Cocaine-Induced Erythrocytosis and Increase in von Willebrand Factor:  Evidence for Drug-Related Blood Doping and Prothrombotic Effects FREE

AS THE MOST frequently used illicit drug among patients presenting with chest pain to hospital emergency departments,¹ cocaine may trigger acute cerebral and myocardial ischemia or infarction.^2- 3 Beyond vasoconstrictive effects shown in human coronary and cerebral circulations,^4- 6 cocaine may also induce a blood doping effect⁷ and alter plasma constituents that influence thrombogenicity.⁸ We therefore measured changes in hematologic parameters in human subjects, before and sequentially after the administration of moderate doses of intranasal and intravenous cocaine sufficient to produce significant changes in blood pressure and heart rate. Hemostatic factors, including von Willebrand factor (vWF), fibrinolytic activity, fibrinogen, plasminogen activator inhibitor (PAI-1) antigen, and tissue-type plasminogen activator (TPA) antigen, were similarly measured after intravenous cocaine or saline placebo to assess changes in thrombotic and fibrinolytic potential. Cardiac troponin subunits T (cTnT) and I (cTnI) were sequentially measured to assess silent injury to the myocardium.

Hemoglobin level (P=.002), hematocrit (P=.01), and red blood cell (RBC) counts (P=.04) significantly increased, from 4% to 6% over baseline following intranasal (n=14) and intravenous (n=7) administration of cocaine in doses of 0.9 mg/kg and 0.4 mg/kg, respectively (Figure 1). Significant elevations in hemoglobin and hematocrit occurred 10 minutes after both routes of cocaine administration and in RBC counts coinciding with peak plasma cocaine concentrations of 476.5±57.8 nmol/L (144.4±17.5 ng/mL) and 774.2±56.8 nmol/L (234.6±17.2 ng/mL) after intranasal and intravenous routes, respectively. There were no changes in white blood cell or platelet counts after either route of cocaine administration. Cardiovascular parameters, including heart rate (P=.001) and systolic (P=.01) and diastolic blood pressures (P=.03), increased significantly following both routes of cocaine administration without cardiorespiratory symptoms or electrocardiographic changes of ischemia.

There was a significant increase (P=.03) in vWF after intravenous administration of cocaine hydrochloride in a dose of 0.4 mg/kg (n=12), with no change after 0.2 mg/kg (n=3) or saline placebo (n=6) (shown as change in mean percent activity±SEM in Figure 2). The increase in vWF peaked at 40%±18.5% over baseline and lasted from 30 to 240 minutes. Changes were similar among 6 male and 6 female subjects matched for age and body mass index, although baseline values were slightly higher in women. Fibrinolytic activity, fibrinogen, PAI-1 antigen, and TPA antigen showed no significant changes, although a circadian increase in fibrinolytic activity and decrease in PAI-1 were uniformly observed. Levels of cTnT remained within normal limits (<0.10 µg/L) and cTnI was undetectable (<0.35 µg/L) up to 4 hours after cocaine administration, with no changes in creatinine or blood urea nitrogen levels.

Subjects aged 21 to 35 years who met criteria for long-term cocaine abuse, as described in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition,⁹ provided informed consent for participation in studies in the Alcohol and Drug Abuse Research Center of McLean Hospital, Belmont, Mass, under protocols approved by the hospital's institutional review board and funded by the National Institute on Drug Abuse. Results of comprehensive medical and laboratory examinations were normal, and urine screens for substances of abuse were negative before participation using urine screening kits (Triage; Biosite Diagnostics, San Diego, Calif).

Double-blind studies were conducted with continuous noninvasive cardiovascular monitoring under direct supervision by a physician, with subjects in a resting semirecumbent position. Cocaine hydrochloride (Mallincrodt, St Louis, Mo) in powder form was used for nasal insufflation or dissolved in sterile water for intravenous injection after processing through a 0.22-µm millipore filter with negative results of testing for pyrogens using a limulus lysate test (Limulus Amebocyte Lysate assay; Whittaker Bioproducts, Walkersville, Md). Subjects received cocaine hydrochloride, either intranasally in a dose of 0.9 mg/kg using a modified snort-stick device,¹⁰ or intravenously in doses of 0.2 or 0.4 mg/kg in a 1-mL bolus for 1 minute, or saline placebo.

