curve for calculation of results. Therefore, reliable results can be obtained with a simple recorder, and more expensive integrators are not required. Paraxanthine, a metabolite of caffeine, is co-eluted with theophylline. Therefore, precise measurement of theophylline in plasma cannot be obtained in subjects exposed to caffeine. This is a known limitation of most reversed-phase chromatographic methods for methylxanthines. Alternative methods have been used to solve this problem. Adequate resolution can be obtained with normalphase chromatography or, preferably, by adding an ion-pair reagent to reversed-phase chromatography (5, 8, 9). We have not addressed this issue because it was not pertinent to our specific application. We have applied this method successfully to measure methylxanthines in only 10 L of plasma. This may have particular application in neonates and small animals. Moreover, we can reliably measure concentrations of caffeine in tissue after its oral administration. The rat is a widely used model to study the behavioral effects of caffeine, as well as its potential toxicology. In addition, methylxanthines have been used as potential pharmacological probes to study the role of adenosine in physiological and pathological processes (10, 11). For each of these applications it is important to determine the concentrations of caffeine achieved at relevant tissue sites, because concentrations in tissue may correlate better with caffeine actions than do caffeine intake or concentrations in plasma. The tissue concentrations of caffeine obtained in this study (slightly <100 ano1fL) approximate the Kd found in in-vitro studies for adenosine receptor antagonism (12). We thank Mrs. Dorothea Boemer and Mr. Bolton Smith for their help in preparing the manuscript and Mr. Bahrat Patel and Mrs. Suzanna Lonce for excellent technical assistance. Tha work was supported by grants HL-34021, HL37961, and RR00095 from the National Institutes of Health. I. B. was a clinical associate physician of the Elliot V. Newman Clinical Research Center. D. R. is a Burroughs Welicome Scholar in Clinical Pharmacology. CLIN.CHEM. 34/11, 2348-2351 References 1. Curatolo P, Robertson D. The health consequences of caffeine. Ann Intern Med 1983;98:641-53. 2. Pickard CE, Stewart AD, Hartley R, Lucock MD. A rapid HPLC method for monitoring plasma levels of caffeine and theophylline using solid phase extraction columns. Ann Clin Biochem 198623:440-6. 3. Hartley R, Smith LI, Cookman JR. Improved high-performance liquid chromatographic method for the simultaneous determinations of caffeine and its N-demethylated metabolites in plasma using solid-phase extraction. J Chromatogr 1985;342:105-17. 4. Robertson D, Frolich JC, Carr RK, et al. Effects of caffeine and plasma renin activity, catecholamines, and blood pressure. N EngI J Med 1978;298:181-6. 5. Christensen HD, Whitsett TL. Measurements of xanthines and metabolites by means of high pressure liquid chromatography. In: Hawk GL, ed. Biological/biomedical applications of liquid chromatography, vol 1. New York: Dekker, 1979:507-37. 6. Kabra PN, Nelson MA, Marten LI. Simultaneous very fast liquid-chromatographic analysis of ethosuximide, primidone, phenobarbital, and carbamazepine in serum. Chin Chem 198329:473-6. 7. Ou C, Frawley VL. Concurrent measurement of theophylline and caffeine in neonates by an interference-free liquid chromatographic method. Clin Chem 1983;29:1934-6. 8. Scott NH, Chakrabokty J, Marks V. Determination of caffeine, theophylline and theobromine in serum and saliva using high performance liquid chromatography. Ann Clin Biochem 1984;21:120-4. 9. Kester MB, Saccar CL, Rocci ML, Mansmann HC. New simplified microassay for the quantitation of theophylline and its major metabolites in serum by high-performance liquid chromatography. J Chromatogr 1976;380:99-108. 10. Evoniuk A, Von Borstel RW, Wurtman RI. Antagonism of the cardiovascular effects of adenosine by caffeine and 8-(p-sulphophyl)theophylline. J Pharmacol Exp Ther 1987;240:428-32. 11. Ohnishi A, Branch RA, Jackson K, Biaggiom I, Deray G, Jackson EK. Chronic caffeine administration exacerbates renovascular, but not genetic hypertension, in rats. J Clin Invest 1986;78:1045-50. 