Skip main navigation

Plasma Proteomics for Epidemiology

Increasing Throughput With Standard-Flow Rates
Originally published Cardiovascular Genetics. 2017;10:e001808



    Mass spectrometry is selective and sensitive, permitting routine quantification of multiple plasma proteins. However, commonly used nanoflow liquid chromatography (LC) approaches hamper sample throughput, reproducibility, and robustness. For this reason, most publications using plasma proteomics to date are small in study size.

    Methods and Results—

    Here, we tested a standard-flow LC mass spectrometry (MS) method using multiple reaction monitoring for the application to large epidemiological cohorts. We have reduced the LC-MS run time to almost a third of the nanoflow LC-MS approach. On the basis of a comparison of the quantification of 100 plasma proteins in >1500 LC-MS runs, the SD range of the retention time during continuous operation was substantially lower with the standard-flow LC-MS (<0.05 minutes) compared with the nanoflow LC-MS method (0.26–0.44 minutes). In addition, the standard-flow LC method also offered less variation in protein measurements. However, 5× more sample volume was required to achieve similar sensitivity. Two different commercial multiple reaction monitoring kits and an antibody-based multiplexing kit were used to compare the apolipoprotein measurements in a subset of samples. In general, good agreement was observed between the 2 multiple reaction monitoring kits, but some of the multiple reaction monitoring–based measurements differed from antibody-based assays.


    The multiplexing capability of LC-MS combined with a standard-flow method increases throughput and reduces the costs of large-scale protein measurements in epidemiological cohorts, but protein rather than peptide standards will be required for defined absolute proteoform quantification.


    The plasma proteome is among the most challenging biological matrices to analyze.1 An untargeted proteomics approach results in oversampling of abundant proteins, leaving peptides from low abundant proteins undetected.2 The selectivity of multiple reaction monitoring (MRM) on the triple quadrupole (QqQ) mass spectrometer (MS) ensures that lower abundant proteins can be targeted using stable isotope-labeled standards (SIS) for absolute quantification.3,4 MRM has been used as the gold standard in small molecule quantification for decades5 and has recently been explored for measuring plasma proteins.610 Other MS approaches for plasma proteomics have also been described, including Stable Isotope Standards and Capture by Anti-Peptide Antibodies6 and data-independent acquisition,7 but these can add several steps to the work flow and increase data analysis time, limiting the application for epidemiological studies. Instead, throughput and costs are critical considerations for applications in clinical research. The latest high sensitive QqQ MS combined with high-performance liquid chromatography (HPLC) offers the potential to eliminate the need for protein enrichment or fractionation, which limits throughput and also introduces technical variability.7,1114 A recent study demonstrated robust analytic performance for MRM analyses across laboratories and instrument platforms,15 a prerequisite for the advancement of MS-based protein quantification for routine clinical applications.

    See Clinical Perspective

    Among the numerous potential biomarkers cited in the proteomics literature, only few have obtained regulatory approval.16 Several clinical studies using targeted plasma proteomics have been published, including both nanoflow8,17 and standard-flow methods,9,18,19 but the sample numbers are low (<100). Currently, the preferred high-sensitivity proteomics workflows use nanoflow LC, which is slow in nature and hampered by poor robustness during continuous operation. A previous comparison of standard-flow and nanoflow LC-MRM was limited to 30 samples.10 The need for validating the applicability of standard-flow MRM method to larger cohorts is apparent.3,20

    In this study, we have applied a standard-flow LC-MS method to quantify 100 proteins in plasma samples from the community-based, prospective Bruneck cohort (n=668 samples, >1500 injections), which is, to our knowledge, the largest targeted proteomics study on human plasma samples to date.


    The data, analytic methods, and study materials will not be made available to other researchers for purposes of reproducing the results or replicating the procedure because of the limited availability of plasma samples. However, the raw files (a subset of 44 samples) and transition list from both nanoflow and standard-flow platforms were deposited into the PeptideAtlas SRM Experiment Library (Data set Identifier: PASS00949). The MRM data are also deposited into Panorama Public: (MacCoss Laboratory Software).

    Human Plasma Samples

    The human plasma samples were from the year 2000 evaluation of the prospective, community-based Bruneck Study.21,22 The Bruneck Study protocol conformed to the Declaration of Helsinki and was approved by the Local Ethics Committee (Bolzano, Italy). All participants provided their written informed consent before entering the study. Blood samples were drawn after an overnight fast and 12 hours of abstinence from smoking. Citrate plasma samples were divided into aliquots and stored at −80°C. In this study, the plasma samples (n=668) were used for observation purposes only. The serotransferrin (TRFE) clinical measurement was conducted using a Behring BNA II nephelometer system immediately after the plasma sample collection.

    Sample Preparation Using PlasmaDive

    Plasma samples were analyzed using the PlasmaDive MRM Panel (beta test version; Biognosys AG, CH). Sample processing was semiautomated on a Bravo liquid handling system (Agilent Technologies). In brief, 10 μL of plasma was denatured, reduced, and alkylated according to the manufacturer’s instruction, followed by spiking-in of SIS peptides and an in-solution digestion overnight using trypsin (Thermo Fisher Scientific). After solid-phase extraction cleanup with a 96-well C18 spin plate (Harvard apparatus), the eluted peptides were dried using a SpeedVac (Thermo Fisher Scientific) and resuspended in 40 μL of LC solution with addition of indexed retention time (RT) peptides (Biognosys, CH) to adjust the RT variability across different LC-MS runs. A pooled sample was used as quality control (QC) sample with repeated injections during continuous operation.

