use of serum myoglobin assays for urine myoglobin measurements

Use of Serum Myoglobin Assaysfor Urine MyoglobinMeasurements Can CauseFalse-Negative Results

To the Editor:

Several conditions, includingrhabdomyolysis, trauma, and sur-gery, are associated with the releaseof large amounts of myoglobininto the circulation. Saturation ofthe salvage system of the kidneywill produce myoglobinuria, acondition associated with acute re-nal failure (ARF). The exact mech-anism is not known, but precipita-tion of myoglobin in the tubulesand myoglobin-mediated forma-tion of free radicals have been pos-tulated (1 ).

Early identification of severemyoglobinuria permits acute treat-ment, thereby avoiding ARF.Quantitative myoglobin assayshave several pitfalls, however, in-cluding discrepancies in myoglo-bin recovery (2 ). The lack of quan-titative information precludesdifferentiating mild myoglobin-uria, which probably does notcause ARF, from severe myoglo-binuria. Therefore, myoglobin im-munoassays developed for serumapplications have been adapted foruse with urine samples (3–5 ). Priorreports have recommended alka-linizing urine specimens beforestorage.

Because quantification of urinemyoglobin might be valuable inthe early evaluation of patientswho might have rhabdomyolysis,a rapid in-house myoglobin assayis desirable. To this end, we evalu-ated 4 different commercial serummyoglobin assays for their poten-tial to measure myoglobin in urine.These 4 assays can be run on theanalyzers present in our laboratory[Integra 700 (Roche), Elecsys 1010(Roche), Aeroset (Abbott Diag-nostics), and BN ProSpec (DadeBehring)]. All 4 assays are based onimmunogenic detection of myo-

globin, but by different techniques,namely turbidimetry, nephelome-try, and a heterogeneous immuno-assay. The assays were evaluatedfor imprecision with control mate-rials provided by the manufactur-ers. For all assays, interassay CVswere �5% for all tested concentra-tions (50 –1100 �g/L).

Although spiking of purifiedmyoglobin into urine samplescould be used to study myoglobinstability in urine, we preferred touse patient-derived material,thereby circumventing the intrin-sic instability of purified myoglo-bin. We collected urine samplesfrom patients with rhabdomyoly-sis, who were selected because oftheir dramatically increased serumactivities of creatine kinase. Thesamples were alkalinized with so-dium hydroxide solution to a pH�8. The measured myoglobin con-centrations in these patients ex-ceeded the linear ranges of the 4 as-says. We therefore diluted theurine samples in myoglobin-freeurine, which was adjusted to pH4.5, pH 7, or pH 8.5. Fig. 1 showsthe results for patient A. The origi-nal myoglobin concentration inthe urine of patient A was approx-imately 30 000 �g/L as measuredwith the Elecsys 1010 kit. This sam-

ple was diluted to a concentrationof approximately 2000 �g/L myo-globin. Acidifying the samples dra-matically decreased the measuredmyoglobin concentrations on all 4platforms (Fig. 1). The lower mea-sured concentration for the Elecsysplatform probably reflects varia-tion in the calibration procedure,which was not investigated further.Similar results were obtained with2 other patient samples, whichshowed a decrease in the measuredmyoglobin concentration of atleast 5-fold after the samples wereacidified. The decrease in mea-sured myoglobin was also observedwith urine samples acidified up topH 6, although to a less pro-nounced extent (data not shown).These findings, which were ob-tained with assays that are cur-rently widely available, are in linewith older studies of myoglobinstability that used spiked myoglo-bin as well as urine from patientswith rhabdomyolysis (4, 5 ). Toour knowledge, only the BNProSpec method has previouslybeen investigated (5 ). We foundthat the measured myoglobin con-centration decreased up to 50%within 1 week of storage of alkalin-ized urine at 4 °C, �20 °C, or�70 °C, with different kinetics

Fig. 1. Measured urine myoglobin concentrations at 3 different pH valueson 4 different assay platforms.

Clinical Chemistry 55:61240–1252 (2009) Letters to the Editor

1240

in different patients (data notshown).

Because alkalinizing urine canstabilize myoglobin, we attemptedto recover myoglobin immunore-activity in acidic urine. We did notrecover any of the measured myo-globin but did observe that myo-globin was stabilized. Immediatealkalinization of urine at the bed-side has been suggested as an ap-propriate sample-collection proce-dure for myoglobin, because itprevents a major decrease in themyoglobin concentration (4, 5 ).Urine resides for varying lengths oftime in the bladder, however,where it might be acidic. To inves-tigate the influence of a physiolog-ical delay in voiding urine from thebladder on the myoglobin concen-tration in acidic urine, we mim-icked this condition by diluting analkaline urine sample containing8000 �g/L myoglobin withmyoglobin-free urine to 300 �g/Lmyoglobin. The urine was thenacidified to pH 4.5 and incubatedaliquots of the acidified urine at37 °C for various times (0.5, 1, 2, 3,and 4 h). The measured myoglobinconcentration was reduced to�10% in �2 h. Similar resultswere obtained with urine from an-other patient (data not shown).Thus, the myoglobin concentra-tion in urine may already have be-come markedly reduced by a phys-iologically low pH and a longretention time in the bladder by thetime a sample can be collected.Such conditions can produce neg-ative results even while the patientis at high risk for ARF.

This process may also be de-pendent on the magnitude of themyoglobin concentration, causingeven more variation in the poten-tial decline in immunoreactivity (5).

Our results show that none ofthe 4 myoglobin assays developedfor serum applications are suitablefor measuring myoglobin in urine,suggesting that probably allimmunoassay-based myoglobin as-

says are unsuitable. This findingshould alert laboratories that haveadapted serum myoglobin assays formeasuring myoglobin in urine to re-evaluate their validation process.

Author Contributions: All authors con-firmed they have contributed to the intellec-tual content of this paper and have met thefollowing 3 requirements: (a) significant con-tributions to the conception and design, ac-quisition of data, or analysis and interpreta-tion of data; (b) drafting or revising the articlefor intellectual content; and (c) final approvalof the published article.

Authors’ Disclosures of Potential Con-flicts of Interest: No authors declared anypotential conflicts of interest.

Role of Sponsor: The funding organiza-tions played no role in the design of study,choice of enrolled patients, review and in-terpretation of data, or preparation or ap-proval of manuscript.

Acknowledgments: We thank Mariannevan Blokland, Maarten van Gool, andAngela Verheijen for excellent technicalsupport.

References

1. Slater MS, Mullins RJ. Rhabdomyolysis and myo-globinuric renal failure in trauma and surgicalpatients: a review. J Am Coll Surg 1998;186:693–716.

2. Loun B, Copeland KR, Sedor FA. Ultrafiltrationdiscrepancies in recovery of myoglobin fromurine. Clin Chem 1996;6:965–9.

3. Loun B, Astles R, Copeland KR, Sedor FA.Adaptation of a quantitative immunoassayfor urine myoglobin. Predictor in detectingrenal dysfunction. Am J Clin Pathol 1996;105:479 – 86.

4. Wu AH, Laios I, Green S, Gornet TG, Wong SS,Parmley L, et al. Immunoassays for serum andurine myoglobin: myoglobin clearance assessedas a risk factor for acute renal failure. Clin Chem1994;40:796–802.