12. Daly JW. Adenosine receptors: targets for future drugs. J Med Chem 1982;25:197-207. (1988) ResultsCompared for TricyclicAntidepressantsas Assayed by LiquidChromatographyand Enzyme Immunoassay R. Cameron Dorey, Sheldon H. Preskom, and Pamela K. Widener2 The tncyclicantidepressantsamitnptyline,nortriptyline,imipramine, and desipramine in serum of patientstaking one of the drugs were quantified in two laboratories by highperformance liquid chromatography (HPLC) and enzymemultiplied immunoassay (EMIT’”; Syva). Results for split sampleswere highlycorrelated,but EMIT gave higherresults in mostcases, and the slopesof the correlationlinesfor each ‘Department KS 67208. of Chemistry, Wichita State University, Wichita, 2Psychiatry Service, Wichita Veterans Administration Center, Wichita, KS 67218. 3Department of Psychiatry, University of Kansas Medicine at Wichita, Wichita, KS 67214. Received May 20, 1988; accepted August 8, 1988. Medical 2348 CLINICALCHEMISTRY,Vol.34, No.11, 1988 School of analyte were >1. Detection limits for the two procedures were such that 18% of the EMIT resultsfor the drug(s) were considered negative, as compared with 4% of the HPLC results.Additionalassay of desmethylor hydroxyantidepressant metabolites by HPLC did not explain the higher EMIT results. The relatively high detection limit for EMIT greatly limits its use in therapeutic drug monitoring, where low concentrationsof tricyclicantidepressantsare as important as highones for dose adjustmentor determinationof compliance. Other problems with EMIT measurement of tricyclic antidepressantsare discussed. Assay of tricyclic antidepressant (TCA) drugs amitriptyline (AM!), nortriptyline (NOR), imipramine (IMI), and desipramine (DM1) in pharmacotherapy is becoming an accepted practice among physicians who prescribe these drugs (1). This is in large part attributable to the recent rapid development of analytical methods for quantifying these drugs, whose therapeutic concentrations in plasma are generally <300 tg/L, whereas toxicity develops at concentrations <500 pg/L (2). TCAs are generally quantified by gas chromatography (CC) or high-performance liquid chromatography (HPLC). In these methods, the parent drug and one or more metabolites can be quantified independently. HPLC is generally more commonly used than CC for TCA analysis, because sample preparation is less time consuming: single-step liquid or solid extractions often suffice for sample cleanup, and derivatization of a secondary amine TCA is not generally needed (3). HPLC procedures, however, cannot separate as many compounds as GC (particularly capillary CC), and most published procedures display interferences, either from some phenothiazine-type antipsychotic drugs (4) or benzodiazepines (5), although some schemes avoid problems with both classes of drugs (6). Interferences depend primarily on the type of HPLC column used; use of an alternative type of column often can eliminate a specific interference (3). In any event, the procedures involved in HPLC or CC are labor-intensive for the toxicology laboratory, because samples must be analyzed serially, and each standard and patient’s sample may require 10 mm for the chromatographic step alone. Moreover, skill is needed in extraction of samples and interpretation of the resulting chromatogram, so the assay is not a technically trivial one. Methods for quantifying TCAs must be able to distinguish between the tertiary amines and their demethylated metabolites-AMI and NOR, IMI and DM1-because both will be present in plasma of patients taking the parent drug (2). Also, all four compounds undergo hydroxylation of the backbone structure, after which the hydroxy and desmethyl metabolites are conjugated to the glucuronide for excretion. At present, the hydroxy metabolites are not often assayed, there being no clear correlation between their concentrations and therapeutic benefit or side effects. This may change, however, given a recent report (7) that hydroxy metabolites of NOR and DM1 are correlated with cardiotoxic symptomatology in some patients. The Syva Co. has recently developed an assay of TCAs based on their proprietary homogeneous enzyme-multiplied immunoassay