC2 binding agent comprises an anti-ApoC2 antibody or binding portion thereof, or an anti-ApoC2 nucleic acid or protein aptamer or binding portion thereof. In other embodiments, the ApoC1 mass spectrometry standard comprises an isotopically labeled or unlabeled ApoC1 protein or protein fragment, and the ApoC2 mass spectrometry standard comprises an isotopically labeled or unlabeled ApoC2 protein or protein fragment.

[0031] In certain embodiments, the systems and compositions further comprise d) a third component comprising i) an apolipoprotein C1 (ApoC1) binding agent, and / or ii) an ApoC1 mass spectrometry standard, and e) a fourth component comprising i) an additional protein binding agent, and / or ii) an additional protein mass spectrometry standard, where the additional protein is selected from the group consisting of ApoM, ApoA1, ApoC2, ApoC4, CLU, SAA4, ApoL1, HP, C3, and PLTP. In other embodiments, the ApoC1 binding agent comprises an anti-ApoC1 antibody or binding portion thereof, or an anti-ApoC1 nucleic acid or protein aptamer or binding portion thereof, and the additional protein binding agent comprises an anti-additional protein antibody or binding portion thereof, or an anti-additional protein nucleic acid or protein aptamer or binding portion thereof. In some embodiments, the ApoC1 mass spectrometric standard comprises an isotopically labeled or unlabeled ApoC1 protein or protein fragment, and the additional protein mass spectrometric standard comprises an isotopically labeled or unlabeled additional protein or protein fragment. In further embodiments, the ApoC1 protein fragment comprises or consists of the amino acid sequence shown in SEQ ID NO: 7 or 8, and the additional protein fragment comprises or consists of the amino acid sequence shown in SEQ ID NO: 23, 24, 9, 10, 20, 31, 32, 18, 19, 1, 2, 12, 36, 37, 27, 28, 25, or 26. In some embodiments, the subject is a human. In certain embodiments, the human has been diagnosed with cardiovascular disease.

[0032] In some embodiments, provided herein are methods of determining an estimated risk of cardiovascular disease (e.g., CAD, atherosclerotic disease, etc.) and / or an estimated reverse cholesterol transport capacity, comprising: a) detecting the level of at least one HDL-associated protein, respectively, of biomarker panels 1-30 (biomarker panels 1-26 are set forth in Table 2, and biomarker panels 27-30 are set forth in Table 54), in a sample from a subject; and b) determining the estimated risk of cardiovascular disease (CVD) in the subject, and / or the estimated cholesterol efflux capacity (CEC) of the sample.

[0033] In some embodiments, the sample is selected from the group consisting of a serum sample, a plasma sample, a blood sample, and a purified high density lipoprotein (HDL) sample. In further embodiments, the determining includes using a first algorithm to generate a cardiovascular disease (CVD) risk score or a cholesterol efflux capacity (CEC) score, where the first algorithm performs an operation including: i) multiplying each HDL-associated protein level by a predetermined coefficient to generate a multiplication value; and ii) summing the multiplication values ​​and adding a panel-specific constant value, thereby generating a CVD risk score or a CEC score. In other embodiments, the method further includes c) generating a report indicative of the CVD risk score and / or the CEC score. In certain embodiments, the report further includes the level of at least one HDL-associated protein. In other embodiments, the determining further includes using a second algorithm to generate a probability of CVD, where the second algorithm applies the CVD risk score to the following formula: CVD probability=1 / (1+exp(-risk score)) or a similar formula. In other embodiments, the method further includes c) generating a report indicative of the probability of CVD (e.g., CAD). In a further embodiment, the report further comprises the level of at least one HDL-associated protein.

[0034] In certain embodiments, the biomarker panel is selected from the group consisting of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, or 30. In some embodiments, the biomarker panel is biomarker panel number 18, consisting of HDL-associated proteins ApoA1, ApoC1, ApoC2, ApoC3, and ApoC4. In other embodiments, the biomarker panel is biomarker panel number 19, consisting of HDL-associated proteins ApoA2, ApoC1, ApoC2, ApoC3, ApoD, and SAA1 / 2. In other embodiments, the biomarker panel is biomarker panel number 5, consisting of HDL-associated proteins ApoA1, ApoC2, and ApoC3. In additional embodiments, the biomarker panel is biomarker panel number 4, which is comprised of HDL-associated proteins ApoC1, ApoC3, CLU, PLTP, and SAA4. In other embodiments, the biomarker panel is biomarker panel number 28, which is comprised of HDL-associated proteins ApoA1, ApoA2, ApoC1, ApoC2, ApoC3, ApoC4, ApoD, ApoE, ApoL1, ApoM, C3, CLU, HP, SAA1 / 2, and SAA4. In further embodiments, the biomarker panel is biomarker panel number 30, which is comprised of HDL-associated proteins ApoA1, ApoA2, ApoC3, ApoC4, ApoD, ApoE, ApoL1, ApoM, C3, HP, PLTP, PON1, and SAA1 / 2. In some embodiments, the CVD is coronary artery disease (CAD). In further embodiments, the CEC is global CEC. In further embodiments, the CEC is ABCA1 CEC.

[0035] In some embodiments, the method further comprises normalizing the level of each of the HDL-associated proteins of the at least one biomarker panel after step a) but before step b), where normalizing is normalizing with the estimated level of total HDL particles, or ApoA1, or HDL cholesterol in the sample to generate a normalized value of HDL protein. In additional embodiments, the method further comprises determining the estimated level of total HDL particles, or ApoA1, or HDL cholesterol in the sample. In further embodiments, determining the estimated level of total HDL particles, or ApoA1, or HDL cholesterol in the sample comprises determining the level of an internal standard added to the sample, where the internal standard comprises a labeled HDL protein.

[0036] In some embodiments, detecting the level of each of the at least one panel of HDL-associated proteins comprises performing an assay on at least a portion of the sample to detect each of the HDL-associated proteins or fragments thereof of the panel of biomarkers. In certain embodiments, the assay is a mass spectrometry assay or an immunoassay. In further embodiments, detecting the level of each of the at least one panel of HDL-associated proteins further comprises adding a labeled or unlabeled protein fragment standard to the sample corresponding to at least one of the detected HDL-associated proteins, and detecting the standard. In other embodiments, the subject is a human. In further embodiments, the human has been diagnosed with cardiovascular disease.

[0037] In other embodiments, the systems and compositions provided herein include: a) a sample from a subject suffering from or suspected of suffering from cardiovascular disease; and b) at least one of the following: i) a binding agent for each of the HDL-associated proteins in at least one of biomarker panels 1-30 (biomarker panels 1-26 are set forth in Table 2, and biomarker panels 27-30 are set forth in Table 54); and / or ii) a mass spectrometry standard protein for each of the HDL-associated proteins in at least one of biomarker panels 1-30.

[0038] In certain embodiments, the system further comprises c) a report indicating the raw or normalized level of HDL-associated protein in the sample for at least one panel of biomarkers. In further embodiments, the system further comprises c) a report indicating a CVD risk score and / or cholesterol efflux capacity (CEC) for the subject based on the level of HDL-associated protein in the sample for at least one panel of biomarkers. In other embodiments, the system further comprises c) a report indicating the probabilistic risk of CVD in the subject. In certain embodiments, the subject is a human.

[0039] In some embodiments, provided herein is a method of reporting a CVD risk score or cholesterol efflux capacity score in a subject, comprising: a) obtaining a subject's value for each of at least one of biomarker panels 1-30 (biomarker panels 1-26 are set forth in Table 2, and biomarker panels 27-30 are set forth in Table 54), and b) determining a CVD risk score and / or cholesterol efflux capacity score in the subject by processing the subject value with a processing system, the processing system comprising: i) a computer processor; and ii) a non-transitory computer memory comprising one or more computer programs and databases, the one or more computer programs being stored in a processing system. The program includes a biomarker panel model algorithm, the database including i) a predetermined coefficient for each HDL-associated protein of the at least one biomarker panel, and ii) a panel-specific constant value for the at least one biomarker panel, and the one or more computer programs, in conjunction with the computer processor, are configured to apply the biomarker panel model algorithm to i) multiply the level of each HDL-associated protein of the at least one biomarker panel by a corresponding predetermined coefficient to generate multiplied values, and ii) sum the multiplied values ​​and add the panel-specific constant values, thereby generating a CVD risk score or a cholesterol efflux capacity (CEC) score for the subject.

[0040] In certain embodiments, the method further comprises c) reporting the CVD risk score and / or CEC score for the subject as determined by the processing system. In some embodiments, the CVD risk score is used to determine a probability of CVD risk for the subject. In further embodiments, the method further comprises d) performing one or more of the following actions: i) performing a coronary catheterization on the subject based on the high probability of CVD risk; ii) treating the subject with a cardiovascular disease (CVD) therapeutic (e.g., a statin, an ACE inhibitor, an aldosterone inhibitor, angiotensin II receptor blocker, a beta blocker, a calcium antagonist, a cholesterol-lowering drug, digoxin, a diuretic, potassium, magnesium, a vasodilator, or warfarin) based on the high or moderate probability of CVD risk; iii) administering a CVD therapeutic to the subject based on the high or moderate probability of CVD risk; prescribing a treatment for the subject; iv) performing at least one additional diagnostic test on the subject based on a moderate or high probability of CVD risk; v) hospitalizing and / or recommending hospitalization of the subject based on a high probability of CVD risk; vi) testing a sample from the subject with one or more CVD risk assays other than a biomarker panel based on a moderate or high probability of CVD risk; vii) discharging the subject from a treatment facility based on a low probability of CVD risk; and viii) performing a stress test on the subject based on a moderate or high probability of CVD risk.

[0041] In certain embodiments, the method further comprises d) performing one or more of the following actions: i) communicating the probability of CVD risk in the subject to a user, ii) displaying the probability of CVD risk in the subject, iii) generating a report indicating the probability of CVD risk in the subject, and iv) generating and / or transmitting a report indicating the probability of CVD risk in the subject. In other embodiments, obtaining the subject value comprises receiving the subject value from a laboratory, from the subject, from an analytical testing system, and / or from a handheld or point-of-care testing device. In some embodiments, the processing system further comprises an analytical testing system and / or a handheld or point-of-care testing device.

[0042] In certain embodiments, obtaining the subject value comprises receiving the subject value electronically. In other embodiments, obtaining the subject value comprises testing a sample from the subject with a detection assay. In further embodiments, the detection assay comprises an immunoassay or a mass spectrometry assay. In additional embodiments, the processing system further comprises a graphical user interface, and the method further comprises entering the subject value via the graphical user interface. In certain embodiments, the graphical user interface is part of a device selected from a desktop computer, a notebook computer, a tablet computer, a smartphone, and a point-of-care analytical device.

[0043] In additional embodiments, the processing system further comprises a sample analyzer. In further embodiments, at least a portion of the computer memory is located within the sample analyzer. In further embodiments, the processing system further comprises a laboratory interface system (LIM). In other embodiments, at least a portion of the computer memory is part of the LIM.

[0044] In some embodiments, the processing system further comprises a processing device selected from the group consisting of a desktop computer, a notebook computer, a tablet computer, a smart phone, and a point-of-care analyzer, hi additional embodiments, at least a portion of the computer memory is located within the processing device.

[0045] In other embodiments, the one or more computer programs further comprise a CVD probability algorithm, the one or more computer programs, in conjunction with the computer processor, are further configured to utilize a CVD probability algorithm and iii) apply the CVD risk score to the following formula: CVD probability = 1 / (1 + exp(-CVD risk score). In certain embodiments, the CVD is coronary artery disease (CAD). In other embodiments, the CECs are overall CECs. In some embodiments, the CECs are ABCA1 CECs.

[0046] In additional embodiments, the subject value for each HDL-associated protein is a normalized level (e.g., relative to the estimated level of total HDL particles, or ApoA1, or HDL cholesterol in the sample, or the estimated level of some other protein in the sample, such as HSA). In further embodiments, the at least one biomarker panel is biomarker panel number 18, which is comprised of HDL-associated proteins ApoA1, ApoC1, ApoC2, ApoC3, and ApoC4. In other embodiments, the at least one biomarker panel is biomarker panel number 19, which is comprised of HDL-associated proteins ApoA2, ApoC1, ApoC2, ApoC3, ApoD, and SAA1 / 2. In additional embodiments, the at least one biomarker panel is biomarker panel number 5, which is comprised of HDL-associated proteins ApoA1, ApoC2, and ApoC3. In another embodiment, the at least one biomarker panel is biomarker panel number 4, which is comprised of HDL-associated proteins ApoC1, ApoC3, CLU, PLTP, and SAA4. In a further embodiment, the at least one biomarker panel is biomarker panel number 28, which is comprised of HDL-associated proteins ApoA1, ApoA2, ApoC1, ApoC2, ApoC3, ApoC4, ApoD, ApoE, ApoL1, ApoM, C3, CLU, HP, SAA1 / 2, and SAA4. In another embodiment, the at least one biomarker panel is biomarker panel number 30, which is comprised of HDL-associated proteins ApoA1, ApoA2, ApoC3, ApoC4, ApoD, ApoE, ApoL1, ApoM, C3, HP, PLTP, PON1, and SAA1 / 2. In a further embodiment, the subject is a human. In another embodiment, the human has been diagnosed with cardiovascular disease.

[0047] In further embodiments, a processing system provided herein comprises: a) a computer processor; and b) a non-transitory computer memory comprising one or more computer programs and a database, wherein the one or more computer programs comprise a biomarker panel model algorithm, wherein the database comprises i) a subject value for each HDL-associated protein in at least one of biomarker panels 1-30 (biomarker panels 1-26 are set forth in Table 2, and biomarker panels 27-30 are set forth in Table 54); ii) a panel-specific predetermined coefficient for each HDL-associated protein in at least one of biomarker panels 1-30; and iii) a panel-specific constant value for the at least one biomarker panel, wherein the one or more computer programs, in conjunction with the computer processor, are configured to apply the biomarker panel model algorithm to: i) multiply the subject value for each HDL-associated protein by a corresponding panel-specific predetermined coefficient to generate multiplied values; and ii) sum the multiplied values ​​and add the panel-specific constant value, thereby generating a CVD risk score or a cholesterol efflux capacity (CEC) score for the subject.

[0048] In certain embodiments, the one or more computer programs further comprise a CVD probability algorithm, the one or more computer programs, in conjunction with the computer processor, are further configured to utilize the CVD probability algorithm and iii) apply the CVD risk score to the following formula: CVD probability = 1 / (1 + exp(-CVD risk score). In some embodiments, the CVD is coronary artery disease (CAD). In further embodiments, the CECs are global CECs. In other embodiments, the CECs are ABCA1 CECs.

