Secondary Logo

Journal Logo

Clinical science

Absorption and Elimination Kinetics of Zidovudine in the Cerebrospinal Fluid in HIV-1-Infected Patients

Rolinski, Boris*†; Bogner, Johannes R.*; Sadri, Ifna*; Wintergerst, Uwe; Goebel, Frank D.*

Author Information
Journal of Acquired Immune Deficiency Syndromes and Human Retrovirology: July 1, 1997 - Volume 15 - Issue 3 - p 192-197
  • Free


Neurologic disease is a common feature in the course of HIV infection. In addition to opportunistic infections or neoplasms of the central nervous system (CNS), it includes HIV encephalopathy (also termed AIDS-dementia complex), HIV encephalitis, and peripheral neuropathy. The improvement and prevention of such neurologic impairment is one of the most striking therapeutic benefits of antiretroviral treatment with zidovudine(ZDV)(1-10). Although low-dose therapy with ZDV (600 mg daily) had been shown to be as effective in reducing mortality and morbidity in HIV-infected patients as the previously used treatment dose (1200 mg daily)(11), there is much debate on the optimal dose of ZDV in the treatment and prophylaxis of HIV-related neurologic disorders(8-10,12).

Until this study, data on ZDV concentrations in cerebrospinal fluid (CSF) in humans were sparse and restricted to single-point measurements, usually in paired CSF and blood samples(13-15). In one study(16), ZDV levels in CSF were evaluated in a larger number of patients and with respect to the time of the last drug intake. In all of these investigations, an extremely high variability in the CSF-plasma ratio was observed. In addition, the CSF-plasma ratio seems to increase with time after drug intake(5,16). However, from these investigations, no conclusions could be drawn regarding the time course of ZDV absorption and elimination in CSF.

To our knowledge, there are no data available on pharmacokinetic parameters of this drug or its major metabolite, glucuronidated ZDV (GZDV) with respect to CSF in humans. Nonetheless, knowledge about peak concentration, half-life, or the time interval of drug level above the 50% inhibitory concentration (IC50)(17) would have direct impact on dosage recommendations and therapy decisions. Therefore, the objective of this study was to investigate the pharmacokinetics of ZDV and GZDV in CSF and plasma in six HIV-infected patients.



All patients were adult male Caucasians (age range, 23-49 years), who were HIV-positive as proven by ELISA and Western blot analysis and had AIDS (Centers for Disease Control and Prevention [CDC] clinical category C3). CD4 counts ranged from 35 to 520 cells/µl. All patients were hospitalized and underwent lumbar puncture for diagnostic purposes; none were receiving ZDV therapy at the time of the study. Clinical diagnoses for puncture included suspicion of cytomegalovirus(CMV) polyradiculitis, extrapulmonary tuberculosis, toxoplasmic encephalitis, cryptococcal meningitis, and cerebral lymphoma. Patients participating in this study were volunteers and gave written informed consent after careful explanation of the procedure. There were no objections to the study by the institutional ethics committee. Patients with psychiatric illness, acute disorientation, or otherwise suspected to be limited in their intellectual perception were excluded from the study. Patients with hemoglobin levels <10 g/dl, neutrophil counts <1000 cells/µl, or platelet count <50,000 cells/µl were also excluded.

Drug Administration

ZDV was administered as a single intravenous (IV) dose of 2.5 mg/kg body weight. According to the manufacturer's recommendation, the appropriate amount of drug was diluted in 100 ml of 5% glucose solution and infused at a constant rate over 1 hour by a peripheral venous catheter.

Collection of CSF and Plasma Samples

Lumbar puncture was performed with a catheter set (Hell, Diespeck, Germany) especially designed for the collection of CSF, derived from catheters used for continuous intraspinal anesthesia and consisting of an 18-gauge needle and a 22-gauge catheter with an inner mandrine. After the lumbar puncture was performed and ∼5 ml CSF was collected through the needle for diagnostic investigations, the catheter was inserted into the needle and placed approximately 5 cm into the intraspinal space. The needle was carefully retracted and the mandrine removed. The end of the catheter was closed and taped to the patient's body.

CSF was collected with a syringe under sterile conditions at baseline, at the end of the ZDV administration, and 1,2,3,4,5, and 6 hours later. To account for the deadspace of the catheter, the first 0.2 ml was discarded, then 0.5 ml of CSF for analysis was removed. Simultaneously, 2-ml heparinized blood samples were drawn before and 1, 2, 3, 4, 5, and 6 hours after the infusion from the indwelling peripheral catheter. The sample at the end of the infusion was taken by venipuncture from the contralateral cubital vein. Plasma was separated by centrifugation, and plasma and CSF samples were frozen at −20°C within 30 minutes of collection.