technology (EMrrTh) (8). This assay, which incorporates a single solid-phase extraction and nonisotopic immunoassa3, is adaptable to different automated analyzers for TCA analysis in laboratories without HPLC or CC equipment (9-11). To investigate local reports that results by EMIT were generally higher than by HPLC, we compared the results by EMIT assays with those by a generally accepted HPLC technique. Also, because published reports often have not compared the results between the two techniques for NOR and DM1 when patients were taking AM! or IMI (11,12), we have done so. Because the total TCA concentration is important therapeutically, increases in the reported metabolite concentrations, from cross-reactions or otherwise, can affect the course of therapy. Nonstandard abbreviations: EMIT, enzyme-multiplied immunoassay technique; TCA, tricyclic antidepressant(s); AM!, amitriptyline; NOR, nortriptyline; IMI, imipramine; DM1, desipramine; CC, gas chromatography. Materials and Methods Specimens. We obtained patients’ specimens from two sources: patients of the Psychiatry Service2 and patients participating in a study involving amitriptyline administration at the University of Kansas Medical School at Wichita. All specimens were collected into 10-mL red-top (no additive) Vacutainer Tubes (Beeton Dickinson, Rutherford, NJ) and allowed to clot, then centrifuged. The serum was then pipetted into polypropylene tubes for storage at -20 #{176}C until assay. Results by either procedure were not made known to the other assay site until after the sample was analyzed by both methods. Statistical analysis. For statistical analysis of the comparison study, we used linear-correlation methods. The statistical program used, the Stats-2 (StatSoft, Tulsa, OK) program, calculates linear correlations of one variable with one or more other variables, so that any cross-reactivity of the EMIT antibodies with metabolites could be factored into the total response of the assay. Samples with results by HPLC that were below the stated detection limit for the EMIT assay (50 g/L for DM1, 25 pgtL for AM!, IMI, or NOR) were not included in the regression analysis. One patient’s sample contained thioridazine and metabolites, which interfere with both procedures; thus it was not included in any analyses of results. HPLC assay. We performed the HPLC assays at the Research Psychopharmacology Laboratory,2 using a modification of the method of Koteel et al. (4), with the extraction step optimized for recovery of hydroxy metabolites. It is a typical HPLC method and has been validated against a GC/ mass spectrometric method (13) at the Medical College of Virginia, where it has been in use for routine clinical monitoring for the past five years. We prepared serum-based standards by adding aqueous solutions of the tertiary and analogous secondary TCAs (prepared from the hydrochloride salts) to drug-free serum to give final concentrations of 50, 100, and 200 gfL. The internal standard used was protriptyline hydrochloride (no patients in this study were taking protriptyline), which was eluted later than either of the TCAs of interest or their hydroxy metabolites. We prepared the samples for HPLC analysis as follows: We first combined 1 mL of serum with 50 L of an aqueous 10 ig/L solution of protriptyline hydrochloride, 1 mL of sodium carbonate solution (2 mollL, pH not further adjusted), and 4 mL of ethyl acetate/hexane/isobutanol (50/49/1 by vol) extraction solvent. We vortex-mixed this mixture for 10 s, centrifuged at 2000 x g for 5 mm, and pipetted the extraction solvent into a second tube. We then evaporated the solvent under reduced pressure at 40 #{176}C in a Model SVC100H vacuum centrifuge (Savant Instruments, Farmingdale, NY), reconstituted the residue in 200 .u.Lof the HPLC mobile phase (see below), and injected 100 L of this into the HPLC. We used a Supelco LC-PCN column, 15 cm x 4.6 mm (i.d.), and a mobile phase consisting of acetonitrile, methanol, and phosphate buffer (10 mmol/L, pH 7.0), 60/15/25 by vol. Drugs were detected by their absorbance at 254 rim, for which we used a Model 440 fixed-wavelength detector (Waters, Millipore Corp., Milford, MA). Calibration curves were