[0049] In certain embodiments, the target value for each HDL-associated protein is a normalized level for total HDL particles, or ApoA1, or HDL cholesterol in the sample. In other embodiments, the system further comprises c) an HDL-associated protein analysis test system and / or a handheld or point-of-care HDL-associated protein point-of-care testing device. In certain embodiments, the HDL-associated protein analysis test system comprises a mass spectrometer or an optical detector. In other embodiments, the system further comprises c) a graphical user interface for inputting the target value for each HDL-associated protein into the computer memory. In other embodiments, the graphical user interface is part of a device selected from a desktop computer, a notebook computer, a tablet computer, a smartphone, and a point-of-care analysis device.

[0050] In some embodiments, the system further comprises a sample analyzer, and at least a portion of the computer memory is disposed within the sample analyzer. In certain embodiments, the system further comprises at least a portion of a laboratory interface system (LIM). In other embodiments, at least a portion of the computer memory is part of the LIM. In other embodiments, the system further comprises c) a processing device selected from the group consisting of a desktop computer, a notebook computer, a tablet computer, a smartphone, and a point-of-care analyzer. In other embodiments, at least a portion of the computer memory is disposed within the processing device. In additional embodiments, the system further comprises a display component configured to display the patient's CVD risk score or CEC score. In other embodiments, the display component is selected from a computer monitor, a tablet computer screen, a smartphone screen, and a point-of-care analyzer screen.

[0051] In some embodiments, the at least one biomarker panel is biomarker panel number 18, which is comprised of HDL-associated proteins ApoA1, ApoC1, ApoC2, ApoC3, and ApoC4. In further embodiments, the at least one biomarker panel is biomarker panel number 19, which is comprised of HDL-associated proteins ApoA2, ApoC1, ApoC2, ApoC3, ApoD, and SAA1 / 2. In other embodiments, the at least one biomarker panel is biomarker panel number 5, which is comprised of HDL-associated proteins ApoA1, ApoC2, and ApoC3. In further embodiments, the at least one biomarker panel is biomarker panel number 4, which is comprised of HDL-associated proteins ApoC1, ApoC3, CLU, PLTP, and SAA4. In some embodiments, the at least one biomarker panel is biomarker panel number 28, which is comprised of HDL-associated proteins ApoA1, ApoA2, ApoC1, ApoC2, ApoC3, ApoC4, ApoD, ApoE, ApoL1, ApoM, C3, CLU, HP, SAA1 / 2, and SAA4. In certain embodiments, the at least one biomarker panel is biomarker panel number 30, which is comprised of HDL-associated proteins ApoA1, ApoA2, ApoC3, ApoC4, ApoD, ApoE, ApoL1, ApoM, C3, HP, PLTP, PON1, and SAA1 / 2. In other embodiments, the subject is a human. In additional embodiments, the human has been diagnosed with cardiovascular disease.

[0052] In some embodiments, the non-transitory computer memory components provided herein comprise one or more computer programs configured to access a database, wherein the one or more computer programs comprise a biomarker panel model algorithm, wherein the database comprises i) a subject value for each HDL-associated protein in at least one of biomarker panels 1-30 (biomarker panels 1-26 are set forth in Table 2, and biomarker panels 27-30 are set forth in Table 54); ii) a predetermined coefficient for each HDL-associated protein in at least one of biomarker panels 1-30; and iii) a panel-specific constant value for the at least one biomarker panel, wherein the one or more computer programs, in conjunction with a computer processor, are configured to apply the biomarker panel model algorithm to: i) multiply the subject value for each HDL-associated protein by a corresponding predetermined coefficient to generate multiplied values; and ii) sum the multiplied values ​​and add the panel-specific constant value, thereby generating a CVD risk score or a cholesterol efflux capacity (CEC) score for the subject.

[0053] In further embodiments, the assay for HDL-associated proteins is performed using a technique selected from the group consisting of mass spectrometry (MS), chromatography, LC-MS, plasmon resonance, and assays involving the use of polyvinylsulfonic acid (PVS) and polyethylene glycol methyl ether (PEGME). In certain embodiments, detection is performed by injecting the purified HDL sample into an instrument that performs both chromatography and mass spectrometry. In some embodiments, the instrument is a liquid chromatography-mass spectrometry (LC / MS) instrument.

[0054] In certain embodiments, the HDL value is a normalized value. In certain embodiments, the normalization is to total HDL cholesterol, or ApoA1 protein, or HDL particles. Determination of total HDL can be performed by measuring HDL cholesterol. Generally, in practice, a "homogeneous" assay using selected reagents added in a specific order is used to "deplete" the serum sample of LDL cholesterol particles containing lipoprotein ApoB. HDL cholesterol is then chemically determined using a conventional enzyme-linked assay. Measurement of total HDL can also be performed using physical methods to isolate HDL particles, usually ultracentrifugation (see, e.g., Warnick et al., Clinical Chemistry September 2001). vol.47 no.9 1579-1596, incorporated herein by reference). In some embodiments, the measured amount of HDL-associated protein may be compared to the total amount of native ApoA1 in the initial sample to determine a normalized ratio. ApoA1 is the major lipoprotein component of each HDL particle. Measurement of HDL cholesterol, but not ApoA1, has been the mainstay of cardiovascular risk assessment, but this view is changing as measurement of ApoA1 is useful for identifying subclinical atherosclerosis (Florvall et al., Journal of Gerontology: BIOLOGICAL SCIENCES 2006, Vol.61A, No.12, 1262-1266, incorporated herein by reference). Total ApoA1 is usually measured using a widely available immunoassay platform assay.

[0055] In some embodiments, the methods provided herein include detecting the levels of at least three (or at least two) HDL-associated proteins (e.g., selected from those listed in Table 2 and Table 54) in a sample from a subject, where the subject has or is suspected of having cardiovascular disease. In certain embodiments, the sample is a purified high density lipoprotein sample.

[0056] In certain embodiments, provided herein are methods for determining an estimated risk of cardiovascular disease and / or an estimated reverse cholesterol transport capacity, comprising: a) detecting the levels of at least three (or at least two) HDL-associated proteins (e.g., selected from those listed in Table 2 and Table 54) in a sample from a subject; and b) determining the estimated risk of cardiovascular disease (CVD) in the subject and / or the estimated cholesterol efflux capacity (CEC) of the sample.

[0057] In some embodiments, the systems provided herein include: a) a sample from a subject suffering from or suspected of suffering from cardiovascular disease; and b) at least one of the following: i) a binding agent for each of at least three (or at least two) HDL-associated proteins (e.g., selected from those listed in Table 2 and Table 54); and / or ii) a mass spectrometry standard protein for each of the at least three (or at least two) HDL-associated proteins.

[0058] In other embodiments, provided herein is a method of reporting a CVD risk score or cholesterol efflux capacity score in a subject, comprising: a) obtaining subject values ​​for each of at least three (or at least two) HDL-associated proteins (e.g., selected from those listed in Table 2 and Table 54); and b) determining a CVD risk score and / or cholesterol efflux capacity score in the subject by processing the subject values ​​with a processing system, the processing system comprising: i) a computer processor; and ii) a non-transitory computer memory comprising one or more computer programs and a database, the one or more computer programs comprising a biomarker algorithm. The database includes: i) a predetermined coefficient for each of the at least three (or at least two) HDL-associated proteins; and ii) a panel-specific constant value for a combination of the at least three (or at least two) HDL-associated proteins; and the one or more computer programs, in conjunction with the computer processor, are configured to apply a biomarker algorithm to: i) multiply the level of each of the at least three (or at least two) HDL-associated proteins by a corresponding predetermined coefficient to generate a multiplied value; and ii) sum the multiplied values ​​and add a panel-specific constant value, thereby generating a CVD risk score or cholesterol efflux capacity (CEC) score for the subject.

[0059] In some embodiments, the processing system provided herein comprises: a) a computer processor; and b) a non-transitory computer memory comprising one or more computer programs and a database, wherein the one or more computer programs comprise a biomarker algorithm, wherein the database comprises: i) a subject value for each of at least three (or at least two) HDL-associated proteins (e.g., selected from those listed in Table 2 and Table 54); ii) a panel-specific predetermined coefficient for each of the at least three (or at least two) HDL-associated proteins; and iii) a panel-specific constant value for a combination of the at least three HDL-associated proteins, wherein the one or more computer programs, in conjunction with the computer processor, are configured to apply the biomarker algorithm to: i) multiply the subject value for each of the at least three (or at least two) HDL-associated proteins by a corresponding panel-specific predetermined coefficient to generate a multiplied value; and ii) sum the multiplied values ​​and add a panel-specific constant value, thereby generating a CVD risk score or a cholesterol efflux capacity (CEC) score for the subject.

[0060] In certain embodiments, the non-transitory computer memory components provided herein comprise one or more computer programs configured to access a database, the one or more computer programs comprising a biomarker algorithm, the database comprising i) a subject value for each of at least three (or at least two) HDL-associated proteins (e.g., selected from those listed in Table 2 and Table 54), ii) a predetermined coefficient for each HDL-associated protein, and iii) a panel-specific constant value for a combination of the at least three (or at least two) HDL-associated proteins, the one or more computer programs, in conjunction with a computer processor, are configured to apply the biomarker algorithm to: i) multiply the subject value for each HDL-associated protein by a corresponding predetermined coefficient to generate a multiplied value, and ii) sum the multiplied values ​​and add a panel-specific constant value, thereby generating a CVD risk score or a cholesterol efflux capacity (CEC) score for the subject.

[0061] In certain embodiments, the biomarker panels described herein are used to diagnose the risk of an HDL-related disease selected from cardiovascular disease, non-alcoholic steatohepatitis (NASH), lupus erythematosus, irritable bowel syndrome (IBS), chronic kidney disease (CKD), rheumatoid arthritis (RA), and Alzheimer's disease.

[0062] The present invention can also be configured as follows. [1] A method comprising detecting a level of at least one HDL-associated protein in a sample from a subject, wherein the at least one HDL-associated protein comprises ApoC3, and the subject has or is suspected of having cardiovascular disease. [2] The method according to [1], wherein the sample is a purified high density lipoprotein sample. [3] The method according to [1], wherein the sample is selected from the group consisting of a serum sample, a plasma sample, and a blood sample. [4] The method of [1], further comprising normalizing the level of the at least one HDL-associated protein to the estimated level of total HDL particles, or ApoA1, or HDL cholesterol to generate a normalized value of at least one HDL protein. [5] The method of [4], wherein the determining of the estimated level of total HDL particles, or ApoA1, or HDL cholesterol in the sample comprises determining the level of an internal standard added to the sample, wherein the internal standard comprises labeled HDL protein. [6] The method of [1], wherein said detecting the level of at least one HDL protein comprises adding a reagent to said sample, said reagent digesting HDL protein in said sample. [7] The method according to [1], wherein said detecting the level of at least one HDL-associated protein comprises performing an assay on at least a portion of said sample to detect ApoC3 or an ApoC3 fragment. [8] The method according to [7], wherein the assay is a mass spectrometry assay or an immunoassay. [9] The method according to [1], wherein the at least one HDL-associated protein further comprises apolipoprotein A1 (ApoA1).

[10] The method of [9], wherein said detecting the level of at least one HDL protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, and ii) ApoA1 or an ApoA1 fragment.

[11] The method according to [1], wherein the at least one HDL-associated protein further comprises apolipoprotein C1 (ApoC1).

[12] The method according to

[11] , wherein said detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, and ii) ApoC1 or an ApoC1 fragment.

[13] The method according to [1], wherein the at least one HDL-associated protein further comprises apolipoprotein C2 (ApoC2).

[14] The method according to

[13] , wherein said detecting the level of at least one HDL-associated protein comprises performing an assay to detect i) ApoC3 or an ApoC3 fragment, and ii) ApoC2 or an ApoC2 fragment.

[15] A method for detecting a cardiovascular disease comprising: a) a sample derived from a subject suffering from or suspected of suffering from a cardiovascular disease; and b) a first component, wherein the first component is i) apolipoprotein C3 (ApoC3) binding agents, and / or ii) A system or composition comprising an ApoC3 mass spectrometry standard.

[16] c) The system or composition described in

[15] , further comprising a second component comprising a composition containing a detectably labeled HDL protein that can serve as a calibrator for normalizing the signal in the sample to the estimated levels of total HDL particles, ApoA1, or HDL cholesterol present in the sample.

[17] The system or composition described in

[15] , wherein the sample is selected from the group consisting of a serum sample, a plasma sample, a blood sample, and a purified high density lipoprotein (HDL) sample.

[18] The system or composition of

[15] , further comprising a second component comprising: i) an apolipoprotein A1 (ApoA1) binding agent; and / or ii) an ApoA1 mass spectrometry standard.

[19] The system or composition of

[15] , further comprising a second component comprising: i) an apolipoprotein C1 (ApoC1) binding agent; and / or ii) an ApoC1 mass spectrometry standard.

[20] The system or composition of

[15] , further comprising a second component comprising: i) an apolipoprotein C2 (ApoC2) binding agent; and / or ii) an ApoC2 mass spectrometry standard.

[21] A method for determining an estimated risk of cardiovascular disease and / or an estimated reverse cholesterol transport capacity, comprising: a) detecting in a sample from a subject biomarker panel 17, 18, and / or 19, said biomarker panel 17 comprising the proteins ApoC1, ApoC2, and ApoC3; The biomarker panel 18 includes the proteins ApoA1, ApoC1, ApoC2, ApoC3, and ApoC4; The biomarker panel includes the proteins ApoA2, ApoC1, ApoC2, ApoC3, ApoD, and SAA1 / 2. detecting the levels of each of the HDL-associated proteins; b) determining the estimated risk of cardiovascular disease (CVD) in said subject and / or the estimated cholesterol efflux capacity (CEC) of said sample.

[22] The method according to

[21] , wherein the sample is selected from the group consisting of a serum sample, a plasma sample, a blood sample, and a purified high density lipoprotein (HDL) sample.

[23] The method of

[21] , wherein the determining includes generating a cardiovascular disease (CVD) risk score or a cholesterol efflux capacity (CEC) score using a first algorithm, the first algorithm performing an operation including: i) multiplying each HDL-associated protein level by a predetermined coefficient to generate a multiplied value; and ii) summing the multiplied values ​​and adding a panel-specific constant value, thereby generating the CVD risk score or the CEC score.