    Nanoflow HPLC-MRM

    On the nanoflow platform (Thermo U3000 RSLCnano; Thermo Fisher Scientific), 2 μL of digested sample was directly injected onto a 15-cm column (Acclaim PepMap 100, C18, 15 cm×75 µm, 3 µm, 100 Å) and separated over a 32-minute gradient at 300 nL/min (0–30 min, 5%–35% B; 30–32 min, 35%–99% B; 32–40 min, 99% B; 40–70 min, 2% B; A=1% acetonitrile, 0.1% formic acid; B=97% acetonitrile, 0.1% formic acid). The nanoflow HPLC was interfaced to a TSQ Vantage MS (Thermo Fisher Scientific). Analysis was performed using the following parameters: Q1 resolution 0.7 Th, Q2 collision gas pressure 1 mTorr, Q3 resolution 0.7 Th. Eight hundred fifty transitions were exported from SpectroDive software version 6 (Biognosys AG, CH) and scheduled with a cycle time of 3 s and a RT window of 4 minutes. The transition list is shown in Table I in the Data Supplement. The SIS peptides were labeled with heavy lysine (+8 Da) or arginine (+10 Da), and the cysteines were carbamidomethylated (+57 Da).

    Standard-Flow HPLC-MRM

    The same samples were also analyzed on a standard-flow platform (Agilent 1290 Infinity II LC). Ten microliters of digest was directly injected onto a 25-cm column (AdvanceBio Peptide Mapping, C18, 2.1 mm×250 mm, 2.7um, 120 Å) and separated over a 23-minute gradient at 350 μL/min (0–0.5 min, 5%–7.5% B; 0.5–18 min, 7.5%–28% B; 18–20 min, 28%–95% B; 20–23 min, 95% B; 23–27 min, 5% B; A=0.1% formic acid in H2O; B=0.1% formic acid in acetonitrile) at 50°C. The standard-flow LC was interfaced to an Agilent 6495 QqQ MS, and both were controlled by MassHunter Workstation software (version B.08.00). Both Q1 and Q3 were set at unit resolution (0.7 Th), and the following parameters were used: Delta EMV 350, Frag 380 V, Cell Acc 4 V, Gas Temp 200°C, Gas Flow 11 L/min, Nebulizer 35 psi, Sheath Gas Heater 250°C, Sheath Gas Flow 12 L/min, Capillary 4 kV, VCharging 300, Ion Funnel Pos High-Pressure RF 180 V, and Pos Low-Pressure RF 90 V. Seven hundred sixty-five transitions were scheduled using Dynamic MRM with a cycle time of 0.5 s and a RT window of 0.8 minutes. The transition list is provided in Table II in the Data Supplement.

    PlasmaDive MRM Data Analysis

    The data were analyzed using SpectroDive, and protein concentrations were calculated using the light/heavy ratio and the concentration of spiked-in SIS peptides. The data with a Q value <0.01 were included in the final results.

    Sample Preparation for Standard Curve

    Pooled human plasma was digested using the PlasmaDive protocol (Biognosys, CH) without SIS peptides. Then 2 pools were made: pool 1, 90 µL of digested plasma +30 µL of SIS peptides; and pool 2, 450 µL of digested plasma+100 µL of LC solution (5% acetonitrile, 0.1% formic acid)+50 µL of 10x iRT standard (Biognosys, CH). Pool 1 represents 1× SIS in plasma matrix, and pool 2 is the background plasma matrix. Serial dilutions of SIS peptides were performed in plasma matrix (1:10, 1:100, 1:1000, 1:10 000, 1:100 000). The same procedure was followed to make serial dilutions of SIS peptides in LC solution (5% acetonitrile, 0.1% formic acid) only. On the nanoflow HPLC system, the injection volume was 3 µL. On the standard-flow HPLC system, the injection volume was 15 µL. The limit of detection was calculated by using the mean response of the blank samples+3×SD. The limit of quantification was calculated by using the mean response of the blank samples+10×SD.

    Enzyme-Linked Immunosorbent Assays

    The levels of platelet factor 4 (PLF4) were measured using the DuoSet ELISA Development kits (DY795) and the DuoSet Ancillary Reagent Kit 2 (DY008, R&D Systems, Minneapolis) according to the manufacturer’s instructions. Absorbance at 450 nm was measured on a Tecan Infinite 200 Pro plate reader (Tecan Group Ltd, Männedorf, Switzerland) using 570 nm as a reference wavelength. Results were calculated using a 4-parameter logistic fit.

    PeptiQuant MRM Assay Biomarker Assessment Kit (BAK-76, MRM Proteomics)

    The plasma samples were processed according to the manufacturer’s instruction. In brief, the individual plasma samples were diluted 1:10 using 25 mmol/L ammonium bicarbonate first, and pooled plasma was used as reference sample. Thirty microliters of diluted plasma sample or pooled plasma (×4 wells) was mixed with 177 μL of 25 mmol/L ammonium bicarbonate and 37 μL of 10% sodium deoxycholate to denature proteins. The proteins were reduced by 5 mmol/L tris(2-carboxyethyl) phosphine and alkylated by 10 mmol/L iodoacetamide. The remaining iodoacetamide was quenched by adding 10 mmol/L dithiothreitol. Trypsin solution (23.3 μL; 0.9 mg/mL; Affymetrix) was added and the samples incubated at 37°C for 16 hours. Six dilutions (from 0.5 to 250 fmol/μL) of the SIS peptide mixture for reference samples were prepared for the reference samples. Fifty microliters of the SIS peptide mixture for reference samples was added to 227 μL pooled reference plasma digest and 277 μL of 1% formic acid. Next, 227 μL of each plasma digest was mixed with 50 μL of 25 fmol/μL SIS peptide mixture for the experimental samples and 277 μL of 1% formic acid. After centrifugation at 12 000g for 10 minutes, 444 μL of peptide supernatant was desalted and concentrated by solid phase extraction (Oasis HLB Extraction Cartridge, Waters). The eluted peptides were freeze-dried and resuspended in 100 μL of 0.1% formic acid. Each reference sample (standard A–F) was injected 3× using 15 μL per injection, and each experimental sample was injected once using 15 μL per sample. The samples were run on the standard-flow platform (Agilent 1290 Infinity II LC with 6495 QqQ MS) and analyzed using MassHunter Quantitative Analysis (Agilent) and Qualis-SIS (University of Victoria, Canada) following the manufacturer’s instructions.