Blood samples were drawn through an indwelling intravenous catheter (Kowarski-Cormed Thrombo-resistant Blood Withdrawal Butterfly Needle and Tubing Set; DakMed, Inc, Buffalo, NY) in the opposite arm. Hematologic parameters were measured at baseline and at 10, 20, 30, 40, and 60 minutes in tubes with 15% EDTA solution and additionally at 2, 4, 8, 12, 16, 20, 120, 180, and 240 minutes for plasma cocaine levels using tubes containing sodium fluoride. Hemostatic factors were assayed at baseline and at 30, 60, 120, 180, and 240 minutes after intravenous cocaine using tubes containing sodium citrate. Blood samples for hemostatic factors and cocaine were centrifuged immediately, and plasma was frozen at −70°C until analyses were performed.

Hematologic parameters were determined using a blood cell analyzer (Miles/Bayer H2 System; SmithKline Beecham Clinical Laboratories, Waltham, Mass). Fibrinolytic activity was assayed using a fibrin plate assay, fibrinogen levels were determined using the Clauss method, commercially available immunoassay kits (Biopool International Inc, Ventura, Calif) were used to measure PAI-1 and TPA antigens, and vWF was measured using an immunoassay method of Penny et al.¹¹ Plasma cocaine levels were measured in duplicate using a solid-phase extraction method with gas chromatography and mass spectroscopy (Hewlett-Packard 5890 Series II; Hewlett-Packard, Palo Alto, Calif).¹² Serum creatinine and blood urea nitrogen were sequentially measured using a commercially available autoanalyzer (Beckman Astra 8; Beckman, Brea, Calif). Cardiac TnI was determined using an immunoassay analyzer (Elecsys 1010 Analyzer; Roche Boehringer Mannheim, Indianapolis, Ind) using Enzymun Troponin T reagents (Roche Boehringer Mannheim). Cardiac TnI was determined using an immunoassay analyzer (Stratus II Analyzer; Dade Behring, Newark, Del). Hematologic parameters, hemostatic factors, and plasma cocaine and cardiovascular measurements were analyzed using repeated-measures analysis of variance. If significant main effects were detected, 1-way analysis of variance was performed to identify the time points that differed significantly.

Although the practice of chewing coca leaf (Erythroxylon coca) spans centuries—as illustrated by the Andean Indians who measure distance across mountains by the number of "cochitas" required for a given journey¹³—the toxic potential of cocaine has emerged in the modern era, with rapid systemic absorption after smoking, nasal insufflation, and intravenous injection.^14- 15 Since cocaine has been shown to alter hematologic parameters in animals and humans,^7,16- 17 we investigated sequential changes in complete blood cell counts in human subjects following moderate intranasal and intravenous doses of cocaine sufficient to produce significant changes in blood pressure and heart rate. The increase in hemoglobin levels, hematocrit, and RBC counts of 4% to 6% over baseline after both routes of cocaine administration was quantitatively similar to infusion of 2 units of packed RBCs,¹⁸ use of erythropoietin every other day for 6 weeks in doses of 20 U/kg,¹⁹ or chewing coca leaf during exercise.²⁰

Cocaine administration has been shown to induce splenic constriction in humans, with a 20% reduction in volume, as assessed by magnetic resonance imaging, which is temporally concordant with altered hematologic parameters.²¹ This constrictive effect contributes to a rapid expansion of the circulating RBC pool, in contrast to gradual splenic emptying during exercise,²² and may account for reported complications such as cocaine-induced intracapsular bleeding²³ and infarction.²⁴ Serum creatinine and blood urea nitrogen levels remained unchanged for 4 hours in our subjects, suggesting a true erythrocytosis, which is consistent with the finding that chewing of coca leaf attenuated dehydration during exercise.²⁵ Although we cannot exclude a contribution of plasma volume contraction in the absence of direct blood volume measurements, the resultant erythrocytosis is similar to other blood doping effects such as RBC transfusion¹⁸ or long-term use of erythropoietin,¹⁹ which are risk factors for intravascular thrombosis during exercise.^26- 27 Changes in whole blood viscosity from cocaine may contribute to exertional collapse and cardiovascular events,²⁸ as reported with sickle cell trait,^29- 30 although we did not directly measure changes in viscosity.