[24] The method of

[21] , further comprising: c) generating a report indicating the CVD risk score and / or the CEC score. [Brief description of the drawings]

[0063] [Figure 1] Panel 1 shows an exemplary workflow corresponding to the steps described in Example 1 below. [Diagram 2] Isolation of ApoA-I-associated serum fractions - Overview schematic of isolating ApoA-I-associated proteins by incubating serum with His6ApoA-I and targeting the His tag using immobilized metal affinity chromatography (IMAC). [Diagram 3]Overlap of the ApoA-I-associated proteome with the HDL proteome - Venn diagram showing the overlap of proteins identified by ApoA-I affinity purification with HDL prepared by gradient ultracentrifugation within the frame of the HDL Proteome Watch list. [Figure 4] Correlation with cell-based cholesterol efflux assays - (A) Correlation of actual cholesterol efflux capacity (CEC) from cAMP-stimulated J774 macrophages with predicted CEC using a panel of six proteins based on targeted analysis of ApoA-I-associated serum proteins, and with serum concentrations of (B) ApoA-I, (C) HDL, and (D) hsCRP. [Diagram 5] Inverse correlation between predicted CECs and CAD - Distribution of CECs predicted by targeted HDL proteome analysis. Comparison of total control and CAD patient populations (with or without MACE). [Figure 6] Correlation between endogenous ApoA-I observed in targeted MRM measurements of ApoA-I-related proteins and serum ApoA-I levels measured by immunoturbidimetry within a referral clinical laboratory. [Figure 7] Comparison of predicted CEC from targeted HDL proteomic analysis of 30 samples (n=15 control, n=15 CAD) with actual cellular cholesterol efflux measured by a cell-based assay of the same sample set in cAMP-stimulated J774 macrophages. [Figure 8] Results from Example 4 are presented comparing predicted cholesterol efflux values ​​between the cohorts, revealing an overall lower cholesterol efflux in the CAD cohort compared to the control cohort. [Figure 9] 1 shows that the results of application of the outcome optimization model of Example 4 stratify patients on a biomarker score scale that correlates with the risk of being diagnosed with CAD. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0064] definition As used herein, "high density lipoprotein" or "HDL" is a non-covalent complex of amphipathic proteins that circulates in the blood and allows lipids such as cholesterol and triglycerides to be transported in the aqueous bloodstream. HDL is composed of approximately 50% by mass of amphipathic proteins that stabilize a lipid emulsion consisting of a phospholipid monolayer (approximately 25%) with embedded free cholesterol (approximately 4%) and a core of triglycerides (approximately 3%) and cholesterol esters (approximately 12%). Subclasses of HDL include HDL2 and HDL3. HDL2 particles are larger and contain more lipid, whereas HDL3 particles are smaller and contain less lipid. Other subclasses, from largest to smallest, include HDL2b, HDL2a, HDL3a, HDL3b, and HDL3c.

[0065] As used herein, "lipoprotein" refers to a type of protein that can be bound or combined with one or more lipid molecules. In some cases, lipoprotein can be "lipid-poor lipoprotein" that has four or fewer phospholipid molecules bound. As used herein, lipoprotein includes proteins (e.g., apolipoproteins) that are not bound to lipid but can be exchanged into HDL particles.

[0066] As used herein, the phrase "reverse cholesterol transport" refers to a multi-step process that results in the actual movement of cholesterol from peripheral tissues back to the liver via the plasma.

[0067] As used herein, the phrase "cholesterol efflux capacity (CEC)" of a sample refers to the ability of HDL in the sample to promote reverse cholesterol transport by accepting cholesterol from lipid-laden macrophages. Methods for measuring CEC include, but are not limited to, those described in de la Llera-Moya et al., Arterioscler Thromb Vasc Biol. 2010 Apr; 30(4): 796-801; Sankaranarayanan et al., J Lipid Res. 2011 Dec; 52(12): 2332-40; and Khera et al., NEJM, 364; 2, 2011, all of which are incorporated herein by reference in their entirety, including the assays disclosed therein. CEC of ABCA1 specifically refers to the reverse cholesterol transport induced specifically by the ABCA1 transporter. For example, this can be measured by subtracting the difference in CEC between macrophages treated with a reagent (e.g., cAMP-treated) and macrophages that are not treated with a reagent. In this case, macrophages are modified such that in the presence of a treatment agent, ABCA1 expression is immediately increased.

[0068] As used herein, "blood sample" refers to a whole blood sample, or a plasma or serum fraction derived therefrom. In certain embodiments, blood sample refers to a human blood sample, such as a whole blood sample, or a plasma or serum fraction derived therefrom. In some embodiments, blood sample refers to a non-human mammalian ("animal") blood sample, such as a whole blood sample, or a plasma or serum fraction derived therefrom.

[0069] As used herein, the term "whole blood" refers to a blood sample that has not been fractionated and contains both cellular and fluid components.

[0070] As used herein, "plasma" refers to the fluid, non-cellular component of whole blood. Depending on the separation method used, plasma may be completely free of cellular components or may contain varying amounts of platelets and / or small amounts of other cellular components. The term "plasma" is distinguished from "serum," described below, because plasma contains various clotting factors, such as fibrinogen.

[0071] As used herein, the term "serum" refers to whole mammalian serum, such as whole human serum, whole serum from test animals, whole serum from pets, whole serum from livestock, etc. Additionally, as used herein, "serum" refers to plasma from which clotting factors (e.g., fibrinogen) have been removed.

[0072] As used herein, the phrase "purified high density lipoprotein sample" refers to a purified sample obtained by purifying (e.g., by ultracentrifugation or ApoA1 exchange methods described below) a blood sample (e.g., serum, or plasma, or whole blood sample) such that at least 90% (e.g., at least 90%...94%...98%...99%...or at least 99.9%) of the total proteins in the purified sample are HDL lipoproteins. In some embodiments, less than 10% of the total proteins in the purified sample are non-HDL lipoproteins (e.g., less than 10%...5%...1%...0.2%). In certain embodiments, the non-HDL lipoprotein is primarily or entirely serum albumin.

[0073] As used herein, the term "cardiovascular disease" (CVD) or "cardiovascular disorder" is a term used to classify a number of conditions that affect the heart, heart valves, and vascular system (e.g., veins and arteries) of the body, including, but not limited to, arteriosclerosis, atherosclerosis, myocardial infarction, acute coronary syndromes, angina, congestive heart failure, aortic aneurysm, aortic dissection, iliac or femoral aneurysm, pulmonary embolism, primary hypertension, atrial fibrillation, stroke, transient ischemic attack, systolic dysfunction, diastolic dysfunction, myocarditis, atrial tachycardia, ventricular fibrillation, endocarditis, arteriopathy, vasculitis, atherosclerotic plaque, vulnerable plaque, acute coronary syndromes, acute ischemic attack, sudden cardiac death, peripheral vascular disease, coronary artery disease (CAD), peripheral arterial disease (PAD), and cerebrovascular disease.

[0074] As used herein, the phrase "suspected of having CVD" refers to a patient who has at least one sign or symptom associated with CVD, such as extreme fatigue, constant dizziness or lightheadedness, fast heart rate (e.g., more than 100 beats per minute at rest), new irregular heartbeat, chest pain or discomfort with activity that goes away with rest, difficulty breathing during normal activity and at rest, worsening respiratory infection or cough, anxiety or confusion, changes in sleep patterns, and loss of appetite or nausea.

[0075] As used herein, the term "arteriosclerotic cardiovascular disease" or "arteriosclerotic cardiovascular disorder" refers to a part of cardiovascular disease that includes atherosclerosis as a component, or a precursor to a particular type of cardiovascular disease, including, but not limited to, CAD, PAD, and cerebrovascular disease. Atherosclerosis is a chronic inflammatory response that occurs in the arterial blood vessel wall. It involves the formation of atheromatous plaques, which can lead to narrowing of the artery ("stenosis") and ultimately to partial or complete closure of the arterial opening and / or plaque rupture. Thus, arteriosclerotic diseases or disorders include the consequences of atheromatous plaque formation and rupture, including, but not limited to, stenosis or narrowing of the artery, heart failure, formation of aneurysms, including aortic aneurysms, aortic dissection, and ischemic events, such as myocardial infarction and stroke. In certain embodiments, the methods, compositions, and systems disclosed herein are used to at least partially diagnose arteriosclerotic CVD.

[0076] The terms "individual," "host," "subject," and "patient" are used interchangeably herein and generally refer to mammals, including, but not limited to, primates, including apes and humans, equines (e.g., horses), canines (e.g., dogs), felines, various farm animals (e.g., ungulates such as swine, pigs, goats, sheep, etc.), as well as household pets and animals kept in zoos. In some embodiments, the subject is specifically a human subject.

[0077] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT Provided herein are methods, systems, and compositions for detecting one or more HDL-associated proteins (e.g., ApoC3; ApoC3 and ApoA1; ApoC3 and SAA1 / 2; or proteins of biomarker panels 1-30) from a sample from a subject suffering from or suspected of suffering from cardiovascular disease (CVD) or other HDL-associated disease. In certain embodiments, such methods, systems, and compositions are used to determine an estimated risk of CVD (or other disease) in a subject, and / or an estimated cholesterol efflux capacity (CEC) of the sample. In certain embodiments, the systems and compositions are comprised of a sample from a subject suffering from or suspected of suffering from CVD, and an HDL-associated binding agent or mass spectrometry standard.

[0078] Provided herein are methods, compositions, and kits for determining predicted cardiovascular disease and / or cholesterol efflux based on measuring protein levels in one or more of the panels shown in Tables 2 and 54 below. From studies conducted during the development of embodiments of the present disclosure, it has been found that either the levels of a single protein (e.g., ApoC3, panel 10) or a combination of multiple proteins (e.g., panels 1-9 and 11-30) can be used to determine CVD and cholesterol efflux capacity. Described below are exemplary, non-limiting methods and systems for use in using biomarker panels with samples from subjects suffering from or suspected of suffering from cardiovascular disease.

[0079] In some embodiments, blood is first drawn from a subject with symptoms of cardiovascular disease (e.g., chest pain) at a hospital, clinic, or other medical facility. The blood sample is processed to generate a serum sample. The serum sample is then processed with a method to purify HDL in the sample, such as ultracentrifugation or purification of tagged HDL-binding particles. The purified serum sample is further treated with an enzyme (e.g., Lys-C, etc.) that digests HDL-associated proteins present in the purified sample. The digested sample is then subjected to one or more detection assays (e.g., mass spectrometry, immunological assays, aptamer binding assays) to detect the levels of HDL-associated proteins present in one or more of biomarker panels 1-30.

[0080] In some embodiments, when using mass spectrometry, the detection protocol may be as follows: A portion of the digested HDL purified sample is diluted and loaded onto an LC column. Peptides are then detected using a triple quadrupole mass spectrometer (e.g., operating in dynamic MRM mode). Peptides are targeted at predefined "transitions" consisting of predefined m / z values ​​for the mass of the intact peptide (filtered in Q1 of the instrument), the collision energy the intact peptide experiences in q2 of the instrument, and fragment (m / z) values ​​for the peptide fragments (filtered in Q3). Transitions are selected to be unique to the peptides targeted within the sample. Two transitions (a "quantifier" used for later quantification and a "qualifier" used for quality control) are monitored per peptide, targeting up to two peptides per protein. A detailed list of peptide targets and their transitions is provided in Table 1 in Example 1 below.

[0081] Peptide signal intensities are obtained by integrating the chromatographic peaks of each peptide's "quantifier" transition. Protein intensities are determined by summing the quantifier peak areas of each of the protein's target peptides and normalizing with the intensities of one or more HDL-associated proteins.

[0082] Additionally, the levels of HDL-associated proteins of the selected biomarker panels are processed through a multivariate algorithm to determine a cholesterol elimination or CVD risk score. The algorithm includes a constant (i) specific to each panel (and specific to the CVD risk score or global elimination or ABCA1 elimination), and a protein and panel specific coefficient (c) that is multiplied by the specific protein level detected. The algorithm is as follows: CVD risk score / predicted overall cholesterol withdrawal = c1 * p1+c2 * p2...c n * p n +i Here, the efflux or CVD risk is determined by adding a constant (i) to the additive sum of coefficients (c) multiplied by the protein level (p) (e.g., normalized peak area) of a given protein. Example 1 below shows the coefficients and constants for each protein in each panel, for both protein and molar protein values ​​determined by mass spectrometry. Example 1 further shows the 95% confidence interval for each of these values. For example, when calculating a CVD risk score and / or cholesterol efflux value, any value within this range can be used for the value of (c) or (p).

[0083] The CVD risk score determined using the algorithm above and the values ​​in Example 1 below can be converted to a probability (risk rate for CVD) using the following formula: Probability=1 / (1+exp(-risk score)). This probability value can then be multiplied by 100 to obtain a risk percentage for CVD. The risk score or probability or risk percentage for CVD can be used by medical professionals to aid in determining treatment for patients suspected of having cardiac disease or to exclude patients who may require treatment or monitoring. For example, in general, patients with a risk percentage for CAD of less than 10% are usually considered low risk (e.g., can be considered as patients who do not require treatment or monitoring for cardiac disease), patients with a risk percentage for CAD of 10-20% are usually considered intermediate risk (e.g., can be monitored for additional signs of cardiac disease), and patients with a risk percentage for CAD of more than 20% are usually considered high risk (e.g., can be treated with therapeutic or surgical intervention).

[0084] In certain embodiments, one or more of the proteins detected as part of panels 1-30 are detected in a purified high density lipoprotein sample. The purified HDL sample can be obtained by any suitable method, including ultracentrifugation or HDL-binding peptide methods.

[0085] For example, ultracentrifugation (105,000xg) of serum with a density of 1.006 kg / L at 18°C ​​for 18 hours can be used to remove very low density lipoproteins that interfere with the precipitation of low density lipoproteins by manganese and heparin. Manganese chloride and heparin are added to a final concentration of 0.0456 molar Mn 2+ When added to the bottom fraction in an amount sufficient to obtain 183 UPS kU / L heparin, non-HDL-C lipoproteins are precipitated. After centrifugation at 1500 x g for 30 min at 4°C, the resulting supernatant is further adjusted to a density of ≥1.21 kg / L with potassium bromide. The density-adjusted solution is then centrifuged again for 24 h (approximately 105,000 x g). After centrifugation, the top layer is carefully removed to obtain the HDL-enriched fraction.

[0086] In certain embodiments, HDL is purified using affinity-tagged HDL-binding peptides (e.g., ApoA1). Such purification methods are described in U.S. Application No. 14 / 713,046, filed May 15, 2015, which is incorporated herein by reference in its entirety as if fully set forth herein. Briefly, affinity-tagged HDL-binding peptides (e.g., ApoA1) are added to a sample. HDL particles incorporate a portion of the bound particles. Such HDL particles are then purified using affinity tags, thereby obtaining a purified HDL sample.

[0087] The present invention is not limited to the method used to detect HDL-associated peptides (eg, ApoC3, SAA1 / 2, ApoE, ApoL1, etc.) from a subject sample.

