Download pdf - Technical Briefs - Semantic Scholar · Incidence of coronary heart disease and lipoprotein cholesterol levels. J Am Med Assoc 1986;256:2835–8. 2. Martin MJ, Hulley SB, Browner WS,

Use of the Friedewald Formula to Estimate LDL-Choles-terol in Patients with Chronic Renal Failure on Dialysis,Roger Johnson,* Prudence McNutt, Stephen MacMahon,1 andRichard Robson2 (Dept. of Clin. Biochem., Natl. Women’sHosp., Private Bag 92189, Auckland 3, New Zealand;1 Clin. Trials Res. Unit, Univ. of Auckland, Auckland; and2 Dept. of Nephrol., Christchurch Hosp., Christchurch,New Zealand; *author for correspondence: fax 64-9-630-9996)

Increased plasma cholesterol, particularly that portionassociated with LDL, is an established risk factor forcoronary heart disease [1, 2]. As a consequence, it iswidely recommended [3] that LDL-cholesterol be deter-mined in individuals with increased total cholesterol.Because isolation of the LDL fraction requires ultracen-trifugation, a technique not generally available in servicelaboratories, the concentration of LDL-cholesterol is usu-ally calculated by the formula of Friedewald et al. [4].

The Friedewald formula provides an adequate estimateof LDL-cholesterol for most fasting specimens but isknown to be less reliable as triglyceride concentrationincreases [5]. Moreover, in patients with chronic renalfailure, in whom an accumulation of remnant particlesand increased concentrations of IDL-cholesterol havebeen found [6, 7], the estimate is possibly further compro-mised [8].

Here we report results on specimens obtained frompatients on dialysis with chronic renal failure that showno greater disparity between measured and estimatedLDL-cholesterol concentrations than those seen in otherpatient groups in which the Friedewald formula is fre-quently used. Additionally we have reexamined the effectof IDL and VLDL composition on the reliability of esti-mations made with the formula.

Two sets of blood specimens taken 6 months apart wereobtained from 106 patients receiving hemodialysis orcontinual ambulatory peritoneal dialysis [9]. Within this6-month period, some patients received an inhibitor ofhydroxymethylglutaryl-CoA reductase, some an inhibitorof angiotensin-converting enzyme, some both treatments,and some placebo only. Blood collected after an overnightfast and separated within 4 h provided serum specimensfor analysis.

Total cholesterol and triglyceride were assayed enzy-mically with reagents from Boehringer Mannheim. HDL-cholesterol was assayed after precipitation of the apoB-containing lipoproteins with phosphotungstate andmagnesium ions [10]. The performance of these assayswas regularly certified by reference to samples providedby the Centers for Disease Control and Prevention, At-lanta, GA. Ultracentrifugation was carried out as de-scribed by others [11] for 15 h at 120 000g with a TFT 80.4fixed-angle rotor in a Centrikon T-2070 centrifuge (Kon-tron Instruments). Cholesterol and triglyceride contents ofsupernatant fractions obtained after the tubes were sliced[11] were measured; the fraction floating at a relativedensity (d) of 1.006 was taken to represent VLDL and thatfloating at d 1.019 obtained by adjustment of 20 volumes

of serum with 1 volume of NaBr solution (d 5 1.27, ;3.9mol/L) was taken to represent VLDL plus IDL. Thecontents of IDL alone were calculated by subtraction.When in other experiments this technique was extendedto permit sequential isolation of LDL and of HDL [11], themean recovery of total cholesterol was 90% (range 86–95%) and of triglyceride was 92% (range 88–95%).

LDL-cholesterol was calculated as total cholesterol 2HDL-cholesterol 2 VLDL-cholesterol; VLDL-cholesterolwas either measured directly (after ultracentrifugation) orcalculated as 0.456 3 total triglyceride concentrationexpressed in mmol/L (Friedewald).

Analysis of 204 specimens with complete data from the106 patients showed concentrations (mmol/L) of totalcholesterol of 3.16–9.25 (median 6.00), triglyceride of0.48–7.74 (median 2.01), VLDL-cholesterol of 0.09–3.78(median 0.92), IDL-cholesterol of 0.15–1.58 (median 0.45),and HDL-cholesterol of 0.36–2.28 (median 0.93). LDL-cholesterol derived from ultracentrifugation data rangedfrom 1.13 to 7.27 (median 3.92) mmol/L.

Thirty-six specimens (18%) had a LDL-cholesterol cal-culated from the Friedewald formula that differed fromthe ultracentrifugation value by .10%. This proportion ofspecimens with significant error is similar to those pro-portions (15–19%) found in much larger series of patientsof more diverse clinical conditions [12, 13]. The propor-tion is substantially less than the 42% of 45 individualswith chronic renal failure reported by others [8], despitethe lower lipid values in their series.

Recognition that the Friedewald formula is unreliablein the presence of chylomicronemia and hypertriglyceri-demia has led to the use of various indices for theexclusion of such specimens. Total triglyceride concentra-tion commonly is used [5], although a low total choles-terol:triglyceride ratio has been proposed as a morerelevant criterion for rejection [14]. Additionally, othershave identified increased cholesterol:triglyceride ratios inVLDL and IDL and increased IDL-cholesterol alone asbeing associated with increased error in the formula [8].

We plotted the number of errors detected against eachindex arranged in centiles (Fig. 1). The median (range) foreach index was: total triglyceride concentration, 2.01(0.48–7.74) mmol/L; total cholesterol:triglyceride ratio,2.92 (0.86–7.86); cholesterol:triglyceride ratio in VLDL,0.75 (0.31–1.28); IDL-cholesterol 0.45 (0.15–1.58) mmol/L;cholesterol:triglyceride ratio in IDL, 2.69 (1.11–6.80). Withtotal triglyceride concentration and total cholesterol:tri-glyceride ratio (by inverse ranking so that increasedincidence of error was positively associated with theindex), error detection was similar with the number oferrors initially small but rising steeply from the 60thcentile. By contrast, with the ratio of cholesterol:triglycer-ide in VLDL and in IDL and with IDL-cholesterol concen-tration, the errors were more evenly distributed amongthe centiles. Quantitatively, 18 errors (50%) were includedwhen triglyceride concentration was 3.51 mmol/L (86thcentile), total cholesterol:triglyceride ratio was 1.91 (81stcentile), cholesterol:triglyceride ratio in VLDL was 0.82(70th centile), IDL-cholesterol was 0.53 mmol/L (65th

Technical Briefs

Clinical Chemistry 43, No. 11, 1997 2183

centile), and cholesterol:triglyceride ratio in IDL was 2.51(38th centile).

The ranking by total triglyceride concentration andtotal cholesterol:triglyceride ratio demonstrates that thesevalues are far better suited as criteria for decidingwhether it is valid to calculate LDL-cholesterol. Moreover,the results confirm that total triglyceride concentrationalone seems to be the most simple and appropriatecriterion to use, with total cholesterol:triglyceride ratiooffering no improvement in discrimination.

In our study, 24 of 41 specimens (59%) with triglycerideconcentration .3.2 mmol/L had an unreliable estimate ofLDL-cholesterol from the Friedewald formula, with allbut two being an overestimate. While such overestimationis typical of abnormal VLDL composition as seen in TypeIII hyperlipoproteinemia [5], there appeared to be no clearassociation between cholesterol:triglyceride ratio in VLDLand the incidence of error (Fig. 1). It seems improbablethat the disparity between our results and those of others[8] could be explained by effects of drug treatment.Indeed, the conclusion was unchanged when the 50specimens from patients on an hydroxymethylglutaryl-CoA reductase inhibitor were removed from analysis.

In summary, use of the Friedewald formula in a groupof patients expected to give aberrant results providedreliable data in the majority of cases. Moreover, triglycer-ide concentration was a suitable index of reliability.However, restricting use of the formula to those speci-mens having a triglyceride concentration of ,4.5 mmol(4 g)/L [13] included a group above the 80th centilein which the error rate exceeded 50%. Limiting use ofthe formula to those below the 80th centile and having atriglyceride concentration of ,3.2 mmol (2.8 g)/L ex-cluded the majority of errors and left 151 of 163 (93%)

with a valid estimate. By contrast with the results ofothers [8, 14], we found little influence of VLDL or IDL onthe reliability of the Friedewald formula and no advan-tage with total cholesterol:triglyceride ratio as a discrimi-nant.

The study of the effects of simvastatin and enalapril onserum lipoprotein concentrations in patients on dialysis,from which this work derives, was supported by a grantfrom Merck, Sharp and Dohme.

References1. Castelli WP, Garrison RJ, Wilson PWF, Abbott RD, Kalousdian S, Kannel WB.

Incidence of coronary heart disease and lipoprotein cholesterol levels. J AmMed Assoc 1986;256:2835–8.

2. Martin MJ, Hulley SB, Browner WS, Kuller LH, Wentworth D. Serum choles-terol, blood pressure, and mortality: implications from a cohort of 361,662men. Lancet 1986;ii:933–6.

3. National Cholesterol Education Program Expert Panel. Summary of thesecond report of the National Cholesterol Education Program (NCEP) ExpertPanel on detection, evaluation, and treatment of high blood cholesterol inadults (Adult Treatment Panel II). J Am Med Assoc 1993;269:3015–23.

4. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration oflow-density lipoprotein cholesterol in plasma, without use of the preparativeultracentrifuge. Clin Chem 1972;18:499–502.

5. Rifai N, Warnick GR, McNamara JR, Belcher JD, Grinstead GF, Frantz ID Jr.Measurement of low-density-lipoprotein cholesterol in serum: a statusreport. Clin Chem 1992;38:150–60.

6. Nestel PJ, Fidge NH, Tan MH. Increased lipoprotein-remnant formation inchronic renal failure. N Engl J Med 1982;307:329–33.

7. Ron D, Oren I, Aviram M, Better OS, Brook JG. Accumulation of lipoproteinremnants in patients with chronic renal failure. Atherosclerosis 1983;46:67–75.

8. Senti M, Pedro-Botet J, Nogues X, Rubies-Prat J. Influence of intermediate-density lipoproteins on the accuracy of the Friedewald formula. Clin Chem1991;37:1394–7.

9. Robson R, Collins J, Johnson R, et al. on behalf of the PERFECT studycollaborative group. Effects of simvastatin and enalapril on serum lipopro-tein concentrations and left ventricular mass in patients on dialysis. JNephrol 1997;10:33–40.

10. Lopes-Virella MF, Stone P, Ellis S, Colwell JA. Cholesterol determination inhigh density lipoproteins separated by three different methods. Clin Chem1977;23:882–4.

11. Mills GL, Lane PA, Weech PK. A guidebook to lipoprotein technique. In:Burdon RH, van Knippenberg PH, eds. Laboratory techniques in biochemis-try and molecular biology, Vol. 14. Amsterdam: Elsevier, 1984:18–116.

12. Warnick GR, Knopp RH, Fitzpatrick V, Branson L. Estimating low-densitylipoprotein cholesterol by the Friedewald equation is adequate for classify-ing patients on the basis of nationally recommended cutpoints. Clin Chem1990;36:15–9.

13. McNamara JR, Cohn JS, Wilson PWF, Schaefer EJ. Calculated values forlow-density lipoprotein cholesterol in the assessment of lipid abnormalitiesand coronary disease risk. Clin Chem 1990;36:36–42.

14. Gonzalez Ga-Estrada M, Rodriguez Ferrer CR, Recalde Astarloa I, MontalvoLahera E. Use of serum cholesterol/triglyceride ratio to discern for whichindividuals the Friedewald formula can be used confidently. Clin Chem1990;36:1673–5.

Rapid Detection of the Fibrinogen Aa16Arg3His Mu-tation, Hayley J. Ridgway,* Stephen O. Brennan, Andrew P.Fellowes, and Peter M. George (Molec. Pathol. Lab.,Christchurch Hosp., Canterbury Health Ltd.,Christchurch, New Zealand; *author for correspondence:fax 64-3-364-0545, e-mail [email protected])

We have now identified mutations in 17 families withdysfibrinogenemias. Over half of these families have theAa16Arg3His mutation. This mutation is the most com-

Fig. 1. Error in LDL-cholesterol determination by Friedewald formula.The 204 specimens were first ranked separately by (a) triglyceride concentration(TRIG), (b) total cholesterol:total triglyceride ratio (CHOL/TRIG), (c) cholesterol:triglyceride ratio in VLDL (VLDL ratio), (d) IDL-cholesterol concentration (IDL), and(e) cholesterol:triglyceride ratio in IDL (IDL ratio). The centile rank for the 36LDL-cholesterol results in error (i.e., those for which the Friedewald formula gavea result that differed from that of ultracentrifugation by .10%) was then found asshown. To provide a consistent comparison, an inverse ranking for totalcholesterol:total triglyceride ratio was used.

2184 Technical Briefs

monly reported cause of dysfibrinogenemia and, likeother dysfibrinogenemias, is readily detected because ofthe associated prolonged thrombin and reptilase times[1-3]. The mutation alters the thrombin cleavage site suchthat release of fibrinopeptide A is delayed. However,fibrinopeptide release assays are difficult and do notdirectly confirm the molecular basis of the impairedfibrinopeptide release. We have therefore designed arapid and technically simple PCR-based method for de-tection of the Aa16Arg3His mutation. This allows reli-able identification of a common dysfibrinogenemia that,in its heterozygous form, is usually asymptomatic anddoes not pose any substantial threat to the health of thepatient. Application of this method will allow clinicallaboratories to determine the molecular defect in many ofthe cases that they detect during coagulation studies.

We examined nine families with the Aa16Arg3Hismutation. These had been referred for further investiga-tion when routine coagulation studies were consistentwith dysfibrinogenemia. All procedures were carried outin accordance with the guidelines of our local ethicscommittee. Blood samples were collected into Na1 citrateVacutainer Tubes (Becton Dickinson), and coagulationstudies were performed by routine clinical tests forthrombin and reptilase times. There was considerablevariation both within and between families in thrombintimes [range 36–70 s (reference range 20 6 2)] andreptilase times [range 45–72 s (reference range 20 6 2)]. Ineach case fibrinopeptide release assays [4] demonstratedreduced fibrinopeptide A concentrations with an addi-tional earlier eluting peak. Either amino acid analysis ofthe abnormal fibrinopeptide or DNA sequence analysisthen confirmed the mutation.

Genomic DNA was isolated from whole blood [5]. Theoligonucleotides Fn1111a (ATT GCT GTT GCT CTC TTTTG) and Fn1309a (AAT CTC CTG CTT CCC CCG CT)were used to amplify a 199-bp region spanning exon 2 ofthe Aa gene by PCR [6]. Each 100-mL amplificationreaction contained 50 mmol/L KCl, 10 mmol/L Tris-HCl,pH 8.3, 1.5 mmol/L MgCl2, 200 mmol/L of each dNTP, 1mmol/L of each primer, 1 mg of DNA template, and 2units of Taq DNA polymerase (Boehringer Mannheim).Amplification was for 30 cycles with denaturation for 30 sat 94 °C, annealing for 30 s at 60 °C, and extension for 1min at 72 °C with a final extension at 72 °C for 7 min. ThePCR products were digested for 4 h at 37 °C with 5 unitsof NlaIII according to the manufacturer’s instructions(New England Biolabs). Typically 7 mL of PCR productwas diluted to 10 mL by the addition of 0.5 mL of 10units/mL NlaIII, 1 mL of NEBuffer 4 (New EnglandBiolabs), 1 mL of 1 mg/mL bovine serum albumin, and 0.5mL of sterile distilled water. Digestion was assayed by gelelectrophoresis in 2% agarose, 50 mmol/L Tris base, 45mmol/L boric acid, 0.5 mmol/L EDTA for 30–40 min at100 V. Products were visualized by staining in 20 mg/mLethidium bromide for 5 min followed by transillumina-tion at 302 nm.

The Aa16Arg3His mutation changes the sequenceCGTG to CATG creating an NlaIII cleavage site near the

middle of the PCR product (Fig. 1, lower panel). Cleavageat this site generated 104-bp and 95-bp products that werenot resolved on the agarose gel, but were clearly sepa-rated from the uncut product. DNA from apparentlyhealthy individuals remained uncut. The upper panel ofFig. 1 shows the restriction pattern produced from appar-ently healthy individuals (lanes 2 and 5) and the patternproduced by individuals heterozygous for the Aa16Arg3His (CGT3CAT) mutation (lanes 3, 4, and 6). Addi-tionally, the assay should be able to detect homozygotesbecause no uncut product should remain; however, ap-propriate controls were not available.

The mutation Aa16Arg3His affects the thrombincleavage site at the N-terminal of the Aa chain. Normalcleavage at this site exposes the Gly-Pro-Arg (A) site,which interacts with a preformed, complementary sitelocated in the C-terminal of the g chain, thereby initiatingpolymerization [7]. The net effect of replacing the arginineat position 16 of the Aa chain is only to delay thethrombin-catalyzed exposure of the A polymerizationsite. Therefore, it is not surprising that this mutation isusually asymptomatic. Despite this, two reported caseshave been associated with mild bleeding tendencies [8, 9].In these cases, the bleeding tendency generally can beattributed to additional abnormalities in other coagula-tion proteins. In fibrinogen Milano VI, the patient showed

Fig. 1. Aa chain exon 2.Upper panel, restriction digest of amplified Aa chain exon 2 from affected andunaffected family members run on a 2% agarose gel for 30–40 min at 100 V.Lane 1, fX174/HaeIII molecular mass marker; lanes 2 and 5, DNA fromapparently healthy individuals; lanes 3, 4, and 6, DNA from heterozygousindividuals. Lower panel, DNA sequence showing the normal and the mutatedsequence of Aa chain exon 2. The mutation produces an NlaIII recognition site(underlined), and the arrow denotes where the enzyme cleaves. Numbersdesignate the amino acid position within the Aa chain.


defective platelet aggregation [8], whereas in fibrinogenBirmingham, abnormalities in von Willebrand factor wereseen [9]. The only reported case of this mutation in itshomozygous form, fibrinogen Giessen I, is associatedwith more severe symptoms and displays a severe bleed-ing tendency and miscarriage [10].

Dysfibrinogenemias with the Aa16Arg3 His mutationare usually detected by prolonged thrombin-clottingtimes. Once detected, the mutation can be characterizedby reversed-phase monitoring of fibrinopeptide release[4]. In patients with the Aa16Arg3His mutation, the Apeptide peak is reduced by half, and there is an additionalearlier-eluting peak that represents the histidine-contain-ing A peptide. Subsequent protein sequencing of theabnormal peptide is required to confirm this mutation.Although the method does provide a definitive result, theapparatus and technical expertise required are well be-yond the scope of most clinical laboratories. With themethod described here, the detection of this mutation,which in our experience accounts for 50% of all cases ofdysfibrinogenemia, is straightforward, requiring only asimple PCR and restriction digest. The absolute identifi-cation of this mutation will enable the clinician to reassurethe patient that their dysfibrinogenemia is unlikely tocause any bleeding disorder.

This work was supported by the Health Research Councilof New Zealand.

References1. Bithell TC. Hereditary dysfibrinogenemia. Clin Chem 1985;31:509–16.