5. Chen-Levy Z, Wener MH, Toivola B, Daum P,Reyes M, Fine JS. Factors affecting urinary myo-globin stability in vitro. Clin Chem 2005;123:432–8.

Harm de Waard*Peter van ’t Sant

Laboratory of Clinical Chemistry and

Haematology

Jeroen Bosch Hospital

‘s-Hertogenbosch

the Netherlands

* Address correspondence to this author

at:

Jeroen Bosch Hospital

nieuwstraat 34

‘s-Hertogenbosch, NA

the Netherlands 5211 NL

Fax �31-(0)73-6992958

E-mail [email protected]

Previously published online atDOI: 10.1373/clinchem.2008.118968

Urinary �-Trace Protein as aNew Renal Tubular Marker

To the Editor:

Recently, lipocalin-type prosta-glandin D synthase, or �-trace pro-tein (BTP),1 has been introducedas a marker to predict the glomer-ular filtration rate. BTP catalyzesthe isomerization of prostaglandinH2 to prostaglandin D2 and is a li-pocalin lipid-transporter proteinthat binds retinoids, thyroid hor-mones, and bile pigments. It hasbeen detected in cerebrospinalfluid, serum, urine, amniotic fluid,and seminal plasma (1 ).

Like other proteins of low mo-lecular mass, BTP is almost com-pletely filtered through the glo-merulus and then completelyreabsorbed through the tubules.Because of its molecular mass(23–29 kDa), its constant produc-tion rate, and its stability, BTPshares many properties of an idealmarker protein of tubular func-tion. We therefore explored the di-agnostic use of BTP as a novel uri-nary marker in comparison withthat of �1-microglobulin (A1M), aconventional indicator of tubularfunction (2– 4 ).

We used a Behring Nephelom-eter II analyzer (Dade Behring) with

1 Nonstandard abbreviations: BTP, b-trace protein;A1M, a1-microglobulin.

Letters to the Editor

Clinical Chemistry 55:6 (2009) 1241

a latex particle–enhanced fixed-timeimmunonephelometric method toassay BTP in urine (measurementrange, 0.21–13.6 mg/L) and ob-tained an interassay imprecision of4.5% at 1.88 mg/L. A Behring Neph-elometer II analyzer was also used tomeasure urinary albumin by immu-nonephelometry. Urinary A1M, cre-atinine, and total protein were mea-sured on a Modular P analyzer(Roche). A compensated rate-blanked Jaffe method (Roche) wasused for creatinine, a pyrogallol redmethod (INstruchemie) was usedfor total protein, and the Tina-quant® reagent set and method(Roche) was used for A1M.

Reference values for BTP wereestablished from samples of firstvoid urine from 120 apparentlyhealthy individuals [60 males (me-dian age, 30 years; range, 17–72years) and 60 females (median age,34 years; range, 22– 60 years)]. Allurine samples were analyzed theday of collection and stored at�80 °C.

We also analyzed freshly col-lected urine samples from 210 pa-tients (130 males, 90 females; agerange, 0.3–91 years) with variousrenal pathologies (upper and lowerurinary tract infections, spinal cordinjuries, nephrotoxic medication).We measured BTP, A1M (upperreference value, 8 mg/L), urinarycreatinine, urinary total protein,and urinary albumin.

In the reference group, a log-normal distribution of urinaryBTP concentration was observed(P � 0.001, Kolmogorov–Smirnovtest). Slightly higher BTP excretionwas found for males (median, 171mg/mol creatinine; range, 31–597mg/mol creatinine) than for fe-males (median, 109 mg/mol creat-inine; range, 17– 474 mg/molcreatinine). This sex-related differ-ence was not statistically signifi-cant (P � 0.07). The upper refer-ence limit for BTP (97.5 percentile)was 7.79 mg/L (452 mg/mol creat-inine) for men and 3.13 mg/L (450

mg/mol creatinine) for women.No age dependency was found.

In the patient group, log-transformed urinary BTP andA1M concentrations (Fig. 1)showed the following correlation:log y � 0.50 � 0.70(log x), where yand x are the BTP concentrationand the A1M concentration, re-spectively, in milligrams per moleof creatinine (n � 196; r � 0.90;P � 0.0001). In contrast, the log-transformed BTP concentrationshowed a poorer correlation withthe log-transformed urinary pro-tein concentration (n � 196; r �0.64; P � 0.0001) and no correla-tion with the log-transformed uri-nary creatinine concentration (n �196; r � 0.14; P � 0.0791). In anonselective subgroup (n � 47),the correlation (r) with the log-transformed urinary albumin con-centration was 0.46 (P � 0.0022).

The mean BTP concentrationsin extracts of kidney and prostate

tissues (n � 5) were 50 �g/g and 10�g/g protein, respectively, suggest-ing that these organs were only mi-nor sources of the BTP in urine. Ananalysis of samples of seminalplasma (n � 15) revealed a medianBTP concentration of 33.9 mg/L(range, 7.7– 65.3 mg/L). This resultmay explain the slightly higher ref-erence values for men.

BTP was stable in urine over a24-h period with different bufferscovering a broad pH range: an ace-tate buffer (pH 4.6, 0.1 mol/L), aphosphate buffer (pH 7.4, 0.1 mol/L), and a Veronal (barbital) buffer(pH 8.9, 0.1 mol/L).

Tubular proteinuria occurswhen glomerular function is nor-mal but the proximal tubules havea diminished capacity to reabsorband catabolize proteins, causing anincreased urinary excretion of pro-teins of low molecular mass thatusually pass through the glomeru-lus, such as A1M, retinol-bindingprotein, and �2-microglobulin (4 ).

10 100 1000 10 000 100 000 1 000 000

100 000

10 000

1000

100

10

1

A1M (mg/mol creatinine)

BT

P (

mg/

mol

cre

atin

ine)

Fig. 1. Plot of the correlation of the logarithms of the BTP and A1Mconcentrations in milligrams per mole creatinine for the group of renalpatients: log y � 0.50 � 0.70(log x); r � 0.90; P < 0.0001.

The inner dashed lines represent the 95% CI, and the outer dotted lines representthe 95% prediction interval. The upper reference limit for A1M is 127 mg/molcreatinine, and that for BTP is 452 mg/mol creatinine for men and 450 mg/molcreatinine for women.


1242 Clinical Chemistry 55:6 (2009)

Our data strongly suggest thatBTP may be useful as a diagnosticmarker for early detection of renaltubular damage. In the patientgroup, BTP correlated well withA1M. Poor correlations were ob-tained with urinary total protein,urinary albumin, and urinary cre-atinine. BTP is not influenced bythe acute-phase reaction (5 ). Nei-ther kidney nor prostate tissue ex-tracts contain high quantities ofBTP. Therefore, in the presence ofdamage to the urinary tract, theseorgans make only a minor contri-bution to the BTP in urine. More-over, BTP is stable over a wide pHrange and, unlike A1M, does notbind other proteins.




Acknowledgments: We thank VincentVandenbroucke for his practical contribu-tion to this study.