[0088] In certain embodiments, the HDL-associated protein is detected using a detection method selected from the following: immunoassay, surface plasmon resonance, in vitro assay, activity assay, co-immunoprecipitation assay, mass spectrometry, fluorescence energy transfer (FRET), bioluminescence energy transfer (BRET), interferometry, biolayer interferometry (BLI), dual polarization interferometry ("DPI"), ellipsometry, and quartz crystal microbalance. EXAMPLES

[0089] Example 1 This example describes the testing of the ability of a panel of 26 single and multiple HDL proteins to predict overall cholesterol efflux and cardiovascular disease risk using various algorithms for overall cholesterol efflux and coronary artery disease risk. HDL proteins were measured by mass spectrometry.

[0090] Purification of HDL The following protocol was used for purification of HDL: in 1x phosphate-buffered saline (PBS) (pH 7.4); 1525 μL of a 0.5 mg / mL solution of N-labeled His-tagged apolipoprotein AI is mixed with 12 μL of human serum. The mixture of serum and ApoA-I is incubated at 37 °C for 15 min during which the His-tagged ApoA-I is incorporated into HDL particles in the serum sample. HDL is then purified from the mixture by immobilized metal affinity chromatography in a pipette tip format suitable for high-throughput automated liquid handling platforms. Briefly, the mixture of serum and ApoA-I is diluted to 200 μL by adding 164 μL of binding buffer (5 mM imidazole, 50 mM sodium phosphate, 300 mM sodium chloride, pH 8.0). The affinity column tip is equilibrated with 400 μL of binding buffer. The affinity column is then used to aspirate 180 μL of the serum / ApoA-I mixture six times at a flow rate of 200 μL / min and dispense onto the column bed. The column bed is then washed once by aspirating and dispensing 200 μL of binding buffer, followed by a second wash with 200 μL of strong wash buffer (10 mM imidazole, 50 mM sodium phosphate, 300 mM sodium chloride, pH 8.0). The bound HDL particles are then eluted by aspirating and dispensing 90 μL of elution buffer (300 mM imidazole, 50 mM Tris-HCl, 25% methanol, pH 9.0).

[0091] HDL digestion Next, 5 μL of 100 mM dithiothreitol and 10 μL of 50 ng / μL endoproteinase Lys-C are added to the eluted HDL and incubated for 4 hours at 37° C. Lys-C cleaves HDL-associated proteins at the C-terminal side of lysine amino acid residues in the sequence, producing target peptides that can be predicted by LC-MS analysis and quantification.

[0092] LC-MS analysis of HDL proteome target peptides 25 μL of Lys-C digested HDL in elution buffer containing 25% methanol is loaded directly onto the LC column after in-line injection and 5-fold dilution using a mixing tee. Dilution and loading are performed using a sample loading pump delivering 0.1% formic acid in water to the mixing tee at 150 μL / min, and a binary pump delivering 99% mobile phase A (0.1% formic acid in water), 1% mobile phase B (0.1% formic acid in acetonitrile) to the tee at 600 μL / min. After loading, the binary pump is switched collinear with the column and peptides are eluted with a linear gradient of mobile phase B from 1% to 65% at 500 μL / min for 5 min. Peptides are detected using an Agilent 6490 triple quadrupole mass spectrometer operated in dynamic MRM mode. Dynamic MRM allows targeted detection and quantification of peptide targets within scheduled retention time windows. Peptides are targeted at predefined "transitions" consisting of a predefined m / z value for the mass of the intact peptide (filtered in Q1 of the instrument), the collision energy that the intact peptide undergoes in Q2 of the instrument, and the fragment (m / z) value for the peptide fragment (filtered in Q3). Transitions are selected to be unique to the peptides targeted in the sample. Two transitions (a "quantifier" used for later quantification and a "qualifier" used for quality control) are monitored per peptide, targeting up to two peptides per protein. A detailed list of peptide targets and their transitions is shown in Table 1. Note that other portions of the peptide sequences than those listed can be used when performing mass spectrometry measurements.

[0093] [Table 1] JPEG2024178265000003.jpg244169JPEG2024178265000004.jpg46169

[0094] 26 HDL Protein Panel Table 2 shows a panel of 26 protein combinations that performed well in measuring global cholesterol efflux from HDL.

[0095] [Table 2]

[0096] Peptide and protein quantification and normalization Peptide signal intensities are obtained by integrating the chromatographic peaks of each peptide's "quantifier" transition. Protein intensities are determined by summing the quantifier peak areas of each of the protein's target peptides and normalizing with the intensities of one or more HDL-associated proteins.

[0097] Algorithm for pullability and CAD risk Based on the normalized intensities from the measured proteins, multivariate algorithms for efflux capacity (discovery cohort and fresh vs. frozen) and CAD risk (total, with or without events) were developed and used to obtain values ​​related to cholesterol efflux capacity or CAD risk of the original serum samples. Model 1 is the overall cholesterol efflux of the discovery samples, model 2 is the overall cholesterol efflux of the fresh and frozen samples, model 3 is the overall coronary artery disease (CAD) risk, model 4 is the CAD risk with events, and model 5 is the CAD risk without events. The general formulas for these models are shown below: Models 1 and 2 - Predicted global cholesterol efflux = c1 * p1+c2 * p2...c n p n +i Model 3~5-CAD (with / without events) risk score = c1p1+c2p2…c n p n +i Here, subtraction is determined by adding a constant (i) to the additive sum of a coefficient (c) multiplied by the normalized peak area (p) of several proteins.

[0098] statistical analysis Statistical analysis of the data for each panel was performed using R (www. followed by r-project.org) and Bioconductor (www. followed by bioconductor.org). Data were normalized by the total intensity of 15N ApoA1 protein. For proteins with two peptides measured by MRM, the protein level was calculated based on the peptide with the highest intensity in the majority of the samples analyzed. Correlation between each protein and global cholesterol efflux was calculated using robust linear regression. The least absolute shrinkage and selection operator (LASSO) was applied to all proteins to select features with non-zero coefficients. Stepwise robust linear regression was applied to the selected features measured in the 70 search samples. For each panel, the model with the lowest Akaike information criterion (AIC) was selected.

[0099] Determination of absolute molar amounts of proteins To obtain molar values, a five-point external calibration curve covering the range of measured signals for each peptide in Table 1 was constructed and monitored by LC-MS using synthetic peptide standards. The calibration curves were linearly fitted and weighted by 1 / X, where X is the peptide concentration. The calculated molar values ​​were then used to calculate the molar mass coefficients.

[0100] Panel 1 Models 1 and 2 were developed to predict the overall cholesterol efflux for Panel 1. The models are shown below. Models 1 and 2 - Predicted Global Cholesterol Efflux = (i) + (c1) * ApoC3+(c2) * ApoE+(c3) * ApoL1+(c4) * PLTP+(c5)* SAA1 / 2 The values ​​for Models 1-5 are shown in Table 3A below. As an example, Model 1 has the following values: Model 1 - predicted global cholesterol withdrawal = 9.25 + (165.75) ApoC3 + (-834.8) ApoE + (-144.26) ApoL1 + (-11548.33) PLTP + (-160.58) SAA1 / 2.

[0101] [Table 3]

[0102] This model was then tested on 35 replicate samples. The correlation between laboratory measured cholesterol efflux and predicted cholesterol efflux was 0.78. A robust linear regression was then applied to this panel of proteins in another cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown above (Table 2) and the correlation between laboratory measured cholesterol efflux and predicted cholesterol efflux was 0.52. As an example, Model 2 yields the following values: Model 2 - predicted overall cholesterol withdrawal = 5.27 + (230.53) * ApoC3+(4786.52) * ApoE+(41.71) * ApoL1+(14368.89) * PLTP+(-93.37) * SAA1 / 2.

[0103] Logistic regression was then applied to this panel of proteins in a separate cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients (Model 3) had an area under the ROC curve (AUC) equal to 0.62 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events (Model 4) had an area under the ROC curve (AUC) equal to 0.64 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events (Model 5) had an area under the ROC curve (AUC) equal to 0.66 based on 5-fold cross-validation. The values ​​for the three models are shown in Table 3A above and the formulas are shown below. CAD (with or without events) risk score = (i) + (c1) * ApoC3+(c2) * ApoE+(c3) * ApoL1+(c4) * PLTP+(c5) * SAA1 / 2 Model 3 – Total CAD risk score = 2.49 + (11.44) * ApoC3+(-1434.48) * ApoE+(17.74) * ApoL1+(-9141.23) * PLTP+(16.45) * SAA1 / 2. Model 4 – Risk score for CAD with events = 2.42 + (7.8) * ApoC3+(-1597.95) * ApoE+(-28) * ApoL1+(-9584.47) * PLTP+(12.03) * SAA1 / 2. Model 5 – Risk score for event-free CAD = 1.08 + (13.1) * ApoC3+(-1221.33) * ApoE+(73.58) * ApoL1+(-8509.05) * PLTP+(24.59) * SAA1 / 2. Note that the values ​​in Table 3A also include the number of 95% confidence interval ranges. For each value in the formula, this range number can be substituted into the formula.

[0104] These formulas and values ​​were then used to calculate the final values ​​for three patients (P1, P2, and P3), the results of which are shown below in Table 3B.

[0105] [Table 4]

[0106] The risk score results from Table 3B can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)), which can then be multiplied by 100 to obtain a risk percentage.

[0107] For example, for patient P1, the risk score result for Model 3 in Table 3A is -1.16. Substituting this into the formula gives: 1 / (1+exp(-(-1.16)))=0.2386673. Multiplying the result by 100 gives a 23.8% risk probability for CAD for patient P1. This same formula can be used for all of the models and patients in Table 3B, as well as the results reported below for the other panels in this Example and Example 2 below.

[0108] The risk score can be used by medical professionals to aid in treatment decisions for patients suspected of having cardiac disease or to exclude patients who may require treatment or monitoring. For example, in general, patients with a risk rate of CAD less than 10% are usually considered low risk (e.g., they may be considered patients who do not require treatment or monitoring for cardiac disease), patients with a risk rate of CAD between 10-20% are usually considered intermediate risk (e.g., they may be monitored for additional signs of cardiac disease), and patients with a risk rate of CAD greater than 20% are usually considered high risk (e.g., they may be treated immediately with therapeutic or surgical intervention).

[0109] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 3C for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used in models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0110] [Table 5]

[0111] Note that the values ​​in Table 3C also include the number of ranges for the 95% confidence interval. For each value in the formula (model), this range number can be substituted into the formula. For example, the value of ApoC3 is not exactly 150.35, but can be any value between 109.91 and 190.8, and this can be used in the model.

[0112] Panel 2 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk in Panel 2. Models 1-5 for Panel 2 are shown below: Models 1-5 in Panel 2 = (i) + (c1) * ApoC2+(c2) * ApoC3+(c3) * ApoE+(c4) * ApoL1+(c5) * CLU+(c6) * PLTP Values ​​for models 1 to 5 are shown in Table 4 below.

[0113] [Table 6]

[0114] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.77. A robust linear regression was then applied to this panel of proteins in a separate cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). The values ​​for this model (Model 2) are shown in Table 4 above, and the correlation between laboratory measured and predicted cholesterol efflux was 0.6.

[0115] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.61 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.7 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.59 based on 5-fold cross-validation. The values ​​for the three models are shown in Table 4. Note that the values ​​in Table 4 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0116] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three patients are shown below in Table 5A.

[0117] [Table 7]

[0118] The results of Table 5A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-risk score)).

[0119] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 5b for Models 1, 3, 4, and 5. Because these coefficients are in molar units, they can be used with models in which protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0120] [Table 8]

[0121] Note that the values ​​in Table 5B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0122] Panel 3 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk in Panel 3. Models 1-5 for Panel 3 are shown below: Models 1-5 in Panel 3 = (i) + (c1) * ApoC2+(c2) * ApoC3+(c3) * ApoL1+(c4) * PLTP Values ​​for models 1 to 5 are shown in Table 6 below.

[0123] [Table 9]

[0124] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.73. A robust linear regression was then applied to this panel of proteins in a separate cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 6 and the correlation between laboratory measured and predicted cholesterol efflux was 0.5.

[0125] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.61 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.68 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.61 based on 5-fold cross-validation. The three CAD models are shown in Table 6. Note that the values ​​in Table 6 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0126] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three patients are shown below in Table 7A.

[0127] [Table 10]

[0128] The results of Table 7A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-risk score)).

[0129] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 7b for Models 1, 3, 4, and 5. Because these coefficients are in molar units, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0130] [Table 11]

[0131] Note that the values ​​in Table 7B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0132] Panel 4 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk in Panel 4. Models 1-5 for Panel 4 are shown below: Models 1-5 in Panel 4 = (i) + (c1) * ApoC1+(c2) * ApoC3+(c3) * CLU+(c4) * PLTP+(c5) * SAA4 Values ​​for models 1 to 5 are shown in Table 8 below.

[0133] [Table 12]

[0134] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.78. A robust linear regression was then applied to this panel of proteins in a separate cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). The values ​​of this model are shown in Table 8 above, and the correlation between laboratory measured and predicted cholesterol efflux was 0.68.

[0135] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.72 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.8 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.66 based on 5-fold cross-validation. The values ​​for the three models are shown in Table 8 above. Note that the values ​​in Table 4 also include the number of ranges for the 95% confidence interval. For each value, this number of ranges can be substituted into the formula.

[0136] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three patients are shown in Table 9A.

[0137] [Table 13]

[0138] The results of Table 9A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-risk score)).

[0139] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 9b for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used in models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0140] [Table 14]

[0141] Note that the values ​​in Table 9B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0142] Panel 5 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 5. Models 1-5 for Panel 5 are shown below: Models 1-5 for panel 5 = (i) + (c1) * ApoA1+(c2) * ApoC1+(c3) * ApoC3 Values ​​for models 1 to 5 are shown in Table 10 below.

[0143] [Table 15]

[0144] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.85. A robust linear regression was then applied to this panel of proteins in a separate cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 10 above and the correlation between laboratory measured and predicted cholesterol efflux was 0.53.

[0145] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.75 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.81 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.68 based on 5-fold cross-validation. The values ​​for the three CAD models are shown in Table 10. Note that the values ​​in Table 10 also include the number of ranges for the 95% confidence interval. For each value, this number of ranges can be substituted into the formula.

[0146] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three patients are shown in Table 11A below.

[0147] [Table 16]

[0148] The results of Table 11A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0149] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 11b for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0150] [Table 17]

[0151] Note that the values ​​in Table 11B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0152] Panel 6 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 6. Models 1-5 for Panel 6 are shown below: Models 1-5 in Panel 6 = (i) + (c1) * ApoC3+(c2) * ApoL1+(c3) * PLTP+(c4) * SAA1 / 2 Values ​​for models 1 to 5 are shown in Table 12 below.