    Milliplex Map Human Apolipoprotein Magnetic Bead Panel Kit

    Apolipoproteins (apo) were measured using Milliplex Map Human Apolipoprotein Magnetic Bead Panel Kit, 96-well plate assay (Cat No. APOMAG-62K, Millipore) following the manufacturer’s instructions. In brief, plasma samples were diluted 1:4000 in the assay buffer in duplicate, and 10 μL of diluted plasma was used per well. One hundred fifty microliters of each antibody-immobilized bead for apoA1, apoA2, apoC2, apoC3, and apoE was mixed and topped up to 3 mL with Bead diluent. QC1, QC2, and the apolipoprotein calibrator cocktail were reconstituted with 250 μL deionized water and mixed thoroughly. The calibrator was further diluted to 1:5, 1:25, 1:125, 1:625, 1:3125, 1:15 625, in Assay buffer. After washing and drying the plate, 65 μL of Assay buffer, 10 μL of Assay buffer/serial dilution of calibrators/QC1/QC2/diluted samples, and 25 μL of the premixed beads were added to each well before the plate was sealed, wrapped in foil, and incubated at 700 rpm for 1 hour at room temperature. A handheld magnetic plate was used to gently remove well contents, and the plate was washed 3× using the Washing buffer. Subsequently, 50 μL of detection antibodies mixture was added to each well, and the plate was incubated at 700 rpm for 30 minutes at room temperature. The wash step was repeated before adding 50 μL of streptavidin–phycoerythrin per well for a final incubation of 30 minutes. After a third wash step, the plate was prepared for analysis on FlexMap 3D by resuspending the beads in 100 μL of sheath fluid on a plate shaker for 5 minutes. The median fluorescent intensity data were analyzed to calculate apolipoprotein concentrations in the samples. A dilution factor of 4000 was applied to the concentrations to correct for the sample dilution, and an average of the 2 technical replicates of each sample was calculated.


    Agreement of nanoflow and standard-flow measurements for all proteins measured was described graphically by overlaying 95% data ellipses on the x=y line indicating perfect agreement. Agreement of nanoflow and standard-flow measurements with independent reference measurements for 2 exemplary proteins (TRFE and PLF4) was analyzed by means of Bland–Altman plots,23 and the dependence of measurement difference on measurement mean was summarized by ordinary least squares regression lines. Both proteins were log transformed toward normality beforehand. Agreement between PlasmaDive, PeptiQuant BAK-76, and Luminex measurements was analyzed using scatter plots, Pearson correlation, and Deming regression, which in contrast to ordinary least squares regression considers error in both x and y. Comparative performance of nanoflow and standard-flow methodology in a clinical context was evaluated by comparing associations with the same clinical variables of measurements produced by either method. Significance of association was derived from Pearson correlation analysis, except for the categorical clinical variables sex and diabetes mellitus, for which Student t tests were used. Analyses were conducted using R 3.3.2.


    Plasma Proteomics in a Community-Based Cohort

    MRM-MS was applied to plasma samples of the prospective, community-based Bruneck study (n=668, year 2000 evaluation) using nanoflow LC-MS (Ultimate3000 RSLCnano interfaced to a TSQ Vantage QqQ MS; Thermo Fisher Scientific) as recommended by the manufacturer (PlasmaDive; Biognosys, CH). The raw files were analyzed using SpectroDive software, which calculated concentrations for the majority of the 100 proteins.

    When using the nanoflow LC-MS platform, each sample required a 32-minute gradient for separation of the peptides and a further 38 minutes for washing and reequilibrating the column. The total run time for the entire cohort (n=668) was 2 months because of the inherent complications of nanoflow LC, for example, blockage of column or spray emitter, prolonged purging/calibration of the nanoflow pump, and shifts in RT.

    To further the application of plasma proteomics to larger cohorts, we transferred the nanoflow LC-MS method onto a standard-flow platform consisting of an Agilent 1290 Infinity II HPLC interfaced to an Agilent 6495 QqQ MS. A 27-minute LC method was successfully established to measure the same 200 peptides (light and heavy peptides for 100 proteins) with a minimum of 3 transitions for each peptide. The lower limit of detection and the lower limit of quantification were evaluated on both the nanoflow and the standard-flow platform (Table III in the Data Supplement and Figure I in the Data Supplement). To obtain a similar sensitivity to the nanoflow method, 5× more sample volume (10 μL) had to be injected using the standard-flow approach compared with nanoflow (2 μL). With the larger sample load, a similar detection range could be achieved despite the switch to standard-flow LC-MS with higher ion suppression. When using the nanoflow system, median SIS peptide intensities in plasma samples were only decreased by one third compared with the intensities in the LC solution (Figure II in the Data Supplement). Using the standard-flow system, the median lower limit of detection was about 2-fold higher in the SIS diluted with plasma than with LC solution (Table III in the Data Supplement).

    Among the proteins quantified by both methods, the results showed a high correlation (r=0.998) in all 668 samples (Figure 1 and Table IV in the Data Supplement) apart from few notable exceptions: for β-2-glycoprotein 1 (APOH_HUMAN) because of chromatography issues with the SIS peptide after digestion, for fibrinogen β chain (FIBB_HUMAN) and for serum paraoxonase/arylesterase 1 (PON1_HUMAN) because of a RT window error, and for α-2-macroglobulin (A2MG_HUMAN) because of a manufacturing error with a missing SIS peptide.

    Figure 1.

    Figure 1. Comparison between nanoflow and standard-flow liquid chromatography (LC) method. Plasma samples from the Bruneck Study (year 2000 evaluation, n=668) were digested, and 100 proteins were measured using multiple reaction monitoring (MRM) with stable isotope-labeled standards (SIS) peptides as spike-in standards. Samples were denatured, reduced, alkylated, and digested with trypsin using PlasmaDive MRM Panel kit (Biognosys, CH) on an automated liquid handling robot (BRAVO, Agilent) following the manufacturer’s instruction. The SIS peptides were spiked-in before tryptic digestion. The digested peptides were purified and analyzed on both nanoflow high-performance liquid chromatography (HPLC) mass spectrometry (MS) and standard-flow HPLC-MS. The results were analyzed using SpectroDive software (Biognosys, CH). Protein concentrations measured by nanoflow and standard-flow LC-MS are highly correlated with the exception of β-2-glycoprotein 1 (APOH_HUMAN) because of a splitting of the chromatographic peak for the reference peptide.

