CCQM-K6.2 Determination of Total Cholesterol in Human Serum

Final Report April 2018

Stephen A. Wise, Karen W. Phinney, and David L. Duewer National Institute of Standards and Technology (NIST) Gaithersburg, MD, USA

With contributions from: Lorna T. Sniegoski and Michael J. Welch National Institute of Standards and Technology (NIST), Gaithersburg, MD, USA

Guiomar Pabello and Marco A. Avila Caldero Centro Nacional de Metrologia (CENAM), Querétaro, México

Liu Qinde and Lee Tong Kooi Health Sciences Authority (HSA), Singapore

Eliane Rego, Bruno Garrido, Gabriella Allegri, Marcia de La Cruz, and Juliana Barrabin Instituto Nacional de Metrologia, Qualidade e Tecnologia (INMETRO), Xerém, Rio de Janeiro, Brazil

Celia Puglisi and Eduardo Lopez Instituto Nacional de Tecnologia Industrial (INTI), Buenos Aires, Argentina

Hwashim Lee and Byungjoo Kim Korea Research Institute of Standards and Science (KRISS), Daejeon, Republic of Korea

Vincent Delatour and Maud Heuillet Laboratoire National de Métrologie et d’Essais (LNE), Paris, France

Jintana Nammoonnoy National Institute of Metrology (Thailand) (NIMT), Pathumthani, Thailand

Ahmet Ceyhan Gören and Gokhan Bilsel National Metrology Institute of Turkey (TÜBİTAK UME), Gebze-Kocaeli, Turkey L. Konopelko, A. Krylov, and E. Lopushanskaya D.I. Mendeleyev Institute for Metrology (VNIIM), St. Petersburg, Russian Federation



SUMMARY

Cholesterol is one of the most frequently measured substances in human blood/serum to assist in assessing the health status of individuals. Because of its clinical significance, CCQM-K6 Determination of Cholesterol in Serum was completed in 2000 as one of the first Key Comparison (KC) studies performed within the Organic Analysis Working Group (OAWG). The first Subsequent KC for cholesterol, CCQM-K6.1, was completed in 2001. Measurements for this second Subsequent, CCQM-K6.2, were completed in 2012. These Subsequent comparisons were conducted to enable CCQM members that had not participated in earlier studies to demonstrate their capabilities to measure a nonpolar (pKow < -2), low molecular mass (100 g/mol to 500 g/mol) metabolite in human serum at relatively high concentrations (1 mg/g to 3 mg/g) found in normal populations. Successful participation in CCQM-K6.2 demonstrated capabilities in analysis of complex biological matrices including sample preparation (extraction, derivatization), LC or GC separation, and quantification using an isotope dilution mass spectrometry approach. Normally in a subsequent KC, no Key Comparison Reference Value (KCRV) would be established and assessment of performance would be via the deviation of participants’ results to the anchor institute’s results, adjusted to account for the anchor’s performance in the original comparison versus its KCRV. Due to the very long-time period since the original key comparison, the OAWG decided that this did not represent the best approach to assess performance in what is a relatively complex measurement. Given the excellent agreement between the anchor institute’s results and robust consensus summary of the participants’ values, the Reference Value for this study was taken as the anchor institute’s result and treated as a “KCRV”. Seven of the nine participants demonstrated agreement with the reference value.

ii



TABLE OF CONTENTS

INTRODUCTION .......................................................................................................................... 1 MEASURAND ............................................................................................................................... 2 STUDY MATERIAL...................................................................................................................... 3

Homogeneity Assessment of Study Material.............................................................................. 3 PARTICIPANTS AND INSTRUCTIONS..................................................................................... 4

Methods Used by Participants .................................................................................................... 4 Methods Used by Anchor Laboratory ........................................................................................ 4 RESULTS ....................................................................................................................................... 5 Participant Results: Reported...................................................................................................... 5 Participant Results: Combined.................................................................................................... 5 Anchor Laboratory Results ......................................................................................................... 7 KEY COMPARISON REFERENCE VALUE (KCRV)................................................................ 7 DEGREES OF EQUIVALENCE ................................................................................................. 10 USE OF CCQM-K6.2 IN SUPPORT OF CMCs.......................................................................... 13 CONCLUSIONS........................................................................................................................... 13 ACKNOWLEDGEMENTS.......................................................................................................... 13 REFERENCES ............................................................................................................................. 13

LIST OF TABLES

Table 1: Previous CCQM Comparisons for Cholesterol, Glucose, and Creatinine....................... 1 Table 2: CCQM-K6.2 Timeline ..................................................................................................... 2 Table 3: Determination of Cholesterol in SRM 1951c Level 2 ..................................................... 3 Table 4: Participants and Anchor in CCQM-K6.2 Cholesterol in Human Serum......................... 4 Table 5: Results for CCQM-K6.2 Cholesterol in Human Serum as Received .............................. 5 Table 6: Participant Results for CCQM-K6.2 Cholesterol in Human Serum as Combined.......... 6 Table 7: Key Comparison Reference Value for CCQM-K6.2....................................................... 7 Table 8: Degrees of Equivalence for CCQM-K6.2 Cholesterol in Human Serum...................... 10 Table A-1: CCQM-K6.2 Sample Size, Extraction, and Cleanup ............................................... A2 Table A-2: CCQM-K6.2 Analytical Techniques ........................................................................ A4 Table A-3: CCQM-K6.2 Calibrants and Standards .................................................................... A6

LIST OF FIGURES

Figure 1: Combined results and robust consensus estimates of location and dispersion............... 6 Figure 2: Participant results for CCQM-K6.2 relative to the KCRV. ........................................... 8 Figure 3: Combined Participant results for CCQM-K6.2 relative to the KCRV........................... 9 Figure 4: Absolute degrees of equivalence for CCQM-K6.2 ...................................................... 11 Figure 5: Relative degrees of equivalence for CCQM-K6, -K6.1 and -K6.2 .............................. 12

LIST OF APPENDICES

Appendix A: CCQM-K6.2 Summary of Analytical Information ............................................... A1 Appendix B: CCQM-K6.2 Summary of Uncertainty Estimation Methods.................................B1

iii



CCQM

CDC CENAM CMC CRM DI GC GC-MS HSA ID INMETRO INTI JCTLM KC KCRV KRISS LC LC-MS LC-MS/MS LNE MADe

NIMT NIST NMI OAWG pKow PTB SRM UME VNIIM



ACRONYMS

Consultative Committee for Amount of Substance: Metrology in Chemistry and Biology Centers for Disease Control and Prevention, USA Centro Nacional de Metrologia, México calibration and measurement capabilities certified reference material designated institute gas chromatography gas chromatography separation with mass spectrometry detection Health Sciences Authority, Singapore isotope dilution Instituto Nacional de Metrologia, Qualidade e Tecnologia, Brazil Instituto Nacional de Tecnologia Industrial, Argentina Joint Committee for Traceability in Laboratory Medicine Key Comparison Key Comparison Reference Value Korea Research Institute of Standards and Science, Republic of Korea liquid chromatography liquid chromatography with mass spectrometry detection liquid chromatography with tandem mass spectrometry detection Laboratoire National de Métrologie et d’Essais, France median absolute deviation from the median (MAD)-based estimate of s: MADe = 1.4826·MAD, where MAD = median(|xi-median(xi)|) National Institute of Metrology (Thailand), Thailand National Institute of Standards and Technology, USA national metrology institute Organic Analysis Working Group logarithm of the octanol-water partition coefficient Physikalisch-Technische Bundesanstalt, Germany Standard Reference Material, a NIST CRM National Metrology Institute of Turkey, Turkey D.I. Mendeleyev Institute for Metrology, Russian Federation



iv



di %di k s ts u(xi)  �   (x) x xi     ̅ U95(x)

Uk=2(x)