[0153] [Table 18]

[0154] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.74. A robust linear regression was then applied to this panel of proteins in a separate cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 12 and the correlation between laboratory measured and predicted cholesterol efflux was 0.49.

[0155] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.6 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.63 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.64 based on 5-fold cross-validation. The values ​​of the three models (models 3 to 5) for CAD risk are shown in Table 12. Note that the values ​​in Table 12 also include the number of ranges of the 95% confidence interval. For each value, this number of ranges can be substituted into the formula.

[0156] The predicted cholesterol withdrawal and CAD risk based on the five models and three patients is shown below in Table 13A.

[0157] [Table 19]

[0158] The results in Table 13A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0159] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 13b for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0160] [Table 20]

[0161] Note that the values ​​in Table 13B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0162] Panel 7 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 7. Models 1-5 for Panel 7 are shown below: Models 1-5 in Panel 7 = (i) + (c1) * ApoC1+(c2) * ApoC3+(c3) * ApoM Values ​​for models 1 to 5 are shown in Table 14 below.

[0163] [Table 21]

[0164] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.84. A robust linear regression was then applied to this panel of proteins in a separate cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 14 and the correlation between laboratory measured and predicted cholesterol efflux was 0.54.

[0165] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.75 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.82 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.73 based on 5-fold cross-validation. The values ​​for the three models are shown in Table 14 above. Note that the values ​​in Table 14 also include the number of ranges for the 95% confidence interval. For each value in the formula, the number of ranges can be substituted into the formula.

[0166] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 15A.

[0167] [Table 22]

[0168] The results in Table 15A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0169] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 15b for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0170] [Table 23]

[0171] Note that the values ​​in Table 15B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0172] Panel 8 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 8. Models 1-5 for Panel 8 are shown below: Models 1-5 for panel 8 = (i) + (c1) * ApoC3+(c2) * SAA1 / 2 Values ​​for models 1 to 5 are shown in Table 16 below.

[0173] [Table 24]

[0174] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.8. Linear regression was then applied to this panel of proteins in another cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 16 and the correlation between laboratory measured and predicted cholesterol efflux was 0.48.

[0175] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.58 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.54 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.53 based on 5-fold cross-validation. The values ​​for the three models are shown in Table 16 above. Note that the values ​​in Table 16 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0176] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 17A.

[0177] [Table 25]

[0178] The results in Table 17A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0179] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 17b for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0180] [Table 26]

[0181] Note that the values ​​in Table 17B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0182] Panel 9 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 9. Models 1-5 for Panel 9 are shown below: Panel 9 Models 1-5 = (i) + (c1) * ApoC1+(c2) * ApoC3 Values ​​for models 1 to 5 are shown in Table 18 below.

[0183] [Table 27]

[0184] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.85. Linear regression was then applied to this panel of proteins in another cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 18 and the correlation between laboratory measured and predicted cholesterol efflux was 0.54.

[0185] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.7 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.77 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.68. The values ​​for the three models are shown in Table 18 above. Note that the values ​​in Table 18 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0186] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 19A.

[0187] [Table 28]

[0188] The results in Table 19A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0189] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 19b for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0190] [Table 29]

[0191] Note that the values ​​in Table 19B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0192] Panel 10 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 10. Models 1-5 for Panel 10 are shown below: Models 1-5 in Panel 10 = (i) + (c1) * ApoC3 Values ​​for models 1 to 5 are shown in Table 20 below.

[0193] [Table 30]

[0194] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.8. Linear regression was then applied to this panel of proteins in another cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 20 and the correlation between laboratory measured and predicted cholesterol efflux was 0.48.

[0195] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.56 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.58 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.49 based on 5-fold cross-validation. The values ​​for the three models are shown in Table 20 above. Note that the values ​​in Table 20 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0196] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 21A.

[0197] [Table 31]

[0198] The results of Table 21A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)). This value can be multiplied by 100 to obtain the risk percentage of CAD.

[0199] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 21b for Models 1, 3, 4, and 5. Because these coefficients are in molar units, they can be used with models in which protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0200] [Table 32]

[0201] Note that the values ​​in Table 21B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0202] Panel 11 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 11. Models 1-5 for Panel 11 are shown below: Panel 11 Models 1-5 = (i) + (c1) * ApoC3+(c2) * ApoL1+(c3) * SAA1 / 2 Values ​​for models 1 to 5 are shown in Table 22 below.

[0203] [Table 33]

[0204] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.8. A robust linear regression was then applied to this panel of proteins in a separate cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 22 and the correlation between laboratory measured and predicted cholesterol efflux was 0.48.

[0205] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.54 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.53 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.48 based on 5-fold cross-validation. The values ​​for the three models are shown in Table 22 above. Note that the values ​​in Table 22 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0206] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 23A.

[0207] [Table 34]

[0208] The results in Table 23A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0209] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 23b for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0210] [Table 35]

[0211] Note that the values ​​in Table 23B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0212] Panel 12 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 12. Models 1-5 for Panel 12 are shown below: Panel 12 Models 1-5 = (i) + (c1) * ApoA1+(c2) * ApoC1+(c3) * ApoC3+(c4) * CLU Values ​​for models 1 to 5 are shown in Table 24 below.

[0213] [Table 36]

[0214] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.86. A robust linear regression was then applied to this panel of proteins in a separate cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 24 and the correlation between laboratory measured and predicted cholesterol efflux was 0.61.

[0215] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.75 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.82 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.63 based on 5-fold cross-validation. The values ​​of the three models are shown in Table 24 above. Note that the values ​​in Table 24 also include the number of ranges of the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0216] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 25A.

[0217] [Table 37]

[0218] The results in Table 25A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0219] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 25b for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0220] [Table 38]

[0221] Note that the values ​​in Table 25B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0222] Panel 13 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 13. Models 1-5 for Panel 13 are shown below: Panel 13 Models 1-5 = (i) + (c1) * ApoC1+(c2) * ApoC3+(c3) * ApoL1+(c4) * HP+(c5) * PLTP Values ​​for models 1 to 5 are shown in Table 26 below.

[0223] [Table 39]

[0224] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.74. A robust linear regression was then applied to this panel of proteins in a separate cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 26 and the correlation between laboratory measured and predicted cholesterol efflux was 0.54.

[0225] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.7 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.79 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.68 based on 5-fold cross-validation. The values ​​for the three models are shown in Table 14 above. Note that the values ​​in Table 26 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0226] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 27A.

[0227] [Table 40]

[0228] The results in Table 27A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0229] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 27b for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0230] [Table 41]

[0231] Note that the values ​​in Table 27B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0232] Panel 14 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 14. Models 1-5 for Panel 14 are shown below: Models 1-5 in Panel 14 = (i) + (c1) * ApoC3+(c2) * ApoD+(c3) * SAA1 / 2 Values ​​for models 1 to 5 are shown in Table 28 below.

[0233] [Table 42]

[0234] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.8. Linear regression was then applied to this panel of proteins in another cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 28 and the correlation between laboratory measured and predicted cholesterol efflux was 0.48.

[0235] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.68 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.71 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.65 based on 5-fold cross-validation. The values ​​for the three models are shown in Table 28 above. Note that the values ​​in Table 28 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0236] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 29A.

[0237] [Table 43]

[0238] The results in Table 29A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0239] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 29b for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used in models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0240] [Table 44]

[0241] Note that the values ​​in Table 29B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0242] Panel 15 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 15. Models 1-5 for Panel 15 are shown below: Models 1-5 for panel 15 = (i) + (c1) * ApoA2+(c2) * ApoC2+(c3) * ApoC3+(c4) * ApoD+(c5) * CLU+(c6) * Values ​​for SAA1 / 2 models 1 to 5 are shown in Table 30 below.

[0243] [Table 45]

[0244] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.73. Partial linear regression was then applied to this panel of proteins in another cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 20 and the correlation between laboratory measured and predicted cholesterol efflux was 0.57.

[0245] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.61 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.7 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.6 based on 5-fold cross-validation. The values ​​for the three models are shown in Table 30 above. Note that the values ​​in Table 30 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0246] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 31A.

[0247] [Table 46]

[0248] The results in Table 31A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0249] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 31b for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0250] [Table 47]

[0251] Note that the values ​​in Table 31B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0252] Panel 16 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 16. Models 1-5 for Panel 16 are shown below: Models 1-5 in Panel 16 = (i) + (c1) * ApoA2+(c2) * ApoC2+(c3) * ApoC3+(c4) * ApoD+(c5) * ApoM+(c6) * SAA1 / 2 Values ​​for models 1 to 5 are shown in Table 32 below.

[0253] [Table 48]

[0254] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.65. Partial linear regression was then applied to this panel of proteins in another cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 32 and the correlation between laboratory measured and predicted cholesterol efflux was 0.51.

[0255] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.65 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.74 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.65 based on 5-fold cross-validation. The values ​​for the three CAD models are shown in Table 32 above. Note that the values ​​in Table 32 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0256] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 33A.

[0257] [Table 49]

[0258] The results in Table 33A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0259] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 33b for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used in models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0260] [Table 50]

[0261] Note that the values ​​in Table 33B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0262] Panel 17 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 17. Models 1-5 for Panel 17 are shown below: Models 1-5 in Panel 17 = (i) + (c1) * ApoC1+(c2) * ApoC2+(c3) * ApoC3 Values ​​for models 1 to 5 are shown in Table 34 below.

[0263] [Table 51]

[0264] This model was then tested on 35 replicate samples. The correlation between laboratory measured cholesterol efflux and predicted cholesterol efflux was 0.81. The elastic net model was then applied to this panel of proteins in another cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 34 above and the correlation between laboratory measured cholesterol efflux and predicted cholesterol efflux was 0.53.

[0265] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.74 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.78 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.72 based on 5-fold cross-validation. The values ​​for the three models are shown in Table 34 above. Note that the values ​​in Table 34 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0266] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 35A.

[0267] [Table 52]

[0268] The results in Table 35A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0269] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 35B for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0270] [Table 53]

[0271] Note that the values ​​in Table 35B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0272] Panel 18 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 18. Models 1-5 for Panel 18 are shown below: Models 1-5 in Panel 18 = (i) + (c1) * ApoA1+(c2) * ApoC1+(c3) * ApoC2+(c4) * ApoC3+(c5) * ApoC4 Values ​​for models 1 to 5 are shown in Table 36 below.

[0273] [Table 54]

[0274] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.86. Linear regression was then applied to this panel of proteins in another cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 36 and the correlation between laboratory measured and predicted cholesterol efflux was 0.55.

[0275] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.77 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.81 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.72 based on 5-fold cross-validation. The values ​​for the three CAD models are shown in Table 36 above. Note that the values ​​in Table 36 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0276] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 37A.

[0277] [Table 55]

[0278] The results in Table 37A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0279] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 37B for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used in models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0280] [Table 56]

[0281] Note that the values ​​in Table 37B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0282] Panel 19 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 19. Models 1-5 for Panel 19 are shown below: Models 1-5 in Panel 19 = (i) + (c1) * ApoA2+(c2) * ApoC1+(c3) * ApoC2+(c4) * ApoC3+(c5) * ApoD+(c6) * SAA1 / 2 Values ​​for models 1 to 5 are shown in Table 38 below.

[0283] [Table 57]

[0284] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.71. A partial linear regression was then applied to this panel of proteins in another cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). The values ​​for this model (Model 2) are shown in Table 38, and the correlation between laboratory measured and predicted cholesterol efflux was 0.51.

[0285] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.74 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.79 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.73 based on 5-fold cross-validation. The values ​​for the three CAD models are shown in Table 38 above. Note that the values ​​in Table 38 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0286] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 39A.

[0287] [Table 58]

[0288] The results in Table 39A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0289] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 39B for Models 1, 3, 4, and 5. Because these coefficients are in units of molar amounts, they can be used in models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0290] [Table 59]

[0291] Note that the values ​​in Table 39B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0292] Panel 20 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 20. Models 1-5 for Panel 20 are shown below: Models 1-5 in panel 20 = (i) + (c1) * ApoA2+(c2) * ApoC3+(c3) * ApoD+(c4) * ApoE+(c5) * ApoL1+(c6) * PLTP+(c7) * SAA1 / 2 Values ​​for Models 1-5 for Panel 20 are shown in Table 40 below.

[0293] [Table 60]

[0294] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.8. The elastic net model was then applied to this panel of proteins in another cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 40 above and the correlation between laboratory measured and predicted cholesterol efflux was 0.52.

[0295] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.64 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.68 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.66 based on 5-fold cross-validation. The values ​​for the three CAD models are shown in Table 40 above. Note that the values ​​in Table 40 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0296] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown in Table 41 below.

[0297] [Table 61]

[0298] The results in Table 41A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0299] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 41B for Models 1, 3, 4, and 5. Because these coefficients are in units of molar amounts, they can be used with models in which protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0300] [Table 62]

[0301] Note that the values ​​in Table 41B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0302] Panel 21 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 21. Models 1-5 for Panel 21 are shown below: Models 1-5 in Panel 21 = (i) + (c1) * ApoC3+(c2) * ApoM+(c3) * PLTP+(c4) * SAA1 / 2 Values ​​for Models 1-5 for Panel 21 are shown in Table 42 below.

[0303] [Table 63]

[0304] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.76. A robust linear regression was then applied to this panel of proteins in a separate cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown above in Table 42 and the correlation between laboratory measured and predicted cholesterol efflux was 0.5.

[0305] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.65 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.71 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.68 based on 5-fold cross-validation. The values ​​for the three CAD models are shown in Table 42 above. Note that the values ​​in Table 42 also include the number of ranges for the 95% confidence interval. For each value in the formula, the number of ranges can be substituted into the formula.

[0306] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 43A.

[0307] [Table 64]

[0308] The results in Table 43A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0309] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 43B for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used in models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0310] [Table 65]

[0311] Note that the values ​​in Table 43B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0312] Panel 22 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 22. Models 1-5 for Panel 22 are shown below: Models 1-5 in panel 22 = (i) + (c1) * ApoC3+(c2) * ApoD+(c3) * PLTP+(c4) * SAA1 / 2 Values ​​for models 1 to 5 are shown in Table 44 below.

[0313] [Table 66]

[0314] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.79. A robust linear regression was then applied to this panel of proteins in a separate cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown above in Table 44 and the correlation between laboratory measured and predicted cholesterol efflux was 0.5.

[0315] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.65 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.71 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.66 based on 5-fold cross-validation. The values ​​for the three CAD models are shown in Table 44 above. Note that the values ​​in Table 44 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0316] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 45A.