2. Henschen AH. Human fibrinogen—structural variants and functional sites.Thromb Haemost 1993;70:42–7.

3. Matsuda M, Yoshidi N, Terukina S, Kensuke Y, Maekawa H. Molecularabnormalities of fibrinogen—the present status of structure elucidation. In:Matsuda M, Iwanaga S, Takada A, Henschen A, eds. Fibrinogen 4: currentbasic and clinical aspects. Amsterdam: Elsevier Science Publishers, 1990:139–51.

4. Brennan SO, Hammonds B, George PM. Aberrant hepatic processing causesremoval of activation peptide and primary polymerisation site from fibrinogenCanterbury (Aa20Val3Asp). J Clin Invest 1995;96:2854–8.

5. Ciulla TA, Sklar RM, Hauser SL. A simple method for DNA purification fromperipheral blood. Anal Biochem 1988;174:485–8.

6. Saiki RK, Gelfand DH, Stoffel S, Scharf SJ, Higuchi R, Horn GT, et al.Primer-directed enzymatic amplification of DNA with a thermostable DNApolymerase. Science 1988;239:487–91.

7. Gorkun OV, Veklich YI, Medved LV, Henschen AH, Weisel JW. Role of the aCdomains of fibrin in clot formation. Biochemistry 1994;33:6986–97.

8. Bogli C, Cofrancesco E, Cortellaro M. Della Volpe A, Hofer A, Furlan M,Zanussi C. Fibrinogen Milano VI: a heterozygous dysfibrinogenaemia (Aa16Arg3His) with bleeding tendency. Eur J Haematol 1990;45:26–30.

9. Siebenlist KR, Prchal JT, Mosesson MW. Fibrinogen Birmingham: a heterozy-gous dysfibrinogenemia (Aa16Arg3His) containing heterodimeric mole-cules. Blood 1988;71:613–8.

10. Alving BM, Henschen AH. Fibrinogen Giessen I: a congenital homozygouslyexpressed dysfibrinogenemia with Aa16Arg3His substitution. Am J Hema-tol 1987;25:479–82.

Urine Pyridinium Cross-Links Determination by Beck-man Cross Links Kit, Isabella Fermo,* Cinzia Arcelloni,Erminia Casari,1 and Rita Paroni (Lab. of Chromatogr.Techniques, Dept. of Lab. Med., IRCCS H. San Raffaele,Milano, Italy; 1 Ist. Clin. Humanitas, Rozzano (MI), Italy;* address for correspondence: Lab. HPLC, H. San Raffaele,Via Olgettina 60, 20132 Milano, Italy; fax 39-2-26432640)

Increased understanding of bone turnover has led to thedevelopment of several biochemical tests of bone metab-olism. Among the biochemical indexes of bone resorptionis measurement of urinary excretion of pyridinoline (Pyr)and deoxypyridinoline (Dpyr), molecules that cross-linkthe collagen chains and are released into the systemiccirculation after the breakdown of mature bone collagen[1–3]. Because they are not metabolized in vivo, they areexcreted directly into urine in free (;40%) and peptide-bound forms (;60%) [4]. Methods to measure cross-linksin urine involve mainly two technical approaches: HPLCanalysis, which, after hydrolytic and purification steps,allows the determination of the total forms of cross-links[5–7], and monoclonal antibody immunoassay methodsable to quantify the sum of free Pyr and Dpyr or only thefree Dpyr form [8, 9]. The quantification of total or freecross-links forms provides, in any case, similar clinicalinformation [9, 10].

Here we report the evaluation of the Cross LinksTM

HPLC kit recently introduced by Beckman Labs. toquantify the total forms of Pyr and Dpyr. We comparedthe performance of this procedure with the Chrom-LinksTM HPLC method from Bio-Rad Labs. In additionwe investigated the determination of free Dpyr by theCross Links method by comparing it with the Pyri-linksTM-D immunoassay (Metra Biosystem) [11].

After the Beckman procedure 0.25 mL of urine washydrolyzed with 0.25 mL of HCl (12 mol/L) (R1 reagent)at 115 °C overnight. After the addition of 0.5 mL of theinternal calibrator solution and 2 mL of 1-butanol (R2reagent), samples were loaded onto solid-phase columnsand washed with 9 mL of R3 reagent (a mixture ofbutanol:water:acetic acid). After the addition of 0.5 mL ofan organic solvent reagent (R4) to clean up the columnsfrom the previous R3 reagent, cross-links were elutedwith 0.5 mL of the extraction reagent (R5) containingheptafluorobutyric acid in water, and 50 mL was injectedinto the HPLC. Unlike the Bio-Rad assay, which employsgravimetric columns, purification of urine is carried outunder vacuum by solid-phase extraction, with a notabletime saving and minimal solvent exposure for the techni-cal staff.

The HPLC equipment consisted of a Beckman SystemGold Model 116 on line with an autosampler Model 507and a Shimadzu Model RF 551 fluorometric detector withexcitation at 295 nm and emission at 395 nm. The chro-matographic separation lasted 10 min (Fig. 1) and wasperformed on a reversed-phase column (100 3 4.6 mm) inisocratic mode, at a flow rate of 1.0 mL/min. The eluting


solution was prepared by mixing the stock buffer (hep-tafluorobutyric acid solution) purchased from Beckmanwith acetonitrile and water in the proportion 0.25:0.65:9.1(by vol). The column was thermoregulated at 37 °C.

The intraassay CVs (n 5 10) for a urine pool with Pyr 5500 pmol/mL and Dpyr 5 89 pmol/mL were 3% and4.9%, respectively, and for Pyr 5 950 pmol/mL, Dpyr 5193 pmol/mL 3.4% and 2.8%, respectively. The interassayvariation was calculated from the analysis of the sameurine pools for 7 consecutive days. The CVs for Pyr andDpyr were 5.5% and 8.7% for the lower concentrationsand 4.0% and 5.2% for the higher concentrations. Themean recoveries were 95% 6 4.6% for Pyr (n 5 3) and99.2% 6 2.0% for Dpyr (n 5 3). The assay was linear (r 50.999) in the tested ranges (0–3000 pmol/mL and 0–500pmol/L for Pyr and Dpyr, respectively). A signal-to-noiseratio of 3 (detection limit) was achieved at 1 pmol injected.

Comparison of the Beckman HPLC kit vs Bio-Rad on 50urine samples from postmenopausal women (57 6 5.7years) who were referred to the Orthopaedic Division ofour Institute yielded the following equations: Bio-Rad 50.970 (6 0.0175) 3 Beckman 1 8.0 (6 14.59) (6SE), r 50.994, Sxuy 5 45.7 for Pyr, and Bio-Rad 5 0.870 (6 0.0274)3 Beckman 1 17.4 (6 5.56), r 5 0.983, Sxuy 5 17.5 for Dpyr.The slight underestimation of the Bio-Rad kit may beimputable to the lack of an internal calibrator to minimizehandling errors.

Comparison of the concentrations of free forms ofcross-links measured with the Cross Links with the meanof duplicate results of the Pyrilinks-D kit in 35 urinesamples yielded: ELISA 5 0.865 (6 0.0519) 3 Beckman 13.71 (4.489), Sxuy 5 10.3 (r 5 0.947). Free Dpyr concentra-tions found with both kits represented ;40% of thecorresponding total form, in agreement with data re-ported in literature [11].

We conclude that the Cross Links HPLC kit providesaccurate results with a short and automated analysis (10

min) that can be performed overnight, thus reducing thetechnician time. We proved also that the detection limit ofthe Beckman kit allows the quantification of the free formsof cross-links (avoiding the time-consuming hydrolysisstep) with a simple open monopump HPLC system.Despite the comparable costs (HPLC ;$29/sample vs theEIA ;$35/sample), the possibility of achieving in a singlerun the quantification of Pyr, Dpyr, and their ratio con-tributed to making this procedure an attractive alternativeto the easily performed immunoassay.

We thank G. Mazzitelli and Beckman Analytical S.P.A.Italy for the gift of the Cross Links kit.

References1. Barnard K, Light D, Sims TJ, Bailey AJ. Chemistry of the collagen cross-links.

Biochem J 1987;244:303–9.2. Demers MD. New biochemical marker for bone disease: is it a break-

through? [Editorial]. Clin Chem 1992;38:2169–70.3. Robins SP, Duncan A, Wilson N, Evans BJ. Standardization of pyridinium

crosslinks, pyridinoline and deoxypyridinoline, for use as biochemical mark-ers of collagen degradation. Clin Chem 1996;42:1621–26.

4. Delmas PD. Biochemical markers of bone turnover. I. Theoretical consider-ations and clinical use in osteoporosis. Am J Med 1993;95:5A–16S.

5. Black D, Duncan A, Robins SP. Quantitative analysis of the pyridiniumcrosslinks of collagen in urine using ion-paired reversed-phase high perfor-mance liquid chromatography. Anal Biochem 1988;169:197–203.

6. Kollerup G, Thamsborg G, Bhatia H, Sorensen OH. Quantitation of urinaryhydroxypyridinium cross-links from collagen by high-performance liquid chro-matography. Scand J Clin Lab Invest 1992;52:657–62.

7. Takahashi M, Hoshino H, Kushida K, Inoue T. Direct measurement ofcrosslinks, pyridinoline, deoxypyridinoline, and pentosidine, in the hydroly-sate of tissue using high-performance liquid chromatography. Anal Biochem1995;232:158–62.

8. Delmas PD, Gineyts E, Bertholin A, Garnero P, Marchand F. Immunoassay ofpyridinoline crosslink excretion in normal adults and in Paget’s disease.J Bone Miner Res 1993;8:643–8.

9. Gomez B, Ardakani S, Evans B, Merrell L, Jenkins DK, Kung VT. Monoclonalantibody assay for free urinary pyridinium cross-links. Clin Chem 1996;42:1168–75.

10. Kamel S, Brazier M, Picard C, Boitte F, Samson L, Desmet G, Sebert JL.Urinary excretion of pyridinolines crosslinks measured by immunoassay andHPLC techniques in normal subjects and in elderly patients with vitamin Ddeficiency. Bone Miner 1994;26:197–208.

11. Kamel S, Brazier M, Neri V, Picard C, Samson L, Desmet G, Sebert JL.Multiple molecular forms of pyridinolines cross-links excreted in humanurine evaluated by chromatographic and immunoassay methods. J BoneMiner Res 1995;10:1385–92.

Convenient, Rapid Test for Lead in Blood with Use ofDisposable Electrodes, Elliot Plotkin,1,2* Jerome F.McAleer,1,2 M. Lucinda Cordeiro,1 Martin R. Ackland,1 Tim-othy M. Sheehan,3 and Robin A. Braithwaite3 (1 EcossensorsLtd., 74 Sunderland Rd., Sandy Bedfordshire SG19 1QY,UK, and 3 Regional Lab. for Toxicol., City Hosp., Birming-ham B18 7QH, UK; 2 present address and * address forcorrespondence: Inverness Medical Ltd., Beechwood ParkNorth, Inverness IV2 3ED, UK, fax 44-1463-724601)

In recent years concern over the adverse effects of lowconcentrations of lead on children has increased. In 1991,the CDC reduced the acceptable blood lead concentrationfrom 250 mg/L to 100 mg/L and recommended screeningof all American children ,6 years old for lead poisoning[1]. Graphite furnace atomic absorption spectroscopy is

Fig. 1. Chromatographic separations of calibrator mixture containingPyr (514 pmol/mL), Dpyr (56 pmol/mL), and internal standard (IS) (A),hydrolyzed urine sample (Pyr 5 820 pmol/mL, Dpyr 5 185 pmol/mL)(B), and nonhydrolyzed urine sample (Pyr 5 260 pmol/mL, Dpyr 553.2 pmol/mL) (C).HPLC analysis was performed on a reversed-phase column at 37 °C, with a flowrate of 1 mL/min and injection volume of 50 mL. Fluorescence detection: lex 5295 nm, lem 5 395 nm.


the most common method of measuring lead in blood, butthe CDC has encouraged the development of other meth-ods that could be used for mass population screening ornear-patient testing. Such methods should be portable,cheap, and easier to use than graphite furnace atomicabsorption spectroscopy as recommended by CDC Pro-gram Announcement 269, 1992.

The electrochemical technique of stripping voltametryat a mercury electrode has also been used for blood leadanalysis. Commercially available instruments based onthis method have been used widely but have insufficientaccuracy and precision for measuring lead at low concen-trations [2]. Recently, improvements have been made inthe electrochemical measurement of blood lead. Ostapc-zuk [3] and Jagner et al. [4] obtained good accuracy andprecision at low concentrations by potentiometric strip-ping analysis. They used a mercury-coated graphite elec-trode that must be cleaned between each analysis, andtheir testing procedure required stirring of the acidifiedblood solution.

An approach to stripping analysis that simplifies thetesting is the use of disposable electrodes, which can beused once and then thrown away. Microarray electrodeshave properties that make them especially suitable for thisapplication. They have high current densities in unstirredsolution, have a high signal-to-background ratio, and arenot affected by dissolved oxygen. Furthermore, they canbe made cheaply by a combination of screen printing andlaser photoablation (international patent WO 91/08474)[5]. In sufficiently large numbers, e.g., 106 electrodes peryear, they could be manufactured for ;$1 each. RecentlyFeldman et al. [6] described the use of disposable micro-electrode arrays for measuring lead in blood. Their testingprotocol required that the electrodes be preplated withmercury. They also found that the electrodes respondeddifferently to different blood types, and they had to usestandard additions to determine the lead concentration.

Here we report a simplified procedure for the use ofdisposable microelectrodes to measure lead in blood.Mercury is not preplated or added to the test solution butis contained in a precoated layer on the electrode andcoplates with the lead. These devices can be precalibrated,removing the need to perform standard additions.

The disposable electrodes were similar to those de-scribed by Feldman et al. [6] but also incorporated acarbon counterelectrode in addition to the Ag/AgCl ref-erence electrode and carbon microelectrode array. Each 2mm 3 3 mm array consists of 13 rows of 19 microelec-trodes, each with a diameter of ;40 mm. The electrodeswere cleaned by soaking for 30 min in 0.1 mol/L HCl. Areagent layer containing ;11 mg of mercury was ink-jetted over each array with the use of a Bio Dot Micro-doser. The ink-jetting solution contained 12.5 g/L mercu-ric nitrate, 15.4 g/L hydroxyethylethylenediaminetriaceticacid (HEDTA) (Aldrich), 30.0 g/L carboxymethylcellulose(CMC) (Aqualon), 1.0 g/L hydroxyethylcellulose (HEC)(Fluka), and 50 mmol/L KCl. The CMC and HEC arefilm-formers, and the HEDTA is a chelating agent that

stabilizes the mercury (international patent applicationPCT/GB96/00301).

The procedure for performing a blood lead test in-volved two steps, acidification and electrochemical mea-surement. The initial acidification step involved the addi-tion of 200 mL of blood to 2.0 mL of 0.9 mol/L HCl in a20-mL polystyrene vial. The vial was then mixed brieflyby hand and put on a bottle roller for ;5 min to ensurecomplete release of the bound lead before the blood wastested. Earlier experiments showed that 1 min was suffi-cient for complete release, but results were unchanged forup to 45 min.

The electrochemical testing was carried out with anAutolab (EcoChemie) in the differential pulse mode.Acidified blood (75 mL) was pipetted to a disposableelectrode, and deposition was effected by polarizing theelectrode at 2800 mV for 165 s. The stripping sequenceused a pulse amplitude of 50 mV, step 5 mV, pulse width3 ms, and trough width 120 ms. Lead-stripping peakheights were measured with use of the Autolab software.

The linear response of the electrodes was tested by theaddition of various amounts of Pb(NO3)2 (Aldrich AAstandard) to portions of a blood sample to give a range oflead concentrations up to 1000 mg/L. The exact leadconcentration of each supplemented blood sample wasmeasured by atomic absorption. Each blood was testedwith 5 disposable electrodes. A graph of peak currentagainst lead concentration was linear up to ;600 mg/Lwith a slope of 0.0070 mA per mg/L and an intercept of0.212 mA. The intercept may result from the presence oflead in the mercury layer on the electrodes. This graphwas used as a calibration curve for subsequent electrodes.

Blood samples were selected from those sent to theRegional Toxicology Laboratory for clinical investigationor occupational monitoring of lead exposure [7]. In themajority of cases the specimens had been anticoagulatedwith potassium EDTA, although some pediatric sampleshad been collected into Li heparin. All the specimencollection tubes used were known to be free of leadcontamination.

Blood lead was measured by electrothermal atomicabsorption spectroscopy with the use of a Varian SpectraAA 44 spectrophotometer equipped with deuterium back-ground correction and a GTA-96 graphite furnace (VarianUK). Blood was assayed directly after suitable dilutionwith an ammonium nitrate/ammonium phosphate mod-ifier solution, with calibration against matrix-matchedcalibrators. Batch accuracy was monitored by the use ofblood reference materials with accurately assigned values[7] (UK Supra Regional Assay Service Trace ElementsLaboratories). Laboratory performance was assessed byparticipation in two national external quality-assessmentschemes: NEQUAS lead and cadmium in blood (WolfsonEQA Laboratories) and TEQAS Trace Elements (RobensInstitute, University of Surrey, Guildford).

The accuracy of the electrodes was determined bydoing a correlation study with 20 venous blood samples,the lead concentrations of which had been determined asdescribed above. Each blood was tested with 2 electrodes.


The lead concentrations were calculated from the lead-stripping peaks with the use of supplemented bloodcalibration. The results of the correlation study (Fig. 1)show an excellent agreement between the disposableelectrodes and the comparison method. All 40 measure-ments were within 20 mg/L of the comparison methodresults. Of the 22 measurements on blood samples withlead concentrations ,200 mg/L, 21 of them were within10 mg/L of the comparison method result. The lowintercept and slope close to 1 demonstrate that the sup-plemented blood calibration works well.

The precision of the electrodes was tested with 3 venousblood samples with lead concentrations of 50, 110, and 250mg/L. Each blood sample was acidified once, and thesolution was tested with 10 electrodes. The observed CVwas 11.9%, 7.2%, and 2.7%, respectively.

In this electrochemical method of measuring lead inblood by disposable electrodes, a minimum of samplepreparation and the use of precalibrated electrodesshould enable relatively unskilled people to use thismethod for screening purposes. Although this lead testwas carried out with 200-mL venous samples, smallervolumes, which would enable fingerstick samples to betested, could be used. The method is sufficiently accurateand precise to meet guidelines recommended by the CDC(Program Announcement 269, 1992). This techniquemight also have application in surveys of blood leadconcentrations in lead-exposed remote populations indeveloping countries where laboratory facilities are notavailable.

We thank Manuel Alvarez-Icaza for helpful discussionsand Enviromed PLC for financial assistance.

References1. US DHHS Public Health Service. Preventing lead poisoning in young children.

Atlanta: Centers for Disease Control, October 1991.2. Braithwaite RA. Interlaboratory and intralaboratory surveys, reference meth-

ods and reference materials. In: Herber RFM, Stoeppler M, eds. Traceelement analysis in biological specimens: techniques and instrumentationin analytical chemistry, Vol. 15. Amsterdam: Elsevier, 1994:213–32.

3. Ostapczuk P. Direct determination of cadmium and lead in whole blood bypotentiometric stripping analysis. Clin Chem 1992;38:1995–2001.

4. Jagner D, Renman L, Wang Y. Determination of lead in microliter amounts ofwhole blood by stripping potentiometry. Electroanalysis 1994;6:285–91.