SYMBOLS

degree of equivalence: xi - KCRV percent relative degree of equivalence: 100·di/KCRV coverage factor: U(x) = k·u(x)

standard deviation of a series of quantity values:      = �∑       =  1(         −     ̅)2⁄(     − 1) Student’s t-distribution expansion factor standard uncertainty of quantity value xi

pooled uncertainty:  �   (    ) = �∑       =  1     2(        )⁄     a quantity value the ith member of a series of quantity values mean of a series of quantity values:     ̅ = ∑       =  1         ⁄     expanded uncertainty defined such that x ±U95(x) is asserted to include the true value of the quantity with an approximate 95 % level of confidence expanded uncertainty defined as Uk=2(x) = 2·u(x)



v



INTRODUCTION



Cholesterol, glucose, and creatinine are three of the most frequently measured substances in human blood/serum to assist in assessing the health status of individuals. Because of their clinical significance, measurements of cholesterol, glucose, and creatinine were three of the first Key Comparison (KC) studies performed within the Consultative Committee for Amount of Substance: Metrology in Chemistry and Biology (CCQM) Organic Analysis Working Group (OAWG). These studies were performed in 2000 and 2001 with the National Institute of Standards and Technology (NIST) as the coordinating laboratory (see Table 1) and were published in Metrologia [1,2,3]. Subsequent Key Comparisons were conducted for each analyte in 2005 with the Korea Research Institute of Standards and Science (KRISS) as the coordinating laboratory for CCQM-K11.1 Glucose and CCQM-12.1 Creatinine and NIST as the coordinating laboratory for CCQM-K6.1 Cholesterol.



Table 1: Previous CCQM Comparisons for Cholesterol, Glucose, and Creatinine



Comparison CCQM-P6 CCQM-K6 CCQM-K6.1 CCQM-P8 CCQM-K11 CCQM-K11.1 CCQM-P9 CCQM-K12 CCQM-K12.1



Name of Comparison Study Cholesterol in Human Serum Cholesterol in Human Serum Cholesterol in Human Serum Glucose in Human Serum Glucose in Human Serum Glucose in Human Serum Creatinine in Human Serum Creatinine in Human Serum Creatinine in Human Serum



Date 1999 2000 2001 2000 2001 2005 2000 2001 2005



Coordinating Laboratory NIST NIST NIST NIST NIST KRISS NIST NIST KRISS



Number of Participants

7 7 2 4 3 3 4 5 3



Since these earlier studies were conducted, additional national metrology institutes (NMIs) or their designated institutes (DIs) are now providing measurement services for one or more of these clinical analytes. At the April 2012 OAWG meeting a proposal was accepted to conduct Subsequent Key Comparison studies for the three analytes with NIST as the coordinating laboratory. These three studies were conducted in parallel as CCQM-K6.2, -K11.2, and K-12.2.

The three studies were designed as Subsequent Key Comparisons, with NIST designated as both the coordinating and anchor laboratory. Therefore, participant results were to be compared with NIST measurements. Due to discordant results in CCQM-K11.2 and CCQM-K12.2 between NIST and the participating laboratories, KRISS and the Physikalisch-Technische Bundesanstalt (PTB) – laboratories that had successfully participated in the original studies – were requested by the OAWG to provide measurements for both glucose and creatinine. At the April 2015 OAWG meeting, the decision was made to treat CCQM-K11.2 and CCQM-K12.2 as modified Track C Key Comparisons rather than as Subsequent Key Comparisons. This report describes only CCQMK6.2. CCQM-K11.2 and CCQM-K12.2 are described in a separate report.



1



The timeline for CCQM-K6.2 study “Determination of Total Cholesterol in Human Serum” is summarized in Table 2.

Table 2: CCQM-K6.2 Timeline



Date April 2012 Nov. 2012 Dec. 2012 April 2013

Nov. 2013 April 2014

April 2015

Oct. 2015 Nov. 2015 June 2016 Sep. 2016 June 2017 April 2018



Action OAWG authorized CCQM-K6.2, -K11.2, and -K12.2 subsequent studies and approved protocols Call for Participation to OAWG members Samples shipped to participants Preliminary results presented to OAWG at Paris meeting. Results for CCQMK6.2 in good agreement; KRISS and PTB asked to provide reference measurements for CCQM-K11.2 and CCQM-K12.2 Reference results for CCQM-K11.2 and CCQM-K12.2 from PTB and KRISS received and discussed at CCQM meeting in South Africa Further discussion of how to assign KCRV for CCQM-K6.2, CCQM-K11.2 and CCQM-K12.2; decision to treat CCQM-K6.2 results as true Subsequent Key Comparison with NIST results as anchor; decision to assign KCRV for CCQMK11.2 and CCQM-K12.2 from participant and reference laboratory results Draft A Report discussed; decision to prepare two Draft A Reports, one for CCQM-K6.2 and a second for CCQM-K11.2 and CCQM-K12.2, which are to be treated as Track C Key comparisons rather than Subsequent Key Comparisons Draft A Report distributed to OAWG Draft B report distributed to OAWG Draft Final report delivered to OAWG Chair Review by CCQM WG chairs Revised Draft Final report delivered to OAWG Chair Final report delivered to OAWG Chair



MEASURAND

The measurands for the three clinical analyte studies were cholesterol, glucose, and creatinine as previously defined in the original studies (CCQM-K6, CCQM-K11, and CCQM-K12). These three clinical health status markers were selected in the original Key Comparison studies to be representative of measurement challenges associated with well-defined and low molar mass organic substances in blood. For CCQM-K6.2 the measurand was the mass fraction of total cholesterol in human serum.

Cholesterol (molar mass 365 g/mol) is a low polarity (nonpolar) analyte that is present in human serum at relatively high concentrations (1 mg/g to 3 mg/g). Cholesterol is predominantly esterified with fatty acids in the blood.



2



STUDY MATERIAL



The study material for CCQM-K6.2 was NIST candidate Standard Reference Material (SRM) 1951c Lipids in Frozen Human Serum (Level 2) [4], prepared as a replacement for SRM 1951b Lipids in Frozen Human Serum and issued in June 2013 after the CCQM-K6.2 results were reported to the OAWG. Participants were provided with three vials of serum for the determination of cholesterol. Each vial contained 1 mL of human serum. Samples were shipped frozen (on dry ice), and participants were instructed that a -20 ºC freezer was adequate for storage up to one week; however, if longer storage time was anticipated, the material should be stored at temperatures of -60 ºC or below.



Homogeneity Assessment of Study Material

Based on nearly two decades of experience with frozen serum samples prepared as SRMs for the determination of cholesterol, there were limited concerns regarding the homogeneity or the stability of the study material. No formal stability study was conducted for the study material. However, studies to assess homogeneity were conducted. For the material used in CCQM-K6.2, cholesterol homogeneity was assessed as part of the certification measurements. A total of 15 vials were selected for analysis based on a stratified sampling plan designed to test for homogeneity across the lot of vials. The 15 vials were analyzed in three sets of five vials per set with duplicate GC-MS injections. The results of the homogeneity assessment are shown in Table 3, where     ̅ designates a mean value, s a standard deviation, and 100 ∙ s/    ̅, the percent relative standard deviation. There was no trend apparent in the data when plotted against the sequence in which the vials were prepared [4].



Table 3: Determination of Cholesterol in SRM 1951c Level 2



Set Statistics



Set Tray Sample Sample  �   



 �   



s              ∙     / �   



1 83



7 241.31 241.19



0.71 0.29 %



1 32



8 241.50



1 14



9 240.10



1 112



10 242.00



1 40



11 241.02



2



2



19 240.24 240.70



0.54 0.23 %



2 31



20 240.57



2 65



21 240.72



2 102



22 240.37



2 81



23 241.62



3 114



31 238.54 240.84



1.89 0.79 %



3 24



32 239.57



3 73



33 243.38



3 119



34 240.89



3 27



35 241.82



3



PARTICIPANTS AND INSTRUCTIONS



Participants were requested to analyze two vials of material for cholesterol; the number of subsamples from each vial was left up to the laboratories. Participants were encouraged to use an appropriate serum-matrix CRM as a control material. Participants were to report the mass of cholesterol per mass of serum (mg/g) in the reporting form provided. The reporting form also included descriptions of methods used, number and order of measurements, reference compounds used as calibrants with purity corrections, control materials used, and method of calculating results. A complete description of their uncertainty calculations was also requested in the reporting form.