References

1. Melegos DN, Diamandis EP, Oda H, Urade Y,Hayaishi O. Immunofluorometric assay of pros-taglandin D synthase in human tissue extractsand fluids. Clin Chem 1996;42:1984–91.

2. Hoffmann A, Nimtz M, Conradt HS. Molecularcharacterization of b-trace protein in human se-rum and urine: a potential diagnostic marker forrenal diseases. Glycobiology 1997;7:499–506.

3. Oda H, Shiina Y, Seiki K, Sato N, Eguchi N,Urade Y. Development and evaluation of apractical ELISA for human urinary lipocalin-type prostaglandin D synthase. Clin Chem2002;48:1445–53.

4. Penders J, Delanghe JR. Alpha 1-microglobulin:clinical laboratory aspects and applications. ClinChim Acta 2004;346:107–18.

5. Kobata M, Shimizu A, Rinno H, Hamada C,Maeda K, Fukui M, et al. Beta-trace protein, anew marker of GFR, may predict the early prog-nostic stages of patients with type 2 diabeticnephropathy. J Clin Lab Anal 2004;18:237–9.

Liesbeth L. Vynckier2

Katelijne M.J. Flore2

Sigurd E. Delanghe2

Joris R. Delanghe2*

2 Department of Clinical Chemistry

Ghent University Hospital

Ghent, Belgium

* Address correspondence to this author at:

Department of Clinical Chemistry

Universitair Ziekenhuis

De Pintelaan 185

B-9000 Gent, Belgium

Fax �32-9-332-4985



Newborn Screening for SickleCell Disease through Use ofTandem Mass Spectrometry

To the Editor:

We read with interest the recentreport by Boemer et al. (1 ) describ-ing the use of tandem mass spec-trometry for screening of new-borns for sickle cell disease. Wecongratulate the authors on an ex-cellent study that follows thepeptide-based philosophy and ap-proach described in our previouspublications (2, 3 ). Unfortunately,we believe that the authors havesubstantially misrepresented ourdata and their own originality.

The authors suggest that ouroriginal communication (2 ) indi-cated that there was insufficientanalytical sensitivity to distinguishhomozygous and heterozygoussamples. Clearly, this assertion isnot the case, because we presented

figures demonstrating heterozy-gous and homozygous signals forhemoglobin S (Hb S), Hb C, Hb E,Hb DPunjab, and Hb OArab. In addi-tion, we pointed out the value ofthe ratio of the signal for the vari-ant peptide to that of the corre-sponding wild-type peptide, par-ticularly with respect to HbDPunjab. The quantitative nature ofour approach and the value of ra-tios were further emphasized byour data demonstrating the valid-ity of the use of the �/� chain ratioas a surrogate measure of Hb A2 indetecting the �-thalassemia trait (3).

One of the most importantfeatures of our sample preparationwas the counterintuitive demon-stration that a rapid 30-min trypsindigest is sufficient to release thepeptides required for accurate de-tection of the clinically significanthemoglobinopathies. The use ofovernight trypsin incubation byBoemer and colleagues, particu-larly in the context of newbornscreening, would appear to be aretrograde step. In addition, thefailure to include transitions forHb DPunjab and Hb OArab markedlyrestricts the universal applicationof their method for newbornscreening of sickle cell disease.

The truly novel feature de-scribed by Boemer and colleaguesis the use of multiple transitions foreach peptide to confer unequivocalspecificity; the authors did notcomment on whether this featurewas necessary. We have pointedout the potential value of eithersubsequent product ion scanningor rapid “real-time” product ionscanning with a linear ion trap in-strument to determine unequivo-cal sequence data (2 ). Specificitywas not a problem, however, eitherin our original series or in our re-cent study of screening dried bloodspots with tandem mass spectrom-etry for clinical hemoglobinopa-thies in 40 000 newborns (4 ).




Authors’ Disclosures of Potential Con-flicts of Interest: Upon manuscript submis-sion, all authors completed the Disclosures ofPotential Conflict of Interest form. Potentialconflicts of interest:

Employment or Leadership: None declared.Consultant or Advisory Role: None declared.Stock Ownership: None declared.Honoraria: None declared.Research Funding: C. Turner, AppleraCorporation/Applied Biosystems Group;R.N. Dalton, UK National ScreeningProgramme/Department of Health andApplera Corporation/Applied BiosystemsGroup.Expert Testimony: None declared.


References

1. Boemer F, Ketelslegers O, Minon JM, Bours V,Schoos R. Newborn screening for sickle cell dis-ease using tandem mass spectrometry. ClinChem 2008;54:2036–41.

2. Daniel YA, Turner C, Haynes RM, Hunt BJ, DaltonRN. Rapid and specific detection of clinicallysignificant haemoglobinopathies using electro-spray mass spectrometry-mass spectrometry.Br J Haematol 2005;130:635–43.

3. Daniel YA, Turner C, Haynes RM, Hunt BJ, DaltonRN. Quantification of hemoglobin A2 by tan-dem mass spectrometry. Clin Chem 2007;53:1448 –54.

4. Daniel YA, Turner C, Farrar L, Dalton RN. Acomparison of IEF and MSMS for clinical hemo-globinopathy screening in 40,000 newborns.From: 50th American Society of Hematology an-nual meeting; 2008 Dec 6–9; San Francisco. http://ash.confex.com/ash/2008/webprogram/Paper9728.html (Accessed November 2008).

Charles Turner*Yvonne DanielR. Neil Dalton

Guy’s & St Thomas’ NHS Foundation Trust

WellChild Laboratory

Evelina Children’s Hospital

St Thomas’ Hospital

London, UK


Guy’s & St Thomas’ NHS Foundation Trust

WellChild Laboratory

1st Floor, Evelina Children’s Hospital

St Thomas’ Hospital

London SE1 7EH, NA, UK

Fax �44-(0)207-1884702



In Reply

We recently validated a tandemmass spectrometry (MS/MS)1 meth-od dedicated to newborn screeningfor clinically relevant hemoglobin(Hb) disorders (1). We providesome explanations for our applica-tion and then present some of ourrecent data collected since the sub-mission and publication of ourreport.

Implementing a large-scaleroutine MS/MS methodology re-quires infallible specificity for identi-fying common Hb variants to mini-mize sample retesting. Limiting atryptic peptide analysis to a uniqueselected reaction monitoring with-out prior chromatographic separa-tion initially seemed hazardous. Thisconcern led us to acquire severaltransitions for each informative pep-tide. Moreover, our choice of thedirect-infusion method, which wasinspired by amino acid and acylcar-nitine profiles, allowed the simulta-neous acquisition of multiple reac-tions comfortably over a 60-s timeperiod. Other investigators also haveidentified tryptic peptides of cerulo-plasmin by monitoring 4 transitions(2), supporting our choice to pro-

ceed with the multiple reactionmonitoring acquisition mode.

Our sample preparation re-quired overnight Hb digestion.This step does not represent anyconstraint, however. Mass spec-trometers in newborn-screeninglaboratories are often monopo-lized overnight for analysis of sam-ples for inborn errors of metabo-lism. Therefore, a sample series forHb testing can be analyzed the fol-lowing day without disturbingroutine testing or causing negativeclinical implications.

