CLIN. CHEM. 38/2, 256-262 (1992)

256 CLINICALCHEMISTRY,Vol.38, No.2, 1992

Improved Detection of Ischemia-Induced Increases in Coronary Sinus Adenosine inPatients with Coronary Artery DiseaseMarc D. Feldman,’4 Caries R. Ayers,’ Marcia R. Lehman,’ Heidi E. Taylor,’ Vicki L. Gordon,’ Peter J.Sabla,’ Don Ras,3 Thomas C. SkaIak,3 and Joel Linden’

Attempts to monitor coronary sinus adenosine as a clini-cal marker of myocardial ischemia in humans have beendisappointing. Accordingly, procedures have been devel-oped for detecting adenosine in blood collected from thehuman coronary sinus. Collection involves using a dou-ble-lumen metabolic catheter, which allows blood to bemixed with a stop solution at the catheter tip, therebyminimizing adenosine formation and degradation. A five-component stop solution almost completely arrests aden-osine formation and degradation. Adenosine analysis isimproved by using both boronate and C18 Sep-Pak col-umns to purify and concentrate adenosine in humanplasma before HPLC. Plasma adenosine in the coronarysinus of patients with and without coronary artery disease,measured before and during peak atnal pacing, showed atwofold atrial pacing-induced increase in adenosine in thepatients with coronary artery disease (n = 9, P <0.001)but no change in the patients with normal epicardialcoronary arteries (n = 6). These preliminary resultsIndicate that coronary sinus adenosine may provide anIndex of myocardial ischemia in patients with coronaryartery disease.

Additional K.yphrasis: myocardlal . ,n&lc oaths-ter - chromatography, liquid

Adenosme has been proposed as a likely metabolicsignal released by the hypoxic or ischemic myocardiumto elicit an increase in coronary blood flow in responsetoa decrease in the oxygen supply/demand ratio (1). How-ever, attempts to monitor coronary sinus adenosine as aclinical marker of myocardial ischemia in humans havebeen disappointing. An early report that adenosine isincreased in the coronary sinus of patients with isch-emic heart disease during pacing (2) was not reproducedin several subsequent studies in which more sensitiveassay techniques were used (3-6). This may be becausethe half4ife of adenosine in hnmsin blood is <1.5 s (7)and because a large fraction of adenosine is taken up byendothelial cells (8) and other cellular blood elements(9). Only when the adenosine transport inhibitor dipy-ridamole was systemically infused before pacing couldSollevi et al. (4, 5) demonstrate a pacing-induced in-

Departments of 1 Internal Medicine and’ BiomedicalEngineer-ing, University of Virginia Health SciencesCenter, Charlottes-ville, VA 22908.

2Addreeacorrespondence to this authorat Division ofCardiol-ogy, Box 158, University of Virginia Health SciencesCenter,Charlottesville, VA 22908.

ReceivedJune 4, 1991; acceptedDecember27, 1991.

crease in coronary sinus adenosine in patients withcoronary artery disease.

We hypothesized that the inability to use blood aden-osine concentrations as a metabolic marker of myocar-dial ischemia in patients might be due in part tomethodological limitations. In current sampling tech-niques, catheters >100-cm long are used to draw bloodfrom the coronary sinus. No solution to stop adenosinemetabolism is included and therefore both degradationand artifactual production of adenosine are likely tooccur during transit in these long catheters. Our objec-tive was to develop methodology capable of detectingischemia-induced increases of coronary artery adeno-sine.

MaterIals and Methods

Sample Collection

We developed a double-lumen metabolic catheter(U.S. patent 5,009,634), which allows addition and mix-ing of a stop solution to blood entering the tip of thecatheter. Figure IA is a schematic diagram of thecatheter, which has two lumens, consisting of an outercatheter (8F multipurpose) and an inner catheter (2.5F)ending 1 cm from the end of the outer catheter (FiguretB). A prototype of this catheter was described previ-ously (10). The outer and inner catheters are connectedby a Y-adaptor. Blood and stop solution movementthrough the outer and inner catheters are coupled by adouble-syringe device. The larger collection syringeholds 6 mL, whereas the smaller infusion syringe holds3 mL. The smaller syringe, being half the cross-sec-tional area of the larger syringe, ensures equivolumemining of blood with stop solution at the tip of themetabolic catheter. During patient studies, the plungeris pulled out of the collection syringe and blood flowsinto the outer catheter and the collection syringe. As theplunger on the collection syringe is being pulled out, thesecond plunger is pushed into the infusion syringe,infusing stop solution through the inner catheter so thatmining between the stop solution and blood occurs at thedistal end of the metabolic catheter. Both catheters areprimed with stop solution before the blood is withdrawn.The total volume of blood plus stop solution in thecollection syringe is 4 mL.