The National Metrology Institutes or Designated Institutes that participated in CCQM-K6.2 are listed in Table 4. NIST was designated as the anchor laboratory.



Table 4: Participants and Anchor in CCQM-K6.2 Cholesterol in Human Serum



NMI/DI CENAM

HSA INMETRO

INTI KRISS LNE NIMT UME VNIIM NIST



CCQM-K6.2 Cholesterol Participant Participant Participant Participant Participant Participant Participant Participant Participant

Anchor



METHODS

Methods Used by Participants For CCQM-K6.2 Cholesterol in Human Serum, results were received from nine participants. The participants used either isotope dilution gas chromatography-mass spectrometry (ID GC-MS) (six labs) or isotope dilution liquid chromatography-tandem mass spectrometry (ID LC-MS/MS) (three labs). The analytical methods used by the participants, including sample preparation, analytical technique, and quantification approach, are summarized in Tables A1 to A3 of Appendix A.

Methods Used by Anchor Laboratory The anchor laboratory used the ID GC-MS procedure published as a definitive method in 1989 [5] and now recognized by the Joint Committee for Traceability in Laboratory Medicine (JCTLM) as a reference measurement procedure. The method as used by the anchor laboratory is summarized in Tables A1 to A3.



4



RESULTS

Participant Results: Reported The results for K6.2 as received from the participants for measurements on each of two vials (as requested) are summarized in Table 5.



Table 5: Results for CCQM-K6.2 Cholesterol in Human Serum as Received



NMI/DI Vial



CENAM



1 2



HSA



1 2



INMETRO



1 2



INTI



1 2



KRISS



1 2



LNE



1 2



NIMT



1 2



UME



1 2



VNIIM



1 2



Mass Fraction, mg/g



x



u(x)



U95(x)



2.443 0.042 0.091



2.486 0.021 0.044



2.357 0.0136 0.027



2.353 0.0135 0.027



2.35



0.016 0.04



2.35



0.020 0.05



1.72



0.12



0.25



1.71



0.12



0.25



2.333 0.017 0.037



2.340 0.017 0.037



2.350 0.026 0.053



2.353 0.028 0.056



2.39



0.039 0.079



2.38



0.045 0.093



2.265 0.031 0.062



2.310 0.033 0.066



2.316 0.029 0.058



2.309 0.029 0.058



Coverage Factor (k)

2.16 2.16 2 2 2.262 2.306 2 2 2.2 2.2 2 2 2.05 2.06 2 2 2 2



Participant Results: Combined Due to an oversight in the study’s design, the report form did not request participants to combine their results for the two vials into a single overall result for the study material. Rather than retrospectively requesting that the participants supply this additional information, the coordinating laboratory calculated the combined results for all participants from their reported results. These combined results are summarized in Table 6 and displayed in Figure 1.

The combined value,     ̅, was estimated as the mean of the two reported results, x1 and x2. The standard uncertainty on this mean,     (    ̅), combined the standard deviation, s, of x1 and x2 and the pooled value of their associated uncertainties,  �   (    ) = �[(    95(    1)⁄2)2 + (    95(    2)⁄2)2]⁄2, yeilding the combined standard uncertainty:     (    ̅) = ��    2 +  �   2(    )�⁄√2. The expanded uncertainty on the combined value,         =2(    ̅), was estimated using the usual k=2 coverage factor:         =2(    ̅) = 2 ∙     (    ̅). Note that  �   (    ) is estimated from the reported expanded uncertainties divided by 2,     95(    )⁄2, to ensure that         =2(    ̅) reflects the participant’s uncertainty expansion policy.

5



Table 6: Participant Results for CCQM-K6.2 Cholesterol in Human Serum as Combined



NMI/DI CENAM

HSA INMETRO

INTI KRISS LNE NIMT UME VNIIM



 �    2.465 2.355 2.350 1.72 2.337 2.352 2.383 2.288 2.313



Mass Fraction, mg/g



s



 �   (    )     ( �   )



0.030 0.071 0.033



0.004 0.027 0.010



0.000 0.045 0.016



0.01



0.25



0.09



0.005 0.037 0.014



0.002 0.055 0.019



0.0045 0.086 0.031



0.032 0.064 0.032



0.005 0.058 0.021



Uk=2( �   ) 0.066 0.020 0.032 0.18 0.027 0.039 0.061 0.064 0.042



Errorbars are 1u



2.4



2.2



Cholesterol, mg/g



2.0



VaVluaelue



PPaarraammeetteerr



9 9NNuummbbeerrvvaalliidd ddaattaa



2.23.1319595%%CCoovveerraaggee ffaacctotor:r:kk==ttss



2.32.53050LoLoccaatitoionn:: MMeeddiiaann



0.0.20020uu(L(Looccaattioionn)): viaa MMAADDee



0.0.40747UU959(5L(Looccaattiioonn) = kk**uu((LLooccaatiotino)n) 0.08.787101000*u*u((LLooccaattiioonn))//LLooccaatitoionn



1.8



1.6



INTI UME VNIIM KRISS INMETRO LNE HSA NIMT CENAM



Figure 1: Combined results and robust consensus estimates of location and dispersion

Dots represent the combined values; the vertical bars on the dots span the k = 2 expanded uncertainties. The black horizontal line represents the median. The red horizontal lines bracket a robust estimate of the 95 % coverage interval about the median, U95. This interval is estimated as the product of the: standard uncertainty, u, estimated as the median absolute deviation from the median scaled to have the same coverage of a normal distribution as provided by the standard deviation (MADe) [6]; a factor of 1.25 reflecting the efficiency of the median as an estimator of the location for normally distributed data; and the 2.31 expansion factor of the Student’s ts distribution for 8 degrees of freedom. The black curve to the right edge is the empirical probability density for the reported results; the blue curve to the right is the Gaussian distribution parameterized with the robust consensus estimates.



6



Anchor Laboratory Results

The serum sample used for CCQM-K6.2 was NIST candidate SRM 1951c Level 2. The anchor laboratory’s certification measurements for this material, (240.91 ± 2.8) mg cholesterol/dL and a density of (102.521 ± 0.016) g/dL serum, were completed in June 2011 [5]. The certified value for this material, (241.41 ± 2.8) mg/dL, combines measurements made at NIST and at the U.S. Centers for Disease Control and Prevention (CDC). The date of issue for SRM 1951c was 27-June-2013, shortly after the CCQM-K6.2 results were revealed to participants. All uncertainties are here stated at an approximate 95 % level of confidence.



KEY COMPARISON REFERENCE VALUE (KCRV)



The certified value for SRM 1951c-2 and the anchor laboratory’s result, both transformed to mass fraction, are shown in Table 7 along with the robust consensus summary of the CCQM-K6.2 participant’s results.



Normally in a subsequent KC no KCRV would be established and assessment of performance would be via the deviation of participants’ results to the anchor lab’s results, adjusted to account for the anchor lab’s performance in the original comparison versus its KCRV. Due to the very long-time period since the original key comparison it was decided that this did not represent the best approach to assess performance in what is a relatively complex measurement.



Considering the excellent agreement between the anchor laboratory’s result and the consensus value, the OAWG at the April 2014 meeting agreed to use the anchor value and its U95 expanded uncertainty a “KCRV” for this comparison.



The participant results, both as reported and as combined, are displayed as x±U95(x) in Figure 2 with KCRV±U95(KCRV) reference lines. The combined results are displayed as x±u(x) in Figure 3 with KCRV±u(KCRV) reference lines.



Table 7: Key Comparison Reference Value for CCQM-K6.2



Source SRM 1951c-2 Certified Value

Anchor Laboratory Result Robust Consensus KCRV



Value 2.355 2.350 2.350 2.350



u(Value)

0.014 0.019 0.024 0.019



U95(Value) 0.027 0.038 0.047 0.038



Units mg/g mg/g mg/g mg/g



7



Total Cholesterol, mg/g



2.4



2.2



2.0



1.8



1.6



Reported Combined



KCRV



KCRV ±U₉₅

1.4



INTI UME VNIIM KRISS INMETRO LNE HSA NIMT CENAM



Total Cholesterol, mg/g



2.5

2.4

2.3

↓

2.2



Reported Combined KCRV KCRV ±U₉₅



INTI UME VNIIM KRISS INMETRO LNE HSA NIMT CENAM



Figure 2: Participant results for CCQM-K6.2 relative to the KCRV.