[0317] [Table 67]

[0318] The results in Table 45A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0319] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 45B for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used in models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0320] [Table 68]

[0321] Note that the values ​​in Table 45B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0322] Panel 23 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 23. Models 1-5 for Panel 23 are shown below: Panel 23 Models 1-5 = (i) + (c1) * ApoA2+(c2) * ApoC3+(c3) * ApoD+(c4) * ApoL1+(c5) * ApoM+(c6) * PLTP+(c7) * SAA1 / 2 Values ​​for models 1 to 5 are shown in Table 46 below.

[0323] [Table 69]

[0324] This model was then tested on 35 replicate samples. The correlation between laboratory measured cholesterol efflux and predicted cholesterol efflux was 0.79. The elastic net model was then applied to this panel of proteins in another cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 46 above and the correlation between laboratory measured cholesterol efflux and predicted cholesterol efflux was 0.5.

[0325] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.65 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.7 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.67 based on 5-fold cross-validation. The values ​​for the three CAD models are shown in Table 46 above. Note that the values ​​in Table 46 also include the number of ranges for the 95% confidence interval. For each value in the formula, the number of ranges can be substituted into the formula.

[0326] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 47A.

[0327] [Table 70]

[0328] The results in Table 47A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0329] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 47B for Models 1, 3, 4, and 5. Because these coefficients are in units of molar amounts, they can be used with models in which protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0330] [Table 71]

[0331] Note that the values ​​in Table 47B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0332] Panel 24 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 24. Models 1-5 for Panel 24 are shown below: Models 1-5 in panel 24 = (i) + (c1) * ApoC1+(c2) * ApoC3+(c3) * SAA1 / 2 Values ​​for models 1 to 5 are shown in Table 48 below.

[0333] [Table 72]

[0334] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.85. Linear regression was then applied to this panel of proteins in another cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 48 above and the correlation between laboratory measured and predicted cholesterol efflux was 0.53.

[0335] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.7 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.75 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.67 based on 5-fold cross-validation. The values ​​for the three CAD models are shown in Table 48 above. Note that the values ​​in Table 48 also include the number of ranges for the 95% confidence interval. For each value in the formula, this number of ranges can be substituted into the formula.

[0336] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 49A.

[0337] [Table 73]

[0338] The results in Table 49A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0339] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 49B for Models 1, 3, 4, and 5. Because these coefficients are in molar units, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0340] [Table 74]

[0341] Note that the values ​​in Table 49B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0342] Panel 25 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for panel 25. Models 1-5 for panel 25 are shown below: Models 1-5 of panel 25 = (i) + (c1) * ApoC1+(c2) * ApoC3+(c3) * C3+(c4) * PLTP Values ​​for models 1 to 5 are shown in Table 50 below.

[0343] [Table 75]

[0344] This model was then tested on 35 replicate samples. The correlation between laboratory measured and predicted cholesterol efflux was 0.79. A robust linear regression was then applied to this panel of proteins in a separate cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 50 and the correlation between laboratory measured and predicted cholesterol efflux was 0.56.

[0345] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.74 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.8 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.7 based on 5-fold cross-validation. The values ​​for the three CAD models are shown in Table 50 above. Note that the values ​​in Table 50 also include the number of ranges for the 95% confidence interval. For each value in the formula, the number of ranges can be substituted into the formula.

[0346] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 51A.

[0347] [Table 76]

[0348] The results in Table 51A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0349] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 51B for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0350] [Table 77]

[0351] Note that the values ​​in Table 51B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0352] Panel 26 Models 1-5 were created to predict overall cholesterol withdrawal and CAD risk for Panel 26. Models 1-5 for Panel 26 are shown below: Panel 26 Models 1-5 = (i) + (c1) * ApoA2+(c2) * ApoC2+(c3) * ApoC3+(c4) * ApoD+(c5) * ApoE+(c6) * ApoL1+(c7) * ApoM+(c8) * CETP+(c9) * CLU+(c10) * PLTP+(c11) * PON1+(c12) * SAA1 / 2 Values ​​for Models 1 to 5 are shown in Table 52 below.

[0353] [Table 78] JPEG2024178265000082.jpg39169

[0354] This model was then tested on 35 replicate samples. The correlation between laboratory measured cholesterol efflux and predicted cholesterol efflux was 0.78. The elastic net model was then applied to this panel of proteins in another cohort of samples consisting of 76 fresh and frozen samples (40 fresh and 36 frozen). This model (Model 2) is shown in Table 52 above and the correlation between laboratory measured cholesterol efflux and predicted cholesterol efflux was 0.62.

[0355] Finally, logistic regression was applied to this panel of proteins in another cohort of samples consisting of 74 healthy controls and 157 CAD patients (83 with events and 74 without events). The model fitted to the comparison of healthy controls and CAD patients had an area under the ROC curve (AUC) equal to 0.63 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients with events had an area under the ROC curve (AUC) equal to 0.71 based on 5-fold cross-validation. The model fitted to the comparison of healthy controls and CAD patients without events had an area under the ROC curve (AUC) equal to 0.6 based on 5-fold cross-validation. The values ​​for the three CAD models are shown in Table 52 above. Note that the values ​​in Table 52 also include the number of ranges for the 95% confidence interval. For each value in the formula, the number of ranges can be substituted into the formula.

[0356] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 53A.

[0357] [Table 79]

[0358] The results in Table 53A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0359] As above, the absolute molar amounts of each protein were determined using the absolute amounts of the internal standards. These molar amounts were then used to determine the coefficients in Table 53B for Models 1, 3, 4, and 5. Because these coefficients are in molar amounts, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0360] [Table 80]

[0361] Note that the values ​​in Table 53B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0362] Example 2 This example describes the use of various algorithms for ABCA1 cholesterol efflux and coronary artery disease risk to test the ability of five panels of single and multiple HDL proteins to predict overall ABCA1 cholesterol efflux (not overall cholesterol efflux) and cardiovascular disease risk. The five panels were run using the same assays and samples as in Example 1 above. As shown in Table 54 below, the five panels are panels 27-30 and panel 10.

[0363] [Table 81]

[0364] Panel 27 Models 1-5 were created to predict ABCA1 cholesterol withdrawal and CAD risk in panel 27. Models 1-5 for panel 27 are shown below: Panel 27 Models 1-5 = Predicted ABCA1 cholesterol efflux = (i) + (c1) * ApoC3+(c2) * ApoE+(c3) * ApoL1+(c4) * HP+(c5) * PLTP Values ​​for Models 1 to 5 are shown in Table 55 below.

[0365] [Table 82]

[0366] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 56A.

[0367] [Table 83]

[0368] The results in Table 56A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0369] The absolute molar amounts of each protein were determined using the absolute amounts of the internal standards, as described above in Example 1. These molar amounts were then used to determine the coefficients in Table 56B for Models 1, 3, 4, and 5. Because these coefficients are in units of molar amounts, they can be used with models in which protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0370] [Table 84]

[0371] Note that the values ​​in Table 56B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0372] Panel 28 Models 1-5 were created to predict ABCA1 cholesterol withdrawal and CAD risk in panel 28. Models 1-5 for panel 28 are shown below: Panel 28 Models 1-5 = Predicted ABCA1 cholesterol efflux = (i) + (c1) * ApoA1+(c2) * ApoA2+(c3) * ApoC1+(c4) * ApoC2+(c5) * ApoC3+(c6) * ApoC4+(c7) * ApoD+(c8) * ApoE+(c9) * ApoL1+(c10) * ApoM+(c11) * C3+(c12) *CLU+(c13) * HP+(c14) * SAA1 / 2+(c15) * SAA4 Values ​​for Models 1 to 5 are shown in Table 57 below.

[0373] [Table 85] JPEG2024178265000090.jpg94169

[0374] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 58A.

[0375] [Table 86]

[0376] The results in Table 58A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0377] The absolute molar amounts of each protein were determined using the absolute amounts of the internal standards, as described above in Example 1. These molar amounts were then used to determine the coefficients in Table 58B for Models 1, 3, 4, and 5. Because these coefficients are in units of molar amounts, they can be used with models in which protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0378] [Table 87] JPEG2024178265000093.jpg69169

[0379] Note that the values ​​in Table 58B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0380] Panel 29 Models 1-5 were created to predict ABCA1 cholesterol withdrawal and CAD risk in panel 29. Models 1-5 for panel 29 are shown below: Panel 29 Models 1-5 = Predicted ABCA1 cholesterol efflux = (i) + (c1) * ApoA1+(c2) * ApoC3+(c3) * ApoD+(c4) * ApoE+(c5) * ApoL1+(c6) * ApoM+(c7) * HP+(c8) * PLTP+(c9) * PON1+(c10) * SAA1 / 2 Values ​​for Models 1 to 5 are shown in Table 59 below.

[0381] [Table 88]

[0382] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 60A.

[0383] [Table 89]

[0384] The results of Table 60A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0385] The absolute molar amounts of each protein were determined using the absolute amounts of the internal standards, as described above in Example 1. These molar amounts were then used to determine the coefficients in Table 60B for Models 1, 3, 4, and 5. Because these coefficients are in units of molar amounts, they can be used with models in which protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0386] [Table 90]

[0387] Note that the values ​​in Table 60B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0388] Panel 30 Models 1-5 were created to predict ABCA1 cholesterol withdrawal and CAD risk in panel 30. Models 1-5 for panel 30 are shown below: Panel 30 Models 1-5 = Predicted ABCA1 cholesterol efflux = (i) + (c1) * ApoA1+(c2) * ApoA2+(c3) * ApoC3+(c4) * ApoC4+(c5) * ApoD+(c6) * ApoE+(c7) * ApoL1+(c8) * ApoM+(c9) * C3+(c10) * HP+(c11) * PLTP+(c12) * PON1+(c13) * SAA1 / 2 Values ​​for models 1 to 5 are shown in Table 61 below.

[0389] [Table 91] JPEG2024178265000098.jpg57169

[0390] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 62A.

[0391] [Table 92]

[0392] The results in Table 62A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0393] The absolute molar amounts of each protein were determined using the absolute amounts of the internal standards as described above in Example 1. These molar amounts were then used to determine the coefficients in Table 62B for Models 1, 3, 4, and 5. Because these coefficients are in units of molar amounts, they can be used with models where protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0394] [Table 93] JPEG2024178265000101.jpg40169

[0395] Note that the values ​​in Table 62B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0396] Panel 10 Models 1-5 were created to predict ABCA1 cholesterol withdrawal and CAD risk in panel 10. Models 1-5 for panel 10 are shown below: Panel 10 Models 1-5 - Predicted ABCA1 cholesterol efflux = (i) + (c1) * ApoC3 Values ​​for models 1 to 5 are shown in Table 63 below.

[0397] [Table 94]

[0398] The predicted cholesterol withdrawal and CAD risk calculated based on the five models and three specific patients are shown below in Table 64A.

[0399] [Table 95]

[0400] The results in Table 64A can be used to derive the probability of CAD based on the following formula: Probability = 1 / (1 + exp(-Risk Score)).

[0401] The absolute molar amounts of each protein were determined using the absolute amounts of the internal standards, as described above in Example 1. These molar amounts were then used to determine the coefficients in Table 64B for Models 1, 3, 4, and 5. Because these coefficients are in units of molar amounts, they can be used with models in which protein values ​​are determined by any type of detection assay (e.g., ELISA, mass spectrometry, etc.).

[0402] [Table 96]

[0403] Note that the values ​​in Table 64B also include the number of 95% confidence interval ranges. For each value in the formula (model), this range number can be substituted into the formula.

[0404] Example 3 Apolipoprotein AI-related proteome is associated with cholesterol efflux capacity and coronary artery disease In this example, we discussed the ApoA-I-associated serum proteome and its relationship to HDL function in relation to cholesterol efflux. Furthermore, we attempted to develop a method for predicting cholesterol efflux capacity (CEC). His-tagged recombinant ApoA-I-associated serum protein fractions were subjected to data-dependent proteomic analysis to develop a targeted quantitative proteomic method for 21 proteins associated with reverse cholesterol transport and / or coronary artery disease (CAD). This targeted method was compared with cell-based CEC measurements (N=105) to derive a proteomic prediction algorithm, which was evaluated in a case / control study of healthy specimens (N=74) as well as CAD specimens with major adverse cardiovascular events (N=83) and CAD specimens without major adverse cardiovascular events (N=74). Significant commonalities were found between the ApoA-I-associated serum proteome and proteins observed in HDL. A proteomic prediction algorithm consisting of ApoA-II, ApoC-I, ApoC-II, ApoC-III, ApoD, and serum amyloid A (panel 19) correlated strongly (R = 0.77) with cell-based CEC assay results. Proteomic predicted CEC measurements in the case / control cohort showed a significant inverse correlation between CEC and CAD diagnosis (P = 0.0032), with a significant decrease in CEC observed between controls and CAD samples without MACE (P = 0.04), with a further decrease in CEC observed in CAD samples with MACE.

[0405] Materials and Methods All study methods were approved by local Institutional Review Boards where necessary. All reagents were purchased from Sigma-Aldrich (St. Louis, MO) or ThermoFisher of the highest grade available unless otherwise stated. Specimens used in this study were obtained from Golden West biologicals, which supplied pooled serum samples. Lipoprotein-free serum prepared by immunoaffinity depletion was purchased from GenwayBio (San Diego, CA). In addition, remnant samples obtained from Cleveland HeartLab were used. Clinical specimens representing cases of coronary artery disease and controls were obtained from the Fairbanks Institute for Healthy Communities.

[0406] Enrichment of ApoA-I-associated serum fractions Recombinantly expressed and purified 0.5mg / mL 15 N-labeled His6-tagged apolipoprotein AI ( 15 24 μL of a 1× PBS (pH 7.4) solution of N-His6ApoA-I (Genscript, Piscataway, NJ) was added to 12 μL of human serum. 15The N-His6ApoA-I / serum mixture was incubated for 30 min at 37°C. After incubation, samples were diluted to a total volume of 200 μL with 5 mm imidazole, 20 mM sodium phosphate, 150 mM sodium chloride, pH 8.0, according to the manufacturer's protocol, and then purified using a Tecan Freedom Evo automated liquid handler (Tecan, Mannedorf, Switzerland) equipped with a PhyTip (Phynexus, San Jose, CA) packed with 5 μL of Ni-NTA HisBind Superflow stationary phase. Briefly, the diluted sample was slowly bound to the Phytip column at 250 μL / min using repeated pipetting cycles, followed by a subsequent wash with 300 μL of 20 mM imidazole, 20 mM sodium phosphate, 150 mM sodium chloride, pH 8.0. Bound His6-ApoA-I and related species were then eluted with 300 mM imidazole, 50 mM Tris-HCl, pH 9.0, 25% methanol. After elution, samples were either used immediately for subsequent analyses or stored at -80°C until required.