5. Wang J, Lu J, Tian B, Yarnitzky C. Screen-printed ultramicroelectrode arraysfor on-site stripping measurements of trace metals. J Electroanal Chem1993;361:77–83.

6. Feldman BJ, D’Alessandro A, Osterloh JD, Hata BH. Electrochemical deter-mination of low blood lead concentrations with a disposable carbon microar-ray electrode. Clin Chem 1995;41:557–63.

7. Braithwaite RA, Girling AJ. Bovine reference materials for accuracy control ofblood lead analysis. Fresenius Z Anal Chem 1988;332:704–9.

Improved Therapeutic Drug Monitoring of Tacrolimus(FK506) by Tandem Mass Spectrometry, Paul J. Taylor,1,2*

Nicholas S. Hogan,2 Stephen V. Lynch,3 Anthony G. John-son,1,2 and Susan M. Pond1,2 (1 Dept. of Clin. Pharmacol.,First Floor Lions Clin. Res. Bldg., 2 Univ. of QueenslandDept. of Med., and 3 Univ. of Queensland Dept. of Sur-gery, Princess Alexandra Hosp., Ipswich Rd., Brisbane,QLD, Australia, 4102; * author for correspondence: fax61-7-3240-5031; e-mail [email protected])

The immunosuppressant drug tacrolimus (FK506), whichexhibits 50–100 times the potency of cyclosporin A, isproving to be highly effective in preventing rejection insolid-organ transplantation [1]. However, because of anarrow therapeutic range, variable pharmacokinetics, andpotential drug interactions, continual therapeutic drugmonitoring (TDM) of tacrolimus is essential [2]. Recently,we published a report that detailed the development of aspecific, sensitive method for quantification of tacrolimusconcentrations in blood [3]. This methodology, whichutilizes HPLC in combination with tandem mass spec-trometry (LC-MS2), was found to have greater specificity,lower detection limits, and a more rapid turnaround timethan existing immunoassays. These attributes make thismethodology ideal for TDM of tacrolimus.

Since our initial report, several modifications that havefurther improved the assay for TDM have been imple-mented. The use of a 100 3 2 mm C8 column (rather thana 30 3 2 mm C4 column), combined with a higher flowrate (400 mL/min, compared with 100 mL/min) andpostcolumn splitting, has reduced the chromatographicassay time from 4 to 2 min. Although the retention timesof the analytes are similar with both methods, intermittentlate-eluting peaks observed previously are no longerpresent, allowing shorter analysis time. In addition, theuse of a single point calibrator (i.e., linear through zero) at10.0 mg/L and two internal quality controls (5.0 and 20.0mg/L), compared with a seven-point calibration curveand three internal quality controls, has reduced the anal-ysis time further. Previously, we reported an overallturnaround time of 2.5 h for a batch of 20 samples (10patient samples). The modified method allows an analysis

Fig. 1. Correlation between blood lead determined by disposablemicroelectrodes and by the atomic absorption method.Duplicate measurements were made on 20 venous blood samples.


time of ;1.75 h for a batch of 20 samples (17 patientsamples). This reduces the analysis time for each patientsample from 15 to 6 min.

Because of the sensitivity (low detection limit) of themethod, we have been able to halve the sample volume ofblood required to perform the assay (i.e., from 1000 to 500mL). The postcolumn split of solvent flow into the massspectrometer (1:12) provides enhanced stability of themass spectrometer response for long periods of time (.12h), without compromising assay sensitivity. For this rea-son a postcolumn split of solvent flow is now routine forall analytical assays conducted in this laboratory.

The imprecision and accuracy of the modified methodwere assessed with internal QCs at three concentrations(0.25, 5.0, and 20.0 mg/L). Over the concentration rangestudied, within-day and between-day imprecision were,11%, and analytical recovery was between 98% and105%. A comparison of these values with those reportedin our original paper, with different concentrations (0.3,8.0, and 80.0 mg/L), revealed similar results (Table 1). Inaddition to internal quality controls, the modified methodwas evaluated against the European Quality AssessmentScheme for tacrolimus. This evaluation involved the com-parison of unknown weighed-in concentrations (n 5 12)with the results determined by our method. The evalua-tion showed good correlation (y 5 0.994x 1 0.398, r 50.996, mean bias 5 0.258 mg/L) and no significant differ-ence (P 5 0.97) between our results and weighed-inconcentrations.

As a cost benefit, the viability of repeated use of the C18solid-phase extraction cartridges was investigated by re-analyzing randomized patient samples on used car-tridges. A comparison of first use and second use of thecartridges revealed that no significant differences existedbetween the concentrations measured by the LC-MS2 (y 51.03x 1 20.127, r 5 0.993, mean bias 5 20.167 mg/L, n 595). As a result of the large number of patient samplesbeing analyzed annually within the laboratory (i.e.,.2000), the 20% reduction in consumable expenditureachieved through a single reuse of the solid-phase extrac-tion cartridges represents substantial cost savings.

These modifications have led to a more time- and

cost-efficient TDM service for tacrolimus, without a com-promise in assay results. The specifications of our method(i.e., specificity, accuracy, and precision) meet the analyt-ical requirements described by Shaw et al. [4] for themeasurement of immunosuppressant drugs. Further, themethod’s specificity for the parent drug lends itself todetailed pharmacokinetic studies and studies of potentialdrug interactions involving tacrolimus. With the in-creased knowledge of tacrolimus treatment in organtransplantation, the proposed therapeutic range for thedrug has been reduced. Our improved method, whichprovides accuracy and precision at these lower concentra-tions, is therefore timely, as an important means ofsupporting the clinical management of patients undergo-ing organ transplantation.

This project was supported in part by the Princess Alex-andra Hospital Research and Development Foundation.

References1. Peters DH, Fitton A, Plosker GL, Faulds D. Tacrolimus. A review of its

pharmacology, and therapeutic potential in hepatic and renal transplanta-tion. Drugs 1993;46:746–94.

2. Jusko WJ, Thomson AW, Fung J, McMaster P, Wong SH, Zylber-Katz E, et al.Consensus document: therapeutic monitoring of tacrolimus (FK-506). TherDrug Monit 1995;17:606–14.

3. Taylor PJ, Jones A, Balderson GA, Lynch SV, Norris RLG, Pond SM. Sensitive,specific quantitative analysis of tacrolimus (FK506) in blood by liquidchromatography-electrospray tandem mass spectrometry. Clin Chem 1996;42:279–85.

4. Shaw LM, Annesley TM, Kaplan B, Brayman KL. Analytical requirements forimmunosuppressive drugs in clinical trials. Ther Drug Monit1995;17:577–83.

Optimization of Nonisotopic PCR–Single-Strand Con-formation Polymorphism Analysis, Helene Blanche,*

Christel Valette, and Christine Bellanne-Chantelot (FondationJean Dausset-CEPH, 27 rue Juliette Dodu, 75010 Paris,France; * author for correspondence: fax 33-1-53-72-50-48;e-mail [email protected])

PCR–single-strand conformation polymorphism (SSCP)analysis is an attractive technique used to screen forunknown mutations because of its simplicity and wide-spread applicability [1]. The technique relies on single-nucleotide variations modifying the conformation of sin-gle-stranded DNA and therefore its mobility inpolyacrylamide gels. The detection of these conformers isperformed either by autoradiography or by nonisotopicmethods such as silver-staining [2], ethidium bromide-staining [3], chemiluminescence [4], or fluorescence [5].

In this paper, we report a nonisotopic PCR-SSCPmethod with the use of the Pharmacia MultiPhorTM

(Pharmacia Biotech) electrophoresis unit for sensitive,reproducible, and cost-effective experiments that can beperformed at high throughput. This method was estab-lished to analyze the 12 exons of the glucokinase (GCK)gene to identify mutations involved in maturity-onset

Table 1. Imprecision and analytical recovery of the originaland modified LC-MS2 whole-blood assays.

Weighed-inconcn., mg/L

CV, %Analytical

recovery, %aWithin-day Between-day

0.30b 7.0 3.3 100.30.25c 5.8 10.9 104.48.0b 1.7 0.42 99.45.0c 2.7 2.5 101.2

80.0b 1.0 1.9 99.420.0c 1.6 1.7 98.5

a Determined from the mean concentration of the between-day controlsdivided by the weighed-in concentration and expressed as a percentage.

b Original LC-MS2, n 5 16.c Modified method, n 5 10.


diabetes of the young, a subset of non-insulin-dependentdiabetes.

The 12 GCK exons [6] were amplified by PCR (frag-ments ranging from 145 to 367 bp), as reported previously[2], either on GeneampTM 9600 (Perkin-Elmer) orPTC100TM (MJ Research) DNA thermal cyclers.

Forty-six known mutations [6–8] were used to establishthe optimal electrophoretic conditions for the 12 PCRfragments. Two types of precast polyacrylamide gels,allowing the analysis of 34 and 23 DNA samples, respec-tively, were used: nondenaturing gels [Cleangel-HP®,10% total acrylamide (T) concentration, 2% total extent ofcross-linking (C); ETC Elektrophorese-Technik] and par-tially denaturing gels (Excelgel®, 7.5% T, 3% C, PharmaciaBiotech). Electrophoresis was performed at temperaturesof 6–20 °C, the length of assays depending on size and GCcontent of the PCR fragments. For all exons except exon 9,samples were run on Cleangel-HP, rehydrated in Delect®

(ETC Elektrophorese-Technik) gel buffer, with electrodewicks soaked in Delect anode and cathode buffers. Elec-trophoresis on Cleangel-HP was carried out in three steps:prerun (200 V, 20 mA, 10 W) and run (375 V, 30 mA, 20W), followed by a last step to refine bands (450 V, 30 mA,20 W). For exon 9, Excelgel and sodium dodecyl sulfatebuffer {200 mmol/L Tricine [N-tris(hydroxymethyl)meth-ylglycine], 200 mmol/L Tris, 5.5 g/L sodium dodecylsulfate} were used, and a one-step run was performed:550 V, 30 mA, 20 W. Specific electrophoretic conditions(assay time and temperature) were determined for eachexon (Table 1). Gels were stained with a silver-staining kit(Silver Staining Kit Plus One®; Pharmacia Biotech) andthen wrapped in cellophane (soaked in a solution of 100mL/L glycerin and 100 mL/L acetic acid) for preserva-tion.

All nucleotide changes previously identified [8] weredetected. Fig. 1 shows the electrophoretic profile of exon10. The effect of different electrophoretic conditions wasevaluated from the number of bands characteristic of agiven PCR product, their sharpness, and resolution. Thepurification of our PCR products tested on exons 6, 9, and10 (data not shown) did not provide an increase insensitivity, contrary to other published results [9]. Indeedwith specific PCR products, in addition to two conform-ers, bands resulting from the interactions between PCRprimers and the single-stranded DNA were visualizedand provided an easier analysis of the electrophoreticprofile.

The protocol used for the GCK gene allowed us toestablish a successful strategy for the development ofPCR-SSCP on other genes such as BRCA1 (breast cancer 1,data not shown) on the basis of GC content and the lengthof PCR products. We found high sensitivity, especially infragments generally difficult to test with the use of thePCR-SSCP technique, i.e., larger than 250 bp and present-ing a high GC content ($60%). Lower detection limits andreliability of the detection of mutations by this PCR-SSCPanalysis may be explained by multiple factors: (a) the useof precast gels; (b) the efficient temperature regulationwith an independent thermostatic circulator that provides

a wide range of precise running temperatures from 6 to20 °C and avoids the addition of glycerol to the gels oftenreported as decreasing the effects of temperature variabil-ity [10]; and (c) the use of partially denaturing conditionswith a temperature varying from 6 to 12 °C, which in-creases the sensitivity of PCR-SSCP analysis, particularlyfor large or GC-rich fragments (exon 9). We hypothesizethat mild denaturing conditions extend the exposed sur-face area of single-stranded DNA, which tends to assumea folded configuration in the complete absence of dena-turing agents. Therefore, detection of locally confinedstructural differences in PCR fragments is improved.

Thus, the lack of detection often encountered with thePCR-SSCP analysis, as compared with other current tech-nologies for the study of large fragments, is overcome inthese conditions. Moreover, screening for mutations onthe MultiPhor system is not expensive considering thegel-loading capacity, apparatus, and reagent prices as

Table 1. Electrophoretic conditions for PCR-SSCP analysison the MultiPhor.

Exon Size, bpGC content,

%Temperature,

°C Assay time

1a 195 57.0 20 Step 1 15 minStep 2 1 h 10 minStep 3 5 min

1b 149 57.0 20 Step 1 15 minStep 2 1 hStep 3 5 min

1c 145 54.5 20 Step 1 15 minStep 2 1 hStep 3 5 min

2 290 62.4 15 Step 1 20 minStep 2 3 hStep 3 15 min



5 195 64.1 20 Step 1 15 minStep 2 1 h 10 minStep 3 5 min




9a 367 67.0 12 4 h10 263 66.3 20 Step 1 15 min

Step 2 1 h 20 minStep 3 15 min

a Run on Excelgel 7.5%; all others analyzed on Cleangel-HP 10%.


compared with fluorescent methods that require an auto-matic sequencer. Contrary to isotopic PCR-SSCP analysis,which is less reproducible, only a single run is routinelyrequired to screen each exon.

In conclusion, PCR-SSCP analysis on the MultiPhorappears to be a useful and reliable tool in screening forunknown sequence variations at a high throughput and isespecially adapted for laboratories that cannot performfluorescent PCR-SSCP analysis on an automatic se-quencer. An easy, nonradioactive detection method isrequired more than ever because of genetic diagnosisprograms, in which routine screening for mutations willbecome more important for healthcare purposes.

We are grateful to M. F. Legrand for technical assistanceand to clinicians and patients who send us blood samples.We thank Howard Cann for helpful discussions about themanuscript. This study was funded by the “Ministere deL’Education nationale, de L’Enseignement superieur et dela Recherche” and the “Conseil Regional d’Ile de France.”

References1. Hayashi K. PCR-SSCP. A simple and sensitive method for detection of

mutations in the genomic DNA. PCR Methods Appl 1991;1:34–8.2. Blanche H, Hager J, Sun F, Dausset J, Cohen D, Froguel P, Cohen N.

Nonradioactive screening of glucokinase mutations in maturity onset diabe-tes of the young. BioTechniques 1994;16:866–75.

3. Yap EP, McGee JO. Nonisotopic SSCP detection in PCR products by ethidiumbromide staining. Trends Genet 1992;8:49.

4. Knoblauch H, Weiss N, Eggersdorfer I, Keller C, Schuster H. A nonisotopicsingle-strand conformation polymorphism protocol using a direct blottingelectrophoresis, a chemiluminescent detection system, and a multiplexapproach. PCR Methods Appl 1994;4:52–5.

5. Iwahana H, Fujimura M, Takahashi Y, Iwabuchi T, Yoshimoto K, Itakura M.Multiple fluorescence-based PCR-SSCP analysis using internal fluorescentlabeling of PCR products. BioTechniques 1996;21:510–9.

6. Stoffel M, Froguel P, Takeda J, Zouali H, Vionnet N, Nishi S, et al. Humanglucokinase gene: isolation, characterization, and identification of twomissense mutations linked to early-onset non-insulin-dependent (type 2)diabetes mellitus. Proc Natl Acad Sci U S A 1992;89:7698–702.

7. Froguel P, Zouali H, Vionnet N, Velho G, Vaxillaire M, Sun F, et al. Familialhyperglycemia due to mutations in glucokinase—definition of a subtype ofdiabetes mellitus. N Engl J Med 1993;328:697–702.

8. Velho G, Blanche H, Vaxillaire M, Bellanne-Chantelot C, Pardini VC, Timsit J,et al. Identification of 14 new glucokinase mutations and description of theclinical profile of 42 MODY2 families. Diabetologia 1997;40:217–25.

9. Cai Q-Q, Touitou I. Excess PCR primers may dramatically affect SSCPefficiency. Nucleic Acids Res 1993;21:3909–10.

10. Teschauer W, Mussack T, Braun A, Waldner H, Fink E. Conditions for singlestrand conformation polymorphism (SSCP) analysis with broad applicability:

a study on the effects of acrylamide, buffer and glycerol concentrations inSSCP analysis of exons of the p53 gene. Eur J Clin Chem Clin Biochem1996;34:125–31.

Low Concentrations of Folate in Serum and Erythro-cytes of Smokers: Methionine Loading Decreases FolateConcentrations in Serum of Smokers and Nonsmokers,Mohammad A. Mansoor,1* Ole Kristensen,1 Tor Hervig,2 PerA. Drabløs,3 Jacob A. Stakkestad,4 Leik Woie,5 Øyvind Het-land,1 and Arve Osland1 [1 Dept. of Clin. Chem. (* addressfor correspondence: fax 47-51519907), 2 Blood Bank, and5 Dept. of Cardiol., Central Hosp. in Rogaland (SiR), 4003Stavanger; 3 Hydro Aluminium AS Karmøy; and 4 Dept.of Clin. Chem., Haugesund Hosp., Rogaland, Norway]

Recent data suggest that hyperhomocysteinemia is asso-ciated with an increased risk for premature vasculardisease. Total plasma homocysteine (tHcy) may be in-creased by impaired activity of enzymes or suboptimalavailability of B vitamins [1].

Homocysteine transsulfuration reactions are catalyzedby the enzymes cystathionine b-synthase and cystathi-onine lyase in the presence of coenzyme vitamin B6, andremethylation of homocysteine is performed by the en-zymes methionine synthase and betaine-homocysteinemethyltransferase with vitamin B12 as a coenzyme for theformer enzyme. 5-Methyltetrahydrofolate (5-methyl-THF) or betaine are methyl donors during remethylation.The enzyme 5,10-methylenetetrahydrofolate reductase(MTHFR) with coenzyme FADH2 synthesizes 5-methyl-THF from 5,10-methylenetetrahydrofolate (5,10-MTHF)[2]. The postmethionine load (PML) abnormal tHcy incre-ments may reflect suboptimal plasma vitamin B6 or defi-ciency of cystathionine b-synthase [3]. The present reportdescribes the effects of methionine loading on serum anderythrocyte folate, tHcy, and related thiols in smokers andnonsmokers.

We recruited 63 apparently healthy smokers [ages37.8 6 0.8 years; body mass index (BMI), 25.4 6 0.5;male/female, 39/24; cigarettes/day, 16.2 6 0.8; smokingperiod (years), 19.7 6 0.7], 44 nonsmokers (ages 37.8 6 1.0years; BMI, 24.5 6 0.5; male/female, 26/18), and 23former smokers who refrained from smoking for $3months (ages 39.5 6 1.0 years; BMI, 27.7 6 0.5; male/female, 18/5) in the study. BMI and age of the participantswere not statistically different. Informed consent wasobtained from the participants, and the study was ap-proved by the regional ethics committee.

Blood sampling and the methionine loading test wereperformed as described previously [4]. Participants wereasked to eat a light breakfast or lunch with low contents offats and proteins. Serum and erythrocyte folate andvitamin B12 were measured by a Dualcount, solid-phase,no-boil RIA developed by Diagnostic Products. tHcy,cysteine, and cysteinylglycine were measured accordingto the method described previously [5].