constructed by correlating the ratio of peak heights of the analyte and internal standard with the concentrations of the corresponding TCA in the plasma standards. EMIT assay. EMIT assays were performed at Roche Biomedical Laboratories, Wichita, KS, with the Syva TCA assay CLINICALCHEMISTRY, Vol. 34, No. 11, 1988 2349 kits used according to the manufacturers directions. The zrr assay involves use of monoclonal antibodies for AM! and 1MI, and polyclonal antibodies for NOR and DM1. Glucose-6-phosphate dehydrogenase is linked to drug derivatives for competitive binding with the drugs found in plasma, and enzyme activity (proportional to drug concentration) is monitored as an absorbance change (NAD to NADH) with time, at 340 nm. Drugs are extracted from plasma before assay with a solid-phase extraction technique (8). Results Of the 94 samples analyzed by both methods, results by for 17 (18%) were below the detection limit for at least one analyte. In contrast, only four samples (4%) had nondetectable concentrations by HPLC. None of the samples negative for TCAs by HPLC gave detectable concentrations by EMIT. These samples were not included in the regression analysis comparing the two methods. Table 1 summarizes the results for the samples used in the single-variable linear regression analysis. Curves are presented separately for patients taking the secondaryamine TCAs (NOR and DM1) and for patients taking the tertiary amine TCAs (AM! and IM1). The slopes and intercepts for these curves are not statistically different between the two groups. This finding shows negligible interference from tertiary-amine TCAs in assays of the corresponding secondary-amine TCAs. One sample from a patient taking IMI had DM1 and total TCA concentrations four times that of the next highest patient. Results for this patient so changed the slope of the calculated regression line that we performed analyses both with and without this patient’s results. The slopes of the two regression lines were, however, statistically equal. We cannot conclude from this single sample whether or not there is any change in the curve at high concentrations. Multiple correlation studies comparing EMIT results of NOR plasma concentrations with HPLC results of NOR and hydroxy-NOR concentrations did not indicate any influence of the hydroxy metabolite on the EMIT results. Similarly, we compared EMIT results for AM! concentrations to HPLC results for AM!, NOR, hydroxy-AMI, and hydroxy-NOR. Factoring in the metabolite concentrations did not increase the correlation coefficient found by using only the parent compound. This suggests no significant interference from metabolites on parent-compound EMIT values. Analysis of EMIT Table 1. Correlations between TCA Plasma Concentrations as Determined by HPLC and EMIT Slope, Drug taken by patIent AMI NOR IMI Analyte AMI NOR AMI #{247} NOR NOR IMI DM1 IMI + DM1 n awr vs HPLC 15 15 15 23 16 (15) 16 (15) 16 1.27 1.19 1.25 1.16 0.97 (15) DM1 39 (1.09) 0.91 (1.02) 0.90 (1.12) 0.97 pg/L -5.6 13.1 3.0 9.6 16.8 (8.1) 33.0 (26.5) 57.9 (25.6) 20.0 Because r 0.96 0.99 0.99 0.94 0.94 (0.95) 0.95 (0.88) 0.95 (0.93) 0.99 the DM1 concentration measured for a patient taking IMI was fourfold the next-highest concentration measured for the group. the results in parentheses exclude this point to show its exaggerated effect on the regression line. 2350 CLINICALCHEMISTRY, Vol.34, No. 11, 1988 the DM1 and IM! correlations between methods produced the same results. Discussion Results for EMIT assay of TCA concentrations in plasma correlate well with HPLC results, but the slopes of the correlation curves often considerably exceed unity, and the intercepts of the correlation curves are often positive. These results are not explained by simple selectivity problems of the antibodies used in the EMiT assays. We do not see any significant direct cross-reactivity between the tertiary amines and their secondary metabolites in these assays. Also, including hydroxy metabolites in the correlation analyses had very little effect. However, EMIT evidently will generally produce higher results than our HPLC method, as happened in 70 of the 77 samples that were reported as positive by both methods. This