    To demonstrate that both high and less abundant proteins were measured reliably, we selected TRFE and PLF4 as examples to be compared with a reference measurement from a routine clinical assay for TRFE or ELISA for PLF4. The high abundant TRFE displayed similar agreement with the reference measurement for both the nanoflow and standard-flow LC-MS measurements (Figure 2A). For less abundant PLF4, standard-flow LC-MS displayed closer agreement with the reference measurement, whereas nanoflow LC-MS measured higher values than the reference measurement in the low concentration range and lower values in the high concentration range, resulting in higher overall deviation from the reference measurement compared with standard-flow method (Figure 2B).

    Figure 2.

    Figure 2. Comparison for high and low abundant plasma proteins. For the high abundant plasma protein TRFE, the nanoflow and standard-flow results were similar when compared with traditional measurements (A). For the less abundant plasma protein PLF4, nanoflow measurements are higher at low concentration end and lower at high concentration end compared with traditional measurements. The standard-flow measurements showed lower variation than nanoflow liquid chromatography (LC) mass spectrometry (MS; B). In the Bland–Altman plots, x axis is the mean of log-transformed protein concentration of subject-wise multiple reaction monitoring (MRM) and clinical measurement; y axis is the difference between these 2 measurements.

    To evaluate the performance of the 2 LC-MS platforms, a pooled QC sample was injected periodically during continuous operation while analyzing hundreds of samples on both platforms. For the peptides quantified on both platforms, the relative SD of the light or heavy peptide peak area on the nanoflow platform was much higher (light 38.07%–83.80%, heavy 37.92%–80.88%) compared with the standard-flow platform (light 5.12%–28.51%, heavy 5.59%–28.14%) with P<0.001. After adjustment to heavy peptides, the variations of the light:heavy ratio in both platforms were more comparable (nanoflow 1.68%–16.56%, standard flow 1.34%–18.52%; P=0.651; Figure 3). Nonetheless, the nanoflow LC-MS system showed higher variation for the absolute signal intensities between different runs.

    Figure 3.

    Figure 3. Comparison of relative SD (RSD) between nanoflow and standard-flow liquid chromatography (LC). The quality control samples on the nanoflow and standard-flow system showed similar RSD of the light/heavy ratio, but the light or heavy peptide peak area showed a higher variation on the nanoflow system.

    When comparing the stability of the RT, in the nanoflow system, the SIS peptides eluted from 3.62 to 27.56 minutes. The SD of the RT of each peptide ranged from 0.26 to 0.44 minutes over 700 injections (Figure 4A). The standard peptides eluted in the standard-flow system from 3.22 to 19.99 minutes but with substantially less variation: the SD range was within 0.003 to 0.05 minutes (Figure 4B).

    Figure 4.

    Figure 4. Comparison of retention time (RT) between nanoflow and standard-flow liquid chromatography (LC). The RT of each peptide showed higher variation across 668 samples using nanoflow LC (A) compared with standard-flow LC (B).

    Comparison of Commercial MRM Kits for Plasma Proteomics

    A different MRM kit (PeptiQuant MRM Assay Biomarker Assessment Kit BAK-76, MRM Proteomics) was used to compare the apolipoprotein measurements with the PlasmaDive (Biognosys, CH) results (n=44). The plasma samples were processed and analyzed on the standard-flow platform with 15 µL injection volume according to the manufacturer’s instructions. The target peptide concentrations were calculated using both single-point calibration and relative response–relative concentration standard curve. The results derived from the single-point calibration for apolipoprotein measurements were compared with the PlasmaDive (Biognosys, CH) method (Figure 5). In general, the results were highly correlated between the 2 commercial kits when using the same proteotypic peptide (apo(a) r=0.928, apoC2 r=0.922, apoE r=0.930). For the same protein with different proteotypic peptides, the MRM-based measurements still showed good correlation (r>0.7). However, the absolute concentrations varied between the 2 kits with extreme ratios for apo(a) (6.01) and apoB (3.22). In both kits, a peptide (GTYSTTVTGR) within the Kringle repeat was selected for the measurement of apo(a) (APOA_HUMAN). The number of Kringle repeats, however, greatly varies between individuals.24,25 Although the peptide is readily detectable in plasma digests, this peptide is not a suitable target for measuring apo(a) concentrations. Instead, both commercial MRM kits provide a readout for Kringle size (Kringle mass). For this reason, apo(a) was excluded from the final version of the PlasmaDive kit (Biognosys, CH).

    Figure 5.

    Figure 5. Comparison between multiple reaction monitoring (MRM)–based kits from 2 different vendors. When comparing the concentration of the same apolipoproteins measured by PlasmaDive (PD) and PeptiQuant BAK-76 (MRM), we observed a high correlation between the 2 kits, especially for those apolipoproteins measured using the same peptide (apo(a) r=0.928, apoC2 r=0.922, apoE r=0.930). However, the absolute concentrations varied between 2 kits with extreme ratios for apo(a) (6.01) and apoB (3.22). The linear regression formula is shown as PD=a+MRM×b on top of each graph.

    Comparison to Antibody-Based Assays

    Next, to evaluate the compatibility of MS-based methods with the traditional antibody-based method, we compared PlasmaDive (n=44) with an antibody-based Luminex assay (Milliplex MAP Human Apolipoprotein Magnetic Bead Panel kit, Millipore) for 5 apolipoproteins (apoA1, apoA2, apoC2, apoC3, and apoE). The PlasmaDive and the Luminex results showed very high correlation for apoC2 (r=0.917), apoC3 (r=0.944), and apoE (r=0.886; Figure 6). However, the MRM measurements returned lower concentrations compared with the antibody-based analysis. This difference was particularly pronounced for apoA2: the concentration of apoA2 was ≈27.7% of that of apoA1 using the Luminex kit. The MRM data, however, estimated apoA2 concentration to be just 3.7% of that of apoA1 (Figure III in the Data Supplement).