[0407] Data-dependent proteomic analysis by nanoLC-MS Dithiothreitol was added to the concentrated ApoA-I-associated serum fraction to a final concentration of 5 mM in addition to 500 ng of endoproteinase LysC (Wako Chemicals USA, Richmond, VA). Samples were digested in an Eppendorf PCR thermal cycler at 37°C for 4 hours and then held at 4°C to stop endoproteinase activity. The resulting peptide mixture was separated using a Waters nanoACQUITY ultra-high pressure LC system (Waters, Inc., Milford, MA). A 10 μL volume containing 500 ng of total peptide material was injected and captured on a reversed-phase Symmetry C18 capture column (180 μm i.d. × 20 mm, 5 μm particles) (Waters, Inc) using mobile phase A (0.1% formic acid, 2% acetonitrile in water) at 10 μL / min for 4 minutes and washed. Peptides were then eluted at 300 nL / min on a nanoACQUITY column (75 μm i.d. × 250 mm) packed with a 1.7 μm BEH130 C18 stationary phase using a gradient of 2%–50% mobile phase B (0.1% formic acid, 2% acetonitrile in water) for 210 min, followed by an increase to 100% mobile phase B for 15 min, followed by re-equilibration with 2% mobile phase B for 15 min. Peptides were detected on a LTQ-Orbitrap Velos mass spectrometer (Thermo Fisher Scientific, Waltham, MA) using a precursor ion scan from m / z 350–1800 in an orbitrap with a resolution of 120,000, followed by ion trapping with collision-induced dissociation for 25 data-dependent MS / MS events. Precursors were separated with a selection window of 2.5 m / z and fragmented at 35% normalized collision energy for 20 ms. Monoisotopic precursor selection was allowed, and precursors with 1+ and empty charge states were excluded from MS / MS selection. Dynamic precursor exclusion with a mass exclusion window of plus or minus 10 ppm was used for 60 seconds. Automatic gain control limits for the ion trap and orbitrap were set to 1 × 10 respectively. 4 and 1×10 6 The raw mass spectrometry data files were analyzed using the integrated Andromeda search algorithm. 15Using MaxQuant (version 1.4.1.2) 14 Proteins were identified by processing the samples with 100% ethanol. In addition to the general contaminant database, a search was performed against the UniProt Human Database (downloaded from the European Bioinformatics Institute on October 14, 2013, consisting of 88,304 entries). The search was performed with a precursor mass tolerance of ±7 ppm and a fragment mass tolerance of 0.6 Da. LysC was set as the cleavage enzyme, and peptides containing at least six amino acids and up to two cleavage errors were analyzed. Methionine oxidation was allowed as a variable modification. The false positive rate for proteins and peptides was set to 1%. Protein identification was further refined to ensure that at least two peptides were identified.

[0408] Targeted Proteome Analysis A reducing agent (5 μL of 100 mM dithiothreitol) and a protease (10 μL of 50 ng / μL endoproteinase LysC (Wako)) were added to 85 μL of the concentrated ApoA-I-associated serum fraction and incubated at 37° C. for 4 hours, at which point the temperature had been lowered to 4° C. 13 C6, 15 5 μL of a defined mixture of N2-lysine-labeled internal standard peptides was added to 75 μL of protein digest, followed by 25 μL injection for analysis by liquid chromatography multiple reaction monitoring mass spectrometry (LC-MRM). The injected sample was loaded and washed onto the column for 1.25 min, and then eluted at 500 μL / min with a linear gradient of mobile phase B. Peptides were detected using an Agilent 6490 triple quadrupole mass spectrometer operated in dynamic MRM mode, which allows targeted detection of peptide targets within a scheduled retention window. Transitions were selected, optimized, and determined to be specific to the targeted peptides in the sample. Two transitions were monitored per peptide, and up to two peptides per protein. A detailed list of peptide targets and their transitions is provided in Table 1 above, and for panel 19 in Table 67 below.

[0409]

Table 97

[0410] The peptide signal intensity was obtained by integrating chromatographic peaks for quantifier transitions using MassHunter Quantitative Analysis software (Agilent). All peaks were manually reviewed using the quantifier ion ratio and internal standard peaks.

[0411] Isolation of HDL by ultracentrifugation HDL (1.063 g / mL < p < 1.21 g / mL) derived from pooled serum was isolated with some modifications to Brewer's method. 16 。

[0412] Evaluation of cholesterol efflux capacity by cell lines Human serum samples were depleted of LDL, and an assay of a cell line that measures the efflux of [H]-labeled cholesterol from J774 macrophages was outsourced to Vascular Strategies, Inc. (Plymouth Meeting, PA) using the method described by de la Llera-Moya. All measurements were reported with the efflux values normalized. 3 H-labeled cholesterol efflux from 17 The serum samples for developing the extraction correlation model were collected from unidentified remnant specimens of CHL collected in two batches with a 6-week interval to prepare training and test sets, respectively. Quantitative analysis of LDL-c, HDL-c, ApoA, ApoB, triglyceride, and high-sensitivity C-reactive protein (hsCRP) was used as a selection guideline for candidate samples for each set, and finally a suitable set of specimens was obtained (Table 65).

[0413] Specimens for CEC algorithm development The serum samples for developing the extraction correlation model were collected from unidentified remnant specimens of CHL collected in two batches with a 6-week interval to prepare training and test sets, respectively. Quantitative analysis of LDL-c, HDL-c, ApoA, ApoB, triglyceride, and high-sensitivity C-reactive protein (hsCRP) was used as a selection guideline for candidate samples for each set, and finally a suitable set of specimens was obtained (Table 65).

[0414]

Table 98

[0415] Quantitative ApoA-I-related proteomic analyses and CEC measurements were collected for 105 specimens with a highly characterized serum pool as a quality control material.

[0416] Clinical specimens Samples were selected from the Fairbanks Institute for Healthy Communities biobank, which consisted of serum samples from 1500 men and women aged 22-87 years, including 750 subjects with coronary angiographic evidence of CAD (≥50% obstruction) and 750 control subjects without evidence of CAD, positive stress test, diabetes, hypertension, or dyslipidemia (LDL-C ≥130 mg / dL, HDL-C ≤40 mg / dL, total cholesterol ≥240 mg / dL, or triglycerides ≥200 mg / dL). Fasting blood samples were collected according to study SOPs and stored at -80°C. Subjects with a diagnosis of CAD were evaluated to determine two groups: cases and cases with events. All CAD patients were classified according to ICD-9 codes defining major adverse cardiovascular events (MACE) as myocardial infarction (410), coronary artery bypass graft or angioplasty (36.1, 45.82), or stroke (434.91). If myocardial infarction was confirmed by ICD9 classification, records were reviewed to select patients with two of the following three conditions: history of ischemic pain, abnormal ECG, or abnormal troponin. In total, 74 CAD subjects without events and 83 CAD subjects with MACE events were selected (Table 66).

[0417] [Table 99]

[0418] A group of 74 matched controls was also selected.

[0419] Targeting algorithm development Prior to computational analysis, the peak area abundance of each target peptide was used in the enrichment process. 15Normalization was performed with the intensity of N-His6ApoA-I. For proteins with more than one peptide measured by LC-MS / MS, the peptide with the highest intensity in the majority of samples analyzed was used to establish the relative abundance of the protein. Protein-level data were used for statistical analysis. An analytical pipeline involving a series of sequential steps for feature selection was applied to the normalized protein-level data to search for proteins associated with CECs. 18 In univariate analysis, robust linear regression was applied to each protein to predict CEC for 70 training samples, and proteins with p-values ​​<0.1 were selected. Further filtering was applied using elastic net model. In multivariate analysis, a biomarker panel was constructed with the filtered proteins using partial least squares regression. To evaluate the performance of the biomarker panel, the Spearman correlation and median absolute deviation (cost) between predicted and measured CEC were calculated. First, the biomarker panel was created using uncalibrated LC-MS peak intensities. The proteomic method was later refined to obtain absolute molar amounts after calibration, and the panel coefficients were refined with the calibrated MRM data using partial least squares regression. Finally, the panel was tested on 35 validation samples. To evaluate whether the CEC predictive biomarker panel can distinguish subjects with and without CAD, biomarker proteins were tested in 74 healthy controls and 157 CAD patients. Statistical analysis of the data was performed using R version 3.2.3 (R Core Team (2013). R: A language and environment for statistical computing, R Foundation for Statistical Computing, Vienna, Austria). Comparison of cumulative distributions of patient cohorts was performed using the Kolmogorov-Smirnov (KS) test.

[0420] result: Rapid preparation of ApoA-I affinity pools To facilitate accurate and high-throughput analysis of lipid-free ApoA-I-associated proteins using mass spectrometry-based proteomics, we devised a strategy for rapid affinity enrichment. This strategy used His-tagged ApoA-I and immobilized metal affinity chromatography (Figure 2). Recombinant lipid-free ApoA-I with an N-terminal polyhistidine tag (His6ApoA-I) was spiked into human serum and incubated at 37 °C. The sample was then passed over a nickel-NTA stationary phase to bind His6ApoA-I and associated proteins. After washing the stationary phase with a low concentration of imidazole to suppress nonspecifically bound proteins, the complexed proteins and excess His6ApoA-I are eluted into a desired recovery buffer containing a high concentration of imidazole. The eluted sample was then subjected to a standard enzymatic digestion workflow for eventual analysis by liquid chromatography-mass spectrometry (LC-MS / MS). It allows specific measurement of exogenous ApoA-I, distinguishing it from endogenous proteins belonging to the analyte being analyzed. 15 N-labeled His6ApoA-I( 15 N-His6ApoA-I) was used.

[0421] Data-driven proteomic analysis of lipid-free ApoA-I affinity pools As revealed by earlier data-driven experiments, the majority of proteins identified in the lipid-free ApoA-I affinity fraction are found to be HDL-associated. We attempted to further clarify this observation by a qualitative proteomic comparison of HDL isolated using conventional ultracentrifugation and the ApoA-I-associated serum fraction using a commercial serum pool of 400 male donors. Three replicate steps were performed per each enrichment technique. Overall, 91 proteins were identified in HDL purified by ultracentrifugation and 162 proteins were identified in the lipid-free ApoA-I affinity pool, of which 78 proteins were common to both preparation methods (see Figure 3).

[0422] We compared the proteins identified in our experiments with the HDL Proteome Watch list, which summarizes the results of 17 publicly available HDL proteome studies. This list contains 229 proteins, of which 95 proteins have been identified in three or more studies or validated by other molecular biology methods. Of the total 229 proteins in the Proteome Watch list, 80 were identified in HDL prepared by ultracentrifugation, 86 were identified in our affinity-purified pool, and 72 (excluding common contaminants such as keratins) were common to both preparation methods (77% common) (Figure 3).

[0423] Targeted quantification by multiple reaction monitoring mass spectrometry To investigate the relationship between ApoA-I affinity pool and cholesterol efflux capacity (CEC), we attempted to use an accurate quantification method based on liquid chromatography multiple reaction monitoring (LC-MRM). Based on a literature search and our data-dependent LC-MS results, 21 proteins known to be associated with lipid transport, reverse cholesterol transport and / or cardiovascular disease were selected for development, including lipid metabolism (apolipoproteins AI, A-II, A-IV, CI, C-II, C-III, C-IV, D, E, F, J, LI, M), enzymes (phospholipid transfer protein-PLTP, cholesteryl ester transfer protein-CETP, lecithin cholesterol acyltransferase-LCAT, paraoxonase 1-PON1), and acute phase response proteins (complement C3, haptoglobin, serum amyloid A1 and 2-SAA1 / 2, and SAA4). Two peptides (when possible) from each protein for assay development were identified and the entire workflow was optimized. Digestion conditions were obtained that produced stable peptide abundances for all proteins within 4 h. Recovery of native ApoA-I from serum stock was estimated to be approximately 55%, which correlated strongly with ApoA-I levels in serum determined by immunoassay (Pearson r = 0.84, Supplementary Figure 6). In comparison, lipid-free 15The average recovery of N-His6ApoA-I was 85% from PBS buffer and 72% from solvent-stripped serum.

[0424] Development and validation of a multivariate algorithm for CEC prediction Using lipid-free ApoA-I affinity enrichment technology, a series of serum samples were analyzed to determine whether associations exist between affinity-enriched proteins and CECs. A set of 70 training test samples and 35 independent test samples (Table 65) were randomly selected without regard to disease diagnosis, but with careful consideration of concordance with lipoprotein measurements. For each specimen, normalized CECs and normalized mass spectrometry data for the complement of 21 proteins were determined. A multi-step informatics pipeline including univariate and multivariate statistical methods was performed to determine whether relative protein abundance correlates with CECs in cell lines. 18 A biomarker panel was developed that correlates with IFN-α. After univariate analysis, 9 proteins (apolipoproteins AI, A-II, CI, C-II, C-III, C-IV, D, CETP, and SAA) with p-values ​​<0.1 were identified by robust linear regression. In a subsequent elastic net regression, 6 proteins (ApoA-II, ApoC-I, ApoC-II, ApoC-III, ApoD, and SAA) were selected and subjected to partial linear regression to determine the final prediction model.

[0425] The prediction panel performed well when accounting for measurement error in the CEC measurements (training set, Spearman r = 0.63, p < 0.001; validation set, Spearman r = 0.70, P < 0.001). The performance of the proteomic prediction CEC algorithm within the full development sample set is shown in Figure 4A (Pearson r = 0.77). Associations between CECs determined by cell line assays and proteomic estimates and other clinical measurements (total cholesterol, HDL-c, LDL-c, non-HDL-c, triglycerides, ApoA-I, ApoB, and hsCRP) were examined. The most significant associations with cell line CECs were found in previous reports. 11、12Similarly, CEC was significantly negatively correlated with hsCRP (Pearson r = -0.23, Fig. 4D). Predicted CEC behaved similarly with total cholesterol, but the relationships with ApoA-I and HDL-c were not calculated because ApoA-I was included in the proteomic panel.

[0426] Predicted CEC is inversely correlated with CAD Hypothesizing that predicted CEC should be inversely correlated with cardiovascular disease similar to reports of CEC measurements in cell lines, we used our multivariate panel to predict CEC for 157 CAD patients (74 without major adverse cardiovascular events (MACE) and 83 with MACE) and 74 age- and sex-matched apparently healthy controls (Table 66).