The blue symbols and vertical bars represent the results as reported; the black symbols and bars represent the results as combined by the coordinating laboratory. The bars are approximate 95 % expanded uncertainties. The horizontal lines represent the KCRV and the KCRV ± U95(KCRV) interval. The lower panel is identical to the upper, but displayed at higher vertical resolution.

8



Total Cholesterol, mg/g



2.4 2.2 2.0 1.8 1.6 1.4

2.5

2.4

2.3

↓

2.2



INTI UME VNIIM KRISS INMETRO LNE HSA NIMT CENAM



Combined KCRV KCRV ±U₉₅

Combined KCRV KCRV ±U₉₅



Total Cholesterol, mg/g



INTI UME VNIIM KRISS INMETRO LNE HSA NIMT CENAM



Figure 3: Combined Participant results for CCQM-K6.2 relative to the KCRV.

The black symbols and bars represent the results as combined by the coordinating laboratory. The bars are standard uncertainties. The horizontal lines represent the KCRV and the KCRV ± u(KCRV) interval. The lower panel is identical to the upper, but displayed at higher vertical resolution.

9



DEGREES OF EQUIVALENCE



The absolute degrees of equivalence for the participants in CCQM-K6.2 are estimated as the signed difference between the combined value and the KCRV: di = xi – KCRV. Since the KCRV is not estimated from the participant values, the 95 % expanded uncertainty on the di, U95(di), is estimated as the square root of the sum of the squares of the expanded uncertainties of the two components:     95(        ) = �     2   =2(        ) +     925(KCRV).

To enable comparison with the degrees of equivalence estimates from CCQM-K6 and –K6.1, it is convenient to express the di and U95(di) as percentages relative to the KCRV: %di = 100·di/KCRV and U95(%di) = 100·U95(di)/KCRV. Table 7 lists the numeric values of di, U95(di), di, and U95(di) for all participants in CCQM-K6.2. Figure 4 displays the absolute di ± U95(di) for CCQM-K6.2; Figure 5 displays the relative %di ± U95(%di) for CCQM-K6, CCQM-K6.1, and CCQM-K6.2.



Table 8: Degrees of Equivalence for CCQM-K6.2 Cholesterol in Human Serum



NMI/DI

CENAM HSA

INMETRO INTI KRISS LNE NIMT UME VNIIM



mg/g



di

0.115 0.004 0.000 -0.635 -0.014 0.002 0.033 -0.063 -0.038



Uk=2(di)

0.076 0.043 0.050 0.181 0.047 0.054 0.072 0.074 0.056



%di

4.9 0.2 0.0 -27.0 -0.6 0.1 1.4 -2.7 -1.6



%

Uk=2(%di)

3.3 1.8 2.1 7.7 2.0 2.3 3.1 3.2 2.4



10



Degree of Equivalence, mg/g



INTI UME VNIIM KRISS INMETRO LNE HSA NIMT CENAM



Degree of Equivalence, mg/g



0.1 -0.1 -0.3 -0.5 -0.7 -0.9

0.2

0.1

0.0

-0.1

↓

-0.2

Figure 4: Absolute degrees of equivalence for CCQM-K6.2

The black symbols and vertical bars represent the di ± U95(di). The horizontal line marks the ideal zero deviation from the KCRV. The lower panel is identical to the upper, but displayed at higher vertical resolution.

11



INTI UME VNIIM KRISS INMETRO LNE HSA NIMT CENAM



Degree of Equivalence, %



10 K6.2

0 K6 K6.1

-10

-20

-30

-40



LGC NIM NIST NMIA NMIJ PTB VSL NMIA VNIIM CENAM HSA INMETRO INTI KRISS LNE NIMT UME VNIIM



Degree of Equivalence, %



12



10



8



6 K6.2

4



2



0



-2 K6

-4



-6



K6.1



↓



-8



LGC NIM NIST NMIA NMIJ PTB VSL NMIA VNIIM CENAM HSA INMETRO INTI KRISS LNE NIMT UME VNIIM



Figure 5: Relative degrees of equivalence for CCQM-K6, -K6.1 and -K6.2

The blue symbols and bars represent %di ± U95(%di) for individual vials distributed in CCQM-K6 and K6.1; the black symbols and vertical bars represent their combined %di ± U95(%di). The red horizontal line marks the ideal zero deviation from the KCRV; the light grey lines are for visual guidance. The lower panel is identical to the upper, but displayed at higher vertical resolution.

12



USE OF CCQM-K6.2 IN SUPPORT OF CMCs

CCQM-K6.2 Cholesterol in Human Serum was designed as a Subsequent Key Comparison for NMIs and DIs that had not participated in earlier studies for determination of cholesterol. The study demonstrates a laboratory’s capabilities to measure a nonpolar (pKow < -2), low molecular mass (100 g/mol to 500 g/mol) metabolite in human serum at relatively high concentrations (1 mg/g to 3 mg/g) found in normal populations. At the time of this study, the OAWG had not formalized the reporting of “core competencies”. However, participation in this study demonstrates calibration and measurement capabilities (CMCs) in analysis of complex biological matrices including sample preparation (extraction, derivatization), LC or GC separation, and quantification using an isotope dilution mass spectrometry approach.

CONCLUSIONS

Intended as a Subsequent Key Comparison, CCQM-K6.2 met the expectations for such a study in that seven of the nine participants demonstrated agreement with the KCRV.

ACKNOWLEDGEMENTS

The study coordinators thank all the participating laboratories for providing the requested information during the course of these studies.

REFERENCES

1 Welch, M.J., Parris, R.M., Sniegoski, L.T., and May, W.E., CCQM-K6: Key Comparison on the Determination of Cholesterol in Serum, Metrologia, 39, Tech. Suppl. 08001 (2002)

2 Welch, M.J., Sniegoski, L.T., Parris, R.M., May, W.E., Heo, G.S., and Henrion, A., CCQMK11: The Determination of Glucose in Serum, Metrologia, 40, Tech. Suppl. 08003 (2003)

3 Welch, M.J., Phinney, C.P., Parris, R.M., May, W.E., Heo, G.S., Henrion, A., O’Conner, G., and Schimmel, H., CCQM-K12: The Determination of Creatinine in Serum, Metrologia, 40, Tech. Suppl. 08005 (2003)

4 Certificate of Analysis, SRM 1951c Lipids in Frozen Human Serum, National Institute of Standards and Technology (2013) (www.nist.gov/srm/index.cfm)

5 Ellerbe P., Meiselman S., Sniegoski L.T., Welch M.J., White E.V. Determination of Serum Cholesterol by a Modification of the Isotope Dilution Mass Spectrometric Definitive Method, Anal. Chem. 61(15), 1718-1723 (1989); Erratum: Anal. Chem. 62(9), 976 (1990)

6 Rousseeuw P.J. and Croux, C., Alternatives to the Median Absolute Deviation, J. Am. Stat. Assoc. 88(424), 1273-1283 (1993)

13



APPENDIX A: CCQM-K6.2 Summary of Analytical Information The following Tables summarize the analytical information provided by the participants in the “Analytical Information” worksheet of the “CCQM-K6.2 Reporting Form” Excel workbook. The summary is provided as three Tables:

Table A-1: CCQM-K6.2 Sample Size, Extraction, and Cleanup, Table A-2: CCQM-K6.2 Analytical Techniques, and Table A-3: CCQM-K6.2 Calibrants and Standards.

DISCLAIMER Certain commercial equipment, instruments, or materials are identified in these Tables to specify adequately experimental conditions or reported results. Such identification does not imply recommendation or endorsement by the National Institute of Standards and Technology or other participant in this Key Comparison, nor does it imply that the equipment, instruments, or materials identified are necessarily the best available for the purpose.

A1 of 6



Table A-1: CCQM-K6.2 Sample Size, Extraction, and Cleanup



Sample



NMI/DI Size (g)



Extraction Method



Post Extraction Cleanup



Anchor 0.1



Basic hydrolysis (KOH) at 37 ºC for 3 Hexane extract evaporated to



Laboratory



h; liquid/liquid extraction with



dryness and derivatized with N,O-



(NIST)



hexane.



bis(trimethyl)acetamide (BSA) at



65 ºC for 30 min.