The stop solution contains the adenosine uptake in-hibitor dipyridamole (0.2 mmol/L; Boehringer In-geiheim, Ridgefield, CT), the adenosine deaminase in-hibitor erythro-9(2-hydroxy-3-nonyl)-adenine (EHNA, 5anol/L; Burroughs Wellcome, Research Triangle Park,NC), the 5’-nucleotidase inhibitors aj3-methylene-aden-

CLINICALCHEMISTRY,Vol.38, No.2, 1992 257

FIg. 1. (A) Diagram ofthemetabolic catheter: blood and stop solutionmovement throughtheouter(1) and Inner(2)cathetersIscoupledbythe double-syringe device ( connectinga collection syringe (4) andan infusion syringe (; ( a magnifiedviewofthedistalendoftheoutercatheter(1), provIdinga cutaway view oftheInner catheter (2)

osine 5’-diphosphate (AOPCP, 62 1.unol/L; Sigma Chemi-cal Co., St. Louis, MO), EDTA (4.2 mmol/L; SterlingDrug, Northridge, CA), and heparin (25 kilo-int. units/L).4 Peripheral venous blood samples from volunteerswere used to validate the assay with use of the abovefive-component stop solution. To eliminate the possibil-ity of accidentally injecting AOPCPand EHNA into pa-tients, we omitted these two components from the stopsolution when we used the metabolic catheter to with-draw blood from the coronary sinus. However, supple-mental stock solutions of AOPCP and EHNA were added tothe collection syringe to mix with the blood as it wascollected. The stop solution in the infusion syringe andthe dead space in the metabolic catheter between theinner and outer catheters and within the inner catheterwere primed with dipyridarnole, EDTA, and heparin atthe concentrations listed above. We injected 0.2 mL ofconcentrated EHNA and AOPCP into the withdrawal sy-ringe, giving final concentrations the same as above,once the blood, dipyridamole, EDTA, and heparin weremixed in the collection syringe.

filtered through a 0.2-sm (pore size) filter (Acrodisc;Gelman Sciences, Ann Arbor, MI), and 106 pmol ofN6-methyladenosine (internal standard; Sigma Chemi-cal Co.) was added to 1 mL of filtrate. Two precolumnsare used to partially purify and concentrate the adeno-sine: a 0.5-mL boronate gel column (Affi-Gel 601; Bio-Rad Labs., Richmond, CA) and a 1.2 x 1.2 cm reversed-phase C18 Sep-Pak column (Waters, Milford, MA).Adenosine and internal standard were eluted from theSep-Pak, warmed to 55#{176}C,dried under N2 in a MultivapAnalytical Evaporator (Organomation Associates,South Berlin, MA), and reconstituted with 400 1zL of2”#{176}4 (5 mxnol/L, pH 3.6).

The HPLC column used was a 4.6 x 250 rum, 5-jmparticle, 0.8-nm pore, C18 column (Beckman Instru-

ments, San Ramon, CA). The mobile phase was KH2PO4(5 mrnol/L, pH 3.6) in methanol (100 mL/L). Adenosine

and the internal standard were eluted isocratically at aflow rate of 1.5 mLfmin. The ultraviolet absorbance (254urn) of the eluent was monitored continuously (0.002absorbance units, full scale).

The amount of adenosine in each sample was deter-mined by comparing the observed peak area with astandard curve. The amount of plasma assayed wascalculated as follows:5

Plasma volume assayed = (1 mL)[(1 - Hct)/(4.44 - Hct)]

Because half of the sample was injected onto the HPLCcolumn, this result was divided by 2. For a normal Hctof 0.40, the assayed plasma volume is 74 ,uL of plasma.The concentration of adenosine in plasma was thencalculated by dividing the amount of adenosine mea-sured by the volume of plasma assayed, with correctionfor recovery of the internal standard, which ranged from50% to 65%.

Assay Verification

Coefficient of variation. A 4-mL peripheral venousblood sample was collected from a volunteer into 12 mLof stop solution through a 2.5-cm, 19-gauge needle. Thesupernate was divided into seven 1-mL portions andprocessed as above. The mean, SD, and CV determinedfrom these seven samples were 0.186 and 0.034 moIfLand 18.3%, respectively. Because the CV was >10%, allresults presented are the means of duplicate determina-tions.

Recovery of aclenosine and internal standard duringsample preparation. To verify that the recoveries of bothadenosine and N6-methyladenosine were the same, weadded 18.7 pmol of adenosine and 106 pmol of internal

Sample Preparationand Analysis

Immediately after collection, the blood sample was

centrifuged at 4800 x g for 2 mm. Each supernate was

4Nonatandard abbreviations: ERNA, e,’ythro-9(2-hydroxy-3-no-nyl)-adenine; AOPCP, a,-methylene-adenosinediphosphate; Re,Reynolds numbers; and Hct, hematocrit.