The Mann–Whitney U-test was applied to investigatethe differences between the groups, and the Wilcoxon

Fig. 1. PCR-SSCP analysis of a GC-rich DNA fragment of the GCK gene.SSCP profiles on exon 10: Lane 1 corresponds to a negative control, lanes 2–4show samples carrying, respectively, mutations Stop466Leu, Ala450Thr, andSer498–1G3C. *, additional conformers present on samples carrying a muta-tion.


signed rank test was used to explore changes within thegroups during methionine loading. Significance of differ-ence between more than two groups was estimated byanalysis of variance with the use of the Bonferroni/Dunnprocedure. StatView for the Macintosh developed byAbacus Concepts was used for statistical calculations.

The concentrations of serum and erythrocyte folate ofsmokers were significantly lower than those of nonsmok-ers (P ,0.05) (Table 1). Lower intake of fruits and vege-tables, free radicals in cigarette smoke, and increasedexcretion of folates may have contributed to a gradualdecline in the concentration of serum and erythrocytefolate [6–9]. An 11–16% reduction in the concentrations ofserum folate was detected during methionine loading insmokers as well as in nonsmokers 2 h (P ,0.01 and,0.001, respectively) and 6 h afterwards (P ,0.01 and,0.001, respectively) (Fig. 1, A). A significant decrease inserum folate of former smokers was observed after 6 h (P,0.01) (not shown in the figure). A drop in the concen-tration of serum folate may have occurred because of thereduced activity of enzyme MTHFR. Methionine loadingincreases the concentration of S-adenosylmethionine, andincreased concentrations of S-adenosylmethionine inhibitthe reductase activity of MTHFR [10].

Women had higher concentrations of serum folate thanmen (P ,0.05) (data not shown). Nonsmokers had higherconcentrations of serum folate than former smokers (P,0.01), but the difference in erythrocyte folate was not

statistically different. Former smokers may not exhibit asignificant change in serum folate until erythrocytes grad-ually accumulate folate to an optimal amount during theperiod of smoking cessation.

The concentrations of erythrocyte folate in smokers andnonsmokers remained unaffected during methionineloading (Fig. 1, B), but a significant reduction in erythro-cyte folate was detected in former smokers after 2 h (P,0.05) (not shown in the figure). Because 5-methyl-THFmakes up ;40–50% of total folate polyglutamates inerythrocytes, minor changes in the concentration of 5-methyl-THF may not influence substantially total eryth-rocyte folate.

Mean tHcy concentrations were significantly lower inwomen than in men (P ,0.01); the significance disap-peared when data were analyzed for male smokers vsfemale smokers (P 5 0.07) and male nonsmokers vsfemale nonsmokers (P 5 0.09). Smokers had slightlyhigher concentrations of tHcy than nonsmokers, but thedifference was not significant even if data were analyzedseparately for gender (data not shown). Previously, it hasbeen reported that heavy smokers have significantlyhigher concentrations of tHcy than nonsmokers [11].

The mean concentration of tHcy increased, and plasmaconcentrations of cysteine and cysteinylglycine decreasedduring methionine loading, as shown previously (Fig. 1,C, D, and E) [5]. Large doses of methionine given to miceand rats seem to perturb activity of enzymes and concen-

Table 1. Concentrations of serum, erythrocyte folate, homocysteine, and related thiols in plasma of smokers, nonsmokers,and former smokers.

Smokers Nonsmokers Former smokers

FastingSerum folate 12.5 (11.2–13.9)a 15.2 (13.5–17.0)b 11.3 (10.1–12.5)Erythrocyte folate 605.5 (556.4–654.6)a 708.9 (654.1–763.7) 649.5 (571.8–727.2)Homocysteine 11.4 (10.4–12.4) 10.7 (9.9–11.5) 11.0 (9.8–12.2)Cysteine 235.1 (226.5–243.6) 240.7 (231.1–250.2) 253.9 (239.6–268.2)Cysteinylglycine 37.2 (34.9–39.5) 34.4 (32.1–36.6) 34.8 (31.7–37.9)

2-h PMLSerum folate 11.5 (10.3–12.6)c 14.0 (12.4–15.6)d 11.7 (10.5–12.8)Erythrocyte folate 614.1 (567.2–661.0)a 700.0 (648.0–751.9)b 576.4 (524.6–628.2)Homocysteine 26.8 (24.7–28.2) 26.0 (23.6–27.5) 26.4 (23.2–29.6)Cysteine 233.7 (224.8–242.5) 236.9 (228.4–245.4) 243.4 (229.6–257.3)Cysteinylglycine 33.4 (31.3–35.5)e 31.3 (28.7–33.9) 28.0 (25.1–31.0)

6-h PMLSerum folate 10.6 (9.6–11.5)a 12.7 (11.1–14.2)d 10.0 (9.1–10.9)Erythrocyte folate 628.9 (580.5–677.0)a 714.4 (661.2–768.2)b 603.7 (558.7–648.7)Homocysteine 33.1 (31.0–35.8) 32.2 (30.1–36.2) 35.0 (31.3–38.7)Cysteine 219.1 (211.5–226.7) 224.2 (216.8–231.6) 229.2 (214.1–244.2)Cysteinylglycine 28.6 (26.7–30.5) 27.3 (25.4–29.2) 27.0 (24.9–29.2)a P ,0.05, smokers (63) vs nonsmokers (44).b P ,0.01, nonsmokers (44) vs former smokers (23).c P ,0.01, smokers vs nonsmokers.d P ,0.05, nonsmokers vs former smokers.e P ,0.05, smokers vs nonsmokers.The concentration units are nmol/L for folate and mmol/L for homocysteine and related thiols. Data are given as mean values with 95% confidence intervals for the

means in parentheses.


trations of amino acids involved in methionine metabo-lism [12, 13].

The 2-h PML response was abnormal (DtHcy concen-tration .90th percentile) in 10% of smokers, 9% of formersmokers, and 7% of nonsmokers. The 6-h PML responsewas abnormal in 10% of smokers, 9% of former smokers,and 9% of nonsmokers. The concentrations of serumvitamin B12 in smokers, nonsmokers, and former smokers(338, 353, and 356 pmol/L, respectively) were not signif-icantly different, and the PML concentrations did notchange significantly in the groups (data not shown).

We conclude that smokers have lower concentrations ofserum and erythrocyte folate than nonsmokers and thatmethionine loading decreases serum folate in all groups.

We thank the following persons for their cooperation andtechnical assistance: Anne Marie Aarstad, Kari Skibeli,Gunn Walmestad, Kirstin Gard, Ingjerd Rasmussen, andOtlo Osvik (Central Hospital in Rogaland), Helga Lie(Hydro Aluminium), and Kirsten Gundersen (HaugesundHospital).

References1. Kang SS, Wong PWK, Malinow MR. Hyperhomocyst(e)inemia as a risk factor

for occlusive vascular disease. Ann Rev Nutr 1992;12:279–98.

2. Wagner CS. Symposium on the subcellular compartmentation of folatemetabolism. J Nutr 1996;126:1228S–34S.

3. Sardarwalla IB, Fowler B, Robins AJ, Komrower GM. Detection of heterozy-gotes for homocystinuria. Study of sulphur-containing amino acids in plasmaand urine after L-methionine loading. Arch Dis Child 1974;49:553–9.

4. Mansoor MA, Svardal AM, Schneede J, Ueland PM. Dynamic relationbetween reduced, oxidized, and protein-bound homocysteine and other thiolcomponents in plasma during methionine loading in healthy men. Clin Chem1992;38:1316–21.

5. Mansoor MA, Svardal AM, Ueland PM. Determination of the in vivo redoxstatus of cysteine, cysteinylglycine, homocysteine, and glutathione in humanplasma. Anal Biochem 1992;200:218–29.

6. Subar AF, Harlan LC, Mattson ME. Food and nutrition intake differencesbetween smokers and nonsmokers in the US. Am J Public Health 1990;80:1323–9.

7. Pryor WA, Stone K. Oxidants in cigarette smoke. Radicals, hydrogen perox-ide, peroxynitrate and peroxynitrite. Ann NY Acad Sci 1993:686:13–28.

8. Gey KF. Prospects for the prevention of free radical disease, regardingcancer and cardiovascular disease. Br Med Bull 1993;49:679–99.

9. Herbert V. Development of human folate deficiency. In: Picciano MF,Stokstad ELR, Gregory JF, eds. Folic acid metabolism in health and disease.New York: Wiley-Liss, 1990:195–210.

10. Kutzbach C, Stokstad EL. Feedback inhibition of methylene-tetrahydrofolatereductase in rat liver by S-adenosylmethionine. Biochim Biophys Acta1967;139:217–20.

11. Nygård O, Vollset SM, Refsum H, Stensvold I, Tverdal A, Nordrehaug JE, etal. Total plasma homocysteine and cardiovascular risk profile. The Horda-land homocysteine study. JAMA 1995;274:1526–33.

12. Finkelstein JD, Martin JJ. Methionine metabolism in mammals. Adaptationto methionine excess. J Biol Chem 1988;261:1582–7.

13. Griffith OW, Bridges RJ, Meister A. Evidence that the g-glutamyl cyclefunctions in vivo using intracellular glutathione: effects of amino acids andselective inhibition of enzymes. Proc Natl Acad Sci U S A 1978;75:5405–8.

Fig. 1. Concentrations of serum and erythrocyte folate and plasmahomocysteine, cysteine, and cysteinylglycine before methionine load-ing (0 h) and 2 and 6 h afterwards.Results are presented as mean 6 SE. F, Smokers; E, nonsmokers. A, serumfolate; B, erythrocyte folate; C, plasma homocysteine; D, cysteine; E, cysteinyl-glycine, ns, nonsignificant.


Rapid Identification of Angiotensin-Converting En-zyme Genotypes by Capillary Electrophoresis, Xiao-Hong Huang,1 Anne Salomaki,1 Riikka Malin,1 Timo Koivula,1

Hannu Jokela,1 and Terho Lehtimaki1,2* (1 Dept. of Clin.Chem., Res. Lab. of Atherosclerosis Genetics, P.O. Box2000, Tampere Univ. Hosp., and 2 Dept. of Med. Biochem.,Tampere Univ. Med. School, FIN-33521, Tampere, Fin-land; * author for correspondence: fax 358-3-247-5554;e-mail [email protected])

The traditional method for DNA separation is slab gelelectrophoresis. Because slab gel electrophoresis, whichinvolves the casting, loading, running, and staining of thegel is labor-intensive and time-consuming, high-resolu-tion capillary electrophoresis (CE) has become an attrac-tive alternative. CE, by contrast to conventional electro-phoretic techniques, is capable of rapid, automated,reproducible, and high-resolution separation of minuteamounts of DNA samples. To separate DNA fragments byCE, polyacrylamide gels or liquid buffers containingsoluble polymers are used. Soluble polymers, such ashydroxyethyl cellulose, hydroxypropylmethyl cellulose,and methyl cellulose act as effective molecular sieves andallow for separation of DNA according to size [1]. CE hasrecently shown promising results for the analysis ofdouble-stranded DNA such as restriction fragment lengthpolymorphisms and polymerase chain reaction (PCR)products [2, 3].

The insertion (I)/deletion (D) of a 287-bp sequencepolymorphism within intron 16 of the gene for angioten-sin-converting enzyme (ACE) is strongly associated withserum ACE concentrations [4]. The D allele has beenidentified as a risk factor for the development of coronaryheart disease and myocardial infarction [5]. In addition,the deletion polymorphism has been associated witheither microalbuminuria or overt nephropathy in diabeticpatients [6, 7]. Here we report a sensitive, simple, rapid,and nonisotopic procedure for identification of ACE I/Dpolymorphism by CE.

Amplification of genomic DNA ACE I/D polymor-phism PCR products of a 190-bp fragment in the absenceof the insertion and a 490-bp fragment in the presence ofthe insertion were generated. The sense oligonucleotideprimer was 59-CTG GAG ACC ACT CCC ATC CTTTCT-39, and the antisense primer was 59-GAT GTG GCCATC ACA TTC GTC AGA T-39 [8]. The PCR mixturecontained 10 mmol/L Tris-HCl (pH 8.8), 50 mmol/L KCl,1.5 mmol/L MgCl2, 1 mL/L Triton X-100, 200 mmol/Leach of the four deoxynucleotides, 1 mmol/L each of theprimers, and 3 U of Dynazyme (Finnzymes) in a finalvolume of 50 mL. To enhance genotyping and to preventmisclassification of heterozygous individuals, we added50 mL/L dimethyl sulfoxide to the reaction mixture [9].Blank controls, containing no genomic DNA, and positivecontrols (ID) were also run with each set of amplification.The amplification cycle was performed on a PTC-100thermal cycler (MJ Research). After an initial denaturation

at 96 °C for 3 min, the DNA was amplified by 30 PCRcycles—each consisting of denaturation at 94 °C for 1 min,annealing at 65 °C for 1 min, and extension at 72 °C for 2min—followed by a final extension at 72 °C for 5 min.PCR products were directly analyzed by CE withoutbeing desalted or cleaned up.

CE separations were carried out with the use of theHP3D Capillary Electrophoresis System (Hewlett-Pack-ard). CEP-coated fused-silica capillaries provided withthe pGEM Double Stranded DNA Analysis Kit (Hewlett-Packard) of 56 cm effective length and 64.5 cm total lengthwere used. The separation/flush buffer consisted of 89mmol/L Tris, 332 mmol/L boric acid, and 2 mmol/LEDTA at pH 7.4 with additives of 15 g/L hydroxyethylcellulose (Hewlett-Packard). The buffer was preparedaccording to the manufacturer’s recommendation, fil-tered, and degassed by sonication. The cathode was set onthe injection side, the anode on the detection side. Sam-ples were introduced into the capillary by electrokinetic(210 kV for 30 s) injection, and the separation wasconducted at constant voltage of 222 kV (340 V/cm). Thecapillary was held at 20 °C throughout the experiment,and ultraviolet absorbance was monitored at 258 nm.Before each injection, a 10-min buffer wash was per-formed to ensure column cleanliness and reproducibility.A typical assay of samples lasted for 30 min. The identi-fication of the various DNA fragments by size wasobtained by plotting log bp vs 1/migration time.

Fig. 1A shows an electropherogram of the separation ofpGEM DNA calibrator with the use of a 15 g/L hydroxy-ethyl cellulose-filled capillary at 222 kV. The separationconditions were optimized in several preliminary experi-ments with electrokinetic and pressure injection; electro-kinetic injection gives higher resolution. The separationconditions in Fig. 1A were sufficient for the rapid andcomplete separation of the pGEM sample, which con-tained 15 fragments ranging from 36 to 2645 bp. The DNAcalibrator was run before the samples to ensure propercalibration of the analysis. Fig. 1B shows an example of asample from a heterozygous individual, who expressesboth 490- and 190-bp fragments. To confirm the accurateidentification of the ACE gene polymorphism, we ran 10samples, and all genotypes determined by CE were con-sistent with the results obtained by agarose gel electro-phoresis.

We have demonstrated excellent separation efficiencyfor PCR products of the ACE gene by CE, in substantiallyshorter analysis time than by slab gel electrophoresis.Several investigators have recently reported that PCRanalysis by CE may be hindered by salt in the reactionmixture, so salt must be removed before the CE analysis[10]. However, in the present study the ACE gene PCRproducts were analyzed directly by CE without furthersample treatment, which saves time and resources. Inaddition, samples can be loaded onto an autosampler andrun overnight, unattended, and analyzed data are directlyentered into computer. In our studies, one coated capil-lary column lasted for .1 month and was used for


hundreds of assays. Furthermore, this method avoids theuse of hazardous materials such as ethidium bromide orradioactivity. Taken together, CE provides a simple,rapid, sensitive, and accurate method for the identifica-tion of the ACE I/D gene polymorphism.

This work was supported by grants from the FinnishFoundation of Cardiovascular Research, Medical Re-search Fund of Tampere University Hospital, and the Elliand Elvi Oksanen Fund of the Pirkanmaan Regional Fundunder the auspices of the Finnish Cultural Foundation.

References1. Baba Y, Tsuhako M. Gel-filled capillaries for nucleic acid separations in

capillary electrophoresis. Trends Anal Chem 1992;11:280–7.2. Martin F, Vairelles D, Benedicte H. Automated ribosomal DNA fingerprinting

by capillary electrophoresis of PCR products. Anal Biochem 1993;214:182–9.

3. Baba Y, Tomisaki R, Sumita C, Tsuhako M. High-resolution separation of

PCR product and gene diagnosis by capillary electrophoresis. BiomedChromatogr 1994;8:291–3.

4. Rigat B, Hubert C, Alhenc-gelas F, Cambien F, Corvol P, Soubrier F. Aninsertion/deletion polymorphism in the angiotensin I-converting enzymegene accounting for half the variance of serum level. J Clin Invest 1990;86:1343–6.

5. Gardemann A, Weiß T, Schwartz O, Eberbach A, Katz N, Hehlern FW, et al.Gene polymorphism but not catalytic activity of angiotensin I-convertingenzyme is associated with coronary artery disease and myocardial infarctionin low-risk patients. Circulation 1995;92:2796–9.

6. Doria A, Warran JH, Krolewski AS. Genetic predisposition to diabeticnephropathy: evidence for a role of the angiotensin I-converting enzymegene. Diabetes 1994;43:690–5.

7. Dudley CRK, Keavney B, Stratton IM, Turner RC, Ratcliffe PJ. U.K. prospec-tive diabetes study XV: relationship of renin-angiotensin system genepolymorphism with microalbuminuria in NIDDM. Kidney Int 1995;48:1907–11.

8. Rigat B, Hubert C, Corvol P, Soubrier F. PCR detection of the insertion/deletion polymorphism of the human angiotensin converting enzyme gene(DCP1) (dipeptidyl carboxypeptidase 1). Nucleic Acid Res 1992;20:1433.

9. Shanmugam V, Sell KW, Sara BK. Mistyping ACE heterozygotes. PCRMethods Appl 1993;3:120–1.

10. Belgrader P, Devaney JM, Del Rio SA, Turner KA, Weaver KR, Marino MA.Automated polymerase chain reaction product sample preparation forcapillary electrophoresis analysis. J Chromatogr B Biomed Appl1996;68:109–14.

Fig. 1. Separation of the pGEM DNA calibratorby CE (A) and electropherogram of a heterozy-gous sample (B).A, capillary: total length 64.5 cm, effective length56 cm; running buffer: 89 mmol/L Tris, 332mmol/L boric acid, and 2 mmol/L EDTA, pH 7.4(1.5% polymer); injection conditions: electrokineticinjection at 210 kV for 30 s; analysis conditions:222 kV at 20 °C; detection: 258 nm. B, two peaksfor I allele (490 bp) and D allele (190 bp) of the ACEgene polymorphism. PCR products were directlyanalyzed by CE without further sample treatment.Separation conditions are as in A.


Very Long Chain Fatty Acid Analysis of Dried BloodSpots on Filter Paper to Screen for Adrenoleukodystro-phy, Kyoko Inoue, Yasuyuki Suzuki,* Shigehiro Yajima,Nobuyuki Shimozawa, Tadao Orii, and Naomi Kondo (Dept. ofPediatrics, Gifu Univ. School of Med., Tsukasa-machi 40,Gifu 500, Japan; * author for correspondence: fax 81-58-265-9011; e-mail [email protected])

Adrenoleukodystrophy (ALD) is an X-linked recessivedisorder characterized by progressive demyelination ofcerebral white matter and adrenal insufficiency. ALD isthe most common peroxisomal disease, afflicting 1 in20 000 male newborns. Patients with childhood ALDmanifest changes of character or behavior and a decline ofintelligence and visual or motor dysfunction as the firstsymptoms during school age. Most of them lapse into avegetative state and die within several years of onset [1].