We also have neglected theidentification of Hb DPunjab and HbOArab, because the prevalence ofsuch variants in our population is in-significant. Nevertheless, in areaswhere these mutants represent a ma-jor public health problem, theiridentification could be part of theMS/MS screening method. Onemust be reminded that the clinicalinterest here in terms of neonatalscreening is limited to sickle cell syn-dromes (i.e., Hb S/DPunjab or HbS/OArab).

Currently, our routine proto-col includes the confirmation ofeach positive sample— on thesame blood spot— by HPLC andby molecular testing of the entire�-globin gene. Sequencing pro-vides essential information to vali-date an anomaly, especially forthalassemic mutations. Moreover,DNA analysis is required to distin-guish between Hb S/S and Hb S/�-thalassemia or between Hb E/E andHb E/�-thalassemia in the neona-tal period (3 ). Considering this re-quirement, multiple additionalheterozygote patients for Hb S, HbC, or Hb E have been identifiedsince our recent publication. TwoHb S/S homozygotes and one com-pound heterozygote for Hb S andHb C have also been diagnosed. In-terestingly, on the basis of our HbA/Hb F ratio, one neonate with asuspected major �-thalassemia hasbeen confirmed by HPLC. DNAtesting of this patient revealed a

1 Nonstandard abbreviations: MS/MS, tandem massspectrometry; Hb, hemoglobin.



homozygous mutation for IVS-I-110 (G�A), confirming the ��-thalassemic syndrome (4 ).

In the future, the availabilityof isotope-labeled synthetic pep-tides at an affordable cost will aidin developing quantitative ap-proaches, and we recommend thatthe contribution of quantitativeMS/MS results for Hb variants beevaluated for the discrimination ofambiguous disorders (i.e., Hb S/Sand Hb S/�-thalassemia).

Our report and the series eval-uated by Daniel and collaborators(5 ) are relevant arguments, dem-onstrating the reliability of thescreening for Hb disorders by MS/MS. We hope that they will providea powerful incentive for the wide-spread implementation of suchapplication in neonatal-screeninglaboratories.




References

1. Boemer F, Ketelslegers O, Minon JM, Bours V,Schoos R. Newborn screening for sickle cell dis-ease using tandem mass spectrometry. ClinChem 2008;54:2036–41.

2. deWilde A, Sadilkova K, Sadilek M, Vasta V,Hahn SH. Tryptic peptide analysis of ceruloplas-min in dried blood spots using liquidchromatography-tandem mass spectrometry: ap-plication to newborn screening. Clin Chem 2008;54:1961–8.

3. Vichinsky E. Hemoglobin E syndromes. Hematol-ogy Am Soc Hematol Educ Program 2007;2007:79–83.

4. Giardine B, van Baal S, Kaimakis P, Riemer C,

Miller W, Samara M, et al. HbVar database ofhuman hemoglobin variants and thalassemiamutations: 2007 update. Hum Mutat 2007;28:206. Database available at http://globin.bx.psu.edu/hbvar/menu.html (Accessed December14, 2008).

5. Daniel YA, Turner C, Farrar L, Dalton RN. A com-parison of IEF and MSMS for clinical hemoglobi-nopathy screening in 40,000 newborns. From: 50thAmerican Society of Hematology Annual Meeting;2008 Dec 6–9; San Francisco. http://ash.confex.com/ash/2008/webprogram/Paper9728.html (Ac-cessed November 2008).

Francois Boemer2*

Olivier Ketelslegers3

Jean-Marc Minon3

Vincent Bours2

Roland Schoos2

2 Centre de Genetique Humaine, CHU

University of Liege

Liege, Belgium3 Service de Biologie Clinique

CHR Citadelle

Liege, Belgium


Biochemical Genetic Laboratory

CHU Sart-Tilman

University of Liege, B35

4000 Liege, Belgium

Fax �32-4-366-8474



Analytical and BiologicalVariation of F2-Isoprostanesduring the Menstrual Cycle

To the Editor:

F2-isoprostanes have been advo-cated as a specific and reliablemarker of oxidative damage (1 )despite a lack of data on their bio-logical and analytical variation. Con-current with a longitudinal studyof oxidative stress during the men-strual cycle [the BioCycle Study(2 )], we have characterized thecomponents of biological and ana-lytical variation of free plasma F2-isoprostanes in the context of the

menstrual cycle. We describe sev-eral of the key characteristics ofF2-isoprostanes as a biomarker.

Samples were collected from9 healthy volunteers who met themajor inclusion criteria of beinga premenopausal woman 18 – 44years of age and having regularmenstrual cycles. Major exclusioncriteria included use of hormonalpreparations or contraceptive de-vices, pregnancy, a history of gyne-cologic abnormalities, recent in-fectious disease, routine intake ofprescription and over-the-countermedications (including vitamin andmineral supplements), and a his-tory of certain chronic diseases orconditions. Fasting blood sampleswere collected between 0700 and0830 on days 2, 7, 12, 13, 14, 18, 22,and 27 of a single menstrual cycle.Collection and handling protocolswere designed to minimize preana-lytical variation. Samples were fro-zen at �80 °C and then analyzed,in duplicate, as a single set of con-secutive samples by GC-MS (1 ) atthe Molecular Epidemiology andBiomarker Research Laboratory atthe University of Minnesota. Sam-ples were analyzed within 6 monthsof collection.

The SAS software package(version 9.1.3; SAS Institute) wasused for all analyses. We log-transformed F2-isoprostane dataand used the following nestedANOVA model (3 ): Yijk � � � subi

� dayj(i) � ek(ji), where Yijk de-scribes the value for the kth repli-cate (k � 1, 2) on the jth day (j � 1,2, 3, . . . , 8) for the ith individual(i � 1, 2, 3, . . . , 9), � is the grandmean, subi is the random effect ofthe ith individual, dayj(i) is the ran-dom effect of the jth day nested inthe ith individual, and ek(ji) is therandom effect of the kth replicatenested in the jth day and the ith in-dividual. Proportions of total vari-ation from sources between (subi)and within (dayj(i)) individualsand from analytical factors (ek(ji))were estimated from the relevant



variance components �B2, �W

2 , and�A

2 , respectively. We determinedCVs for each variance componentand calculated the acceptability

index (AI)1 �AI �

CVA

CVW� for qual-

ity specification.The intraclass correlation

coefficient (ICC) was defined as:

ICC ��B

2

�B2��W

2 ;

the index of individuality (II) wasdefined as:

II ��CV W

2 � CV A2

CVB.

Assuming homogeneity of vari-ances, we defined the critical differ-ence (CD) as:

CD � 2.77 � �CV W2 � CV A

2 .

Finally, we defined the number ofsamples, k, necessary to estimatethe individual-specific mean as:

k � �1.96 ��CV W

2 � CV A2

D� 2

,

where D represents the acceptabledeviation from the mean.

Participants had a mean (SD)age of 38.5 (4.7) years, and themean menstrual cycle length was28.4 (1.3) days. One individualwas African American, and the re-maining 8 (88.9%) were white. F2-isoprostanes were measured in 120of 144 possible samples (9 partici-pants � 8 time points � 2 replicatemeasurements). A duplicate sam-ple was unavailable for 1 individualon days 2, 7, and 12; for 1 individ-ual on days 2 and 22; and for 1 in-dividual on days 7, 18, and 22. Fig.1A presents F2-isoprostane con-centrations for each study partici-pant, and Fig. 1B presents the datawith respect to each day in themenstrual cycle.