5Th1s equation was derived as follows: The total volumeofblood/stop solution in the collectionsyringe is 4 mL, of which 0.2mL is AOPCP/EHNA, 2.9 mL is stop solution,,and 0.9 rnL is blood.The plasma volume collected = 0.9mL (1- Hct). Thetotalvolumecollected minus erythrocytes =4 mL - (0.9 mLXHct). Therefore,each 1 mL of collected sample contains a volumeof plasma given by1 mL{0.9mL(1 - Hct)44 mL - (O.9mLXHct)]}, or 1 mL((1 - Hct)/(4.44 - Hot)].

25$ CLINICALCHEMISTRY,Vol.38, No.2, 1992

standard to 1 mL of stop solution and processed andanalyzed this sample as above. The chromatographicpeak areas were compared with those for identicalamounts of adenosine and internal standard injecteddirectly onto the HPLC column. Duplicate samples wereanalyzed on three separate occasions (n = 6).

Linear recovery of adenosine added to plasma. Wecollected 4 mL of blood in 12 mL of stop solution asdescribed above. The supernatant fluid from this samplewas divided into six 1-mL portions, to which we addedadenosine (0, 4.7, 9.4, 18.7, 28.1, and 37.4 pmol) andprocessed the samples as above. The amount of adeno-sine detected was plotted as a function of adenosineadded.

Effect of adenosine deaminase on HPLC profiles. Wecollected blood as in the preceding section but dividedthe supernate into two portions. One portion was proc-essed as described above. The second portion was proc-essed similarly except that the dried samples werereconstituted with 200 iL of adenosine deaminase (2kU/L added to HPLC-grade H20; Sigma Chemical Co.).After letting the samples sit at room temperature for 20mm, we added 200 p.L of KH2PO4 (10 mmol/L, pH 3.6) toeach sample. The deaminase was inactivated by boilingfor 3 mm, and the sample was filtered as before andinjected onto the HPLC column. We did not add internalstandard to the sample to be deaminated because it wasa substrate for the enzyme.

Stop-Solution Verification

To determine the extent to which adenosine break-down is arrested by the metabolic catheter when dipy-ridamole, EDTA, and heparin are added at its tip andEENA and AOPCPare added in the collection syringe, weperformed the following experiment. Peripheral venousblood (3.6 mL) was drawn from a volunteer into heparin(25 kilo-int. unita/L). The blood was immediately dis-pensed into 11.6 mL of dipyridamole and EDTA solution(at concentrations given in Sample Collection) contain-ing 16 L of [14C]adenosine (New England Nuclear,Boston, MA; specific activity 57.2 Ci/mol; 58 159dpm/mL in blood plus stop solution). At 9 and 60s afterwithdrawal of blood from the volunteer, 0.8 mL of EHNA

and AOPCP were added (9 s is the average transit timefor blood in the metabolic catheter). Blood, five-compo-nent stop solution, and PC]adenosine were centrifugedat 2500 x g for 1.5 ruin, and the supernate was removed.We added 50 L of trichloroacetic acid solution (finalconcentration in supernate, 50 g/L) to 950 pL of super-nate and centrifuged at 2500 x g for 1 mm. We thenmixed 0.5 mL of this supernate with 0.5 mL of ti-n-octylamine (0.5 mmol/L) in Freon and centrifuged againat 2500 x g for 1 mm. We injected 50 1zL of the top(aqueous) layer into the HPLC column and continuouslymonitored the radioactivity with a Beckman 171 radio-isotope detector. To determine the extent to which AMPis degraded, we substituted 483 pL of [“CUJ)]AMP(specific activity 590 Ciimol; 71713 dpni/mL in bloodplus stop solution) for [‘4C]adenosine.

The metabolism of adenosine and AMP was markedly

inhibited by the stop solution at 9s. [‘4C]Adenosine and[14CJAMP were recovered in HPLC peaks correspondingto these purines at the following rates (mean ± SD):98.5% ± 0.8% (n = 6) and 90.5% ± 1.5% (n = 8),respectively. At 60 s, although the metabolism of [14C]-adenosine was still markedly inhibited [95.5% ± 1.2%was still present (n = 2)], the metabolism of [14C]AMPwas not: 1.2% ± 0.5% remained (n = 2).