Patients with all clinical subtypes, including childhoodALD, adolescent ALD, adult ALD (cerebral type), adreno-myeloneuropathy, and Addison disease without neuro-logical symptoms, have defects in the ALD gene [2].Defective activity of peroxisomal lignoceroyl-CoA ligaseis considered to lead to the accumulation of very longchain fatty acid (VLCFA) in various tissues and fluids asa secondary phenomenon [3, 4].

Dietary erucic acid (C22:1) therapy may help preventneurological deterioration in presymptomatic boys [5, 6].Patients in the early or presymptomatic stages may becandidates for bone marrow transplantation [5, 7]. Genetherapy may also become possible. Screening for ALD,especially in the presymptomatic stage, will be importantwhen these therapeutic methods are established, as in thecases of phenylketonuria or maple syrup urine disease.

ALD is usually diagnosed by plasma VLCFA analysis[1, 8], sometimes followed by measurement of b-oxidationactivity in cultured skin fibroblasts [9] and mutationanalysis of the ALD gene [2, 10–16]. However, screeningrequires a simpler and more economical method to dealwith a large number of subjects. Utilization of a driedblood spot on filter paper for VLCFA analysis was re-ported previously [17]. Here, we describe an easy methodof VLCFA analysis of dried blood spots on filter paperand discuss the possibility of screening for ALD.

Sep-Pak® silica cartridges for solid-phase extraction(VAC/1cc, part no. 23595) were purchased from Waters.n-Hexane and methyl-tert-butyl ether (MTBE) were pre-pared to make solution A (n-hexane/MTBE, 96:4, by vol)and solution B (n-hexane/MTBE, 200:3, by vol). Sep-Pakcartridges were preactivated by 1 mL of solution Afollowed by 3 mL of n-hexane.

A dried blood spot on Guthrie filter paper, equivalentto 100 mL of blood, was added to a tube containing 0.25mL of distilled water and 2.5 mL of chloroform/methanol(1:1, by vol) and soaked. The sample was then left for 1 hat room temperature after being shaken for a few minutes.After centrifugation, the eluent containing lipids wasplaced in a fresh tube and 1.25 mL of chloroform and 0.75mL of distilled water were added. These contents were

shaken for 3 min at room temperature and centrifuged for3 min. The lowest fraction, containing total lipids, wasaspirated and concentrated by evaporation. The residuewas heated with 1 mL of 50 mL/L concentrated HCl inmethanol at 100 °C for 1 h [17]. There was no significantdifference in the efficiency of methanolysis for 1 h or 2 hat 100 °C. Fatty acid methyl ester was extracted with 2 mLof n-hexane and concentrated by evaporation. The residuewas dissolved in 0.5 mL of solution B and passed throughthe cartridge followed by 2.5 mL of solution B. Extractionof VLCFA took ,4 h. After concentration, the extract wasdissolved in 100 mL (minimal quantity for the autosam-pler) of n-hexane containing 0.05 g/L butyl hydroxytolu-ene and analyzed on a Hewlett-Packard 5890A gas chro-matograph equipped with a splitless capillary injectionsystem, a flame ionization detector, a fused silica capillarycolumn (25 m 3 0.32 mm, Model HP-1) and an autosam-pler (Model HP7673). The injection volume was 2 mL, andthe temperatures at the injection and detection ports were250 °C and 285 °C, respectively. The column temperaturewas increased from 60 °C to 180 °C at 15 °C/min, to250 °C at 4 °C/min, to 280 °C at 15 °C/min, and main-tained at 280 °C for 5 min. Helium was the carrier gas.Peaks were identified by comparison of retention timeswith those of authentic standards and measured by area.

The ratios of lignoceric acid (C24:0) to behenic acid(C22:0) in dried blood spots on filter paper are shown inFig. 1, top. The mean 6 SD values were 1.6 6 0.2 in the 21controls between ages 1 week and 1 month, 1.3 6 0.1 inthe 19 control infants of age 1 year, 1.2 6 0.1 in the 27controls between ages 2 and 5 years, and 1.3 6 0.1 in the25 controls between ages 6 and 15 years. The ratio was.1.5 in all ALD patients (1.8 6 0.2) (10 with childhoodALD and 2 with adrenomyeloneuropathy) and in 6 of 7carrier mothers (1.6 6 0.2) (Fig. 1, bottom). The C24:0/C22:0ratios did not change at room temperature within 10 days.There were no significant sex- or age-related differencesin the ratios of C24:0/C22:0 in serum sphingomyelin [8]. Inour previous data, the ratio of C24:0/C22:0 in the controlgroup was 0.6 6 0.1, whereas that of ALD patients was1.4 6 0.2 [8]. About one-half of the control newbornsshowed C24:0/C22:0 ratios higher than the cutoff point.There may be an age-dependent difference in fatty acidcomposition of erythrocytes. It would be difficult toscreen ALD patients in the neonatal period with the use ofGuthrie paper. All children between ages 2 and 5 yearsshowed ratios ,1.5. ALD patients and many of thecarriers could be distinguished. Although our method isnot good for the screening of children under age 2 years,patients are in the presymptomatic stage during thatperiod. The ratio of hexacosanoic acid (C26:0) to C22:0 wasnot useful for diagnosis because C26:0 produced a verysmall peak, and the SD was very wide. Therefore, it isappropriate to screen children between ages 2 and 5 yearswith a cutoff value of 1.5. Blood sampling is feasiblebecause the health check for infants is performed at ages4, 10, 18, and 36 months in Japan.

Compared with the previous VLCFA analysis of serum,our method with dried blood spots on filter paper has


some merits. First, it requires only 100 mL of blood, andthe separation of serum is not necessary. Second, it is easyto use, and specimens can be sent by mail. Third, theextraction of fatty acids with the use of disposable pre-packed silica Sep-Pak columns [18] is easy and takes ashort time, so we can analyze samples from a largenumber of subjects. The gas chromatograph took ;35 minto analyze one sample and 10 min to be ready for the next.At that rate, 32 subjects/day, i.e., 11 680 subjects/year canbe analyzed theoretically with one gas chromatograph.Taking into account routine checks and maintenance,;100 gas chromatographs would be enough to analyze allmale infants in Japan, 600 000 every year, and the time tomeasure a sample would be shortened. Although theextraction of fatty acids by our method is more compli-cated than that of Nishio et al. [17], the analysis by gas

chromatography is easier to deal with and more econom-ical than the gas chromatography–mass spectrometrythey used. Therefore, despite some shortcomings such asthe age dependence and the false-positive rate at ayounger age, our method is useful for the screening ofALD in children. With our method, treatment in the earlyand presymptomatic stages is possible.

References1. Moser HW, Smith KD, Moser AB. X-linked adrenoleukodystrophy. In: Scriver

CR, Beaudet AL, Sly WS, Valle D, eds. The metabolic and molecular basesof inherited disease. New York: McGraw-Hill, 1995:2325–49.

2. Mosser J, Douar A-M, Sarde C-O, Kioschis P, Feil R, Moser H, et al. PutativeX-linked adrenoleukodystrophy gene shares unexpected homology with ABCtransporters. Nature 1993;361:726–30.

3. Hashmi M, Stanley W, Singh I. Lignoceroyl-CoASH ligase. Enzyme defect infatty acid b-oxidation system in X-linked childhood adrenoleukodystrophy.FEBS Lett 1986;196:247–50.

4. Wanders, RJA, van Roermund CWT, van Wijland MJA, Schutgens RBH, vanden Bosch H, Schram AW, et al. Direct demonstration that the deficientoxidation of very long chain fatty acids in X-linked adrenoleukodystrophy isdue to an impaired ability of peroxisomes to activate very long chain fattyacids. Biochem Biophys Res Commun 1988;153:618–24.

5. Moser HW, Moser AB, Smith KD, Bergin A, Borel J, Shankroff J, et al.Adrenoleukodystrophy: phenotypic variability and implications for therapy.J Inherit Metab Dis 1992;15:645–64.

6. Moser HW, Kok F, Neumann S, Borel J, Bergin A, Mostafa SD, et al.Adrenoleukodystrophy update: genetics and effect of Lorenzo’s oil therapy inasymptomatic patients. Int Pediatr 1994;9:196–204.

7. Aubourg P, Blanche S, Jambaque I, Rocchiccioli F, Kalifa G, Naud-SaudreauC, et al. Reversal of early neurologic and neuroradiologic manifestations ofX-linked adrenoleukodystrophy by bone marrow transplantation. N EnglJ Med 1990;322:1860–5.

8. Suzuki Y, Shimozawa N, Yajima S, Inoue K, Orii T, Kondo N. Incidence ofperoxisomal disorders in Japan. Jpn J Hum Genet 1996;41:167–75.

9. Inoue K, Suzuki Y, Yajima S, Shimozawa N, Tomatsu S, Orii T, et al. Carrieridentification of X-linked adrenoleukodystrophy by measurement of very longchain fatty acids and lignoceric acid oxidation. Clin Genet 1996;50:348–52.

10. Uchiyama A, Suzuki Y, Song X-Q, Fukao T, Imamura A, Tomatsu S, et al.Identification of a nonsense mutation in ALD protein cDNA from a patientwith adrenoleukodystrophy. Biochem Biophys Res Commun 1994;198:632–6.

11. Matsumoto T, Kondoh T, Masuzaki H, Harada N, Matsusaka T, Kinoshita E,et al. A point mutation at ATP-binding region of the ALD gene in a family withX-linked adrenoleukodystrophy. Jpn J Hum Genet 1994;39:345–51.

12. Fanen P, Guidoux S, Sarde C-O, Mandel J-L, Goossens M, Aubourg P.Identification of mutations in the putative ATP-binding domain of theadrenoleukodystrophy gene. J Clin Invest 1994;94:516–20.

13. Song X-Q, Fukao T, Suzuki Y, Imamura A, Uchiyama A, Shimozawa N, et al.Identification of a novel frameshift mutation in a Japanese adrenoleukodys-trophy patient. Hum Mol Genet 1995;6:1093–4.

14. Ligtenberg MJL, Kemp S, Sarde C-O, van Geel BM, Kleijer WJ, Barth PG, etal. Spectrum of mutations in the gene encoding the adrenoleukodystrophyprotein. Am J Hum Genet 1995;56:44–50.

15. Braun A, Ambach H, Kammerer S, Rolinski B, Stockler S, Rabl W, et al.Mutations in the gene for X-linked adrenoleukodystrophy in patients withdifferent clinical phenotypes. Am J Hum Genet 1995;56:854–61.

16. Feigenbaum V, Lombard-Platet G, Guidoux S, Sarde C-O, Mandel J-L,Aubourg P. Mutational and protein analysis of patients and heterozygouswomen with X-linked adrenoleukodystrophy. Am J Hum Genet 1996;58:1135–44.

17. Nishio H, Kodama S, Yokoyama S, Matsuo T, Mio T, Sumino K. A simplemethod to diagnose adrenoleukodystrophy using a dried blood spot on filterpaper. Clin Chim Acta 1986;159:77–82.

18. Hamilton JG, Comal K. Rapid separation of neutral lipids, free fatty acidsand polar lipids using prepacked silica Sep-Pak columns. Lipids1988;23:1146–9.

Fig. 1. The ratios of lignoceric acid (C24:0) to behenic acid (C22:0) in thedried blood spots on filter paper (top) and chromatograph of a samplefrom a control (bottom left) and from a patient with adrenoleukodys-trophy (bottom right).Top: ALD, patients with adrenoleukodystrophy; Carrier, heterozygous mothers;1W-1M, controls between ages 1 week and 1 month; 1Y, controls age 1 year;2Y-5Y, controls between ages 2 and 5 years; 6Y-15Y, controls between ages 6and 15 years. Mean 6 SD in each group is expressed by a bar. Bottom, thepeaks of C22:0 and C24:0 were identified by the retention time.


Serum Concentrations of Interleukin-6 Are IncreasedWhen Sampled Through an Indwelling Venous Cathe-ter, Adalsteinn Gudmundsson,* William B. Ershler, BrianGoodman, Stephanie J. Lent, Steven Barczi, and Molly Carnes(Dept. of Med., Univ. of Wisconsin-Madison, 600 High-land Ave., Madison, WI 53792, and Geriatric Res., Educa-tion and Clin. Center, William S. Middleton Memorial VAHosp., 2500 Overlook Terrace, Madison, WI 53705; * ad-dress correspondence to this author, at Middleton Veter-ans Hospital: fax 608-262-7648, e-mail [email protected]).

Circadian and ultradian variation is a prominent featureof most endogenous biodynamic processes and has beenextensively studied in neuroendocrine systems [1 ]. Inter-leukin-6 (IL-6) is a multifunctional cytokine that plays animportant role in many age-related diseases, includingpostmenopausal osteoporosis [2, 3 ]. Estrogen has beenshown to directly inhibit IL-6 production [4, 5 ]. Very littleinformation on the circadian rhythms and no informationon ultradian rhythms of IL-6 or other pro-inflammatorycytokines exists. Recent research has demonstrated astrong reciprocal interaction between IL-6 and the hypo-thalamic–pituitary–adrenal (HPA) axis [6, 7 ]. Specifically,recombinant IL-6 has been shown to stimulate the HPAaxis in humans [8 ]. The recent availability of highlysensitive assays for measuring IL-6 provides the opportu-nity to study the rhythmic pattern of this cytokine. Astudy in men, in which samples were collected every 3 hby direct venipuncture, showed a large circadian varia-tion in circulating IL-6 with a peak at 0100 and a nadir at1000 [9 ]. More frequent sampling is required to elucidateultradian fluctuations, which typically have periods of 60to 90 min [10 ].

For practical reasons, indwelling venous cathetersrather than direct venipunctures have been used forstudying frequently sampled time series. We designed astudy, using an indwelling venous catheter for continu-ous integrated 15-min sample collection, to establish thenormal variations of circulating IL-6 over 24 h in healthypostmenopausal women, and to assess the effect of estro-gen replacement therapy on these patterns. In studyingthe first subject, it became apparent that serum IL-6reached higher values than expected. Here, we demon-strate that an indwelling peripheral venous catheter leadsto local tissue production of IL-6. We also demonstratethat this local production of IL-6 does not affect endoge-nous cortisol concentrations.

The study protocol was approved by the University ofWisconsin Human Subjects Committee, and the subjectgave written informed consent. The study was conductedat the General Clinical Research Center at the Universityof Wisconsin, Madison.

The subject in this study was a healthy 59-year-oldwoman who was 6 years postmenopause. A thrombo-resistant blood withdrawal and tubing set (DakMed,Buffalo, NY) was inserted in a peripheral vein at 0800.Integrated blood samples were collected at 15-min inter-vals over 25 h by continuous withdrawal with a peristaltic

pump (DakMed). Clotted blood samples were centrifugedat 1520g for 10 min, then separated and frozen at 270 °Cuntil IL-6 concentrations were measured in duplicate by ahighly sensitive ELISA (Quantakine HS; R & D systems,Minneapolis, MN). This assay kit is validated for measur-ing IL-6 concentrations in the range 0.094–10 ng/L. In ourlaboratory the calibration curve loses linearity at 40 ng/L.The intraassay and interassay CVs for this assay are 3.8%and 7.1%, respectively. The study was repeated after 7weeks of oral estrogen replacement therapy with 0.625 mgof conjugated estrogen (Premarin) daily. Serum cortisolconcentrations were measured in duplicate by RIA (Coat-A-Count, Diagnostic Products Corp.) with an intraassayCV of 10.5%. The reported limit of detection of the assaywas 2 mg/L (5.5 nmol/L).

Inspection of the first 25-h time series (Fig. 1a) revealedconsiderable variation in IL-6 concentrations. IL-6 concen-trations remained constant at 2.8–4.7 ng/L for 3 h aftercatheter insertion, but this was followed by a steepincrease to 35 ng/L, almost 10 times the baseline value,after 20 h of sample collection. The IL-6 concentration at0900 on the second day was much higher than at 0900 onthe first day.

In a repeat study, after 7 weeks of estrogen replacementtherapy, a sample was drawn from the contralateral armby direct venipuncture 11 h after the first sample (Fig. 1b).The IL-6 in this sample measured 2.6 ng/L, while asimultaneously drawn sample from the indwelling cath-eter measured 24.2 ng/L. After 12 h of sampling, a newvenous catheter was placed in contralateral arm. Theconcentrations of IL-6 from this new site were initially lowbut again showed an abrupt increase at 3 h. After 22 h, anew catheter was again placed in an arm vein proximal tothe site of the first catheter. This time, the IL-6 concentra-tions remained high (8–27 ng/L).

The cortisol concentrations measured in the same se-rum samples as IL-6 demonstrated a normal diurnalvariation, with the peak values at 0530 after a nadir at0200. No increase in cortisol was seen at the time of therapid increase in IL-6. After 24 h the cortisol concentra-tions had returned to baseline value (Fig. 1c).

The use of indwelling venous catheters for studyingtime series of hormones is the accepted experimentalmodel. In most instances, hormones are secreted fromglandular tissue and the measurements represent theconcentrations in the systemic circulation. This case studyindicates that the onset of increasing IL-6 concentrations3 h after catheter insertion most probably represents localproduction of IL-6 and not the circulating concentra-tions—which indicates that this sampling method mayhave limitations when time series of immune mediators(e.g., IL-6) are being assessed. The local production of IL-6is supported by several lines of evidence. First, the valuesfor IL-6 were markedly lower when drawn by directvenipuncture or through a newly placed catheter from theopposite arm. Secondly, the blood concentrations of IL-6were higher than values reported in other studies ofpostmenopausal women that used the same assay system[11, 12 ]. Finally, the morning concentrations of IL-6 in


samples drawn upon initial catheter insertion were muchlower than those at the same time of day but drawn afterthe catheter had been in the vein for 24 h.

The increased concentrations of IL-6 found when thethird catheter was placed in a vein proximal tothe previous site of insertion supports the contention thatthe high IL-6 values derive from local tissue productionrather than from the catheter system itself. Importantly,this local tissue production of IL-6 did not appear toincrease the systemic concentrations of the cytokine for at

least 12 h and did not seem to stimulate the HPA axis, asevidenced by measured cortisol concentrations.

One can speculate that interruption of the vascularendothelial layer by a venipuncture mediates a cytokinecascade sequence that results in increased IL-6 concentra-tions. Endothelin, which is produced by endothelial cells,has recently been shown to be a potent stimulator of IL-6production [13 ]. Other cells at the site of injury, such asvascular smooth muscle cells [14 ], have also been shownto secrete IL-6. One previous study suggested that theincreased concentrations of IL-6 in bolus samples col-lected via indwelling venous catheters every 3 h for 9 hreflect local production of IL-6 rather than circulatingconcentrations [15 ]. The continuous sampling, more-fre-quent measurements, and use of a sensitive IL-6 assay inour study confirm this finding and provide a picture ofthe timing of this local effect, which in this subjectoccurred at ;3 h.