CENAM 0.5



Basic hydrolysis and liquid/liquid Free cholesterol in alcoholic



extraction using cyclohexane for 30 medium was derivatized with



min.



N-methyl-N-(trimethylsilyl)



trifluoroacetamide in



cyclohexane; heat at 60 ºC for 1 h;



after cooling to room temperature



adding 0.5 mL pyridine.



HSA



0.1



Basic hydrolysis at 50 ºC for 3 h



GC-MS: Ethanolic solution



followed by liquid/liquid extraction derivatized using



using cyclohexane; extract evaporated N,O-bis(trimethylsilyl)acetamide



to dryness and reconstituted in



(BSA).



ethanol.



LC-MS: Ethanolic solution



diluted with MeOH/water for



injection to LC-MS.



INTI



0.2



Basic hydrolysis for 3 h followed by Hexane extract derivatized using



liquid/liquid extraction using hexane N,O-bis(trimethylsilyl)acetamide



for 1 min



INMETRO 0.035 Basic (KOH) hydrolysis for 1 h



Extract residue derivatized using



followed by liquid/liquid extraction MSTFA at 60 ºC for 15 min



using hexane; extract evaporated to



dryness



KRISS 0.075 Basic (KOH) hydrolysis for 3 h



No further cleanup



followed by liquid/liquid extraction



using hexane; extract evaporated to



dryness and reconstituted with ethanol



LNE



0.08 Basic (aqueous KOH and ethanol) Hexane extract derivatized using



hydrolysis for 2 h followed by



MSTFA/pyridine.



liquid/liquid extraction using hexane



NIMT



0.1



Basic (KOH) hydrolysis at 50 ºC for 3



h followed by liquid/liquid extraction



using hexane for 3 h. Hexane extract



evaporated to dryness and



reconstituted in MeOH



A2 of 6



Table A-1: CCQM-K6.2 Sample Size, Extraction, and Cleanup (Continued)



NMI/DI UME

VNIIM



Sample



Extraction Method



Post Extraction Cleanup



Size (g)



0.2



Basic (aqueous KOH/ethanol)



Cyclohexane extract filtered



hydrolysis at 50 ºC for 4 h followed through 0.2 µm membrane filter



by liquid/liquid extraction using



cyclohexane (5 min)



0.1



Alkaline (NaOH) hydrolysis followed Derivatization with BSTFA +



by liquid/liquid extraction using



10 % TMCS



hexane



A3 of 6



Table A-2: CCQM-K6.2 Analytical Techniques



Analytical Chromatographic



Chromatographic and



NMI/DI Method



Column



Mass Spectrometry Conditions



Anchor GC-MS DB5-MS 30 m



Split mode injection (8:1); 200 ºC 0.5 min hold, 20



Laboratory



capillary column ºC/min to 300 ºC 5 min hold. MSD quadrupole at



(NIST)



200 ºC, source at 230 ºC



Ions monitored: m/z 458 cholesterol



trimethylsilyl ether and m/z 461 labeled



cholesterol trimethylsilyl ether



CENAM GC-MS HP-1MS capillary Split mode injection: 190 ºC 1 min, 30 ºC/min to



column, 30 m × 0.32 280 ºC hold 10 min; He carrier gas at 0.9 mL/min



id, 0.25 µm film constant flow. MSD: transfer line at 270 ºC,



thickness



quadrupole at 150 ºC, source at 250 ºC



Ions monitored: m/z 458 cholesterol



trimethylsilyl ether and m/z 464 deuterated



cholesterol trimethylsilyl ether



HSA



GC-MS; GC-MS: DB5-MS, GC-MS: Inlet at 280 ºC, 70 ºC to 300 ºC at 50



LC-MS 15 m × 0.25 mm id ºC/min then hold 6 min; Flow at 1.0 mL/min;



× 0.25 µm film transfer line at 270 ºC



thickness



Ions monitored: m/z 458 and m/z 460 (IS)



LC-MS: Hypersil (quantifying ions) and m/z 368 and m/z 370 (IS)



GOLD Phenyl, 100 (confirmatory ions) LC-MS: 93 % methanol/7 %



mm × 2.1 mm, 3 µm 10 nmol/L ammonium formate at 0.5 mL/min



particles



Ions Monitored: m/z 369 and m/z 371 (IS)



(quantifying ions)



INMETRO GC-MS VF1ms, 10 m × 110 170 ºC for 1 min, 30 ºC/min to 280 ºC and hold



mm id × 0.1 µm 10 min. Split mode injection; helium carrier gas



film thickness



Mass Selective Detector: Ions monitored: m/z



458.4 cholesterol trimethylsilyl ether and m/z



464.4 deuterated cholesterol trimethylsilyl ether



INTI



GC-MS HP5-MS capillary Inlet at 280 ºC, 180 ºC to 250 ºC at 40 ºC/min



column, 30 m × 0.25 then to 320 ºC (hold 2 min) at 20 ºC/min. Flow at



id, 0.25 µm film 0.5 mL/min; split injection (1:100). Ions



thickness



monitored: m/z 129, 329, 368, and 458



(quantifying ions) and m/z 131, 333, 374, and



464 (IS) (quantifying ions)



KRISS LC-MS/MS Hypersil ODS 100 Mobile phase: (A) 1 % acetic acid in water, (B)



mm × 2.1 mm, 3 µm 0.05 % acetic acid in MeOH; 1 % A and 99 % B



particles



isocratic



LNE



GC-MS DB5-MS capillary Initial 100 ºC, then 20 ºC/min to 280 ºC and hold



column, 30 m × 250 8 min; split injection (20:1) at 270 ºC; Mass



µm id, 0.25 µm film Selective Detector at 230 ºC



thickness



Ions monitored: m/z 458 and m/z 460



A4 of 6



Table A-2: CCQM-K6.2 Analytical Techniques (Continued)



NMI/DI NIMT UME

VNIIM



Analytical Chromatographic



Chromatographic and



Method



Column



Mass Spectrometry Conditions



LC-MS/MS Haisil C18, 100 mm Isocratic mobile phase 20 % isopropanol with



× 0.3 mm id, 5 µm 0.1 % formic acid and 80 % methanol with 0.1 %



particles



formic acid at 0.8 mL/min



LC-MS/MS Kintex C18 100 mm Mobile phase: isocratic at 80 % acetonitrile and



× 2.1 mm, 2.6 µm 20 % methanol at 0.25 mL/min



particles



Ions Monitored: m/z 369.0 and 161.0 and m/z



372.3 and 161.0 (IS)



GC-MS Rtx-5MS 20 m × Initial 70 ºC, then 15 ºC/min to 270 ºC and hold 10



0.18 mm × 0.18 µm min



film thickness



Ions monitored: m/z 368 and m/z 370



A5 of 6



Table A-3: CCQM-K6.2 Calibrants and Standards



NMI/DI Anchor Laboratory (NIST) CENAM

HSA

INMETRO

INTI KRISS LNE NIMT

UME VNIIM



Quantification Type of



Source of



Method Calibration



Internal Standard



Traceability



IDMS



bracketing Cholesterol-25,26,27-13C3 (99 NIST SRM 911c atom % 13C, 99 % CP) (Isotec,



Miamisburg, OH)



IDMS



bracketing Labeled cholesterol of 99 % Purity assessed at



purity added before hydrolysis CENAM using



HPLC, DSC, and



Karl Fischer



IDMS



6-point



13C2-cholesterol (Cambridge NIST SRM 911c



calibration Isotopes) of 99.6 % purity



added during gravimetric



preparation of sample and



calibrants



IDMS



One standard Deuterium-labeled (6)



NIST SRM 911c



point-to- cholesterol (Cambridge



point



Isotopes) of 99.8 % purity,



isotopic enrichment 98.3 %



IDMS



bracketing Deuterium-labeled (6)



NIST SRM 911c



cholesterol (Cambridge



Isotopes)



IDMS/MS



bracketing Deuterium-labeled (4)



SRM 911c



cholesterol (CDN Isotopes) (NIST)



IDMS



5-point



13C2-cholesterol (Cambridge NIST SRM 911c



calibration Isotopes) of 99 % purity added



prior to hydrolysis



IDMS/MS



Exact



13C2-cholesterol (Cambridge Calibration blend



matching Isotopes)



prepared from



IDMS with



matrix matched



single-point



NIST SRM 1951b



calibration



IDMS/MS



2-point



13C3-cholesterol added prior to NIST SRM 1951b



calibration hydrolysis



NIST SRM 968e



IDMS



Single-point 13C2-cholesterol (Cambridge NIST SRM 911c



Isotopes)



A6 of 6



APPENDIX B: CCQM-K6.2 Summary of Uncertainty Estimation Methods The following are pictures of the uncertainty-related information provided by the participants in the “Analytical Information” worksheet of the “Reporting Form” Excel workbook. Information is grouped by participant and presented in alphabetized acronym order.