[0427] The distribution of predicted CEC among these cohorts is shown in Figure 5A. We found that CAD patients had a lower median predicted CEC (9.54% withdrawal / 4 hours; controls 9.98% withdrawal / 4 hours; p=0.0032). When CAD cases are divided into those without MACE and those without specific adverse events, a significantly lower predicted CEC is observed in CAD patients without MACE compared to controls (P=0.04). CAD patients who underwent revascularization (n=29) showed a lower predicted CEC than CAD patients without MACE (P=0.027, P=0.0003 compared to controls). CAD patients who had experienced a myocardial infarction (n=38) or stroke (n=16) showed a lower median predicted CEC, but did not differ significantly from CAD patients without MACE. Comparison of CAD samples with age- and sex-matched controls was repeated after 6 weeks and demonstrated support for the trends observed in the initial analysis. We also performed a comparison of the performance of predicted CECs with cell-based CEC measurements in a small number of selected samples with clinical outcomes. After identifying the 15 highest and lowest predicted CEC values ​​in the control and CAD cohorts, respectively, cell-based CEC measurements were obtained. As shown in Supplementary Figure 7, a subset of CAD specimens had significantly lower CECs whether determined by proteomic prediction or cell-based CEC measurements (P value < 0.001). Furthermore, when comparing predicted values ​​with cell-based measurements in subsets with the same outcomes, no significant differences were evident (P = 0.39 for controls and P = 0.08 for CAD).

[0428] It is clear that cell-based assays for determining CECs are effective in investigating cholesterol metabolism, and a growing number of studies have demonstrated that serum-based CEC measurements are independent predictors of cardiovascular risk.Unfortunately, this method is labor-intensive, has complex workflows, and is relatively inaccurate, which hinders the translation of this technology into clinical testing, where scale, throughput, precision, calibration, and ultimately analytical validation are required.

[0429] Performance of proteome prediction CEC The performance of the proteomic predictive CEC on the test set (r=0.71) was promising, suggesting an important role for proteins other than ApoA-I in regulating cholesterol efflux. A recent clinical study of CEC performing modified cell line assays with fluorescently labeled cholesterol showed significant predictive value (R=0.54) that was somewhat consistent with normalized efflux using the predicated radioisotope assay. 12 Other correlations with withdrawal have been confirmed. As expected, most studies have reported positive univariate correlations with ApoA-I in the range of r = 0.22–0.64 (r = 0.57 in this study). 11、19 It was demonstrated that the sphingomyelin / phosphorethanolamine ratio was significantly associated with withdrawal (r=0.64) in high HDL-C (>68 mg / dL) samples, but not in normal HDL-C levels. Withdrawal is also known to be dependent on HDL subclass, further suggesting the complexity of a series of structural and functional interactions. 20 Although these variations remain to be fully understood, it is possible that additional biochemical and biophysical data can be integrated into models to improve CEC predictions.

[0430] Mechanistic insights into CEC prediction panel proteins The proteomic predictor of CECs obtained in our study consists of proteins mechanistically related to cholesterol transport (apolipoproteins A-II, CI, C-II, C-III, D, and SAA; panel 19). Among the proteins positively correlated with efflux, ApoA-I and ApoA-II function as the primary receptors for cellular cholesterol. 1、21 Similarly, ApoC-I also activates lecithin-cholesterol acyltransferase LCAT, which is a minor but efficient cholesterol acceptor and plays a role in the maturation of HDL. 22 ApoC-II and ApoC-III regulate lipoprotein lipase, which is involved in HDL remodeling, and ApoD plays a role in the formation of pre-β3-HDL, the potent cellular cholesterol-accepting form of HDL. 22、23Serum amyloid A (SAA) is an acute phase response protein that has been extensively characterized to negatively affect cholesterol efflux by displacing ApoA-I from HDL. 24、25 We also noted that the functional role of SAA in regulating efflux is reflected in the negative correlation between efflux and hsCRP, another biomarker commonly used to measure systemic inflammation (although we did not identify the relevance of ApoA-I in this study). Interestingly, the interactions of lipid-free ApoA-I with serum proteins elucidated by proteomic analysis show a high degree of overlap with the canonical HDL proteome classified in the Proteome Watch list. References for this example: 1. Kontush A, Chapman MJ. High-Density Lipoproteins: Structure, Metabolism, Function, and Therapeutics. John Wiley & Sons, Inc;2012. 2.Drew et al.,High-density lipoprotein and apolipoprotein AI increase endothelial NO synthase activity by protein association and multisite phosphorylation.Proc Natl Acad Sci.2004;101(18):6999-7004. 3.Ansell et al.Inflammatory / Antiinflammatory Properties of High-Density Lipoprotein Distinguish Patients from Control Subjects Better Than High-Density Lipoprotein Cholesterol Levels and Are Favorably Affected by Simvastatin Treatment.Circulation.2003;108(22):2751-2756. 4.Gaidukov L,Tawfik DS.High Affinity,Stability,and Lactonase Activity of Serum Paraoxonase PON1 Anchored on HDL with ApoA-I.Biochemistry.2005;44(35):11843-11854. 5.Huang et al.Myeloperoxidase,paraoxonase-1,and HDL form a functional ternary complex.J Clin Invest.2013;123(9):3815-3828. 6.Cavigiolio et al.,Exchange of apolipoprotein A-I between lipid-associated and lipid-free states: a potential target for oxidative generation of dysfunctional high density lipoproteins.J Biol Chem.2010;285(24):18847-18857. 7.Fredolini et al.Immunocapture strategies in translational proteomics.Expert Rev Proteomics.2016;13(1):83-98. 8.Gavin et al.Functional organization of the yeast proteome by systematic analysis of protein complexes.Nature.2002;415(6868):141-147. 9.Gundry et al.,Investigation of an albumin-enriched fraction of human serum and its albuminome.Proteomics-Clin Appl.2007;1(1):73-88. 10.Poulsen et al.,Serum Amyloid P Component(SAP)Interactome in Human Plasma Containing Physiological Calcium Levels.Biochemistry.2017;56(6):896-902. 11.Khera,et al.Cholesterol efflux capacity,high-density lipoprotein function,and atherosclerosis.N Engl J Med.2011;364(2):127-135. 12.Rohatgi et al.HDL cholesterol efflux capacity and incident cardiovascular events.N Engl J Med.2014;317(25):2383-2393.13.Mody et al.,Beyond Coronary Calcification,Family History,and C-Reactive Protein.J Am Coll Cardiol.2016;67(21):2480-2487. 14.Cox J,Mann M.MaxQuant enables high peptide identification rates,individualized p.p.b.-range mass accuracies and proteome-wide protein quantification.Nat Biotechnol.2008;26(12):1367-1372. 15.Cox et al.,a peptide search engine integrated into the MaxQuant environment.J Proteome Res.2011;10(4):1794-1805. 16.Duverger et al.,Biochemical characterization of the three major subclasses of lipoprotein A-I preparatively isolated from human plasma.Biochemistry.1993;32(46):12372-12379. 17.de la Llera-Moya et al.,The ability to promote efflux via ABCA1 determines the capacity of serum specimens with similar high-density lipoprotein cholesterol to remove cholesterol from macrophages.Arterioscler Thromb Vasc Biol.2010;30(4):796-801. 18.Sin et al.,Biomarker Development for Chronic Obstructive Pulmonary Disease.From Discovery to Clinical Implementation.Am J Respir Crit Care Med.2015;192(10):1162-1170. 19.Saleheen et al.Association of HDL cholesterol efflux capacity with incident coronary heart disease events: a prospective case-control study.Lancet Diabetes & Endocrinol.2015;3(7):507-513. 20.Du et al.HDL particle size is a critical determinant of ABCA1-mediated macrophage cellular cholesterol export.Circ Res.2015;116(7):1133-1142. 21. Rotllan et al., Overexpression of Human Apolipoprotein A-II in Transgenic Mice Does Not Impair Macrophage-Specific Reverse Cholesterol Transport In Vivo. Arterioscler Thromb Vasc Biol. 2005;25(9). 22. von Eckardstein A, Nofer J-R, Assmann G. High Density Lipoproteins and Arteriosclerosis: Role of Cholesterol Efflux and Reverse Cholesterol Transport. Arterioscler Thromb Vasc Biol. 2001;21(1):13-27. 23. Francone OL, Fielding CJ. Initial steps in reverse cholesterol transport: the role of short-lived cholesterol acceptors. Eur Heart J. 1990;11(suppl E). 24. Vaisar et al. Inflammatory remodeling of the HDL proteome impairs cholesterol efflux capacity. J Lipid Res. 2015;56(8):1519-1530. 25. Banka et al., Serum amyloid A(SAA): influence on HDL-mediated cellular cholesterol efflux. J Lipid Res. 1995;36(5):1058-1065. Example 4 In this example, panel 18 (ApoA1, ApoC1, ApoC2, ApoC3, and ApoC4) was applied to samples from the Fairbanks Institute for Healthy Communities and a subset of the Abcodia Biobank.

[0431] The Fairbanks Institute for Healthy Communities (Fairbanks Institute) is creating an extensive annotated biospecimen repository for hypothesis-driven research. These biospecimens are linked to longitudinal clinical and disease-specific health record information extracted from the Indiana Network for Patient Care (INPC), a statewide electronic clinical data repository. The Fairbanks Institute CAD collection used includes serum and plasma from individuals diagnosed with coronary artery disease, as well as matched controls. Using this sample set, a cholesterol efflux prediction model based on molar protein abundance measured by mass spectrometry (Table 37B, Model 1) was applied to a sample set consisting of previously healthy (control) patients (n=74) or patients with a history of a coronary artery disease (CAD) diagnosis (n=157). As shown in Figure 8, when comparing predicted cholesterol efflux values ​​between each cohort, it is clear that cholesterol efflux is overall lower in the CAD cohort compared to the control cohort (P=0.03 by Mann-Whitney test). Running a logistic regression on the protein components of panel 18 to optimize the branching of clinical outcomes (control vs. CAD) results in the parameters of Table 37, Model 3. We confirm that when applied to the Fairbanks clinical set, it improves the stratification of patient cohorts based on outcomes. As shown in Figure 9, application of the outcome optimization model stratifies patients on a biomarker score scale that correlates with the risk of being diagnosed with CAD. Here, the CAD cohort shows significantly higher biomarker scores compared to the control group (P<0.0001 by Mann-Whitney test).

[0432] We also applied the Panel 18 model to a subset of the Abcodia Biobank. This subset consisted of 200,000 postmenopausal women aged 50-74 years, of whom 50,000 were sampled longitudinally over several years. In this study, control patients were matched to cases with confirmed nonfatal myocardial infarction on sampling location, age, and sampling date and time. Both Models 1 and 3 (Table 37B) show persistent and statistically significant divergence between these cohorts over the range 0-8 years prior to the event.

[0433] [Table 100]

[0434] All publications and patents mentioned in this application are incorporated herein by reference. Various modifications and variations of the described methods and compositions of the invention will be apparent to those skilled in the art that do not depart from the scope and spirit of the invention. Although the invention has been described in connection with certain preferred embodiments, it should be understood that the invention as claimed should not be unduly limited to such specific embodiments. Indeed, various modifications of the described embodiments of the invention that are apparent to those skilled in the relevant art are intended to be within the scope of the appended claims.

Claims

1. 1. A composition comprising a detectably labeled ApoA1 protein or a fragment thereof for use in a method for determining an estimated risk of cardiovascular disease (CVD) in a subject, an estimated cholesterol efflux capacity (CEC) of a sample from a subject, or a combination thereof, said method comprising: a) detecting a level of one or more HDL associated proteins in a sample, wherein said one or more HDL associated proteins comprise ApoA1 protein or a fragment thereof; and b) determining an estimated risk of CVD in the subject, an estimated CEC for the sample, or a combination thereof, by generating a CVD risk score and optionally a CEC score using a first algorithm, wherein the first algorithm performs operations including: i) multiplying each HDL-associated protein level by a predetermined coefficient to generate a multiplied value; and ii) summing the multiplied values ​​and adding a panel-specific constant value, thereby generating the CVD risk score and optionally the CEC score.

2. 2. The composition for use of claim 1, wherein step b) further comprises generating a probability of CVD using a second algorithm, said second algorithm applying said CVD risk score to the following formula: CVD Probability = 1 / (1 + exp(-Risk Score)), and optionally further comprising multiplying said CVD probability by 100 to obtain a risk percentage of CVD.

3. 3. The composition for use according to claim 2, wherein a risk rate of CVD less than 10% is considered low risk, a risk rate of CVD between 10% and 20% is considered intermediate risk, and a risk rate of CVD greater than 20% is considered high risk.

4. The composition for use according to any one of claims 1 to 3, further comprising: c) generating a report indicating said CVD risk score and optionally said CEC score.

5. 4. The composition for use according to any one of claims 1 to 3, wherein the one or more HDL associated proteins further comprise an HDL associated protein selected from the group consisting of apolipoprotein (Apo) C1, ApoC2, ApoC3, ApoC4, ApoD, ApoE, ApoL1, ApoM, CLU, HP, phospholipid transfer protein (PLTP), PON1, SAA1, and SAA2, or fragments thereof.

6. The composition for use according to any one of claims 1 to 3, wherein said one or more HDL associated proteins are proteins listed in biomarker panels 1 to 30.

7. 2. The composition for use according to claim 1, wherein step a) comprises detecting the levels of two or more HDL-associated proteins.

8. 2. The composition for use of claim 1, wherein step a) comprises adding a reagent to the sample, said reagent digesting said one or more HDL-associated proteins in said sample.

9. The composition for use according to claim 1 , wherein step a) comprises mass spectrometry or immunoassay.

10. The composition for use according to any one of claims 1 to 3, wherein the sample is selected from the group consisting of a serum sample, a plasma sample, a blood sample, and a purified high density lipoprotein sample.

11. After step a) and before step b), normalizing the level of said one or more HDL associated proteins to the estimated level of total HDL particles, or ApoA1, or HDL cholesterol to generate a normalized value for at least one HDL associated protein. The composition for use according to any one of claims 1 to 3, further comprising:

12. The composition for use of claim 11, wherein the estimated level of total HDL particles, or ApoA1, or HDL cholesterol in the sample is determined by measuring the level of an internal standard added to the sample, the internal standard comprising a labeled HDL-associated protein.

13. The composition for use according to any one of claims 1 to 3, wherein step a) comprises adding an internal standard to the sample and detecting said internal standard, said internal standard being a labeled or unlabeled HDL-associated protein.

14. 13. The composition for use of claim 12, wherein the labeled HDL associated protein is selected from the group consisting of Apo1, ApoC1, ApoC3, ApoD, ApoE, ApoL1, ApoM, PLTP, SAA1, and SAA2.

15. The method of claim 1 , wherein the first algorithm comprises a logistic regression process.

Images