Because of the local tissue production of IL-6, our studydid not allow an assessment of the diurnal variation ofIL-6. Although there appears to be a faster increase in IL-6concentrations during estrogen replacement in this onesubject, its exact magnitude is likely to vary, both fromtime to time and from one vein to another. The mainreason for this is variation in the rate of blood flow, whichultimately will dilute the locally secreted IL-6 to variousextents.

Our results suggest that circulating concentrations ofIL-6 may be more reliably assessed by repeated directvenipunctures at a progressively more-distal venous siteeach time. However, discomfort and inconvenience asso-ciated with this method would obviate frequent samplingin human subjects. Researchers and clinicians measuringIL-6 for diagnostic or monitoring purposes should takeinto account the sample collection method used, particu-larly if high concentrations are encountered. This localproduction of IL-6 occurred as early as 3 h after insertionof a catheter at a fresh site and almost immediately froma catheter inserted downstream (proximal) to a previouscatheter insertion site. Whether production of other cyto-kines is similarly triggered after insertion of a catheter isa matter for future investigation. Several previous reportsof increased IL-6 in clinical settings may need reevalua-tion in view of our findings [16, 17 ].

This work was supported by a Merck/American Federa-tion for Aging Research Fellowship in Geriatric ClinicalPharmacology to A.G.; NIH grants RR03186, RR29-DK40759, R01 AG-1071, and K07 AG-0451; the Universityof Wisconsin Medical School Research Fund; and theDepartment of Veterans Affairs.

References1. Van Cauter E, Turek FW. Endocrine and other biological rhythms. In: Degroot

LJ, ed. Endocrinology, 3rd ed. Philadelphia: Saunders, 1994:2487–548.2. Ershler WB. Interleukin-6: a cytokine for gerontologists. J Am Geriatr Soc

1993;41:176–81.3. Manolagas SC, Jilka RL. Bone marrow, cytokines, and bone remodeling.

Emerging insights into the pathophysiology of osteoporosis. N Engl J Med1995;332:305–11.

Fig. 1. Twenty-five-hour profile for serum IL-6 and cortisol in a healthypostmenopausal woman: (a) Samples collected continuously, every 15min, through an indwelling venous catheter; (b) a repeat study, with asecond catheter placed in the contralateral arm and a third catheterplaced in an arm vein proximal to the first catheter (arrows indicateplacement of catheters and also a time when serum IL-6 was mea-sured by direct venipuncture from the opposite arm); (c) serum cortisolconcentrations, determined in the same samples as in a (0.1 mg/Lcortisol 5 275.9 nmol/L).


4. Jilka RL, Hangoc G, Girasole G, Passeri G, Williams DC, Abrams JS, et al.Increased osteoclast development after estrogen loss: mediation by inter-leukin-6. Science 1992;257:88–91.

5. Pottratz ST, Bellido T, Mocharia H, Crabb D, Manolagas SC. 17b-Estradiolinhibits expression of human interleukin-6 promoter–reporter constructs bya receptor-dependent mechanism. J Clin Invest 1994;93:944–50.

6. Kennedy RL, Jones TH. Cytokines in endocrinology: their roles in health anddisease. J Endocrinol 1991;129:167–78.

7. Hermus A, Sweep F. Cytokines and the hypothalamic–pituitary–adrenal axis.J Steroid Biochem Mol Biol 1990;37:867–71.

8. Mastorakos G, Chrousos GP, Weber JS. Recombinant interleukin-6 activatesthe hypothalamic–pituitary–adrenal axis in humans. J Clin Endocrinol Metab1993;77:1690–4.

9. Sothern RB, Johnson BR, Kanabrocki EL, Yager JG, Roodell MM, WeatherbeeJA, et al. Circadian characteristics of circulating interleukin-6 in men. JAllergy Clin Immunol 1995;95:1029–35.

10. Carnes M, Goodman B, Lent SJ. High-resolution spectral analysis of plasmaadrenocorticotropin reveals a multifactorial frequency structure. Endocrinol-ogy 1991;128:902–10.

11. Kania DM, Binkley N, Checovich M, Havighurst T, Schilling M, Ershler WB.Elevated plasma levels of interleukin-6 in postmenopausal women do notcorrelate with bone density. J Am Geriatr Soc 1995;43:236–9.

12. McKane WR, Khosla S, Peterson JM, Egan K, Riggs BL. Circulating levels ofcytokines that modulate bone resorption: effects of age and menopause inwomen. J Bone Miner Res 1994;9:1313–8.

13. Xin X, Cai Y, Matsumoto K, Agui T. Endothelin-induced interleukin-6 produc-tion by rat aortic endothelial cells. Endocrinology 1995;136:132–7.

14. Aarden L, Helle M, Boeije L, Pascual-Salcedo D, De Groot E. Differentialinduction of interleukin-6 production in monocytes, endothelial cells andsmooth muscle cells. Eur Cytokine Netw 1991;2:115–20.

15. Seiler W, Muller H, Hiemke C. Interleukin-6 in plasma collected with anindwelling cannula reflects local, not systemic, concentrations [Tech Brief].Clin Chem 1994;40:1778–9.

16. Bauer J, Hohagen F, Ebert T, Timmer J, Ganter U, Krieger S, et al.Interleukin-6 serum levels in healthy persons correspond to sleep–wakecycle. Clin Invest 1994;72:315.

17. Gudewill S, Pollmacher T, Vedder H, Schreiber W, Fassbender K, Holsboer F.Nocturnal plasma levels of cytokines in healthy men. Eur Arch Psychiatry ClinNeurosci 1992;242:53–6.

Interference of Anthracycline Derivatives with Mea-surement of Proteins with BCA, Abdul Kader and Hu Liu*(School of Pharmacy, Memorial Univ. of Newfoundland,St. John’s, NF A1B 3V6, Canada; *author for correspon-dence: fax 709-737-7044, e-mail [email protected]. mun.ca)

Common methods for the assay of proteins in biologicalfluids include Lowry [1], Coomassie Brilliant Blue G-250protein dye-binding assay (Bradford) [2], and the 2,29-bicinchoninic acid (BCA) assay [3]. Among them, theLowry method has been used widely for protein quanti-fication in biological samples. However, many substancescommonly used during protein purification interfere withthe Lowry assay. To overcome this, the BCA method wasdeveloped. This assay is similar to the Lowry method inthat both rely on the biuret reaction for generation of acolored complex between peptide bonds and cuprousions when protein is placed in an alkaline environmentcontaining Cu21. Unlike the Lowry method, the BCAmethod has been found to have exceptional tolerance tononionic detergents and simple buffer salts. However,sucrose and detergents, biogenic amines, chlorpromazine,penicillins, vitamin C, paracetamol, hydrogen peroxides,or any compound that can reduce Cu21 to Cu1 willproduce the characteristic purple color associated withthe binding of Cu1 with BCA and interfere with this

method [4, 5]. It is not surprising that anthracyclines,which are easily oxidized in alkaline media in the pres-ence of metal ions, would produce the necessary reduc-tion of Cu21 for the formation of a Cu1-BCA complex. Inthe Bradford method, the Coomassie blue dye binds toprimarily basic and aromatic amino acid residues, espe-cially arginine, and forms colored complexes. In thismethod, interferences may be caused by drug–protein and(or) drug–dye interactions. Because the mechanism of colorformation is different in the BCA and the Bradford methods,both methods were considered for interference studies.

The primary objective of this report was to determinewhether an anthracycline, e.g., doxorubicin (Dox), mayinterfere with the measurement of protein concentrationswith the BCA and the Bradford assays. We examined thesensitivity and specificity of the BCA and CoomassieBrilliant Blue for Dox in this study. Such findings mayhave significant implications in the assay of protein whenan anthracycline is present.

Dox showed a significant interference in the BCA assay,whether assayed in the presence or absence of a bovineserum albumin (BSA) protein calibrator (Fig. 1). Theinterference was found to be linear over the standardprotein range, with an immediate appearance of typicalassay color without turbidity (even after further dilu-tions). The reaction of Dox with the BCA reagent at roomtemperature (RT) for 2 h suggested that a wide range ofanthracyclines will react with the BCA reagent to producethe characteristic purple color if the reduction of Cu21 toCu1 by anthracyclines is dependent on the anthracyclinenucleus rather than on the side-chain substitution. The

Fig. 1. Dox interference in the BCA protein assay.Aliquots of 20 mL of sample containing 0.25, 0.50, 1.0, 2.5, 5.0 3 1021 g/L ofeach of the following were assayed: Dox (M), BSA protein calibrator (E), and BSAcalibrator containing Dox (BSA:Dox 5 1:1) (‚). The data points are the mean 6SD of quadruplicate determinations (where vertical lines are missing the SD wasless than the size of the symbol).


detection limits were 2 3 1024 g/L for Dox and 6 3 1023 g/Lfor BSA. Although protein shows a nonlinear color responsewith BCA with increasing concentration, at low concentra-tion the response is linear (Fig. 1). Dox did not interfere withthe Bradford method to quantify protein (not shown).

The development of color associated with the reactionof protein with the BCA reagent is slow at RT andenhanced by increasing the temperature and incubationtime [3]. The color developed after 2 h at RT is similar tothat developed over 30 min at 37 °C. Dox concentrationsof 0.01 to 0.10 g/L with different protein concentrationshave produced color intensities that are significantlyhigher than those due to the protein alone (data notshown). These results indicate that anthracyclines willsignificantly enhance color development produced byproteins at anthracycline concentrations that exceed 2 31024 g/L per gram of protein.

Measurement of the total protein concentration of theplasma used to monitor the nutritional status of patientsreceiving chemotherapy [6], and changes in serum andurine protein may reflect chemotherapy-associated neph-rotoxicity [7]. When calculated for a healthy individual,the plasma concentration of Dox or its metabolites isexpected to be .35 mg/L, as .75% of Dox is bound toplasma proteins [8, 9]. Therefore, plasma concentrationsof anthracyclines, including both the parent and themetabolites, are expected to be .0.2 mg/mg of protein. Atthis low drug-to-protein ratio Dox would not increase theprotein concentration significantly when measured by theBCA method.

Anthracycline derivatives have been widely used inexperimental therapeutics. For example, anthracyclinessuch as daunomycin have been loaded into lipoproteinsand their distribution in plasma proteins has been studied[10]. To quantify the drug associated with lipoprotein, it isessential to determine the protein concentrations of thelipoproteins. In all drug loading experiments, the anthra-cycline concentration was found to be .0.10 g/L [11] andas high as 0.29 g/L [12]. In these cases, the ratio of Dox toprotein was high enough (for example, 25 mg drug/mg ofprotein) to interfere in the color development of protein inthe BCA method by .50%. The protein concentrationmeasured by the BCA method would be 1 g/L with 25 mgof Dox/mg of protein when the actual protein concentra-tion was ,0.5 g/L. The magnitude of interference ishigher when the drug-to-protein ratio is higher (Fig. 1).Protein concentration is used as a bench mark to quantifycells and to express activity of an enzyme. In manycytotoxicity and enzymatic studies, the Dox concentrationwas found to be significantly high (7 mg to 0.29 mg ofDox/mg of protein) [12]. The results from our studiesindicate a significant increase in the color development ofthe BCA reagent with Dox in a concentration-dependentmanner in the BCA method but not in the Bradford assay.

This study was supported in part by a research grant toH.L. by the Banting Research Foundation.

References1. Peterson GI. Review of the folin phenol protein quantitation method of Lowry,

Rosebrough, Farr and Randall. Anal Biochem 1979;100:201–20.2. Bradford MM. A rapid and sensitive method for the quantitation of micro-

gram quantities of protein utilizing the principle of protein dye binding. J BiolChem 1976;72:248–54.

3. Smith PK, Krohn RI, Hermanson GT, Mallia AK, Gartner FH, Provenzano MD,et al. Measurement of protein using bicinchoninic acid. Anal Biochem1985;150:76–85.

4. Tracy LS, Jean DD. Interference of biogenic amines with the measurement ofproteins using bicinchoninic acid. Anal Biochem 1991;195:14–7.

5. Marshal T, Williams KM. Drug interference in the Bradford and 2,29-bicinchoninic acid protein assays. Anal Biochem 1991;198:352–4.

6. Mayhew SL, Thorn D. Enteral nutrition support: an overview. Am Pharm1995;35:47–62.

7. Skinner R, Pearson DJA, Coulthard MG, Skillen AW, Hodson AW, GoldfinchME, et al. Assessment of chemotherapy-associated nephrotoxicity in chil-dren with cancer. Cancer Chemother Pharmacol 1991;28:81–92.

8. Lentner C. Geigy scientific tables, 8th ed., Vol. 3. Basel, Switzerland:Ciba-Geigy, 1984:pp 65 and 329.

9. Chassany O, Urien S, Claudepierre P, Bastian G, Tillement JP. Binding ofanthracycline derivatives to human serum lipoproteins. Anticancer Res1994;14:2353–6.

10. Iwanik MJ, Shaw KV, Ledwith BJ, Yanovich S, Shaw JM. Preparation andinteraction of a low-density lipoprotein:daunomycin complex with P 388leukemic cells. Cancer Res 1984;44:1206–15.

11. Bose R, Verheij M, Friedman AH, Scotto K, Fuks Z, Kolesnick R. Ceramidesynthase mediates daunorubicin-induced apoptosis: an alternative mecha-nism for generating death signals. Cell 1995;82:405–14.

12. Mangiapane EM. The effect of adriamycin on glycerophosphate acyltrans-ferase and lipid metabolism in rat hepatocytes in monolayer culture.Biochem Pharmacol 1990;40:1577–82.

Stability of a Control Material Suitable for QuantitativeMeasurement of Urine Myoglobin, Kenneth R. Copeland,Bounthon Loun, and Frank A. Sedor* (Dept. of Pathol., Div.of Clin. Labs., Duke Univ. Med. Center, Durham, NC27710; *author for correspondence: fax 919-681-7786, e-mail [email protected])

Myoglobin is a 17-kDa single-chain hemoprotein found inskeletal and cardiac muscle. This heme protein facilitatesthe movement of oxygen into cells and provides for localstorage of oxygen. Myoglobin is found in the circulationas a result of muscle damage. Several conditions areassociated with the release of myoglobin into the circula-tion, including myocardial infarction, trauma, ischemia,surgery, exercise, rhabdomyolysis, and other myopathy-associated disease states [1–3].

The quantitative measurement of myoglobin in urine isclinically important in diagnosing myoglobinuria, whichcan subsequently induce acute renal failure, particularlyin posttrauma, surgery, and rhabdomyolysis patients.Recent studies have suggested that patients with a urinemyoglobin concentration .20 000 mg/L, particularly witha decreased myoglobin clearance rate (,4 mL/min), areat increased risk for decreased renal function [4, 5]. Al-though the mechanism for myoglobin-induced acute re-nal failure has not yet been elucidated, large amounts ofmyoglobin present in the tubules may precipitate, partic-ularly under acidic conditions, resulting in increasedintratubular pressure and, subsequently, the decreasedglomerular filtration rate [6] and (or) free-radical genera-tion from inorganic iron may cause renal damage [7]. Theidentification of the early clinical sequelae of myoglobin-


uria is important for enabling administration of prophy-lactic treatment for acute renal failure [8].

Quantitative methodologies, including automated im-munoassays, have been advocated and are becomingmore widespread for the measurement of urine myoglo-bin [4, 5, 9] because previous qualitative methods havebeen shown to be unreliable [9, 10]. As with all quantita-tive assays, it is imperative that appropriate controlspecimens be analyzed in parallel with all patient speci-mens. Ideally, such materials should be matrix-matchedand treated in a manner identical to that of all patientspecimens. The concentration of the material should be ator near those relevant to clinical decision concentrations.In addition, this material should be stable over an ex-tended period of time. To date, a suitable commercialquality-control (QC) material that meets these criteria isnot available. The purpose of this study was to evaluatethe suitability and long-term stability of an in-houseprepared urine pool for use with quantitative urine myo-globin assays.

A pool was prepared from urine obtained from severalpatients with increased urine myoglobin concentrations.The pool was diluted with buffer [100 mmol/L phosphatebuffer, pH 9.0, containing 70 g/L bovine serum albumin(BSA) and 0.1 g/L sodium azide] to a final myoglobinconcentration of ;20 000 mg/L. To avoid repetitive freez-ing/thawing, 100-mL aliquots of the urine myoglobin poolwere stored in polyethylene vials at 280 °C. The targetconcentration of the pool was determined to be 18 800mg/L as determined by analysis of five aliquots in dupli-cate over 5 days. A tentative range (mean 1 2SD) of16 200–20 900 mg/L was established with this data andwas subsequently used as a target value in assessment ofthe stability of the pool over the following 6 months. Themean and SD of all QC data were evaluated monthly andat completion of the study and compared with this initialrange.

Urine myoglobin concentration was determined bymodification of the serum Stratus II immunoassay (BaxterHealthcare Corp.) as previously described [4]. As part ofour routine workload, the in-house QC material wasanalyzed in parallel with at least one concentration of acommercial creatine kinase (CK)-MB/myoglobin immu-noassay control (Dade International). Both patient urineand urine QC were treated in an identical manner by

diluting 1:101 as follows: Twenty microliters of eitherurine myoglobin QC or patient sample was pipetted into2 mL of buffer (described above) and vortex-mixed for30 s before the analysis. This dilution was previouslydetermined to be most effective in detection of clinicallyrelevant urine myoglobin concentrations. Each vial of QCwas used once and the remainder discarded. The Dadeserum control material was analyzed without dilution.

The values obtained with the in-house urine myoglobincontrol (n 5 199) as well as 194 runs of the commercialserum material over a period of 179 days are shown inTable 1. The overall mean 1 SD of the urine myoglobinQC was 19 200 1 2020 mg/L (CV 5 10.5%). There was nodifference between the mean of the initial target determi-nations and the monthly means obtained over this timeperiod (data not shown). Similarly, no difference wasobserved with the serum control material over this pe-riod.

Over the 6-month study period a total of 16 outlierswere obtained with the in-house QC material and 15 withthe Dade material. In 13 of these 16 cases, the results werewithin allowable limits upon repeat analysis. In manycases, the initial cause for rejection may have been adilution error by one of 38 technologists who were re-sponsible for specimen analyses. Further corrective actionwas required in the remaining three cases. In two cases,the outlier was caused by a reagent problem. The final onerequired recalibration.

The CV of the in-house material was double that of thecommercial serum material (10.5% vs 5%). This increasedCV may result from the 101-fold dilution before analysis.The fact that the QC was diluted in a manner identical topatient specimens is beneficial in detection of possibledilution errors before analysis by the technologist.

The use of this material offers several advantages forthis application. The target concentration of the material isclinically relevant (;20 000 mg/L). Perhaps the greatestadvantage lies in the fact that the material is treatedidentically to patient specimens (as discussed above). Thismatrix-matched material is useful for detection of errorsin dilution and for detection of problems that may occurwith the diluent. Secondly, the stability of the materialprepared in this manner (at least 6 months) allows forpreparation of a large lot, thus obviating the need forweekly preparation of commercially available serum ma-

Table 1. Performance characteristics of in-house and commercial myoglobin QC materials.n Mean, mg/L SD, mg/L CV, % Target range,a mg/L Outliers

In-house control 199 19 200 2020 10.5 18 600 6 1200 16(16 200–20 900)

Dade Level I 68 47 2.1 5 44 6 4.5 5(35–53)

Dade Level II 59 233 8.9 4 227 6 18.5 7(190–264)

Dade Level III 67 463 18 4 432 6 40 3(351–512)

a Target mean 6 SD and 2 SD range for the commercial material was that stated by the manufacturer.


terials, as well as providing consistency for the detectionof long-range trends. Finally, the preparation of such anin-house pool has economic advantages.