Metabolic Catheter Verification

A model of the metabolic catheter was built to verifythat the stop solution and blood are mixed at its tip. Weused a precision-bore glass tube to simulate the outercatheter, to allow a direct microscopic view of flow at thetip. An inner catheter was inserted into the glass tube,and the coupled syringe system was connected to aninfusion-withdrawal pump to control the flow rate. Theopen end of the glass tube was located in a smallreservoir with a pressure of 1-3 cm of H20 (-100-300Pa), simulating the coronary sinus. Dextran solutionswith viscosities equal to those of blood at 37#{176}Cand stopsolution at 25#{176}Cwere used. The dextran stop solutionalso contained cresyl violet (20 g/L) to provide contrastfor flow visualization. For all observations we used amicroscope with a 3.5x Leitz long-working-distanceobjective and photographed the flow patterns for lateranalysis. The Reynolds numbers (Re) that indicatedadequate miring (11) were then determined. Flow ratesof 0.045, 0.090,0.180,0.270,0.315, and 0.405 mIIs wereexamined, corresponding to Re of 7.7, 15.5, 31.0, 46.5,54.2, and 69.7, respectively. Analysis of flow data in-volved determination by digital densitometry of theblack-and-white negative micrographa of the space be-tween the glass model of the outer and inner catheters.A uniform abeorbance (optical density) between thesecatheters was interpreted as adequate mixing. A grada-tion of optical densities was interpreted as indicatingregions of laminar flow and inadequate mixing of sim-ulated blood and stop solution.

Figure 2 shows micrographs of the flow patterns inthe glass model of the outer catheter, at the location ofthe inner catheter tip. The black rectangular shapesarethe tips of the inner catheters. The left and rightboundaries of the micrographs are the inner walls of theglass model of the outer catheter. The dashed linesdrawn on each micrograph indicate where the opticaldensity was measured. Panels A-C show an unmixed(clear) region of simulated blood at the left wall of theouter catheter and a dark (blue dye) region of stopsolution near the inner catheter body. This gradation ofrnirtng is reflected in the optical density measurements,which show a continuous decrease from the inner cath-eter body to the wall of the outer catheter. In panel D themicrograph appears to show uniform mixing, but theoptical density measurements show that a concentra-tion gradient is still present. In panels E and F theoptical density is uniform across the lumen of the outercatheter, indicating that complete mixing has beenachieved. Thus, adequate mixing occurs at the tip of themetabolic catheter if Re are maintained at �54.2. This

CLINICALCHEMISTRY, Vol. 38, No. 2, 1992 259

(I,z

-J

C-)I-00

- l.4 0.7 0.0 0.7

RADIAL DISTANCE, R (mm)

FIg.2. MixIng ofartificial blood with stopsolution inthemodelofthemetabolic catheter as shown in micrographs (right) and densitomet-nc measurements (Iel offlowpatternsThe blackrecfangularshapesarethe tipsof the Innercatheters, and the leftand right boundaiiesof themlcrographsaretheInnerwallsof the glassmodelof theoutercatheter.Adequate mixing of simulatedblood and stop solutionIsachievedat a Reynoldsnumberof 54.2 (panelE), correspondingtoa flow rateof 0.315mLJs

corresponds to withdrawal of a 4-mL sample (of bloodplus stop solution) in 12.78(0.315 mIJs). For all studieswith patients, 4 mL of blood plus stop solution waswithdrawn in 12 s, which exceeds the minimum flowrate required for adequate mixing.

Oxygen and Lactate AnalysisOxygen saturation of blood samples was determined

by an electrochemical fuel-cell method (Co-oximeter;Corning Medical, Medfleld, MA) in the catheterizationlaboratory. Blood lactate concentrations were deter-mined with the aca (E.I. du Pont de Nemours and Co.,Wilmington, DE), by a modification of the Marbach andWeil method, which involves the oxidation of lactate topyruvate. Lactate production was defined as coronarysinus lactate exceeding aortic lactate at peak pacing.

Studieswfth PatientsFifteen patients with a history of stable angina pec-

tons or atypical chest pain were chosen from the elec-tive cardiac catheterization schedule. Six of these pa-tients were found to have normal epicardial coronaryarteries and left ventriculography. The other nine had

two- or three-vessel coronary artery disease, with atleast one stenosis >70%. Patients were identified 18 hbefore catheterization and were approached for consent.All medications were stopped at this time.

We waited 30 miii between the routine diagnosticcatheterization and the research protocol. The atrialpacing stages were 100,120, and 140 beats/ruin, each for2 miii, with a final stage of 160 beats/ruin for 5-7 mm.We measured baseline and peak pacing coronary sinusadenosine, simultaneous coronary sinus and aortic lac-tate and oxygen saturation, heart rate, and blood pres-

sure, and obtained a 12-lead electrocardiogram.