B1 of 13



Uncertainty Information from CENAM



wx



=



   







m2 mI 2



 ⋅



w2



⋅ (Rx



−



R1 ) R2



− −







m1 mI1



R1



 ⋅



w1



⋅ (Rx



−



R2 )







 



⋅







mIx mx











w1 w2 R1 R2 mi (1)



Mass fraction of the solution calibration standard (low level) (mg/g) Mass fraction of the solution calibration standard (high level) (mg/g) Response relationship of low level solution Response relationship of high level solution Mass of the isotope solution added to the low level solution calibration standard (g).



m1 mi (2) m2

mx m Ix Rx



Mass of the analyte standard solution of low level calibration (g) Mass of the isotope solution added to the high level solution calibration standard (g). Mass of the analyte standard solution of high level calibration (g)

Mass of sample to be measured (g). Mass isotope of the solution added to the sample (g). Instrument response relationship (GC or LC) between the analyte in the sample and its isotope added (dimensionless).



Symbol



Value



Units



w1



2.3512 mg/g



WNA (w2)



2.7385 mg/g



R1



1.6887



R2



2.0109



mi (1)



0.49931 g



m1



0.5004 g



mi (2)



0.5009 g



m2



0.5024 g



mx



0.4950 g



m Ix



0.4946 g



Rx



1.7792



mathematical model uncertainty Repeatibility between subsamples Combined Uncertainty

Expanded Uncertainty



0.0172

0.0383 0.0420

0.0908



k(95%)



2.16



Uncertainty source experimental experimental experimental experimental experimental experimental experimental experimental experimental experimental experimental



Type of distribution normal type A

normal type A normal type A normal type A normal type B normal type B

normal type B normal type B normal type B

normal type B

normal type A



Standard uncertainty Units

0.0036 mg/g 0.0034 mg/g 0.006570 0.004795 0.00003 g 0.00003 g 0.00002 g 0.00003 g 0.000051 g 0.000053 g 0.0089



Relative uncertainty ui(y)

0.1515% 0.1255% 0.3891% 0.2385% 0.0061% 0.0054% 0.0050% 0.0051% 0.0103% 0.0107% 0.4995%



0.7%



5.1%



B2 of 13



Uncertainty Information from HSA



The mass fraction of total cholesterol in serum was calculated based on the IDMS calibration curve as follows:



CX



=



(mRB



+



b)×



WY MX



=



(mRB



+



b)×



M Y CY MX



(1)



where C X = mass fraction of total cholesterol in the serum sample M X = mass of serum sample (determined by weighing) M Y = mass of isotope standard solution (determined by weighing) W Y = mass of the isotope labeled standard spiked into the serum sample (equals to M Y × C Y ) R B = peak area ratio of sample blend (determined by GC-MS or LC-MS measurements) C Y = concentration of isotope labeled standard solution (determined by weighing and from purity of the isotope labeled standard) m = gradient of the slope of linear regression plot (determined by the linear fit of the isotope mass ratio and the peak area ratio of the calibration blends) b = intercept on y axis of the linear regression plot (determined by the linear fit of the isotope mass ratio and the peak area ratio of the calibration blends)



For the estimation of uncertainty, considering R M = mR B + b , and let R M = R M ´C Y /C Z , Equation (1) is converted to:



CX



=



RM



×



MY CZ MX



(2)



where R M = isotope mass ratio in sample blend C Z = concentration of cholesterol in the calibration standard solution



A standard uncertainty was estimated for all components of the measurement in Equation (2), which were then combined using respective derived



sensitivity coefficients to estimate a combined standard uncertainty in the reported result of total cholesterol in the serum samples. A coverage factor k



with a value of 2 was used to expand the combined standard uncertainty at a 95 % confidence interval. Possible sources of biases [method precision (F P ), choice of different ion pair (F I ), and other factors during sample extraction (F C1 ) and derivatisation (F C2 )] are accounted for in the final uncertainty



budget with the use of the measurement equation:



CX



=



FP × FI



×



FC1



×



FC



2



×



RM



'×



M Y CZ MX



(3)



The sensitivity coefficients of each component can be expressed as follows:



∂C X = C X ∂RM ' RM '



∂C X = C X



∂M Y



MY



∂C X = − C X



∂M X



MX



∂C X = C X ∂C Z C Z



∂C X = C X



∂FP



FP



∂CX = CX



∂FI



FI



∂CX = CX ∂FC1 FC1



∂CX = CX ∂FC 2 FC 2



The standard uncertainty of each component was calculated as follows: (1) M Y and M X : The standard uncertainty was calculated based on the calibration report using the standard weights calibrated by the National Metrology Centre, A*STAR. (2) F P : The pooled standard deviation of the mean of the GC-MS and LC-MS results for each sample was used as the standard uncertainty of method precision. (3) F I : The standard deviation of the difference of the results using two ion pairs divided by the square root of the number of samples (for insignificant difference using t-test) or the average of the difference of the results using two ion pairs divided by 2 (for significant difference using t-test). (4) F C1 and F C2 : A relatively standard uncertainty of 0.2 % was employed for each of these two factors. (5) C Z : The certified purity and uncertainty of NIST SRM 911c in combination with the uncertainty of weighing for the preparation of the calibration standard solution. (6) R M ' : Consider R M = R M ' ×C Z /C Y , the conversion of equation R M = mR B + b leads to:

R B = (C Z ×R M ') / (C Y ×m) - b/m Let m' = C Z /(C Y ×m) and b' = - b/m , we have: R B = m'R M ' + b' The standard uncertainty of R M ' was calculated using the following equation:



B3 of 13



Uncertainty Information from HSA (continued)



µ RM



=



1 m'



×



sy/x



×



1 +1+



( )2

RB − RBc



∑ N



n



n

m'2 (RMc − RMc )2



(4)



i =1



where



s y/x = standard deviation of the regression



R B = peak area ratio of sample blend



= average peak area ratio of calibration blends



nRB=c number of calibration blends used for the linear regression plot



N = injection time for each sample



R Mc = isotope mass ratio in calibration blends



= average isotope mass ratio in calibration blends



TheRcMocmbined standard uncertainty was calculated using the equation below:



where



∑ u =



ci2uxi 2



(5)



i



u = combined standard uncertainty



c i = sensitivity coefficient of each component



u xi = standard uncertainty of each component



The expanded uncertainty (U ) was calculated by mutiplying the combined standand uncertainty (u ) with a coveragy factor (k = 2) for a confidence level of



95 %.