In summary, we found that urine myoglobin preservedwith 70 g/L BSA phosphate buffer, pH 9.0, with 1 g/Lsodium azide is stable for at least 6 months when stored at280 °C. This allows preparation of a QC material suitablefor long-term use with newer quantitative urine myoglo-bin assays.

References1. Bywaters EGL, Bead D. Crush injuries with impairment of renal function. Br

Med J 1941;1:427–32.2. Grossman RA, Hamilton RW, Mores BM, Penn AS, Goldberg M. Nontraumatic

rhabdomyolysis and acute renal failure. N Engl J Med 1974;291:807–11.3. Rasmussen HH, Ibels LS. Acute renal failure. Multivariate analysis of causes

and risk factors. Am J Med 1982;73:211–8.4. Loun B, Astles R, Copeland KR, Sedor FA. Adaptation of a quantitative

immunoassay for urine myoglobin: predictor in detecting renal dysfunction.Am J Clin Pathol 1996;105:479–86.

5. Wu AHB, Laios I, Green S, Gornet TG, Wong SS, Parmley L, et al.Immunoassays for serum and urine myoglobin: myoglobin clearance as-sessed as a risk factor for acute renal failure. Clin Chem 1994;40:796–802.

6. Flamenbaum W, Gehr M, Gross M, Kaufman J, Hamburger R. Acute renalfailure associate with myoglobinuria and hemoglobinuria. In: Brenner BM,Lazarus JM, eds. Acute renal failure. Philadelphia: Saunders, 1983:269–82.

7. Shah SV. Oxidant mechanisms in glomerulonephritis. Semin Nephrol 1991;11:320–6.

8. Ron D, Taitelman U, Michaelson M, et al. Prevention of acute renal failure intraumatic rhabdomyolysis. Arch Intern Med 1984;144:277–80.

9. Hamilton RW, Hopkins MB, Shihabi ZK. Myoglobinuria, hemoglobinuria, andacute renal failure. Clin Chem 1989;35:1713–20.

10. Loun B, Copeland KR, Sedor FA. Ultrafiltration discrepancies in recovery ofmyoglobin from urine. Clin Chem 1996;42:965–9.

Rapid Identification of HLA DQA1*0501, DQB1*0201,and DRB1*04 Alleles in Celiac Disease by a PCR-BasedMethodology, Lucia Sacchetti,1 Claudia Sarrantonio,1 LucioPastore,1 Valeria Carlino,2 Giuseppe Calcagno,1 Anna Ferra-jolo,1 and Francesco Salvatore1* (1Dipto. di Biochim. e Bio-tecnol. Mediche, Facolta di Med. e Chirurg., and CEINGE-Biotecnol. Avanzate, Univ. “Federico II”, via S. Pansini 5,80131 Napoli, Italy; 2Ist. policattedra di Biochim. e Chim.Medica, Univ. di Bari, p.zza Giulio Cesare, 70124 Bari,Italy; *address for correspondence: Dipto. di Biochim. eBiotecnol. Mediche, Facolta di Med. e Chirurg., Univ.“Federico II”, via S. Pansini 5, 80131 Napoli, Italy; fax139-81-7463650)

Celiac disease (CD) is an autoimmune disorder associatedwith a small bowel lesion induced by toxic gliadin com-ponents [1, 2]. In this condition, an antigen peptide froma-gliadin, corresponding to the amino acid sequencebetween residues 31 and 49, initiates the cellular immuneresponse, which is mediated by gliadin-specific T lym-phocytes [3]. Antigen recognition by T lymphocytes in CDmainly occurs if the gliadin-derived peptides are carriedby the HLA class II molecules HLA-DQ2 or HLA-DR4[4, 5]. Thus, the genetic susceptibility towards CD derivesfrom inheriting certain HLA class II alleles encoding forthe above-mentioned specific molecules. Among the CD-HLA associations that have been described so far, the one

caused by the presence of alleles DQA1*0501/DQB1*0201(encoding for the DQ2 molecule) is present in most cases,whereas the DRB1*04 alleles (encoding for the DR4 mol-ecule) occurs almost invariably in the other cases [5, 6]. Infact, these two associations characterized .95% of theaffected celiac patients in European populations [7].

The aim of our study was to improve a PCR-basedmethodology for the rapid typing of the DQA1*0501,DQB1*0201, and DRB1*04 alleles, and thus provide anadditive simple tool for the diagnosis of CD.

DNA was extracted from 5 mL of fresh whole blood inEDTA by proteinase K treatment followed by phenol–chloroform extraction [8]. Isolated DNA samples werequantified by UV spectrophotometry at 260 nm absor-bance and diluted with distilled water to the concentra-tion required for PCR (100 ng/mL). We examined theDNA of three subjects whose HLA serological specifici-ties, previously studied, were: (a) DQ2/DQ5, (b) DQ5/DQ7, and (c) DR4/DR16. The first haplotype showed theDQ2 molecule that is strongly associated withDQA1*0501, DQB1*0201 genotype and so represented apositive DNA control for these alleles. The second haplo-type lacked the DQ2 and DR4 molecules and so we usedit as a negative control for the other alleles. The thirdhaplotype showed the DR4 molecule encoded byDRB1*04 alleles, thus representing a positive DNA controlfor these alleles.

Two reaction mixtures were used, one for the amplifi-cation of the DQA1*0501 and DQB1*0201 alleles, and theother for the identification of the DRB1*04 alleles, with theallele-specific primers described in Table 1. The sequencesof these primers differed in 2 or 3 nucleotides from thoseof the other known alleles at the same HLA loci. Toincrease primer specificity, we introduced a mismatch attheir 39 end [9]. We used exon 10 of the pyruvate kinasegene [10] in the first mix and exon 10 of the cystic fibrosisgene [11] in the second mix (see Table 1). These twoprimers did not match the HLA allelic sequences and sotheir PCR products (respectively 238 bp and 491 bplength) represented internal positive amplification con-trols. Both PCR reaction mixtures contained in a finalvolume of 50 mL: 100 ng of genomic DNA; 200 mmol/Leach of dATP, dCTP, dGTP, and dTTP (Pharmacia Bio-tech); PCR buffer [83 mmol/L (NH4)2SO4, 335 mmol/LTris-HCl pH 8.8, 33.5 mmol/L MgCl2, 50 mmol/L b-mercaptoethanol, 100 g/L bovine serum albumin, 34mmol/L EDTA]; 2.5 U of AmpliTaq DNA polymerase(Perkin-Elmer–Roche Molecular Systems); and the allele-specific primers and the internal positive control primers.

The first mix consisted of 0.6 mmol/L DQA1*0501primers, 0.4 mmol/L DQB1*0201 primers, and 0.8 mmol/Linternal positive control primers. The second mix con-tained 0.4 mmol/L DRB1*04 primers and 0.2 mmol/Linternal positive control primers. A DNA thermal cycler(Gene Amp PCR System 9600, Perkin-Elmer) was used forPCR amplification. Double-stranded DNA was denaturedinitially by heating to 95 °C for 5 min, followed by a30-cycle profile of 30 s at 94 °C for denaturation, 10 s ofannealing at 64 °C, and 20 s of extension at 72 °C. The last


PCR cycle was followed by an additional 10 min at 72 °Cto complete the final extension.

PCR products were separated by agarose gel electro-phoresis. Thirty microlitres of the PCR reaction mixtureswere added to 5 mL of loading buffer (300 mL/L glycerolstained with bromphenol blue and xylene cyanol) andloaded in a 3% agarose gel prestained with ethidiumbromide (0.5 mg/mL gel) (Sigma Chemical Co.). Electro-phoresis was performed at 100 V for 40 min in 13 TBEbuffer (89 mmol/L Tris base, 89 mmol/L boric acid, 0.2mmol/L EDTA). PCR products on gel were visualized byUV illumination as bands of 144 bp (DQA1*0501), 110 bp(DQB1*0201), and 177 bp (DRB1*04) and then photo-graphed. Fig. 1 shows an example of HLA typing ob-tained with our procedure on DNA samples from thethree previously HLA-characterized (serologically) indi-viduals. All three HLA-DQA1*0501, DQB1*0201, andDRB1*04 alleles appeared as well-separated bands.

In the diagnostic iter of gastrointestinal malfunction,the finding of the HLA DQA1*0501, DQB1*0201, andDRB1*04 alleles increases the possibility of having CD [7].Various methods have been described for DNA typing at

the HLA level, mainly for histocompatibility testing.These procedures involve a combination of PCR amplifi-cation with restriction fragment length polymorphism[12], sequencing, oligonucleotide probing, or sequence-specific primers [13–16], all of which are time consumingand expensive when used for CD diagnosis. We havedevised a rapid, accurate, and cost-effective PCR-basedmethod to type, with a high level of resolution, theHLA-DQA1*0501, DQB1*0201, and DRB1*04 alleles. Theadvantage of our method is that it does not requireradioactive material and that, after the same PCR pro-gram, the three most frequently CD-associated alleles arevisualized on the same gel, within ;2 h from DNAsample collection. We have used this method to examine.100 celiac patients and ;30 patients affected by othergastrointestinal syndromes clinically similar to CD. Theresults are similar to those obtained in other populations,with other methods (manuscript in preparation). Wesuggest that this procedure could represent an adjunctivetool in the diagnostic approach to CD; it is suitable forfamily screening, epidemiological studies, and for diag-nosis particularly when histological and immunologicalpatterns are ambiguous.

This study was supported by grants from the MURST,Rome, Italy; the Progetto Finalizzato Biotecnologie, CNR(Rome); and from Ricerca Sanitaria Finalizzata (RegioneCampania, Italy).

References1. Trier JS. Celiac sprue. In: Sleinsenger MH, Fordtran JS, eds. Gastrointestinal

disease pathophysiology, diagnosis, management, 5th ed. Philadelphia: WBSaunders Co., 1993:1078–96.

2. Ferguson A, Arranz E, O’Mahony S. Clinical and pathological spectrum ofcoeliac disease—active, silent, latent, potential [Review]. Gut 1993;34:150–1.

3. Gjertsen HA, Lundin KEA, Sollid LM, Eriksen JA, Thorsby E. T cells recognizea peptide derived from a-gliadin presented by the celiac disease-associatedHLA-DQ (a1*0501, b1*0201) heterodimer. Hum Immunol 1994;39:243–52.

4. Lundin KEA, Scott H, Hansen T, Paulsen G, Halstensen TS, Fausa O, et al.Gliadin-specific, HLA-DQ (a1*0501, b1*0201) restricted T cells isolatedfrom small intestinal mucosa of celiac disease patients. J Exp Med1993;178:187–96.

Table 1. Primers for identification of HLA-DQA1*0501, DQB1*0201, and DRB1*04 alleles.Primer Sequence Length of PCR products

DQA1*050159-AGCAGTTCTACGTGGACCTGGGG-39

144 bp59-GGTAGAGTTGGAGCGTTTAATCAGA-39

DQB1*020159-CGCGTGCGTCTTGTGAGCAGAAG-39

110 bp59-GGCGGCAGGCAGCCCCAGCA-39

DRB1*0459-GGTTAAACATGAGTGTCATTTCTTAAAC-39

177 bp59-GTTGTGTCTGCAGTAGGTGTC-39

AC-159-CGAAACCCAGTCCCCACCCG-39

238 bp59-CCTTCCATACCCCAGTGCCC-39

AC-259-GCAGAGTACCTGAAACAGGA-39

491 bp59-CATTCACAGTAGCTTACCCA-39

AC-1 and AC-2, amplification controls: exon 10 of pyruvate kinase gene and exon 10 of cystic fibrosis gene, respectively.Primers used as internal positive amplification controls AC-1 and AC-2 are also reported. The first oligo is located upstream and the second one downstream along

the genes.

Fig. 1. Examples of HLA DQA1*0501, DQB1*0201, and DRB1*04alleles typing by PCR amplification of DNA from three subjects thatshow the presence of both DQA1*0501 and DQB1*0201 (lane 1),only DQA1*0501 (lane 2), and only DRB1*04 (lane 3) alleles,respectively.SM, size marker (72–1353 bp); AC-1 and AC-2, internal positive amplificationcontrols, 234 and 491 bp long, respectively.


5. Fernandez-Aequero M, Figueredo MA, Maluenda C, de la Concha EG.HLA-linked genes acting as additive susceptibility factors in celiac disease.Hum Immunol 1995;42:295–300.

6. Tighe R, Ciclitira PJ. Molecular biology of coeliac disease [Review]. Arch DisChild 1995;73:189–91.

7. Sollid LM, Thorsby E. HLA susceptibility genes in celiac disease: geneticmapping and role in pathogenesis [Review]. Gastroenterology 1993;105:910–22.

8. Sambrook J, Fritsch EF, Maniatis T, eds. Molecular cloning. A laboratorymanual, 2nd ed. Cold Spring Harbor, NY: Cold Spring Harbor LaboratoryPress, 1989:9.16–9.

9. Newton CR, Graham A, Heptinstall LE, Powell SJ, Summers C, Kalsheker N,et al. Analysis of any point mutation in DNA. The amplification refractorymutation system (ARMS). Nucleic Acids Res 1989;7:2503–17.

10. Baronciani L, Beutler E. Analysis of pyruvate kinase-deficiency mutationsthat produce nonspherocytic hemolytic anemia. Proc Natl Acad Sci U S A1993;90:4324–7.

11. Zielenski J, Rozmahel R, Bozon D, Kerem B, Grzelczak Z, Riordan JR, et al.Genomic DNA sequence of cystic fibrosis transmembrane conductanceregulator (CFTR) gene. Genomics 1991;10:214–28.

12. Nomura N, Ota M, Tsuji K, Inoko H. HLA-DQB1 genotyping by a modifiedPCR-RFLP method combined with group-specific primers. Tissue Antigens1991;38:53–9.

13. Bugawan TL, Erlich HA. Rapid typing of HLA-DQB1 DNA polymorphism usingnon radioactive oligonucleotide probes and amplified DNA. Immunogenetics1991;33:163–70.

14. Scharf SJ, Griffith RL, Erlich HA. Rapid typing of DNA sequence polymor-phism at the HLA-DRB1 locus using the polymerase chain reaction andnonradioactive oligonucleotide probes. Hum Immunol 1991;30:190–201.

15. Buyse I, Decorte R, Baens M, Cuppens H, Semana G, Emonds MP, et al.Rapid DNA typing of class II HLA antigens using the polymerase chainreaction and reverse dot blot hybridization. Tissue Antigens 1993;41:1–14.

16. Olerup O, Aldener A, Fogdell A. HLA-DQB1 and -DQA1 typing by PCRamplification with sequence-specific primers (PCR-SSP) in 2 h. TissueAntigens 1993;41:119–34.

Model for Establishing Biological Variation in Non-healthy Situations: Renal Posttransplantation Data,Carmen Biosca,1* Carmen Ricos,2 Carlos Vıctor Jimenez,3

Ricardo Lauzurica,1 and Roman Galimany1 [1Hosp. “Ger-mans Trias i Pujol”, Badalona (Barcelona), Spain; 2Hosp.Gen. “Vall d’ Hebron”, Barcelona, Spain; 3Centro deAsistencia Primaria “Dr. Robert”, Badalona, Spain; *ad-dress for correspondence: Dept. of Biochem., Hosp. “Ger-mans Trias i Pujol”, Ctra. de Canyet s/n, 08916 Badalona(Barcelona), Spain; fax 343 395 42 06]

Renal transplantation has come to be accepted as standardtreatment for patients with terminal kidney failure. Be-cause the physiological state of the renal transplant recip-ient is unstable, he or she is monitored with a well-defined protocol that is strictly followed by clinicians.Renal dysfunction is a common complication due tovarious problems, especially drug toxicity, and rejection isan ever-present danger [1]. To detect episodes of rejection,clinicians use empirical criteria derived from experiencethat is mainly based on changes in creatinine concentra-tions. Posttransplantation monitoring, which includes fre-quent analysis of a number of constituents, generateslarge amounts of data. With this information, findingobjective, early indicators that predict trends towardcomplications before the patient’s condition is seriouslycompromised would be beneficial, so that preventiveactions could be taken.

Data from biological variation (BV), the normal fluctu-ation around the homeostatic set point, has been used to

evaluate the significance of changes in serial results [2].Therefore, it can provide clinicians with an indication offuture patient status: A change between two consecutiveobservations higher than the established variation aroundthe homeostatic set point could signal the beginning of acomplication. However, the components of BV, sensitiveand specific enough to characterize a “certain state ofhealth” from the start of crisis, can be investigated onlywhen a stable situation that denotes equilibrium has beendemonstrated in the specific pathology.

The aims of this work were to delineate the stableperiod after renal transplantation for six serum analytesexpected to reflect instability/rejection, and to calculatewithin- and between-subject BV, indices of individuality(II), and critical differences (CDs) between serial resultsand compare outcome with published data in healthysubjects to determine whether BV data can predict crisesin this population.

We studied 19 patients (12 men and 7 women), 19 to 64years old, with chronic renal insufficiency who had re-ceived an orthotopic kidney graft. Permission for enroll-ment in the study was obtained from all patients, asrequired by the Helsinki II protocol.

Serum specimens were collected according to the usualhospital follow-up protocol designed by the nephrologistsfor these patients, as summarized: (a) First week post-transplantation, daily; (b) from the second to fourth week,twice a week; (c) from the first to the third month, weekly;(d) from the third to the sixth month, every 15 days; (e)from the sixth to the twelfth month, once a month; (f) fromthe first year on, every 2 or 3 months, indefinitely.

Stability in posttransplantation patients is routinelyverified through a combination of clinical, analytical, andimaging parameters: clinical normality as indicated bysymptomatology, physiological constants, diuresis,weight, physical examination, etc.; analytical profile withparticular attention directed to the stability of creatinineresults (expected to differ ,25% between two consecutivesamplings); and doppler echography. The data from our19 patients was studied for 2 years posttransplantation,and during this time there was no evidence of crisis orrejection according to the nephrologists’ protocol.

The conditions of specimen collection were standard-ized to minimize the effect of collection. Specimens werecollected into evacuated blood-collection tubes withoutanticoagulant. The specimens were allowed to clot atroom temperature and were centrifuged at 3000g for 15min. We separated the serum, and quantities were ana-lyzed. The study was conducted in real time throughout.

Six serum biochemistry analytes were determined: cre-atinine, urea, urate, sodium, potassium, and chloride.

Establishment of the homeostatic point: The homeo-static point for each analyte was derived from the periodof maximum stability.

1) Determination of the beginning of the stable period:The analytic results from each analyte in each patientwere represented graphically from the beginning of theposttransplantation period. Visual inspection of thegraphs showed high results in the first determinations


that decreased to a point after which results remainedconstant over time (from negative slope to horizontallines) in some of the analytes. This point was consideredto be the beginning of the stable period and was called“point zero.” Point zero and the stable pattern were seenvery clearly in the creatinine analyses and were confirmedin the urea graph; however, in the remaining analytes aclear inflection point was not observed. Therefore in eachpatient, point zero for all the quantities studied wasderived from the creatinine results.