Results

Assay Verification

Recovery of adenosine and internal standard duringsample preparation. In three paired samples (ii =6) thepercentage recoveries for adenosine and N6-methylade-nosine through all steps of the assay, from preparationfor boronate column application to reconstitution inphosphate buffer, were 61.9 (SD 7.0)% and 61.1 (SD5.5)%, respectively (not significantly different).

Analytical recovery of adenosine added to plasma. Theamount of adenosine detected by the assay was linearlyrelated (r2 = 0.97) to the amount ofexogenousadenosineadded to plasma. In addition, the slope of the regressionline was near unity (0.973), and the sum of endogenousand added adenosine was close to the predicted value (y= 17.61 + 0.973x).

Effect of adenosine deaminase on HPLC profiles. Fig-ure 3 depicts chromatograms for paired samples ob-tained from the same volunteer. One sample of each pairwas incubated with adenosine deamunase; the other wasnot. Adenosine deaminase e1iminited the adenosinepeak, indicating that this peak was indeed adenosine.(We did not add internal standard to the sample to bedenminted because it was a substrate for the enzyme.)

A.Control BAdenoueDeominase

Fig. 3. Effectof adenosinedeamlnase on HPLCprofilesforpairedsamplesfromthesamevolunteer,before (A) and after ( treatmentwith adenosinedeaminaseAdenoelne(black arrow) andInternalstandard(wl,Ite arrow)wereelutedat 9.7and28.8 mm,respectively.Internalstandardwas not added to the sample tobedeaminated

I0.

,J.u

Table 1. Results of Patients’ Studies (Mean ± SD)CAD (n =9)

Bessllns82 ± 14

11.3 ± 3.7

Pacing150 ± ga

19.8 ± 3.8a

68 ± 18b

8.5 ± 3.7

Normal(n = 6)

Bas.tin.71 ± 13

10.1 ± 1.9Heart rate,beats/mmRPP, x103ECG changes, no. 7 1

Anginascore 0±0 7±4 74b 0±0Ao-CSO2dIff,mLIL 108±16 114±21 5±12 118±15LactateExF +0.48± 0.15 +0.01± 0.36a -0.48 ± 0#{149}36b 0.48± 0.33Adenoslne, funol/L 0.132± 0.051 0.265± 0.071a 0.133± 0042b 0.140± 0.066

P <0.05, baselInevs pacing.b P <0.05. CAD vs normal.CAD, coronary artery disease;normal,normalepicardlalcoronaryarteries;pacing,peak pacing stage; APP, rate-pressureproduct; ECG changes.

pacing-InducedIschemlcelectrocardiogramchanges (>1.5mm ST-segmentdepression);anginascore,reproductionof typicalsymptomsona scale of 0-10.10being severe discomfort;Ao-CSO2dtff,aortocoronarysinusoxygencontentdifference;lactateExF,myocardlallactateextractIonfraction;adenoeIne. coronarysinus adenoelne.

Pasing

160 ± 622.0 ± 3.8a

1±3120 ± 14

0.42 ± 0.320.141 ± 0.070

87 ± 1411.9 ± 4.3

1±32±6

-0.06 ± 0.110.001 ± 0.012

260 CLINICALCHEMISTRY,Vol.38,No.2,1992

Studieswith PatientsPatients’ results are displayed in Table 1. There was

no difference in baseline heart rates, but the normalgroup attained a higher peak heart rate during atrialpacing. There was no difference in the baseline or peakatrial pacing rate-pressure products between the nor-mal and coronary artery disease groups. Seven of ninecoronary artery disease patients had normal baselineelectrocardiograms and developed >1.5 mm ST-seg-ment depression. The other two patients also developed>1.5 mm ST-segment depression, but the electrocardio-gram changes with atnial pacing were nondiagnosticbecause their baseline electrocardiograms were abnor-mal, with ST-segment depression. One of six normalpatients with ST-segment depression at baseline devel-oped >1.5 mm ST depression with atnial pacing. Eightof nine coronary artery disease patients developed re-production of their typical angina with atnial pacing,whereas only one of six normal patients developedreproduction of their typical chest pain symptoms withpacing.

There was no difference at baseline or change withpacing in the aortocoronary sinus oxygen content differ-ence between the coronary artery disease and normalgroups. Only three of the nine patients with coronaryartery disease had lactate production, defined as themyocardial lactate extraction fraction becoming nega-tive,although the coronary artery disease group as awhole had a decrease in the extraction fraction. Therewas no change in the myocardial lactate extractionfraction forthenormal group. There was no difference inthe baseline coronary sinus adenosine concentrationbetween the two groups; every patient in the coronaryarterydiseasegroup had an increase in coronary sinusadenosine of �1.5-fold (Figure4).There was no changein coronary sinus adenosine with atrial pacing in thenormal group.