is not clinically significant in cases involving TCA toxicity or long-term routine monitoring. Therapeutic guidelines for TCA concentrations in plasma cover wide ranges, and there is no immediate risk of toxicity or rapid relapse at moderately high concentrations. However, there are two situations in which EMiT is at a distinct disadvantage. First, the EMIT assay is inappropriate for low TCA concentrations in plasma, because of its high detection limits and the positive intercepts of some of the EMIT vs HPLC correlation curves. This situation is likely in three common settings: when a physician starts a patient at a low TCA dose and wishes to increase the dosage on the basis of the concentration in plasma; when there is a question of compliance or rapid metabolism in a nonimproving patient; or where a “test dose” of a TCA is given, with the TCA concentration in plasma measured after 24 h, from which daily doses are extrapolated. Second, the EMIT assay cannot be modified to eliminate interference from other drugs. In the assay, there is significant interference by chlorpromazine in therapeutic concentrations (8), and the package insert with the assay kit describes a similar interference with thioridazine. Two other phenothiazines and the benzodiazepines do not interfere at therapeutic concentrations (8). Thus, TCA as measured in plasma by EMIT in patients who are taking phenothiazine antipsychotics will often be erroneously high. The tertiary-amine TCAs cross-react in each other’s assay, as do the secondary-amine TCAs (8), although we have seen minimal cross-reactivity between the two classes. This makes the ordering of the correct test (not just “tricyclics”) imperative for the requesting physician, and will cause problems when a patient is being switched from one TCA to another without “washing out” the first drug before checking the concentration of the second drug in the patient’s plasma. Laboratories where EMIT is to be the routine method for quantifying TCAs should have a chromatographic method on hand for backup where these cases are suspected. We gratefully acknowledge partial support of this study from the Veterans’ Administration Merit Review research program. References 1. American Psychiatric Association Task Force. Tricycic antidepressanta-blood level measurements and clinical outcome. Am J Psychiatry 1985;142:155-62. 2. Preskorn SH, Dorey RC, Jerkovich GS. Therapeutic monitoring of tricyclic antidepressants. Cliii Chem 1988;34:822-8. 3. Wong SHY. Measurement of antidepressants by liquid chromatography: a review of current methodology [Review]. Cm Chem 1988;34:848-55. 4. Koteel P, Mullins RE, Gadsden RH. Sample preparation and liquid-chromatographic analysis for tricycic antidepressants in serum. Cliii Chem 198228:462-6. 5. Breutzmann DA, Bowers LD. Reversed-phase liquid chromatography and gas chromatography/mass fragmentography compared for determination of tricydic antidepressant drugs. Cm Chem 198127:1907-11. 6. Yang S, Evenson EA. Simultaneous liquid chromathgraphic determination of antidepressant drugs in human plasma. Anal Chem 1983;55:994-8. 7. Pollock BG, Perel JM. Hydroxynietabolites of tricyclic antidepressants: evaluation of relative cardiotoxicity. Cliii Pharmacol Psychiatry. 8. Pankey In press. 5, Collins C, Jaklitsch CLIN. CHEM. 34/11, 2351-2354 A, et al. Quantitative homoge- neous enzyme inununoassays for amitriptyline, nortriptyline, imiprainine, and desipramune. Clin Chem 1986;32:768-72. 9. Orsulak PA, Haven MC, Huth JA, Studts DJ. EMIT#{174} quantitive tricyclic antidepressant assays applied to the Hitachi 705#{174} chemistry analyzer [Tech Brief]. Cliii Chem 1987;33:1471. 10. Haven MC, Orsulak PA, Huth JA, Markin ES. EMIT#{174} quanti- tive tricycic antidepressant assays applied to the Encore” chemistry system [Tech Brief]. Cm Chem 1987;33:1472. 11. Frye R, Mathews S. Evaluation of Syva#{174} EMIT” enzyme immunoassays for the tricyclic antidepressants on the Cobas#{174} centrifugal analyzer [Abstract]. Clin Chem 1986;32;1053. 12. Foglia JP, Perel JM, Stiller RL, et al. Effect of specimen collection on immunoassay techniques for selected tricyclic antidepressant drugs [Abstract]. Clin Chem 1986;32:1052. 