Factor M X (g) M Y (g) C Z (µg/g) RM' F P (µg/g) F I (µg/g) F C1 (µg/g) F C2 (µg/g)



Table 1. Uncertainty Budget for Sample 1



Value x



Relative



Sensitivity



Uncertainty Uncertainty Coefficient ( c )



Contribution



u(x)



u(x)/(x)



δCx/δx



c2 . u(x)2



%



0.0960 0.000099 0.103%



24545.16 5.9042



3.2%



0.6538 0.000099 0.015%



3605.11 0.1274



0.1%



1519.3



5.1353 0.338%



1.55 63.4705



34.5%



1.1505



0.0034 0.299%



2048.71 49.6015



27.0%



2357



2.2651 0.096%



1.00 5.1308



2.8%



2357



3.9038 0.166%



1.00 15.2398



8.3%



2357



4.7140 0.200%



1.00 22.2220



12.1%



2357



4.7140 0.200%



1.00 22.2220



12.1%



Factor M X (g) M Y (g) C Z (µg/g) RM' F P (µg/g) F I (µg/g) F C1 (µg/g) F C2 (µg/g)



Table 2. Uncertainty Budget for Sample 2



Value x



Relative



Sensitivity



Uncertainty Uncertainty Coefficient ( c )



Contribution



u(x)



u(x)/(x)



δCx/δx



c2 . u(x)2



%



0.0992 0.000099 0.100%



23708.42 5.5085



3.0%



0.6565 0.000099 0.015%



3582.97 0.1258



0.1%



1519.3



5.1353 0.338%



1.55 63.2069



34.6%



1.1505



0.0034 0.299%



2044.45 49.3955



27.0%



2352



2.2604 0.096%



1.00 5.1095



2.8%



2352



3.8957 0.166%



1.00 15.1765



8.3%



2352



4.7042 0.200%



1.00 22.1297



12.1%



2352



4.7042 0.200%



1.00 22.1297



12.1%



B4 of 13



Uncertainty Information from INMETRO



All factors from the measurement equation were considered in the uncertainty estimation.



All of the evaluated uncertainties were of type B except for the R'B and R'Bc repeatabilities.



Hence their standard uncertainties were obtained by dividing the expanded uncertainties



by the coverage factors encountered in the certificates.



For the repeatabilities, standard uncertainties were obtained by the standard errors of the means



The standard uncertainties were multiplied by their sensitivity coefficients using the



GUM methodology and then combined using the square root of the squared sum of the components.



Effective degrees of freedom were calculated and the coverage factors for 95 % probability



were taken for the expanded uncertainties.



Factor



% contribution



The full uncertainty budget is presented below as for sample 1:



mfinal msolute

P



0,00001 2,14496 8,70135



mz



0,13319



myc



0,03935



my



0,04002



mx R'B R'Bc Total



0,15553 37,89210 50,89350

100



Method was validated by the preparation by two different analysts of the CRM from NIST 909c. These results showed that both analysts were capable of generating results equivalent to the

certified property values for the CRM by comparison of the Δm (absolute difference between the mean measured value and the certified value) and the UΔ (expanded uncertainty of the difference between the measurement result and the certified value), obtaining Δm < UΔ which means the measured value and the certified value have no significant differences according to ERM Application Note 1.

These experiments demonstrated repeatability, intermediate precision and trueness (bias) evaluations of the method.



B5 of 13



Uncertainty Information from INTI



The picture can't be displayed.



wx



=



mIx ⋅ Rx ⋅ m0 mx ⋅ R0 ⋅ mI 0



⋅ w0



w0 Fracción de masa de la disolución estándar de calibración. (concentracion STD) R0 Relación de respuesta del instrumento (CG o CL) entre el analito en el patrón de calibración y el isótopo adicionado (adimensional). mI0 Masa de disolución de isótopo adicionado a las disoluciones patrón de calibración (g). (masa STDi al STD) m0 Masa de disolución de analito patrón de calibración (g) (masa STD agregada a los viales) mx Masa de muestra problema a medir (g). mIx Masa de la disolución de isótopo adicionado a la muestra (g). Rx Relación de respuesta del instrumento (CG o CL) entre el analito en la muestra y su isótopo adicionado (adimensional).



Parámetro (simbolo)

w0 R0

mI0

m0



Descripción



Valor



Fracción de masa de la disolución estándar de

calibración. (concentracion STD) Relación de respuesta del instrumento (CG o



1.602



CL) entre el analito en el patrón de calibración 1.576



y el isótopo adicionado (adimensional). Masa de disolución de isótopo adicionado a las



disoluciones patrón de calibración (g). (masa 0.156



STDi al STD) Masa de disolución de analito patrón de



calibración (g) (masa STD agregada a los 0.315



viales)



mx



Masa de muestra problema a medir (g).



0.1487



m Ix Rx



Masa de la disolución de isótopo adicionado a la muestra (g). Relación de respuesta del instrumento (CG o CL) entre el analito en la muestra y su isótopo adicionado (adimensional).



0.1453 1.631



Bias measuring SRM 909c



1.398



Unidades mg/g



Origen de la



incertidumb Tipo de



Incetidumbr



re



distribution e estandar



certificado/e xperimental



normal tipo A



0.016825



Units mg/g



1 experimental normal tipo A 0.001795



g



certificado/e xperimental



normal tipo B



0.000035



g



g



certificado/e xperimental



normal tipo A



0.000038



g



g



experimental normal tipo A 0.000122 g



g



experimental normal tipo A 0.000088 g



experimental normal tipo A 0.0039



experimental normal tipo A



0.1



Incertidumbr e relativa ui(y)

1.05% 0.11%

0.02%

0.01% 0.08% 0.06% 0.24% 7.15%



u(Cmtra) Inc. combinada



U exp



Inc expandida



k(95%) 2



0.1241 0.072 0.2482



B6 of 13



Uncertainty Information from KRISS

  =(  (      −        ,                 )∙  (    −        ))/       ∙[((〖    〗              −〖    〗1)/(〖    〗2−〖    〗1 ))∙(〖    〗(        , 2)−〖    〗(        , 1) )+〖    〗(        , 1) ]

Here, Mis-sol,spiked is the weight of the cholesterol-d4 solution spiked in the sample, Cs-sol is the concentration of the cholesterol standard solution (mg/kg), and Ws is the weight of the sample, . ARsample is the observed area ratio of cholesterol/cholesterol-d4 of the sample from the LC/MS/MS measurement, ARi is the observed area ratio of cholesterol/cholesterol-d4 of the calibration standard

mixture i (i=1,2) from the LC/MS/MS measurement, and MRmix,i is the weight ratio of the cholesterol solution/cholesterol-d4 solution in the calibration standard mixture i (i=1,2) from the LC/MS/MS measurement.

Measurement protocol: each subsample was separately measured by LC/MS/MS in comparison with Isotope ratio standard

B7 of 13



Uncertainty Information from LNE C = (aR 458/460+ b) x ((mLabCLab)/ mser ))



C = mass fraction of cholesterol in the serum sample (mg/g) mLab = mass of labeled cholesterol solution (g) CLab = concentration of labeled cholesterol solution (mg/g) a = gradient of the slope for linear regression plot b = intercept on y axis for the linear regression plot R 458/460 = unlabeled/labeled ion peak area ratio of serum sample mser = mass of serum sample (g)



sample1



Component



Type (A or B) relative Uncertainty (%)



Purity of primary standard



B



2.45%



preparationof sample blends (weighings)



B



6.66%



Calibration model



B



5.63%



Preparation of calibration blend (weighings)



B



1.79%



Precision



B



83.47%



sample2



Component



Type (A or B) relative Uncertainty (%)



Purity of primary standard



B



2.06%



preparationof sample blends (weighings)



B



5.88%



Calibration model



B



5.63%



Preparation of calibration blend (weighings)



B



1.51%



Precision



B



84.92%



B8 of 13



Uncertainty Information from NIMT



Expanded measurement equation:



wx



=



FP .FE .FI .wz ,c



⋅



my mx



⋅ mzc ⋅ myc



⋅ R'b R'bc



wz,c is the mass fraction of analyte in the calibration solution used to prepare the calibration blend



my



is the mass of spike solution added to sample blend



my,c is the mass of spike solution added to calibration blend



mx



is the mass of sample added to sample blend



mz,c is the mass of standard solution added to calibration blend



R'B and R'B,C are the observed isotope amount ratios in the sample blend and the calibration blend, respectively



FE



is the extraction efficiency factor



FP



is the method precision factor



FI



is the interference effect factor



u(wx ) = wx



 



u(wZ ,C wZ ,C



)



2 



+



 



u(mY m

Y



)



2   



+



 



u(mY ,C mY ,C



)



2 



+







u(mX mX



)



2  



+







u(mZ ,C mZ ,C



)



2  



+







u ( FP FP



)



2



+







u ( FE FE



)



2



+







u ( FI FI



)



2



u(wz,c) is the standard uncertainty of the mass fraction of analyte in the calibration solution used to prepare the calibration blend. The value was estimated from the certified mass fraction value of matrix-matched calibration standard, masses weighed for preparation of calibration standard and uncertainty using different standards (standard comparison).

u(my), u(my,c), u(mx) and u(mz,c) are standard uncertainties of the masses. These values were estimated from the bias and precison effect of the balance.

u (FP) is the standard uncertainty of the precision factor. This value was estimated from standard deviation of the multiple IDMS results. u(FE) is the standard uncertainty of the extraction efficiency factor which was estimated from the extraction and protein precipitaion u(FI) is the standard uncertainty of the interference effect. This value was estimated from potential bias between primary ion pair and secondary ion pair of the MRM program.