Discussion

These results suggest that adenosine in coronarysinus blood is consistently increased during myocardialischemia. The equivocal results of previous studies in

0.4

0.3

0.2

:

CAD (n=9)

1Ba..lln. Pacing

Normal (n=6)

0.3

: fiBasslIn. Pacing

FIg.4. Coronarysinusadenosineconcentrations (pmol/L) at base-lIneandpeak atrial pacingfor Individualpatients In the coronaryarterydisease (CAD)andnormalgroupsMeans ± SD are shown for each group. The IncreaseIn the CAD group wassignifIcant(* <0.001)

which attempts were made to correlate blood adenosinewith myocardial ischemia in patients may have beendue to methodological limitations. We used a double-lumen metabolic catheter and a modified stop solutionto prevent artifactual production of adenosine by the useof inhibitors of 5’-nucleotidase. Using these techniques,we obtained plasma samples that showed that nine ofnine patients with coronary artery disease responded topacing-induced ischemia with a �1.5-fold increase incoronary sinus adenosine, whereas only three of thenine had lactate production.

Adenosine has a half-life of <1.5 a in human blood (7).To detect increases in coronary sinus adenosine result-ing from myocardial ischemia, one must quickly arrest

Betoreacs

6Thisstudy

43

13

CLINICALCHEMISTRY,Vol.38, No.2, 1992 261

adenosine metabolism. A double-syringe device wasintroduced by Ontyd and Schrader (12) to address thisneed. This device allowed stop solution to be added atthe entrance of the syringe collecting blood. However,such a device will not arrest the adenosine metabolismthat occurs as the blood travels through a catheter.Because long catheters must be used to sample bloodfrom the human coronary sinus,the only practical wayto arrest the metabolism of adenosine is by injectingmetabolicinhibitorsat the tip of the sampling catheter.The utility of such a metabolic catheter has been vali-dated in vitro by Shryock et al. (10) and applied clini-cally in the present study. We built a model of themetaboliccatheterto ensure that miring of blood andstop solution occurs at its tip. Moreover, in all patients’studies, we used a flow rate at which adequate miTingoccurred (0.33 mLds). The use of this catheter mayexplain why this study demonstrates ischemia-inducedincreases in coronary sinus adenosine in the absence ofsystemicdipyridamole, even though no such changescould be detected in earlierstudies (3-6).

The coronary sinus adenosine concentrations reportedhere are lower than in studieswhere inhibitors of5’-nucleotidaae were not incorporated into the stop so-lution (Table 2). Hamm et al. (6) did not use an adeno-sine deaininase inhibitor, which may have contributedto the low concentrations of adenosine they detected.The low concentrations of coronary sinus adenosine wedetected here suggest that artifactual production ofadenosine is an important source of error in studiesdealing with whole human blood. Studies in which thehighest concentrations of coronary sinus adenosine werereported used stop solutions designed to prevent loss butnot production of adenosine. The stop solutions in thesestudies included dipyridamole (3,4, 13), EHNA (3,4), orMnC12 (13) to prevent cellular uptake of adenosineor toinhibitadenosine deaminase. However, for accurateassessment of adenosine concentrations in blood, onemust prevent adenosine formation. Because adenosinemight be formed from adeninenucleotidesreleased as aresultof platelet aggregation(14), Solleviet al. (4, 5)added indomethacin to prevent platelet aggregates.However, adenine nucleotides may also be derived fromeven slight lysis of erythrocytes, which may occur dur-ing collection of coronary sinus blood samples. The

Table 2. Coronary Sinus Adonosine ConcentrationCompared with Other StudIes

Adenoslns Relativecanon, gimol/L adsnosln.b Stop solutlon

0.106±0.049 0.8 -

0.132 ± 0.051 1.0 AOPCP,EDTA0.22±0.02 1.7 -

0.29 2.2 -

0.674 ± 0.333 5.1 -

a The only study in whichan adenosinedeaminase inhibitor was not used.b Ratio of adenoslne concentration reported In cited reference dMded by

that reported in thisstudy.#{176}Inhibitorsof artifactual adenosine production were in thestopsolution.

enzymes necessary to degrade adenine nucleotides toadenosine also are present in human blood (15). In thepresent study, we incorporated inhibitors of 5’-nucle-otidease (AOPCP and EDTA) into the stop solution.Although, to ensure patient safety, we could placeAOPCP only in the collectionsyringe, EDTA could beinfused at the tip of the metabolic catheter.