Because cocaine-induced coronary vasospasm^4- 5 and a decrease in cerebral blood flow³¹ may trigger ischemic events as a prelude to thrombosis, we assessed the effects of cocaine on hemostatic parameters related to thrombotic and fibrinolytic potential. There was a significant increase (P=.03) in vWF from 30 to 240 minutes, peaking at 40%±18.5% over baseline following intravenous cocaine hydrochloride in a dose of 0.4 mg/kg (n=12), using the same assay for vWF as shown to predict recurrence of myocardial infarction in coronary heart disease.³²

Cocaine-induced coronary vasoconstriction may disrupt unstable atherosclerotic plaques,³³ causing release of vWF from damaged vascular endothelium. This process may be mediated by stimulants such as catacolamines, thrombin, vasopressin,³⁴ and cocaine, including production of high-molecular-weight multimers of vWF,³⁵ which were not measured in our study. Cocaine may promote the adhesion and aggregation of platelets that release vWF from α granules^36- 37 to form bridges between exposed collagen fibrils and glycoprotein Ib/IX receptors on platelets.³⁸ Pharmacokinetic analysis of plasma cocaine and adrenocorticotrophic hormone suggests that the increase in vWF has a direct effect on vascular endothelium³⁹ similar to the release of immunoreactive endothelin.⁴⁰

Fibrinolytic activity, fibrinogen, TPA antigen, and PAI-1 antigen were unchanged after cocaine administration, indicating a lack of a compensatory rise in endogenous fibrinolysis, which has been shown to accompany exercise-enhanced levels of vWF.^41- 43 Such an imbalance in hemostatic factors, occurring as a consequence of cocaine, has been observed as an independent risk factor for the development of coronary artery disease⁴⁴ and acute ischemic cardiovascular events.^45- 47

Recent evidence supports the improved detection of ischemic myocardial injury using cTnT and cTnI.^48- 50 The diagnostic utility of cardiospecific troponins for myocardial injury has been demonstrated in patients with unstable coronary artery disease^51- 53 and in cocaine-associated chest pain.^54- 55 Levels of cTnT remained within normal limits (<0.10 µg/L), and cTnI was undetectable (<0.35 µg/L) up to 4 hours after cocaine in these asymptomatic subjects, providing biochemical evidence against silent myocardial injury. Recent reports that an early rise in vWF predicts adverse outcome in patients with unstable angina and that enoxaprin has protective effects by reducing its release^56- 57 suggest a valve for low-molecular-weight heparin in cocaine-induced myocardial ischemia in addition to conventional treatments.⁵⁸

Based on the small sample size of our study, the hypothesis that cocaine promotes thrombogenesis mediated by altered blood viscosity and an increase in vWF should be regarded as preliminary. Areas for further study regarding cocaine include the measurement of changes in whole blood viscosity, in vitro effects on platelets, vascular endothelium, metabolism of vWF multimers,³⁵ and assessment of drug tolerance.⁵⁹ Clinical studies of enoxaprin in patients with cocaine-associated cerebral or myocardial ischemia may be of value. Cocaine-induced changes in blood viscosity from erythrocytosis, and a selective increase in vWF following vasoconstriction, may contribute to the abrupt and transient increase in risk for acute MI associated with its use.²

Accepted for publication January 9, 1999.

This study was supported in part by grants DA09448, DA04059, DA00064, DA00329, DA00343, DA03994, and DA10757 from the National Institute on Drug Abuse, National Institutes of Health, Bethesda, Md.

We thank Susan Currier for her excellent assistance in preparation of the manuscript.

Reprints: Arthur J. Siegel, MD, Department of Internal Medicine, McLean Hospital, 115 Mill St, Belmont, MA 02478 (e-mail: AJSIEGEL@mclean.harvard.edu).