    Figure 6.

    Figure 6. Comparison of mass spectrometry (MS)–based measurements (PlasmaDive [PD]) with antibody-based assays (Luminex). Measurements for apoC2, and apoE revealed a good correlation between the MS- and antibody-based measurements, but the absolute concentration is lower when measured using liquid chromatography (LC) MS (PD; Biognosys, CH). A poor correlation was observed for apoA1, and apoA2 concentrations were substantially lower compared with antibody-based measurements. The linear regression formula is shown as PD=a+Luminex×b on top of each graph.

    Association of Protein Measurements With Clinical Variables

    Figure 7 shows the associations of protein measures with clinical variables. As expected, most proteins showed a similar strength of association when using either the nanoflow or standard-flow method, and the measurements of apoA1 and apoB were strongly associated with HDL-C and LDL-C, respectively (Figure IV in the Data Supplement). Nonetheless, 51.1% of comparisons returned lower P values with the standard-flow method, but this was not significant (P=0.5 with t test und P=0.4 with Wilcoxon test). Importantly, there was less interference with the MRM transitions in the standard-flow method. The most discordant protein measurements are highlighted in Figure V in the Data Supplement.

    Figure 7.

    Figure 7. Associations of clinical variables with proteins measured by either nanoflow or standard-flow liquid chromatography (LC) method. The associations of protein measurements from both the nanoflow and standard-flow LC method were compared with clinical variables including age, body mass index, C-reactive protein, diastolic blood pressure, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), sex, systolic blood pressure, and type 2 diabetes mellitus.


    In this study, we have developed a standard-flow LC-MS workflow to analyze 100 proteins in plasma samples from a community-based, prospective cohort (n=668 samples, >1500 injections). The optimized standard-flow LC-MS method improved reproducibility, robustness, and throughput in comparison to the traditional proteomics method using nanoflow LC. Using the standard-flow LC-MS method, the Bruneck cohort (n=668) could be analyzed in just over 2 weeks rather than 2 months without the need for enrichment, protein depletion, or fractionation of peptides.

    The PlasmaDive LC-MS method recommended by the manufacturer is based on a nanoflow LC method, and the protocol requires minimal user optimization. For large cohorts, however, analysis time becomes prohibitive. For example, for the samples of the Bruneck study, the total run time extended to 2 months because of the inherent complications that arose from nanoflow LC applications. The extended run time also required cleaning and calibration of the QqQ MS in between. Thus, nanoflow LC-based workflows did not allow the throughput required for clinical research applications. On the other hand, standard-flow LC offers greater stability and robustness and thus less variability but has only been described in small-scale studies.610,17 By applying MRM-MS to a community-based cohort, we demonstrate the advantages of shifting from a nanoflow LC to a standard-flow LC method, ideally suited to epidemiological studies: The HPLC run time was shortened from 70 minutes in nanoflow LC-MS to 27 minutes. Accordingly, the cycle time was reduced to 0.5 s, and 3 transitions per peptide were used to accommodate the 200 peptides into the narrower peak width (0.2 minutes) and maintain a reasonable peak shape (≈15 data points). Because of the different ionization properties between the nanospray source and Jet Stream ESI source, the most abundant precursor charge state of some peptides had to be changed to get the best signal intensity. These transitions were also optimized independently using a quadrupole time-of-flight MS/MS (Agilent) system, which confirmed that the selection of the 3 transitions did not decrease the accuracy or sensitivity for reliable MRM quantification (data not shown). In fact, some of the peptide measurements were even improved by removal of transitions with a saturated peak or of an interfering peak.

    Reduced sensitivity and increased ion suppression are the main limitations of standard-flow LC compared with nanoflow LC.26,27 We have tested the limit of detection and ion suppression on both platforms. With the increased injection volume (5×), the standard-flow LC system reached the same sensitivity as the nanoflow LC system. However, this may also be in part because of the better sensitivity of the QqQ MS coupled to standard LC. The increased injection volume on the standard-flow platform is not problematic when the sample source is plasma, as the high protein concentration means that only a small volume (10 µL) is required to perform the analysis. Our ion suppression test revealed that some of the SIS peptides have a lower limit of quantification in LC solution compared with plasma, whereas others are not influenced as much. The ion suppression depends in part on the ionization properties of the peptides and coeluting abundant peptides. For low abundant proteins, the effects of ion suppression can be compensated by increasing the concentrations of spike-in SIS peptides as long as they remain with similar concentration to the target peptides.

    Once the standard-flow LC-MS method was optimized, we analyzed 668 samples plus regular QCs, back-to-back with a total run time of 15 days. Overall, we demonstrate the advantages of shifting from a nanoflow LC to a standard-flow LC method. This is because of 2 main reasons:

    1. The 2.1 mm inner diameter column had no blockages during continuous operation for a few months.

    2. The better RT stability means smaller and less overlapping RT windows in a scheduled MRM, therefore, longer dwell times for each transition resulting in better signal intensities and less missing data.