2) Determination of the end of the stable period: Resultswere normalized according to the start value [3]: the ratioof each result from a patient with respect to his value atpoint zero was calculated.

The CV of the ratios of the 19 patients (normalized CV)was calculated and the difference between the CV for eachsampling day and the CV for point one (the CV at pointzero is 0) were depicted on a graph, against the analyticalvariation of the method for each constituent. When thedifference between the normalized CV was found to behigher than the analytical CV, stability was considered tohave ended.

BV calculations: Analytical imprecision (s2a) was calcu-

lated through control materials, averaging the routinedata for 12 months, and using the control concentrationclosest to the mean of the concentration values found inthe 19 patients studied.

Before performing calculations with the patients’ re-sults, the Cochran test [4] was applied to exclude outlyingvalues from the individual subjects, and the Reed test [5]to eliminate mean outlying values.

The ANOVA test [6] was used to estimate within-subject (s2

i) plus analytical variation (s2i1a), expressed as

the weighted mean of variances from the 19 patients.Within-subject (intraindividual) BV (s2

i) was calculated bya subtraction step with the two previous variables (s2

i1a 2s2

a). Between-subject (group) BV (s2g) was obtained by

subtracting the within-subject plus analytical variationfrom the total variation (s2

t) found by using all data fromall patients:

s2g 5 s2

t 2 s2i 1 a

II were calculated with the formula:

II 5 s2i 1 a/s2

g

CDs between consecutive results were calculated afterconfirming with the Kolmogorov–Smirnov test [6] thatthe (s2

i1a) data were normally distributed; subsequently,the one-tail formula (a 5 0.05) was used to detect onlysignificant increases:

CD 5 1.65 3 2(s2i 1 a)1⁄2 5 2.33si 1 a (P , 0.05)

To derive BV data, the stable period within the pathol-ogy has to first be defined: The more precisely the stableperiod is delineated, the more robust will be the indicatorsproduced. The beginning of posttransplantation stabilitywas determined by examining the slopes of the quantities

studied. We considered that a clear inflection (and subse-quent maintenance) of the slope in more than one quan-tity simultaneously would signal the beginning of stabil-ity. For all 19 patients, creatinine and urea showed thispattern and were considered valid for our purposes.

The beginning of the stable period was found betweenthe first and second month posttransplantation, althoughthe exact moment when it occurred was not necessarilythe same in all patients. The fact that at this time thenephrologists’ protocol reduces the required analyses to afrequency of one per week indicates that the empiricallybased criteria also perceive stability at this time.

To minimize interindividual variation and to facilitatethe detection of relative changes among patients, allresults were normalized [3, 7, 8]. Fig. 1 shows the differ-ence between the normalized CV at each sampling dayand the CV at point one, in relation to the analytical CV(parallel line to the x-axis) for creatinine. The period ofstability was considered to conclude when the CV of thedifference was higher than the analytical CV. This oc-curred within an interval of eight determinations (thatstarts between 1 and 2 months after surgery), whenfollow-up protocol analyses are performed once a week,and is maintained for an average of 3 months, dependingon the patient.

Table 1 shows the components of analytical and bio-logical variation found in this study, expressed in terms ofCVs. Data from healthy subjects obtained by averagingresults from previous works [9, 10] are also exhibited.

When studying the components of BV, the analyticalcomponent (CVa) is lower than half the within-subjectcomponent (CVi) for all constituents studied (except chlo-ride and sodium), demonstrating that the use of “realtime” analytical data is appropiate for studying BV.Browning et al. [11] recommended that when studyingBV, the analytical component should be ,20% of the totalvariance found. The analytical difficulties for chloride and

Fig. 1. End of the stable period determination.Each point represents the difference between normalized CV at each samplingday and CV at point one, in relation to the analytical CV (CVa 5 5.4% forcreatinine).


sodium are similar to those found by other authors insimilar studies [12, 13].

BV data from healthy subjects have been compiled intwo well-known articles [9, 10]. Few works have dealtwith BV in pathological status. Fraser and colleaguesreported that the within-subject and between-subject vari-ation for certain analytes in patients with renal dysfunc-tion and cardiac infarction are the same as in healthysubjects [13, 14]. Holzel [15, 16], however, shows discrep-ancies in specific pathological situations. Our data showthat within-subject variation was higher in the kidneygraft recipients than in the healthy population, being mostevident in the creatinine, potassium, urate, and urearesults. We found no differences in the between-subjectvariation, except for potassium.

We studied BV to determine if analytical results fromthe routine monitoring protocol could be used as predic-tors of functional alteration in renal transplant recipients.To know which constituents are suitable as early indica-tors of negative evolution in pathological situations, the IIare determined. Creatinine, urate, and urea, with II of;0.6, are suitable for monitoring [2].

Harris and colleagues proposed a formula derivedfrom within-subject variation to interpret CDs in serialresults [17, 18]. We found that CDs were ;28% in creati-nine, urate, and urea. This figure is very close to the 25%criteria used by the nephrologist in their protocol. More-over, none of the 19 patients surpassed this differenceover the study period, indicating that no significantchanges occurred.

The other constituents provided no information for thepurpose of predicting functional alteration. Thus, aninterpretation of serial results from the combination ofcreatinine, urate, and urea analyses could be a method forearly detection of possible crises in posttransplantationpatients.

In conclusion, this work describes a model for studyingBV in a nonhealthy state by using available laboratorydata. It shows a method for demonstrating stability withinthe pathology and for deriving the components of BV.

We thank Per Hyltoft Petersen for his invaluable help inthe normalization section of the study, and in improvingthe orientation of this work.

References1. Yatscoff RW. Laboratory support for transplantation. Clin Chem 1994;40:

2166–73.2. Fraser CG, Harris EK. Generation and application of data on biological

variation in clinical chemistry. Crit Rev Clin Lab Sci 1989;27:409–37.3. Hyltoft Petersen P, Feldt-Rasmussen U, Horder M, et al. Variability of plasma

proteins according to molecular size. Long-term and short-term intra-individ-ual variation. Scand J Clin Lab Invest 1981;41:143–50.

4. Cochran WS. The distribution of the largest of a set of estimated variancesas a fraction of their total. Ann Eugen 1941;11:47–51.

5. Reed AH, Henry RJ, Mason WB. Influence of statistical method used on theresulting estimate of normal range. Clin Chem 1971;17:275–9.

6. Domenech Massons JM. Bioestadıstica, metodos estadısticos para inves-tigadores. Barcelona: Ed Herder, 1977:322–4.

7. Kancir CB, Petersen PH, Wandrup J. The effects of plasma volume variationson the calcium concentration during epidural anaesthesia. Acta Anaesthe-siol Scand 1987;31:338–42.

8. Hyltoft Petersen P, Horder M. Influence of analytical quality on test results.Scand J Clin Lab Invest Suppl 1992;208:65–87.

9. Fraser CG. The application of theoretical goals based on biological variationdata in proficiency testing. Arch Pathol Lab Med 1988;112:404–15.

10. Fraser CG. Biological variation in clinical chemistry. An update: collateddata; 1988–1991. Arch Pathol Lab Med 1992;116:916–23.

11. Browning MCK, Ford RP, Callaghan SJ, Fraser CG. Intra- and interindividualbiological variation of five analytes used in assessing thyroid function:implications for necessary standards of performance and the interpretationof results.Clin Chem 1986;32:962–6.

12. Fraser CG, Hearne CR. Components of variance of some plasma constitu-ents in patients with myocardial infarction. Ann Clin Biochem 1982;19:431–4.

13. Fraser CG, Williams P. Short-term biological variation of plasma analytes inrenal disease. Clin Chem 1983;29:508–10.

14. Fraser CG, Hearne CR. Components of variance of some plasma constitu-ents in patients with myocardial infarction. Ann Clin Biochem 1982;19:431–4.

15. Holzel WGE. Intraindividual variation of some analytes in serum of patientswith cronic renal failure. Clin Chem 1987;33:670–3.

16. Holzel WEG. Influence of hypertension and antihypertensive drugs on thebiological intra-individual variation of electrolytes and lipids in serum. ClinChem 1988;34:1485–8.

17. Harris EK, Yasaka T. On the calculation of a “reference change” forcomparing two consecutive measurements. Clin Chem 1983;29:25–30.

18. Boyd JC, Harris EK. Utility of reference changes values for the monitoring ofinpatient laboratory data. In: Zinder O, ed. Optimal use of the clinicallaboratory. 5th Int. Meet. Clin. Lab. Org. Management, Haifa, Israel, 1985.Basel: Karger, 1986:111–22.

Table 1. Biological variation components in posttransplantation patients vs healthy subjects.

Analyte CVa, %

CVi 1 a, % CVi, % CVg, %

Transplanted Healthy Transplanted Healthy Transplanted Healthy

Creatinine 5.4 12.9 7.2 11.8 4.8 14.7 15.1Urea 5.1 14.2 11.4 13.9 11.1 21.1 20.7Urate 3.8 12.3 8.3 11.7 7.5 21.4 25.0

Sodium 1.7 2.0 1.9 1.1 0.7 1.1 0.7Potassium 2.4 7.1 4.1 6.7 3.5 4.9 2.6Chloride 2.6 3.0 1.6 1.4 1.2 1.6 1.4


Concurrent Determination of Second-Generation Anti-depressants in Plasma by Using Gas Chromatographywith Nitrogen–Phosphorus Detection, Nicole H. Jourdil,*Philippe D. Fontanille, and Germain M. Bessard (Lab. dePharmacol. et Analyses Toxicol., Centre Hospitalier Uni-versitaire de Grenoble, BP 217, F-38043 Grenoble Cedex 9,France; *author for correspondence: fax 00-33-4-76–76-56-55, e-mail Germain.Bessard @ ujf-grenoble.fr)

Antidepressant (ADP) drugs are among the most com-monly involved compounds in voluntary intoxicationsbecause of their large prescription to people with majordepression and their benefical effect in some associatedpsychiatric disorders [1]. The second-generation com-pounds have clear advantages in comparison with tricy-clic ADPs, but numerous studies report the frequentoccurrence of interaction when new ADPs are associatedwith tricyclics, antipsychotics, or anxiolytics [2, 3]. Meth-ods of analysis in biological fluids include gas chroma-tography (GC), GC–mass spectrometry (MS), HPLC, andimmunoassays [4]. Chromatographic methods usuallyappear to be complex and unable to provide chemicalidentification under screening conditions. We present inthis study a simple and rapid GC procedure with anitrogen–phosphorus detector (NPD) for the simulta-neous determination of seven frequently requested newADPs in plasma. As far as we know, no previous reporthas been published on the concurrent determination ofADPs by using a common extraction procedure and GCanalysis. This assay was developed for the analysis ofamoxapine, dothiepin, fluoxetine, fluvoxamine, medifox-amine, mianserin, and viloxazine.

The reagents, apparatus (Varian-Star 3400 CX coupledto a Fisons-Chromcard for Windows Software), and pro-cedure for the separation and measurement of the ana-lytes have been previously described, according to themethod for fluoxetine [5]. Briefly, the patient’s alkalinizedspecimen (1 mL of a 20% ammonia solution mixed withan equal volume of serum), with added protriptyline asinternal standard (IS) (15 mL of a 10 mg/L workingsolution in ethanol), is extracted in a one-step procedurewith hexane:dichloromethane:isoamyl alcohol (57:42:1 byvol). The residue is dissolved in 50 mL of ethanol anddirectly chromatographed on a 25 m 3 0.32 mm (i.d.) OV1capillary column (25-mm-thick film) with helium as thecarrier gas. The method gives a linear response to at least2000 mg/L. The limits of detection ranged between 0.2and 2.0 mg/L and the limit of quantification was estab-lished as between 0.5 and 5.0 mg/L (n 5 5). Intraassay andbetween-day CVs were #10% (n 5 10). The mean overallabsolute recoveries (n 5 5) ranged from 65% (medifox-amine) to 97% (fluoxetine). Plasma from drug-free pa-tients did not exhibit significant peaks in the 30-minretention range. Calibrators and controls were preparedwith drug-free plasma of healthy volunteers.

We studied 57 compounds that can potentially interferein plasma. A complete list, with retention times (tRs), isgiven in Table 1. ADP metabolites may also be detected inplasma patients. So whenever possible, we tested the

Table 1. Retention times of coextracted drugs in ADP GC-NPD assay.

Drug Retention time, min Concn. tested, mg/L

Caffeine 9.43 5000

Medifoxaminea 10.42 50

Norfluoxetine 10.77 250

Viloxazinea 11.05 500

Fluoxetinea 11.61 250

Meprobamate 12.07 20 000

Fluvoxaminea 12.65 50

Mianserina 17.88 50

Cocaine 17.94b 500

Amitriptyline 17.96b 100

Dextropropoxyphene 17.98 500

Nortriptyline 18.05 100

Trimipramine 18.16 100

Imipramine 18.16 100

Medazepam 18.18 20

Desipramine 18.26b 100

Desmethyltrimipramine 18.33c 100

Protriptylinea 18.33 100

Prometazine 18.52 100

Oxazepam 18.83 1000

Maprotiline 18.98 100

Codeine 19.20b 250

Dothiepina 19.28 50

Lorazepam 19.37 20

Northiaden 19.37 20

Clomipramine 19.50 100

Codethyline 19.50 250

Morphine 19.52 100

Diazepam 19.54 100

Tetrazepam 19.54 100

Desmethylclomipramine 19.61 100

Dothiepin sulfoxide 19.69 20

Methadone 19.70 150

Desmethyldiazepam 19.87 500

Amineptine 19.95 100

Chlordiazepoxide 19.98 20

Clotiazepam 20.02 20

Levomepromazine 20.13 100

Clobazam 20.15 100

Benzoylecgonine 20.21 500

Paroxetine 20.32 50

Midazolam 20.39 10

Flunitrazepam 20.43 10

Bromazepam 20.43 100

Amoxapinea 20.60 50

Prazepam 20.71 10

Acepromazine 20.74 100

Temazepam 20.78 20

Lormetazepam 21.02 10

Nitrazepam 21.18 50

Zolpidem 21.39 100

Clozapine 21.86 400

Alprazolam 22.21 10

Haloperidol 22.27 10

Estazolam 22.49 20

Zopiclone 23.08 50

Amphetamine 23.94 500

Pholcodine 23.96 250

Pipotiazine 24.20 100

Tetrahydrocannabinol 24.37 10

Amisulpiride 25.16 50

Buprenorphine 26.04 10

Triazolam 26.59 10

Lysergic acid diethylamide 27.08 10

a Assay was developed for concurrent analysis of these drugs.b Interference occurs above high concentrations.c Interference occurs at any concentration.


metabolites of the various analytes. No interference camefrom endogenous plasma constituents eluting during thecomplete analysis set. The potential for interference wasjudged by the presence of an interfering peak for coex-tracted drugs, the retention time of the peak, and peakheight in relation to the drug’s plasma concentration.Many psychoactive drugs were tested on the OV1 capil-lary column because they could be associated with ADPs.Of all the drugs tested, only trimipramine coadministra-tion could modify the results, since its metabolite, thedesmethyltrimipramine, has the same tR as the IS (18.33min). If a peak appeared at tR 18.16 min, it is highlyprobable that a tricyclic ADP (trimipramine or imipra-mine) was administered. No quantification is available inthe case of trimipramine coadministration. Therefore onlyGC-MS appeared to be able to provide unequivocalchemical identification under screening conditions fortoxicological samples. Fours drugs also showed peaksthat would partially overlap within 60.1 min tR of analytepeaks: cocaine and amitriptyline with mianserin, codeinewith dothiepin, and desipramine with protriptyline. TheGC-NPD method was applied to plasma samples fromdepressed patients, chronically treated with ADPs, whohad intentionally ingested high doses of their treatment.In the case of a patient overdosed with dothiepin, ourmethod allowed a successful resolution of dothiepin (tR19.28 min) with its two major metabolites, the northiaden(tR 19.37 min) and the dothiepin sulfoxide (tR 19.69 min),in spite of close tRs.

The main advantages of this technique are based on:1) Use of NPD detector: The introduction of new

generations of NPD detectors allows signals of highstability during long-term operation, and the nitrogen-containing structure of ADPs is compatible with a highsensitivity for NPD detection. The method’s measurementrange (0.5–2000 mg/L for amoxapine and dothiepin and5–4000 mg/L for fluvoxamine) makes it possible to avoidthe potential need for sample dilution in cases of intoxi-cation, and it also has a low detection limit (#2 mg/L).

2) Simplicity of the procedure: The simple addition ofan ammonia solution before the initial introduction ofplasma and other reagents for extraction minimizes ad-sorption of ADPs on glass and also makes the biologicalmaterial slightly alkaline (pH .10) for the extraction.Protriptyline was chosen as IS because this compoundeither has been withdrawn from the market in some

countries, or is not as widely used as tricyclic ADPs. Thedecrease in the proportion of hexane (57%), the additionof a more polar solvent, dichloromethane (42%), andisoamyl alcohol (1%) to break up emulsions was shown tobe preferable for recovery of tricyclic ADPs and fluox-etine, as we previously described in recent reports [5, 6].Overall recoveries ranged between 65% and 97%.

3) Good resolution: The benefit of capillary GC essen-tially resides in a high resolution, allowing for peakresolution of peaks with tRs within .0.1 min of eachother.

4) Powerful detection tool for ADPs: Automated immu-noassays are unable to detect ADPs unrelated to theimipramine structure [4]. Chromatographic proceduresare the most viable techniques used to identify anddetermine these drugs. The benefits of monitoring ADPshave been much discussed, but in many cases, the infor-mation obtained has proved useful for the management ofpsychiatric patients [7].

In conclusion, the present assay is one of the mostsensitive and specific assays for second-generationADPs. Its ability to measure several new ADPs simul-taneously makes it particularly useful for toxicologicalemergency and for managing psychiatric patientstreated with ADPs.

We are grateful to Anphar-Rolland, Eli-Lilly, Lederle,Minden, Organon, Solvay-Pharma, and Zeneca compa-nies for providing analytical calibrators.

References1. Hudson J L, Harrison G, Pope JR. Newer antidepressants in the treatment of

bulimia nervosa. Psychopharmacol Bull 1987;23:52–7.2. Richelson E. Pharmacology of antidepressants—characteristics of the ideal

drug. Mayo Clin Proc 1994;69:1069–81.3. Van Harten J. Clinical pharmacokinetics of selective serotonin reuptake

inhibitors. Clin Pharmacokinet 1993;24:203–20.4. Gupta RN. Drug level monitoring antidepressants. J Chromatogr Biomed Appl

1992;576:183–211.5. Fontanille P, Jourdil N, Villier C, Bessard G. Direct analysis of fluoxetine and

norfluoxetine in plasma by gas chromatography–nitrogen–phosphorus detec-tion. J Chromatogr Biomed Appl 1997;692:337–43.

6. Jourdil N, Pinteur B, Vincent F, Marka C, Bessard G. Simultaneous determi-nation of trimipramine and desmethyl- and hydroxytrimipramine in plasmaand red blood cells by capillary gas chromatography with nitrogen-selectivedetection. J Chromatogr Biomed Appl 1993;613:59–65.

7. Preskorn SH, Fast GA. Therapeutic drug monitoring for antidepressants:efficacy, safety and cost effectiveness. J Clin Psychiatry 1991;52/6:23–33.