Previous studies with HPLC were hampered by thepresence in human plasma of several substances thatare eluted near adenosine. Although some early reportsquantified adenosine directly (16, 17), more recent in-vestigators have used peak-shift methodology (10, 12).In the latter technique, an HPLC fraction containingadenosine is collected, deaminated, and injected for asecond HPLC run; the amount of inosine is then quan-tified. McCann and Katholi (18) recentlydescribed pre-paring fluorescent etheno-derivatives of adenosine butstill rechromatographed the human blood samples toremove interfering substances. If adenosine is to be usedin patient care, simplification of methods for quantify-ing adenosine in human blood would be advantageous.We combined the use of boronate and C18 Sep-Pakprecolunms to partially purify and concentrate theadenosine in plasma before HPLC; thus, we could quan-tify the adenosine accurately as an overnight procedurewith a single isocratic HPLC step. Although boronate(4,17) and C18 Sep-Pak columns (7,19) have been usedindividually in assays of adenosine, their combined usewas not previously reported.

An outgrowth of this study may be the physiologicalassessment of coronary artery disease at the time ofcardiac catheterization. Clinical decisions regarding re-vascularization vs drug therapyarelargelyaffectedbythe percentage of stenosis revealed by coronary angiog-raphy, currently the “gold standard” by which thefunctional severity of coronary disease is evaluated.Many studies, however, document marked variability ininterpretations of the severity of a stenosis (20,21), andautopsy studies have demonstrated a lack of correlationwith arteriography (22). Thus new methods are neededfor better physiological assessment of coronary arterystenoses.

There are several limitations to the methodology wehave presented. First, we have not accounted for theeffect of pacing-induced changes in arterial adenosineconcentration or coronary sinus blood flow on the con-centration of coronary sinus adenosine. Because pacingis associated with increased coronary sinus blood flow(23), the increase in adenosine plasma concentration inthe coronary artery disease patients probably underes-timates the true change in the magnitude of adenosineproduction. A second limitation is that we cannot dis-tinguish between adenosine derived from ischemic my-ocardium and that from other sources such as thevascular endotheliuni (24, 25), blood cells (15), andnerves (26). Nonetheless, this study establishes adeno-sine as a marker of myocardial ischemia, regardless ofthe precise mechaniam by which it is released. A thirdlimitation is the inability of the metabolic catheter to

262 CLINICAL CHEMISTRY, Vol. 38, No. 2, 1992

sample regional myocardial adenosine production, espe-cially from the right coronary artery. This cathetersamples coronary sinus blood, a reflection of globalmyocardial adenosine release, whereas coronary arterydisease and ischemia affect the myocardium in a focalmanner. Adenosine released from nonischemic myocar-dium will, to a certain extent, be mixed with adenosinefrom ischemic myocardium and sampled together whenblood is drawn from the coronary sinus. Another limi-tation is a time delay for blood transit from the myocar-dial capillaries to the metabolic catheter in the coronarysinus, during which time metabolism of adenosine canoccur. This may explain why we cannot detect pacing-induced increases in coronary sinus adenosine in pa-tients with normal epicardial coronary arteries and whywe detect only a twofold increase in patients withcoronary artery disease, even though greater than six-fold increases were detected in interstitial myocardialadenosine concentration in animal models of ischemia(27). Finally, because of the small sample size-ninepatients with coronary artery disease and six patientswith normal epicardial coronary arteries-we must con-sider the results of this study to be only preliminary.

In summary, by using a double-lumen metabolic cath-eter, modified stop solution, and a simplified HPLCassay, we are able to detect an ischemia-induced twofoldincrease of coronary sinus adenosine in the absence ofsystemic dipyridamole. This is the first convincing dem-onstration that adenosine produced by ischemic myocar-dium in patients can escape from the interstitial space,cross the major metabolic endothelial barrier, and accu-mulate and be detected in coronary sinus blood.

WethankDrs.Rafael RubioandGeorge Belier for their advice,Jonquil Feldman for artwork, and the members of the cardiaccatheterization staff for their support. Dr. Luiz Belardinelli de-signed a prototype of the metabolic catheter usedin this study.Supportedby the Bayer Fund for Cardiovascular Research, Ster-ling Drug, New York, NY (M.D.F.);the American Heart Associa-tion, Virginia Affiliate, Grant-in-Aid (grant G880010) (M.D.F.);and the Cordis Corp.,Miami, FL, which supplied components tobuild themetaboliccatheter. Portionsof this workwere presentedin abstract form at the 39th Annual Scientific Session of theAmerican CollegeofCardiology,New Orleans, LA, March 18-22,1990.