Note: For the uncertainty contributing to the R'B and R'B,C ,the precision in measuring the isotope amount ratios of the analyte and the internal standard in the sample and calibration blends was assumed to be incorporated in the overall method precision.The effect of any biases on these ratios was assumed to be negligible as any systematic biases should cancel out since the calibration blends and sample blends were exact-matched for analyte concentration and isotope ratio. Other biases that may arise from extractions are captured in other factors.



B9 of 13



Uncertainty Information from NIMT (Continued)



Uncertainty budget of Cholesterol (sample I)



Factor



Values



Uncertainties



x



u(x)



u(x)/(x)



Parameter (unit)



Method Precision, FP(1) mz,c (g) my (g)



1.0000 0.08372 0.07478



0.01004 0.000049 0.000049



1.004% 0.0591% 0.0662%



my,c (g)



0.07489



0.000049



0.0661%



mx (g) wz,c (ug/g)



0.10035 0.2295



0.000049 0.0016



0.0493% 0.6944%



Additional Factors



Extraction effects, FE (1) Interference from two different ion pairs, FI (1)



1.000 1.000



0.0100 0.0028



1.000% 0.283%



Uncertainty Analysis Results



wx=



2.386



ug/g



u(x) =



0.038



ug/g



u(x)/x =



1.61%



Veff(total) =



27.151



k=



2.05



(@ 95% level)



U(x) = 0.079 ug/g %U(x) = 3.30%



Uncertainty budget of Cholesterol (sample II)



Factor



Values



Uncertainties



x



u(x)



u(x)/(x)



Parameter (unit)



Method Precision, FP(1)



1.0000



0.01407



1.407%



mz,c (g)



0.08372



0.000049



0.0591%



my (g) my,c (g)



0.07478 0.07489



0.000049 0.000049



0.0662% 0.0661%



mx (g)



0.10035



0.000049



0.0493%



wz,c (ug/g)



0.2295



0.0015



0.6617%



Additional Factors



Extraction effects, FE (1) Interference from two different ion pairs, FI (1)



1.000 1.000



0.0100 0.0036



1.000% 0.357%



Uncertainty Analysis Results



wx=



2.380



ug/g



u(x) =



0.045



ug/g



u(x)/x =



1.89%



Veff(total) =



25.181



k=



2.06



(@ 95% level)



U(x) = 0.092 ug/g %U(x) = 3.89%



B10 of 13



Uncertainty Information from UME



RF = AABx xC ISx



RF : Response factor



AISx xC ABx



CABx : Concentration of native compound (mg/g)



AABx : Peak area of native compound



AISx : Peak area of labelled compound



CISx : Concentration of labelled compound (mg/g)



1-Mass of sample



Value



Standard Uncertainty



Mass of compound Calibration

Mass of Tare Calibration



mCompound mtare



uCmCompound uCmtare



u (mCompound ) =



uCmCompound



+ u 2



2



CmTare



2-Mass of Labelled STD



Mass of labelled compound Calibration

Mass of Tare Calibration



Value mCompound

mtare



Standard Uncertainty uCmCompound uCmtare



u (mCompound ) 13 C 3 = uCmCompound



+ u 2

13 C 3



2 CmTare



3-Labelled Compounds Stock Solution



Mass of Compound13C3

Calibration Mass of Tare

Calibration Mass of Solvent

Calibration



Value m13C3

mtare

msolvent



Standard Uncertainty uCmC13C3 uCmtare uCmsolvent



u (mstock ) 13C 3 =



u + u + u 2 stock 13 C 3



2 CmSolvent



2 CmTare



5-Method Precision



where, u (rep ): Uncertainty of repeatability SD : Standard deviation



u(rep ) = SD n



n : Number of sample



6-Instrument Repeatability



where,



u (rep ): Uncertainty of repeatability u(rep ) = SD



SD : Standard deviation



n



n : Number of sample



7-Calibration Graph



u(c0 )



=



S B1



1 + 1 + (c0 − c )2



pn



S xx



n

∑ S xx = (ci − c)2 i=1



S: Residual standard deviation



B1: Slope



p: number of measurement to determine c0 n: number of measurement for the calibration



c 0: determined concentration

: mean value of the different calibration standards i : index for the number of calibration standards



4- Uncertainty of calibration standard



Value



Standard Uncertainty



Mass of calib Calibration

Mass of Tare Calibration



mCompound mtare



uCmCompound uCmtare



u (mCalib ) =



u + u 2 CmCalib



2 CmTare



B11 of 13



Uncertainty Information from UME (Continued)



CCQM SAMPLE 1



Parameter Mass of sample (mg) Mass of labelled std(mg) Labelled stock solution (mg/kg)

Uncertainty of calibration standard level 2 (mg) Uncertainty of calibration standard level 3 (mg) Method Precision Instrument repeatability



Value(X) 2.046E+02 7.566E+01 4.000E+03 2.070E+02 2.063E+02 1.000E+02 1.000E+02



Calibration curve Relative Combined Uncertainty Result (mg/g)



2.265E+00 2.265E+00



Combined Standard Measurement Uncertainty Expanded Uncertainty (k=2)



Relative Uncertainty



u(x) 2.860E-05 3.910E-06 8.303E-03 2.927E-05 2.907E-05 1.887E-01 1.770E-01 3.036E-02

3.092E-02 6.185E-02 2.730E+00



CCQM SAMPLE 3



Parameter Mass of sample (mg) Mass of isotopic standard (mg) Labelled stock solution (mg/kg)

Uncertainty of calibration standard level 2 (mg) Uncertainty of calibration standard level 3 (mg) Method Precision Instrument repeatability Calibration curve (mg/g) Relative Combined Uncertainty



Value(X) 2.053E+02 7.651E+01 4.000E+03 2.070E+02 2.063E+02 1.000E+02 1.000E+02 2.310E+00



Result (mg/g)



2.310E+00



Combined Standard Measurement Uncertainty



Expanded Uncertainty (k=2) Relative Uncertainty



u(x) 2.878E-05 3.999E-06 8.303E-03 2.927E-05 2.907E-05 1.887E-01 1.556E-01 3.248E-02

3.297E-02 6.594E-02 2.855E+00



u(x)/X 1.398E-07 5.168E-08 2.076E-06 1.414E-07 1.409E-07 1.887E-03 1.770E-03 1.340E-02 1.365E-02

u(x)/X 1.402E-07 5.227E-08 2.076E-06 1.414E-07 1.409E-07 1.887E-03 1.556E-03 1.406E-02 1.428E-02



B12 of 13



Uncertainty Information from VNIIM

W=(San*mis)/(Sis*m*F) W - mass fraction of the creatinine in the sample, mg/g; mis - mass of internal standard added to sample before sample preparation, mg; m - mass of sample, g; F - response factor; F=(Sancal*Cis)/(Siscal*Can) Cancal- concentration of creatinine in calibration solution; Cis - concentration of internal standard in calibration solution Sancal - peak area for the creatinine; Sis - peak area for the internal standard



Source of uncertainty mass of sample (m) mass of internal standard added to sample before extraction (mIS) response factor (F) purity of referense standard preparation of calibration solution RSD of F determination RSD of results, %



u, % 0.29 0.58 0.85 0.12 0.82 0.19 0.47



comb.std uncertainty



1.16



expanded uncertainty (k=2)



2.32



B13 of 13