    We have recently applied this standard-flow method to compare the associations of the different apolipoproteins with cardiovascular risk.22 We also measured the same apolipoproteins in an intervention study using antisense therapy for apoC3. Plasma apoC3 levels were decreased by 75%, whereas apoC2 and apoE levels showed a 50% drop on antisense treatment. These apolipoproteins are predominantly associated with very-low-density lipoproteins, and there is increasing evidence for a role of very-low-density lipoproteins in addition to low-density lipoproteins in atherosclerosis.28

    MRM Versus Alternative MS Quantification Methodologies

    Different MS approaches for plasma proteomics have been described in recent years. Stable Isotope Standards and Capture by Anti-Peptide Antibodies uses antipeptide antibodies to enrich peptides of interest after addition of stable isotope-labeled internal peptide standards to the protein digest. Light and heavy peptides are enriched simultaneously, resulting in a sensitive and accurate quantification of the peptides of interest.29,30 Data-independent acquisition analysis offers a hypothesis-free approach and allows the analysis of a larger number of proteins. The samples are converted into digital archive as all detectable peptides in the samples are fragmented and MS/MS spectra are acquired. However, sensitivity and dynamic range are limited in nondepleted plasma samples. In a subcohort of the UK Twin registry (n=232), 342 plasma proteins could be analyzed by Sequential Windowed Acquisition of All Theoretical Fragment Ion Mass Spectra analysis, but essential steps including depletion of the 14 most abundant proteins and fractionation into 6 fractions were needed.7 Complicated analytic workflows or demanding data analysis will limit the wider application of proteomics in epidemiology. Throughput and costs are critical considerations for applications in clinical research. Thus, we focused on a standard-flow LC-MRM method for plasma proteomics to meet the requirements for epidemiological studies in terms of reducing costs and increasing throughput. Further improvements may include parallel reaction monitoring with comparable linearity, dynamic range and precision compared with MRM,31,32 and less method development time for new targets as only the accurate precursor masses of the target peptides are required. For all MS-based methods, the choice of the proteotypic peptide is critical for meaningful quantification of a protein and must not only be based on criteria for LC-MS analysis, such as peptide length, uniqueness, post-translational modifications, physiochemical properties, and abundance, but also biological function as exemplified by the reference peptide selected for apo(a).

    LC-MRM-MS Versus Antibody-Based Protein Measurements

    Antibody-based methods require lower investment in equipment and offer automation and high throughput. Compared with LC-MS, however, immunoassays rely on the specificity and selectivity of the binders. The necessity of good antibodies results in high running costs, gives rise to batch to batch inconsistencies, and limits the multiplexing capability. Moreover, only external standard but no internal standards are used. LC-MS has a higher initial cost for instrumentation and maintenance and offers a lower throughput at this moment. However, LC-MS methods are faster and cheaper to develop; they offer high multiplexing capability within a dynamic range for the simultaneous detection of low and high abundant analytes over 4 to 5 orders of magnitude; and stable isotope-labeled internal standards reduce variability while increasing robustness.3

    Both MRM- and antibody-based absolute quantification methods rely on the accuracy of the standard. In general, lower concentrations were detected by the MRM method than the antibody-based method. We have investigated the digestion efficiency by measuring known amount of human TRFE in a fetal bovine serum background using the PlasmaDive kit. The PlasmaDive results returned about 88% of the spike-in concentration when the transferrin concentration is high and dropped to 60% when the concentration is low (data not shown). These differences may come from the weighting error, the incomplete digestion, the peptide loss during the experiment procedure, and the ion suppression especially when measuring at low concentrations. An inherent risk of using peptide standards in a relatively low concentration is that every sample preparation step may cause losses, that is, because of the hydrophobic peptides binding to the surface of plastic ware.33 This may also be the reason for the variation of the late eluting peptides (eg, VSFLSALEEYTK from apoA1). Ideally, SIS proteins should be used instead of SIS peptides for MRM-based measurements. Stable isotope-labeled proteins, however, are significantly more expensive than using SIS peptides.34 On the other hand, ELISA kits tend to use protein standards in physiological buffers or in reference samples but without the biological matrix of the actual sample. Differences in the matrix between samples may interfere with the binding of the antibody to the antigen and thus influence the results. With MRM-MS, proteins can be measured directly and the multiplexing capability allows for measurements of biomarker panels in a single run. Although MRM-MS is routinely applied clinically for measuring small molecules,3537 the use of MRM-MS could be expanded to protein measurements in future.


    For large epidemiological studies, experimental variation has to be well controlled. We have successfully applied a standard-flow LC-MRM method to obtain tens of thousands of protein measurements in a community-based cohort. The standard-flow LC-MRM method resulted in significant time savings and offered higher precision and better reproducibility. Although the limitations of antibodies are well recognized, questions should also be raised about discordance between peptide measurements using LC-MS and the true protein concentration. For absolute quantification, protein standards may be required to adjust for tryptic digestion efficiency and to account for peptide losses, which might not always be similar for spike-in SIS and endogenous peptides.


    Luminex Analysis was performed using Luminex FlexMap 3D at The iLab, Flow Cytometry Core, the National Institute of Health Research (NIHR) Biomedical Research Centre based at Guy’s and St Thomas’ NHS Foundation Trust, and King’s College London in partnership with King’s College Hospital.


    The Data Supplement is available at

    Correspondence to Manuel Mayr, MD, PhD, King’s British Heart Foundation Centre, King’s College London, 125 Coldharbour Ln, London SE5 9NU, United Kingdom. E-mail