References1. Berne RM. Cardiac nucleotides in hypoxia: possible role inregulation of coronary blood flow. Am J Physiol 1963;204:317-22.2. Fox AC, ReedGE, GlasamanE, Kaitman AJ, Silk BB. Releaseof adeno8ine from human hearts during angina induced by rapidatrial pacing. J Clin Invest 1974;53:1447-57.3. Edlund A, Berglund B, van Dome D, et al.’ Coronary flowregulation in patients with ischemic heart disease:release ofpurines and prostacyclin and the effect of inhibitors of proetaglan-din formation. Circulation 1985;71:1113-20.4. Sollevi A. Cardiovascular effects of adenosinein man; possible

clinical implications. Prog Neurobiol 1986;27:319-49.5. Sollevi A, Torssell L, Owall A, Edlund A, Lagerkranaer M.Levels and cardiovascular effects of adenosine in humans. In:Gerlach E, Becker BF, eds. Topics and perspectives in adenosineresearch. Berlin: Springer-Verlag, 1986:599-613.6. Hamm CW, Kupper W, Bredehorst R, Hilz H, Bleifeld W.Quantitation of coronary venous adenosine in patients: limitationsevaluated by radioimmunoassay. Cardiovasc Res 1988;22:236-43.7. Moser GH, Schrader J, Deussen A. Turnover of adenosineinplasma of human and dogblood. AmJ Phyaiol1989;256:C799-806.8. Nees S, Herzog V, Becker BF, Bock M, Des RosiersC, GerlachE. The coronary endothelium: a highly active metabolic barrier foradenosine.Basic Res Cardiol 1985;80:515-29.9. Bunag RI), Douglas CR, Imai S, Berne RM. Influence of apyrimidopyrimidine derivative on deamination of adenosine byblood. Circ Res 1964;15:83-8.10. Shryock JC, Boykin MT, Hill JA, Belardinelli L. A newmethod of sampling blood for measurement of plasma adenosine.Am J Physiol 1990;258:H1232-9.11. BatchelorGK. An introduction to fluid dynamics. Cambridge:Cambridge UniversityPress, 1987:255-63.12. Ontyd J, Schrader J. Measurementof adenosine,inosine, andhypoxanthine in human plasma. J Chroznatogr 1984;307:404-9.13. Haneda T, Ichihara K, Abiko Y, Onodera S.Release of aden-oaine and lactate from human hearts during atrial pacing inpatients with ischemic heart disease. Clin Cardiol 1989;12:76-82.14. Fukami MH, Holinsen H, Salganicoff L. Adenine nucleotidemetabolism of bloodplatelets. Part IX. Time course of secretionand changesin energy metabolismin thrombin-treated platelets.BiochimBiophysActa 1976;444:633-43.15. CoadeSB,Pearson JD. Metabolism of adenine nucleotidesinhuman blood.Circ Res1989;65:531-7.16. CapogrossiMC, HoldinessMB, Israili ZH. Determination ofadenosinein normal human plasma and serum by high-perfor-manceliquid chromatography. J Chromatogr 1982227:168-73.17. PfadenhauerEH, Tong SD. Determination of inosine andadenosine in human plasma using high-performance liquid chro-matography and a boronate affinity gel. J Chromatogr 1979;162:585-90.18. McCann WP, Katholi RE. Control of artifacts in plasmaadenosinedeterminations.ProcSocExp BiolMed 1990;194:314-9.19. Deussen A, Borst M, Kroll K, Schrader J. Formation ofS-adenosylhomocysteine in the heart. II. A sensitive indexforregional myocardial underperfusion. Circ Res 1988;63:250-81.20. Do Rouen TA, Muray JA, Owen W. Variability in the analysisof coronary arteriograms. Circulation 1977;55:324-8.21. Zir LM, Miller SW,Dinsmore RE,Gilbert JS,Hawthorne JW.Intraobserver variability in coronary angiography. Circulation1976;53:627-32.22. Arnett EN, Isner JM, Redwood DR, et al. Coronary arterynarrowing in coronary artery disease:comparison of cineangio-graphic and necropsy findings. Ann Intern Med 1979;91:350.-6.23. MarcusML. The coronary circulation in health anddisease.New York: McGraw Hill, 1983:65-92.24. Nees 5, Gerlach E. Adenine nucleotideand adenosine metab-olism in cultured coronaryendothelialcells:formation and releaseof adenine compounds and possiblefunctional implications. In:Berne RM, Rail TW, Rubio R, ode.Regulatory function of adeno-sine.The Hague: MartinusfNjjhoff, 1983:347-55.25. Sparks HV, BardenheuerH. Regulation of adenosineforma-tion by the heart. Circ Res 1986;58:193-201.26. Su C. Purinergic neurotransmissionand neuromodulation.Annu Rev Pharmacol Toxicol 1983;23:397-411.27. Berne RM, RubioR, Duling BR. Vasoactive substances affect-ing the coronary circulation.In: RossRS, Hoffman F, eds. Myocar-dial ischemia. Leiden:AW Sijthoff, 1971:28-43.