13. Narasimhachari N, Sandy J, Friedel RO. Quantitative mapping of metabolites of imipramine and desipramine in plasma samples by gas chromatography-mass spectrometry. Biol Psychiatry 1981;16:937-44. (1988) “Flipped”Patterns of Lactate Dehydrogenaselsoenzymesin Serum of Elite College Basketball Players hi Rotenberg,’ Richard Seip,2 Larry A. Wolfe, and DavidE. Bruns4 We kinetically measured total lactate dehydrogenase (LD, EC 1.1.1.27), total creatine kinase (CK, EC 2.7.3.2), and aspartate aminotransferase (AST, EC 2.6.1.1.) in 16 elite college basketball players, before the competition season and not in close temporal relation to near-maximal exercise, and in 17 healthy non-athlete controls. LD isoenzymes were determined by both electrophoretic and immunoprecipitation methods. CK-MB isoenzyme was measured electrophoretically. We found significantlly higher mean LD-1 values and LD-1/LD-2 ratios in the players than the controls: 31.6 (SD 3.7)% vs 25.8 (SD 3.2)% (P <0.005) and 1.1 (SD 0.13) vs 0.87 (SD 0.16) (P <0.001), respectively. A “flipped” LD pattern (LD-1 > LD-2) was found in half the players and in six of the eight black athletes, but in only two of the control group and in none of the black controls. Mean CK activity in serum exceeded normal values in the serum of the athletes and was higher in comparison with the control group [274 (SD 156) vs 103 (SD 82) U/L]. Mean CK was significantly higher in the eight athletes with the flipped LD pattern than in those with LD-1 <LD-2 [322 (SD 163) vs 180 (SD 98) U/L; P 0.05], and also in comparison with CK in the two controls with flipped LD pattern. We saw no significant difference in mean CK between the nine players with normal immunochemical LD-1/LD ratios and the seven players with above-normal ratios. CK-MB was not detected in either athletes or controls. Departments of Pathology and2 Health and Physical Education, Schools of Medicine and Education, University of Virginia, Charlottesville, VA 22908. Present address: ‘Massada Center for Heart Diseases, Beiinson Medical Center, Petah Tikva 49100, Israel; and3 School of Physical and Health Education, Queens University, Kingston, Ontario, Canada, K7L 3N6. ‘Address correspondence to this authoi- Box 168, Department of Pathology, University of Virginia Medical School, Charlottesville, VA 22908. Received June 29, 1988; accepted August 8, 1988. None of the players had any clinical or electrocardiographic evidence for myocardial ischemia or infarction. Evidently the flipped LD pattern usually found in patients with acute myocardial infarction and reported in some athletes after extreme exercise such as ultra-marathon running may also be found in athletes who are in their “basal fitness shape” but who are not involved in competitive physical activity. Numerous reports have been published on the increase in serum enzyme activities following various types of exercise (1-10). The most commonly described increase is that in serum creatine kinase (CK, EC 2.7.3.2) (1-4, 6-9, 11). Increases in aspartate aminotransferase (AST, EC 2.6.1.1) (1, 7) and lactate dehydrogenase (LD, EC 1.1.1.27) (2, 6, 7, 11) have also been seen, but the changes were generally smaller than for CK. Most studies have reported increased LD only after vigorous exercise, such as marathon running. Some (4, 9, 12) have even described a “flipped” LD pattern (LD-1 > LD-2) in certain athletes after long-distance running, similar to that described in patients with acute myocardial infarction (13-16). The present study was prompted by the unexpected finding of abnormalities of both CK and LD isoenzymes in college basketball players. In particular, five of 16 players demonstrated a ffipped LD ratio when the athletes underwent routine examinations on the second day of team practice. The present prospective study was undertaken in the same group a year later, to determine a “basal” proffle of LD isoenzymes before the competition season. Patients and Methods We studied 16 members of a highly ranked U.S. basket(mean age 20.3, SD 1.3 years), and 17 healthy, tall ball team 5Nonstandard abbreviations: dehydrogenase; AST, aspartate CK, creatine kinase; LD, lactate aminotransferase. CLINICALCHEMISTRY, Vol. 34, No. 11, 1988 2351