    • 1. Anderson NL, Anderson NG. The human plasma proteome: history, character, and diagnostic prospects.Mol Cell Proteomics. 2002; 1:845–867.CrossrefMedlineGoogle Scholar
    • 2. Gillette MA, Carr SA. Quantitative analysis of peptides and proteins in biomedicine by targeted mass spectrometry.Nat Methods. 2013; 10:28–34. doi: 10.1038/nmeth.2309.CrossrefMedlineGoogle Scholar
    • 3. Carr SA, Abbatiello SE, Ackermann BL, Borchers C, Domon B, Deutsch EW, et al. Targeted peptide measurements in biology and medicine: best practices for mass spectrometry-based assay development using a fit-for-purpose approach.Mol Cell Proteomics. 2014; 13:907–917. doi: 10.1074/mcp.M113.036095.CrossrefMedlineGoogle Scholar
    • 4. Kusebauch U, Campbell DS, Deutsch EW, Chu CS, Spicer DA, Brusniak MY, et al. Human SRMAtlas: a resource of targeted assays to quantify the complete human proteome.Cell. 2016; 166:766–778. doi: 10.1016/j.cell.2016.06.041.CrossrefMedlineGoogle Scholar
    • 5. Grebe SK, Singh RJ. LC-MS/MS in the clinical laboratory - where to from here?Clin Biochem Rev. 2011; 32:5–31.MedlineGoogle Scholar
    • 6. Razavi M, Leigh Anderson N, Pope ME, Yip R, Pearson TW. High precision quantification of human plasma proteins using the automated SISCAPA Immuno-MS workflow.N Biotechnol. 2016; 33(5 pt A):494–502. doi: 10.1016/j.nbt.2015.12.008.CrossrefMedlineGoogle Scholar
    • 7. Liu Y, Buil A, Collins BC, Gillet LC, Blum LC, Cheng LY, et al. Quantitative variability of 342 plasma proteins in a human twin population.Mol Syst Biol. 2015; 11:786.CrossrefMedlineGoogle Scholar
    • 8. Hüttenhain R, Soste M, Selevsek N, Röst H, Sethi A, Carapito C, et al. Reproducible quantification of cancer-associated proteins in body fluids using targeted proteomics.Sci Transl Med. 2012; 4:142ra94. doi: 10.1126/scitranslmed.3003989.CrossrefMedlineGoogle Scholar
    • 9. Domanski D, Percy AJ, Yang J, Chambers AG, Hill JS, Freue GV, et al. MRM-based multiplexed quantitation of 67 putative cardiovascular disease biomarkers in human plasma.Proteomics. 2012; 12:1222–1243. doi: 10.1002/pmic.201100568.CrossrefMedlineGoogle Scholar
    • 10. Percy AJ, Chambers AG, Yang J, Domanski D, Borchers CH. Comparison of standard- and nano-flow liquid chromatography platforms for MRM-based quantitation of putative plasma biomarker proteins.Anal Bioanal Chem. 2012; 404:1089–1101. doi: 10.1007/s00216-012-6010-y.CrossrefMedlineGoogle Scholar
    • 11. Anderson L, Hunter CL. Quantitative mass spectrometric multiple reaction monitoring assays for major plasma proteins.Mol Cell Proteomics. 2006; 5:573–588. doi: 10.1074/mcp.M500331-MCP200.CrossrefMedlineGoogle Scholar
    • 12. Anderson NL, Anderson NG, Haines LR, Hardie DB, Olafson RW, Pearson TW. Mass spectrometric quantitation of peptides and proteins using Stable Isotope Standards and Capture by Anti-Peptide Antibodies (SISCAPA).J Proteome Res. 2004; 3:235–244.CrossrefMedlineGoogle Scholar
    • 13. Martosella J, Zolotarjova N, Liu H, Nicol G, Boyes BE. Reversed-phase high-performance liquid chromatographic prefractionation of immunodepleted human serum proteins to enhance mass spectrometry identification of lower-abundant proteins.J Proteome Res. 2005; 4:1522–1537. doi: 10.1021/pr050088l.CrossrefMedlineGoogle Scholar
    • 14. Geyer PE, Kulak NA, Pichler G, Holdt LM, Teupser D, Mann M. Plasma proteome profiling to assess human health and disease.Cell Syst. 2016; 2:185–195. doi: 10.1016/j.cels.2016.02.015.CrossrefMedlineGoogle Scholar
    • 15. Addona TA, Abbatiello SE, Schilling B, Skates SJ, Mani DR, Bunk DM, et al. Multi-site assessment of the precision and reproducibility of multiple reaction monitoring-based measurements of proteins in plasma.Nat Biotechnol. 2009; 27:633–641. doi: 10.1038/nbt.1546.CrossrefMedlineGoogle Scholar
    • 16. Polanski M, Anderson NL. A list of candidate cancer biomarkers for targeted proteomics.Biomark Insights. 2007; 1:1–48.MedlineGoogle Scholar
    • 17. Cohen Freue GV, Borchers CH. Multiple reaction monitoring (MRM): principles and application to coronary artery disease.Circ Cardiovasc Genet. 2012; 5:378. doi: 10.1161/CIRCGENETICS.111.959528.LinkGoogle Scholar
    • 18. Razavi M, Anderson NL, Yip R, Pope ME, Pearson TW. Multiplexed longitudinal measurement of protein biomarkers in DBS using an automated SISCAPA workflow.Bioanalysis. 2016; 8:1597–1609. doi: 10.4155/bio-2016-0059.CrossrefMedlineGoogle Scholar
    • 19. Blankley RT, Fisher C, Westwood M, North R, Baker PN, Walker MJ, et al. A label-free selected reaction monitoring workflow identifies a subset of pregnancy specific glycoproteins as potential predictive markers of early-onset pre-eclampsia.Mol Cell Proteomics. 2013; 12:3148–3159. doi: 10.1074/mcp.M112.026872.CrossrefMedlineGoogle Scholar
    • 20. Anderson L. Six decades searching for meaning in the proteome.J Proteomics. 2014; 107:24–30. doi: 10.1016/j.jprot.2014.03.005.CrossrefMedlineGoogle Scholar
    • 21. Stegemann C, Pechlaner R, Willeit P, Langley SR, Mangino M, Mayr U, et al. Lipidomics profiling and risk of cardiovascular disease in the prospective population-based Bruneck study.Circulation. 2014; 129:1821–1831. doi: 10.1161/CIRCULATIONAHA.113.002500.LinkGoogle Scholar
    • 22. Pechlaner R, Tsimikas S, Yin X, Willeit P, Baig F, Santer P, et al. Very-low-density lipoprotein-associated apolipoproteins predict cardiovascular events and are lowered by inhibition of APOC-III.J Am Coll Cardiol. 2017; 69:789–800. doi: 10.1016/j.jacc.2016.11.065.CrossrefMedlineGoogle Scholar
    • 23. Altman DG, Bland JM. Measurement in medicine - the analysis of method comparison studies.Statistician. 1983; 32:307–317.CrossrefGoogle Scholar
    • 24. Marcovina SM, Albers JJ, Scanu AM, Kennedy H, Giaculli F, Berg K, et al. Use of a reference material proposed by the International Federation of Clinical Chemistry and Laboratory Medicine to evaluate analytical methods for the determination of plasma lipoprotein(a).Clin Chem. 2000; 46:1956–1967.CrossrefMedlineGoogle Scholar
    • 25. Kronenberg F, Utermann G. Lipoprotein(a): resurrected by genetics.J Intern Med. 2013; 273:6–30. doi: 10.1111/j.1365-2796.2012.02592.x.CrossrefMedlineGoogle Scholar
    • 26. Annesley TM. Ion suppression in mass spectrometry.Clin Chem. 2003; 49:1041–1044.CrossrefMedlineGoogle Scholar
    • 27. Gangl ET, Annan MM, Spooner N, Vouros P. Reduction of signal suppression effects in ESI-MS using a nanosplitting device.Anal Chem. 2001; 73:5635–5644.CrossrefMedlineGoogle Scholar
    • 28. Khetarpal SA, Zeng X, Millar JS, Vitali C, Somasundara AVH, Zanoni P, et al. A human APOC3 missense variant and monoclonal antibody accelerate apoC-III clearance and lower triglyceride-rich lipoprotein levels.Nat Med. 2017; 23:1086–1094. doi: 10.1038/nm.4390.CrossrefMedlineGoogle Scholar
    • 29. Whiteaker JR, Zhao L, Anderson L, Paulovich AG. An automated and multiplexed method for high throughput peptide immunoaffinity enrichment and multiple reaction monitoring mass spectrometry-based quantification of protein biomarkers.Mol Cell Proteomics. 2010; 9:184–196. doi: 10.1074/mcp.M900254-MCP200.CrossrefMedlineGoogle Scholar
    • 30. Whiteaker JR, Zhao L, Lin C, Yan P, Wang P, Paulovich AG. Sequential multiplexed analyte quantification using peptide immunoaffinity enrichment coupled to mass spectrometry.Mol Cell Proteomics. 2012; 11:M111.015347. doi: 10.1074/mcp.M111.015347.CrossrefMedlineGoogle Scholar
    • 31. Ronsein GE, Pamir N, von Haller PD, Kim DS, Oda MN, Jarvik GP, et al. Parallel reaction monitoring (PRM) and selected reaction monitoring (SRM) exhibit comparable linearity, dynamic range and precision for targeted quantitative HDL proteomics.J Proteomics. 2015; 113:388–399. doi: 10.1016/j.jprot.2014.10.017.CrossrefMedlineGoogle Scholar
    • 32. Gallien S, Kim SY, Domon B. Large-scale targeted proteomics using Internal Standard Triggered-Parallel Reaction Monitoring (IS-PRM).Mol Cell Proteomics. 2015; 14:1630–1644. doi: 10.1074/mcp.O114.043968.CrossrefMedlineGoogle Scholar
    • 33. Goebel-Stengel M, Stengel A, Taché Y, Reeve JRThe importance of using the optimal plasticware and glassware in studies involving peptides.Anal Biochem. 2011; 414:38–46. doi: 10.1016/j.ab.2011.02.009.CrossrefMedlineGoogle Scholar
    • 34. Lebert D, Dupuis A, Garin J, Bruley C, Brun V. Production and use of stable isotope-labeled proteins for absolute quantitative proteomics.Methods Mol Biol. 2011; 753:93–115. doi: 10.1007/978-1-61779-148-2_7.CrossrefMedlineGoogle Scholar
    • 35. Tang WH, Wang Z, Levison BS, Koeth RA, Britt EB, Fu X, et al. Intestinal microbial metabolism of phosphatidylcholine and cardiovascular risk.N Engl J Med. 2013; 368:1575–1584. doi: 10.1056/NEJMoa1109400.CrossrefMedlineGoogle Scholar
    • 36. Chen Z, Caulfield MP, McPhaul MJ, Reitz RE, Taylor SW, Clarke NJ. Quantitative insulin analysis using liquid chromatography-tandem mass spectrometry in a high-throughput clinical laboratory.Clin Chem. 2013; 59:1349–1356. doi: 10.1373/clinchem.2012.199794.CrossrefMedlineGoogle Scholar
    • 37. Zoller M, Maier B, Hornuss C, Neugebauer C, Döbbeler G, Nagel D, et al. Variability of linezolid concentrations after standard dosing in critically ill patients: a prospective observational study.Crit Care. 2014; 18:R148. doi: 10.1186/cc13984.CrossrefMedlineGoogle Scholar