Most of the variation in F2-isoprostane concentrations oc-curred between study individuals(approximately 85%; �B

2 � 0.145),whereas within-individual varia-tion (approximately 13%; �W

2 �0.022) and analytical variation(approximately 2%; �A

2 � 0.004)were smaller. Within-individualimprecision (CVW � 3.7%) ex-ceeded that for analytical factors(CVA � 1.5%), for an AI value of0.41; thus analytical variation in-creased within-individual varia-tion by only 8% (i.e., total CVW ��0.4 � CVW)2 � (CVW)2 � 1.08),thereby meeting “desirable” per-formance specifications (4 ). Thehigh ICC value of 0.97 (95% CI,0.96 – 0.98) ensured that decreasesin study power due to analyticalfactors would necessitate a com-pensatory increase in sample size of11%, at most.

The II value of 0.42 suggeststhat population reference limits forthe detection of “abnormal” values

1 Nonstandard abbreviations: AI, acceptability in-dex; ICC, intraclass correlation coefficient; II, indexof individuality; CD, critical difference.

Fig. 1. Mean (SD) concentrations of F2-isoprostanes measured in plasma by individual (A) and by day of themenstrual cycle when the sample was collected (B).



are likely to be of limited utility.Use of a CD criterion suggests thata minimum difference of 1.0 ng/Lbetween serial samples might benonstochastic in nature. Estima-tion of the individual-specific meanF2-isoprostane concentration with-in 5%, 10%, or 20% of the actualmean value will require 13.8, 3.5,or 0.9 simultaneously assayed sam-ples, respectively.

The number of participants inthis study was small; however,studies on biological variation havedemonstrated that estimates ofintraindividual and interindividualvariation are similar regardless ofthe number of individuals studied(5 ). Our estimates of variationagree with those of other studiesthat did not span the menstrual cy-cle and suggest that plasma F2-isoprostanes are sufficiently reli-able for use as a biomarker ofoxidative damage in epidemiologicstudies of women.



Employment or Leadership: None declared.Consultant or Advisory Role: M. Trevisan,Johnson & Johnson.Stock Ownership: None declared.Honoraria: M. Trevisan, Johnson & Johnson.Research Funding: National Institute ofChild Health and Human Development(Contract no. ADB-N01-HD-4-3394).Expert Testimony: None declared.

Role of Sponsor: The funding organiza-tions played a direct role in the design of thestudy, in the choice of enrolled patients, inthe review and interpretation of data, and inthe preparation and final approval of themanuscript.

References

1. Milne GL, Yin H, Brooks JD, Sanchez S, JacksonRoberts L 2nd, Morrow JD. Quantification ofF2-isoprostanes in biological fluids and tissuesas a measure of oxidant stress. Methods Enzy-mol 2007;433:113–26.

2. Wactawski-Wende J, Schisterman EF, Hovey KM,Howards PP, Browne RW, Hediger M, et al., forthe BioCycle Study Group. BioCycle study: designof the longitudinal study of the oxidative stressand hormone variation during the menstrual cy-cle. Paediatr Perinat Epidemiol 2009;23:171–84.

3. Fraser CG. Biological variation: from principlesto practice. Washington (DC): AACC Press;2001. 151 p.

4. Cotlove E, Harris EK, Williams GZ. Biological andanalytic components of variation in long-termstudies of serum constituents in normal subjects.3. Physiological and medical implications. ClinChem 1970;16:1028–32.

5. Sebastian-Gambaro MA, Liron-Hernandez FJ,Fuentes-Arderiu X. Intra- and inter-individualbiological variability data bank. Eur J Clin ChemClin Biochem 1997;35:845–52.

Richard W. Browne2,3*

Michael S. Bloom4

Enrique F. Schisterman5

Jean Wactawski-Wende3

Kathy Hovey3

Maurizio Trevisan3,6

Myron Gross7

Departments of 2 Biotechnical and Clinical

Laboratory Sciences and3 Social and Preventive Medicine

University at Buffalo

State University of New York

Buffalo, NY4 Department of Environmental

Health Sciences

University at Albany


Albany, NY5 Epidemiology Branch

Division of Epidemiology,

Statistics & Prevention Research

Eunice Kennedy Shriver National Institute

of Child Health and Human Development

Bethesda, MD6 Health Sciences System of the

Nevada System of Higher Education

and the Department of Laboratory Medicine

and Pathology

Las Vegas, NV7 School of Medicine

University of Minnesota

Minneapolis, MN


Department of Biotechnical and

Clinical Laboratory Sciences

University at Buffalo


26 Cary Hall

Buffalo, New York 14214

Fax 716-829-3601



Serum 25-Hydroxyvitamin DImmunoassays:Recommendations for CorrectClinical Interpretation

To the Editor:

We read with interest the report byLeino et al. (1 ) concerning the an-alytical evaluation of the 25-hydroxyvitamin D3 (25-OH-D3)1

assay on the Roche Modular ana-lyzer. They used clinical samplesthat were almost free of 25-hydroxyvitamin D2 (25-OH-D2)(�10 nmol/L) as confirmed byliquid chromatography–tandemmass spectrometry (LC-MS/MS)and found a good overall agree-ment between the results obtainedwith the Roche assay and thosemeasured with the DiaSorin RIAand by LC-MS/MS. Although theyacknowledged that the Roche as-say, which does not measure 25-OH-D2, may underestimate vita-min D status in patients receivingvitamin D2, they considered thatunderestimating was probably nota clinical problem in their country(Finland), because vitamin D sup-plementation in adults is mostcommonly provided as vitamin D3.We briefly discuss this point by re-

1 Nonstandard abbreviations: 25-OH-D3, 25-hydroxyvitamin D3; 25-OH-D2, 25-hydroxyvitaminD2; LC-MS/MS, liquid chromatography–tandemmass spectrometry.



porting our own experience inFrance, where surveys performed inour units found that approximately10%–15% of adult patients treatedwith vitamin D receive vitamin D2.

As individuals working in uni-versity reference laboratories, wereceive several telephone calls eachweek from physicians who are puz-zled by the fact that the serum 25-OH-D concentration for their pa-tients has not increased, or haseven decreased, during treatmentwith vitamin D (sometimes largedoses). These patients invariablyreceived vitamin D2 and weremonitored for their 25-OH-D con-centration with the Roche assay.Every time a verification measure-ment was done in our laboratorieswith the DiaSorin RIA, the concen-tration was typical (�75 nmol/L)and sometimes quite high (�200nmol/L). As we recently reported(2 ), this situation not only gener-ates useless and costly explorationof results but also produces a cer-tain degree of anxiety in the pa-tients. We are even aware of 2 pa-tients in whom a malabsorptionsyndrome was suspected and anupper gastrointestinal endoscopyprocedure was planned. Fortu-nately, the procedures were notperformed after we explained tothe physician that the absence of anincrease in the serum 25-OH-Dconcentration was due to an ana-lytical problem. Furthermore, alow 25-OH-D concentration mea-sured with the Roche assay in a pa-tient treated with vitamin D2 mayprompt a physician to prescribelarge doses of vitamin D in a pa-tient already replete with vitaminD, thus potentially causing toxic25-OH-D concentrations to be at-tained. Finally, after discussing thisissue with physicians, we came torealize that many physicians pre-scribing vitamin D are unaware ofwhether they have prescribed adrug containing vitamin D2 or vi-tamin D3. We also realized thatmany of these physicians thought

that vitamin D2 was in fact 1,25-dihydroxyvitamin D. This confu-sion highlights the urgent need forproviding clear and simple infor-mation about vitamin D immuno-assays to the medical community.

Because some 25-OH-D as-says do not measure 25-OH-D2,one can argue that vitamin D3

should be the only vitamin D com-pound to use in clinical practice.Because other commercial assays,such as the DiaSorin assay and,to a lesser extent, the Immunodiag-nostic Systems kit (50%–75%cross-reactivity with 25-OH-D2),measure both 25-OH-D2 and 25-OH-D3, our opinion is that thisrecommendation would be validonly if vitamin D2 is clearly dem-onstrated to be less effective thatvitamin D3. To our knowledge,apart from a shorter half-life for25-OH-D2, which must be takeninto account when vitamin D isprescribed in large, spaced-outdoses (3 ), vitamin D2 seems as po-tent as vitamin D3 when prescribedas daily doses (4 ). We thus believethat in countries where vitamin D2

is prescribed (even in a low pro-portion of patients, as in France),25-OH-D assays should measureboth 25-OH-D2 and 25-OH-D3

and that the only interesting infor-mation to be provided to physi-cians in clinical practice is the sumof the 25-OH-D2 and 25-OH-D3

concentrations. Separating the re-porting of the 2 compounds maybe misleading, as previously re-ported in this journal (5). This ratio-nale is why, on behalf of the FrenchSociety of Clinical Biology and theGroup of Specialized Biology (GBS)of the French Society of NuclearMedicine, we recommend the use ofa 25-OH-D assay that measures both25-OH-D2 and 25-OH-D3.

Author Contributions: All authors con-firmed they have contributed to the intellec-tual content of this paper and have met thefollowing 3 requirements: (a) significant con-

tributions to the conception and design, ac-quisition of data, or analysis and interpreta-tion of data; (b) drafting or revising the articlefor intellectual content; and (c) final approvalof the published article.



References

1. Leino A, Turpeinen U, Koskinen P. Automatedmeasurement of 25-OH vitamin D3 on the RocheModular E170 analyzer. Clin Chem 2008;54:2059–62.

2. Cavalier E, Wallace AM, Knox S, Mistretta VI,Cormier C, Souberbielle JC. Serum vitamin Dmeasurement may not reflect what you give toyour patients. J Bone Miner Res 2008;23:1864–5.

3. Armas LA, Hollis BW, Heaney RP. Vitamin D2 ismuch less effective than vitamin D3 in humans.J Clin Endocrinol Metab 2004;89:5387–91.

4. Holick M, Biancuzzo RM, Chen TC, Klein EK,Young A, Bibuld D, et al. Vitamin D2 is aseffective as vitamin D3 in maintaining circulatingconcentration of 25-hydroxyvitamin D. J ClinEndocrinol Metab 2008;93:677–81.

5. Binkley N, Drezner M, Hollis BW. Laboratoryreporting of 25-hydroxyvitamin D results: poten-tial for clinical misinterpretation. Clin Chem2006;52:2124–5.

Catherine Massart2,3*

Jean-Claude Souberbielle4

2 Unite Fonctionnelle d’Hormonologie

CHU de Rennes

Rennes, France3 INSERM 0203 Centre d’Investigation

Clinique

Universite de Rennes 1

Rennes, France4 Laboratoire d’Explorations Fonctionnelles

Hopital Necker-Enfants – Malades

Paris, France


Unite Fonctionnelle d’Hormonologie

CHU de Pontchaillou

Rennes, France





Effect of a Variable MagneticField on Clinical LaboratoryTesting

To the Editor:

The performance of clinical analyz-ers is commonly assessed under a va-riety of environmental conditions toset acceptable limits (e.g., tempera-ture and humidity) for operation.We recently faced an unusual en-vironmental challenge within ourautomation laboratory, namely avariable magnetic field emanat-ing from a Niobe� Magnetic Nav-igation System (Stereotaxis; http://www.stereotaxis.com) being installedin the cardiac catheter laboratoryimmediately above our laboratory.The system contains 2 strong perma-nent magnets that generate magneticfields of 0.08–0.1 T (800–1000 G) inany direction. When the magnets arein the stowed position, the maxi-mum magnetic field is 0.2 T (2000G). Although more prominent forother medical devices, the permissi-ble magnetic field strength is not anenvironmental variable usually con-sidered by manufacturers of clinicalanalyzers (1, 2).

We thus faced a multifacetedproblem: On the day of the magnetdelivery, we had to assess its effectson all of the analyzers within ourautomation laboratory, and wehad to assess its effects on replace-ment analyzers due to be installedin the following weeks. Accord-ingly, we needed to assess the ef-fects of a variable magnetic field onthe accuracy and precision of thecurrent and future clinical analyz-ers within our automation labora-tory (Beckman Coulter DxI, DxC800, LH 780, and LH 755; OrthoClinical Diagnostics/Johnson &Johnson Vitros 950AT; Roche Di-agnostics Modular Analytics PModule; Siemens UF-100; bio-Merieux MDA; Siemens SMS Im-mulite 2500; Abbott DiagnosticsAxSYM; bioMerieux mini VIDAS),in particular those that used mag-

netic particle– based reagents (Im-mulite 2500). We had replaced allcathode ray tube displays in thelaboratory with liquid crystal dis-play screens before magnet instal-lation because the expected mag-netic field was known to interferewith cathode ray tube operation.

Our assessment study includedrepeated replicate analyses of QCmaterials (high, medium, and lowconcentrations of each analyte) be-fore installation of the magnet, in thepresence of the relatively low andconstant magnetic field when themagnets were stowed, and in themuch higher but transient magneticfield when the magnets were in mo-tion. To eliminate any bias due to thevested interests of either the clinicallaboratory or the catheter labora-tory, we had an independent exter-nal consultant supervise the evalua-tion and analyze the data.

Our goal was to determinewhether a statistically significantchange in results occurred after themagnet was installed. Given thenumber of experiments to be evalu-ated in a short period, we based ourassessment on a 99% confidencelevel (rather than 95%). The highestmagnetic field measured at any loca-tion in the laboratory when the mag-nets were stowed or moving was�1.26 G (�0.126 mT). This rela-tively weak field [occupationalexposure levels designated by the International Commission on Non-Ionizing Radiation Protection are200 mT for continuous exposureand 2000 mT for short-term whole-body exposure (3)] indicated thatthe shielding installed under themagnetic system was effective inprotecting the analyzers in our labo-ratory from the magnetic field.