    Despite the ongoing genetic sequencing efforts, the impact of most genetic variation on cardiovascular disease is still poorly understood. Attention has shifted to postgenomic technologies, including proteomics and lipidomics, to refine the cardiovascular disease phenotypes at a molecular level. For omics measurements in epidemiological cohorts, experimental variation has to be well controlled to assure data comparability over time and between studies. We have successfully applied mass spectrometry to obtain plasma protein measurements in a community-based cohort using a standard-flow liquid chromatography method. The standard-flow method resulted in significant time savings and offered higher precision and better reproducibility. The multiplexing capability of mass spectrometry combined with the standard-flow liquid chromatography method increased throughput and reduced the costs of large-scale protein measurements in epidemiological cohorts. For example, apolipoproteins can be measured with high specificity and precision using mass spectrometry. Our study demonstrates the use of mass spectrometry for simultaneously measuring a broader panel of apolipoproteins than the current antibody-based apolipoprotein measurements used in the clinic.


    eLetters should relate to an article recently published in the journal and are not a forum for providing unpublished data. Comments are reviewed for appropriate use of tone and language. Comments are not peer-reviewed. Acceptable comments are posted to the journal website only. Comments are not published in an issue and are not indexed in PubMed. Comments should be no longer than 500 words and will only be posted online. References are limited to 10. Authors of the article cited in the comment will be invited to reply, as appropriate.

    Comments and feedback on AHA/ASA Scientific Statements and Guidelines should be directed to the AHA/ASA Manuscript Oversight Committee via its Correspondence page.