Data were captured and ana-lyzed with EP Evaluator (David G.Rhoads Associates) and customsoftware. We evaluated changes inprecision with the SD ratio (ratioof the postmagnet SD to the pre-magnet SD). We compared the SDratio with a cutoff value, and if the

ratio exceeded the cutoff, we con-cluded that a statistically signifi-cant change had occurred. The cut-off level was based on the sizes ofthe 2 samples being compared andthe desired false-rejection rate (4 ).Smaller sample sizes have a largercutoff value. For example, for 100premagnet replicate tests and 20postmagnet replicate tests, the 99%cutoff was 1.45. For 20 premagnetreplicate tests and 20 postmagnetreplicate tests, the cutoff was 1.74.Accuracy changes were judged fromLevey–Jennings control charts cov-ering the total testing period. Thezero (target) line on this chart was setat the mean value observed duringthe first phase of testing, and the SDwas the pooled within-batch SD overall phases.

We observed a significant dif-ference based on the cutoff level inonly 19 of 900 precision data sets;however, the observed statisticallysignificant changes (measured inrelation to the SD of the process) inmean test results before and afterinstallation of the magnet were fortests with a very small SD (hencethe statistically significant change).Although the change was statisti-cally significant, it was not clini-cally important and thus not ex-pected to have an adverse effect onpatient safety or care. For example,results for replicate analyses of acontrol material for potassiumgave a mean of 2.80 mmol/L and anSD of 0.02 mmol/L (CV, 0.7%). Af-ter installation of the magnet, weobtained a mean potassium valueof 2.85 mmol/L and an SD of 0.05mmol/L (CV, 1.7%). Although thischange is statistically significant, itwas not clinically meaningful.

Our studies confirmed thatthis external magnetic field had noclinically meaningful effect on theassays performed in our clinicallaboratory, and the results pro-vided a basis for continued opera-tion in the presence of an unusualenvironmental factor.





Employment or Leadership: None declared.Consultant or Advisory Role: David G.Rhoads, David G. Rhoads Associates, Inc.Stock Ownership: None declared.Honoraria: None declared.Research Funding: None declared.Expert Testimony: None declared.


References

1. Sykes S, ed. Electromagnetic compatibility formedical devices: issues and solutions. Food andDrug Administration/Association for the Ad-vancement of Medical Instrumentation (FDA/AAMI) Conference; 1995 May 24–25; Anaheim,CA. Arlington (VA): Association for the Advance-ment of Medical Instrumentation; 1996.

2. Memorandum: electromagnetic compatibility formedical devices: issues and solutions. ODE 63903/05/1995. Federal Register 2006;71:15445.http://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfTopic/topicindex/topindx.cfm?alpha�e (AccessedFebruary 2009).

3. International Commission on Non-Ionizing Radi-ation Protection. Guidelines on limits of expo-sure to static magnetic fields. Health Phys 1994;66:100–6.

4. Crow EL, Davis FA, Maxfield MW. Statistics man-ual. New York: Dover Publications; 1960. p 76–7.

Larry J. Kricka1*

Michael Milone1

Stephen R. Master1

Leslie M. Shaw1

Donald S. Young1

Rodellia Fontanilla1

Treasa Smith1

JoAnn Gardiner1

David Cardamone1

Marilyn Fleming2

David G. Rhoads2

1 Department of Pathology &

Laboratory Medicine

University of Pennsylvania Medical Center

Philadelphia, PA2 David G. Rhoads Associates, Inc.

Kennett Square, PA

* Address correspondence to this author

at:

Hospital of the University of Pennsylvania

Department of Pathology &

Laboratory Medicine

3400 Spruce St.

Philadelphia, PA 19104-4283

Fax 215-662-7529

E-mail [email protected].


More on Methanol-AssociatedMatrix Effects in ElectrosprayIonization Mass Spectrometry

To the Editor:

Based on the observations reportedby Annesley regarding methanol-associated matrix effects (1), Ichanged from the use of Burdick andJackson (B&J)1 to EMD Chemicals(EMD) methanol in the mobilephases for tacrolimus, cyclosporine,and sirolimus procedures that usedelectrospray ionization on singleLC-MS and tandem quadrupoleLC-MS (LC-MS/MS) instrumentsystems. Presented herein are obser-vations that supplement Annesley’sobservations regarding the influencethat brand of methanol can have onanalytical performance.

I routinely performed tacroli-mus and cyclosporine proceduresusing LC-MS/MS and sirolimususing LC-MS. The tacrolimus andcyclosporine procedures have beendescribed previously (2, 3 ). The

sirolimus procedures (unpub-lished data) were similar to the oth-ers. The same sample preparationwas used for both LC-MS and LC-MS/MS analysis. Importantly,methanol supplemented with am-monium acetate and formic acid(prepared in and used directlyfrom the manufacturer’s bottle)was the LC-MS/MS mobile-phaseorganic component, in which am-monium adducts were monitored.For LC-MS methods we usedmethanol (directly from the man-ufacturer’s bottle), and the aque-ous component, water supple-mented with sodium formate andformic acid, ensured sodium ad-duct signals for monitoring.

The change in brand of meth-anol from B&J to EMD did not af-fect the performance of the tacroli-mus or cyclosporine LC-MS/MSassay, although cyclosporine andcyclosporin D (internal standard)peak areas increased moderately(15%). However, there was animmediate adverse effect on thesirolimus LC-MS procedure; per-formance worsened progressivelyuntil, on the sixth analytical run,the assay failed. Of note was theeffect on the internal standard(ascomycin), for which the meanwithin-run peak height unexpect-edly decreased between 10% and36% from values in previous runswith B&J methanol. Peak heightwas used for tacrolimus and siroli-mus LC-MS procedures because itreduced the need for manual rein-tegration, especially for sampleswith low concentration of analyte.Peak height was shown to provideperformance specifications equiv-alent to peak area during methodvalidation.

The within-run variation, a cri-terion for analytical run acceptance(i.e., �5%), was 9.1% initially, then4.5%–7.3% subsequently. Freshwhole blood standard curve param-eters were acceptable; however, theaccuracy of the QC materials (freshwhole blood supplemented, divided

1 Nonstandard abbreviations: B&J, Burdick andJackson; EMD, EMD Chemicals; LC-MS/MS, tan-dem quadrupole LC-MS.



Table 1. Influence of source of methanol in mobile phase on mass spectrometer response.

Sample, parameterExp SRLConcc

LC-MS/MS (EMD vs B&J)

EMDa (lot 47229) B&Jb (lot CU703) EMD vs B&J

SRL peakarea

IS peakarea

Calc SRLConcc

SRL peakarea

ISd peakarea

Calc SRLConc

SRL arearatio

IS arearatio Pe,f

Calibrator 2 36 2154 1.86 19 1128 2.04 1.89 1.91

5 97 2215 4.90 47 1159 5.13 2.06 1.91

10 209 2235 10.5 93 1187 9.94 2.25 1.88

20 410 2259 20.4 182 1214 19.2 2.25 1.86

30 610 2243 30.2 284 1238 29.5 2.12 1.81

40 769 2172 39.9 339 1034 42.2 2.27 2.10

50 1034 2369 49.2 461 1121 49.0 2.24 1.96

Mean 2235 1167 2.16 1.92 0.003

Standard curve slope 0.0089 0.0077

QC 3.76 76 2370 3.59 35 1273 3.37 2.17 1.86

11.9 232 2504 10.4 114 1333 10.9 2.04 1.88

32.6 707 2450 32.5 330 1456 29.2 2.14 1.68

Mean 2441 1354 2.12 1.81 0.014

Pe,f 0.003 0.007 0.678 0.134

LC-MS (EMD vs B&J)

EMD (lot 47192) B&J (lot CT606) EMD vs B&J

SRL peakareag

IS peakareag

Calc SRLConc

SRL peakareag

IS peakareag

Calc SRLConc

SRL arearatio

IS arearatio P

Calibrator 2 1.39 26.2 1.69 1.41 41.6 2.06 0.99 0.63

5 3.31 26.8 5.41 3.57 42.0 4.67 0.88 0.64

10 5.57 27.0 10.6 8.19 41.6 10.4 0.68 0.65

20 10.6 26.7 21.7 16.2 41.8 20.1 0.66 0.65

30 14.4 27.0 29.7 24.0 41.6 29.9 0.60 0.65

40 18.3 27.4 37.3 31.8 42.2 39.0 0.57 0.65

50 24.2 27.1 50.6 41.0 41.6 50.8 0.59 0.65

Mean 26.9 41.8 0.71 0.64 0.002

Standard curve slope 0.0172 0.0195

QCg 4.00 3.22 31.6 4.50 2.98 41.5 4.00 1.08 0.76

11.9 9.06 32.3 14.9 8.86 41.3 11.3 1.02 0.78

32.2 24.4 29.4 42.8 25.6 42.6 31.1 0.95 0.75

Mean 32.0 41.8 1.02 0.76 0.003

P �0.001 0.951 0.015 0.001

a Omnisolv® high-purity solvent for gas chromatography, HPLC, spectrometry, and gradient analysis.b High purity solvent for HPLC, gas chromatography, pesticide residue analysis and spectrometry; B&J.c Expected (Exp) sirolimus (SRL) concentration (Conc) and calculated (Calc) SRL Conc (�g/L).d IS, internal standard.e P value determined by independent samples t-test.f Bold italicized type used for QC value exceeding established range, P value �0.05.g SRL peak area, IS peak area � 10�5.



into aliquots, and frozen until use)decreased progressively. On thesecond run the high-QC material(mean 32.6 �g/L � 4%) exceededthe established range, and on subse-quent runs both medium (mean11.9 �g/L�5%) and high QC failed.By the sixth run, all materials failed(low, mean 3.76 �g/L � 5%). Im-portantly, reanalysis of the samplesfrom each of these runs using LC-MS/MS (with EMD in the mobilephase) met the established accep-tance criteria. Subsequent reversionto B&J for the sirolimus LC-MSprocedure normalized assay perfor-mance immediately!

To document this phenome-non more clearly, a sirolimus ana-lytical run, consisting of a standardcurve and QC materials, was ana-lyzed on the LC-MS/MS system byusing EMD and then B&J. Anotherset of samples was analyzed on theLC-MS system in similar fashion.

When EMD was compared toB&J, peak areas for sirolimus in-creased more than 2-fold, whereasthose of the internal standard in-creased somewhat �2-fold (P �0.05) with the use of LC-MS/MS.In calibrator and QC materials,peak areas were increased similarly(Table 1, upper panel). Thus theEMD methanol appeared to im-prove assay sensitivity.

With the use of LC-MS withEMD compared to B&J, sirolimusand internal standard peak areaswere decreased, except for siroli-mus in QC materials (Table 1,lower panel). Interestingly, theEMD-to-B&J peak-area ratios forsirolimus in the standard materialsdecreased in relation to concentra-tion. Moreover, with EMD but notB&J methanol, the internal stan-dard peak areas were significantlyhigher in QC than calibrator mate-rials. These adverse responses ofsirolimus and internal standardwith the EMD methanol, as well ascalibrator and QC materials, leadto procedural failure (all QC valuesexceeded the established ranges).

After these procedures wereperformed, EMD was applied rou-tinely to LC-MS/MS proceduresand B&J to LC-MS. This protocolprovided acceptable performanceof all procedures until similar, butnot identical, phenomena were ob-served when an old lot of B&Jmethanol was used inadvertently.Failure of a sirolimus LC-MS runwas due to a concentration-dependent effect on the peak re-sponses for sirolimus in the calibra-tor but not QC materials. The poorquality of the old B&J methanol wasconfirmed through separate infu-sions of old and more recent B&Jwith acquisition of mass scans (m/z15–1600). Old B&J methanol evi-denced strong signals at approxi-mately 290, 440, and 600 m/z andbackground noise across the rangewas approximately 10-fold higherthan the more recent lot of B&Jmethanol. Reanalysis of the samplesusing the recent B&J methanol pro-duced a successful run.

In summary, differential re-sponses of sirolimus and ascomy-cin, but not tacrolimus, to methanolin the mobile phase in fresh and pre-viously frozen whole blood matricesaffected control of analytical proce-dures. In no case did methanol com-promise the linearity of the calibra-tion, although the slopes certainlydiffered. Moreover, sirolimus exhib-ited an unexpected and unaccept-able concentration-dependent re-sponse. These experiences supportthe need for continual assessment ofanalytical performance, with the ex-pectation that aberrations may arisefrom unexpected sources. Also, asrecommended by Annesley and oth-ers (1, 4, 5), rigorous evaluation ofmethod performance (including in-ternal standard) is required for pro-cedures that are intended for clinicalservice.

Author Contributions: All authors con-firmed they have contributed to the intellec-tual content of this paper and have met the

following 3 requirements: (a) significant con-tributions to the conception and design, ac-quisition of data, or analysis and interpreta-tion of data; (b) drafting or revising the articlefor intellectual content; and (c) final approvalof the published article.



References

1. Annesley TM. Methanol-associated effects inelectrospray ionization tandem mass spectrom-etry. Clin Chem 2007;53:1827–34.

2. Napoli KL. Is microparticle enzyme-linked im-munoassay (MEIA) reliable for use in tacro-limus TDM? Comparison of MEIA to liquidchromatography with mass spectrometric de-tection using longitudinal trough samplesfor transplant recipients. Ther Drug Monit2006;28:491–504.

3. Napoli KL. Comparison of 12-hour AUC cyclo-sporine concentrations determined by liquidchromatography-mass spectrometry and fluores-cence polarization immunoassay reveals siroli-mus effect on cyclosporine pharmacokinetics.Ther Drug Monit 2006;28:726–36.

4. Taylor PJ, Brown SR, Cooper DP, et al. Evalua-tion of 3 internal standards for the measurementof cyclosporin by HPLC-mass spectrometry. ClinChem 2005;51:1890–3.

5. O’Halloran S, Ilett KF. Evaluation of a deuterium-labeled internal standard for the measurementof sirolimus by high-throughput HPLC electro-spray ionization tandem mass spectrometry. ClinChem 2008;54:1386–9.

Kimberly L. Napoli

Division of Immunology and

Organ Transplantation

University of Texas

Medical School at Houston

Houston, TX,

retired

Address correspondence to the author at:

31 Joci Court

Durham, NC 27704





use of serum myoglobin assays for urine myoglobin measurements

Documents