Drug Determination

ZDV and GZDV concentrations in plasma and CSF were determined by high-performance liquid chromatography (HPLC). A 0.5-ml plasma or CSF sample was extracted on a C18 500-mg solid-phase extraction column (Bond Elut. Analytichem Int., Frankfurt, Germany)(13). Separation was performed on a 250- × 4.6-mm Superspher 4-µm RP 18 endcapped column (Merck, Darmstadt, Germany) with an isocratic eluent of 100-mmol/l ammonium phosphate. pH 3.3, containing 14%(v/v) acetonitrile(18). Flow rate was set to 1 ml/minute, and column effluent was monitored by ultraviolet (UV) absorbance at 267 nm. Concentrations of ZDV and GZDV were calculated by peak height, referring to external standards and an internal standard (BW A22U). In this system, retention times for GZDV, BW A22U, and ZDV were approximately 5.3, 7.9 and 8.5 minutes, respectively. The method was linear for both GZDV and ZDV from 0.1 to 100 µmol/l. Limits of quantitation (lowest calibrated standard) were 0.1 µmol/l for ZDV and 0.2 µmol/l for GZDV. The overall intra- and interassay coefficients of variation for extracted plasma samples (spiked to 10 µmol/l) were 4.4% and 5.9% for ZDV and 5.3% and 9.2% for GZDV, respectively. CSF calibration curves were made up with pooled CSF left over from the clinical chemistry laboratory.

All chemicals used were of analytical grade or better. GZDV was purchased from Sigma (Deisenhofen, Germany). Pure ZDV and the internal standard BWA22U were a gift of G. Land (Wellcome Research Laboratories, Beckenham, U.K.).

Pharmacokinetic Analysis

The following pharmacokinetic parameters were determined using a noncompartmental model for plasma, CSF, or both: Cmax, time to peak concentration (tmax). t1/2, AUC, total body clearance (Cl) and the total apparent volume of distribution (Vd). The AUC was calculated using the linear-trapezoideal rule and approximation of the last data point without (AUCd) or with (AUC∼) extrapolation to infinity(Clast/β); t1/2 was calculated by least square linear regression analysis after visual inspection of the curve, and generally the last 5 data points were used. Cl was calculated by the formula Cl = dose/AUC and Vd by the formula Vd = dose/AUC ⋅ (ln2/t1/2).

Calculations were done using the pharmacokinetic software package Topfit 2.0(19). Correlations were calculated by Pearson's correlation coefficient using the SPSS software package (SPSS/PC 4.0, Chicago, U.S.A.).

Blood-CSF Barrier

In the CSF and serum samples taken for diagnostic purposes, albumin and IgG were analyzed by routine clinical chemistry methods. In addition, cell count and oligoclonal bands were determined. CSF-serum ratio was calculated by (concentration CSF/concentration serum) × 1000. The blood-CSF barrier was assumed to be intact if the CSF-serum ratio of albumin was <7.4. A CSF-serum ratio of IgG >4.2 indicated intracerebral IgG production.


Procedure of the Lumbar Puncture

The lumbar puncture and the insertion of the intraspinal catheter was well tolerated in all patients. Once in place, the indwelling catheter did not cause any pain or discomfort to the patients and was removed without complications at the end of the study period. Two patients complained of mild headache the day after the investigation that resolved within 2 days. We did not encounter any clinical signs of infectious complications following the lumbar puncture. In the first two patients, the catheter tip was investigated for bacterial growth with negative results.

With the CSF collection device used in this investigation, sufficient amounts of CSF could easily be gathered. The flow rate through the 22-gauge catheter was high enough to allow the collection of the CSF sample within 30 seconds.

ZDV infusion was well tolerated in all patients without any side effects.

Blood-CSF Barrier

Mean albumin and IgG concentrations in CSF were 154 ± 106 mg/l and 108 ± 92 mg/l, respectively. The mean CSF-serum ratio for albumin was 5.6 ± 2.9 (range, 1.1-8.8) and thus within the normal range. In contrast, the mean CSF-serum ratio for IgG was 7.1 ± 6.0 (range, 1.5-15.3) indicating an autochthonous production of IgG in most of the patients. Accordingly, oligoclonal bands could be detected in four of the six patients. Cell count was elevated in two patients. These data indicate a mild inflammatory involvement of the CNS in five of six patients. Blood-CSF barrier, however, was normal in four of six patients and was only slightly impaired in the remaining two patients.

Pharmacokinetic Analysis

Mean Cmax in plasma at the end of the infusion was 6.7 ± 2.7 µmol/l (range, 4-10.7 µmol/l). The mean AUCd in plasma was 448 ± 213 µmol ⋅ minutes/l. Extrapolating the AUC to infinity (AUC∼) increased the AUCd in plasma on average only by 6% (data not shown).

After the infusion ZDV was cleared rapidly from plasma, with a mean t½ of 75.5 ± 4.9 minutes as seen inFigure 1. Mean Cl from plasma was 0.024 ± 0.012 l/kg⋅ minute, and mean Vd was 2.6 ± 1.1 l/kg.

In CSF, Cmax was reached in all patients 1 hour after the end of the infusion (Fig. 1). Mean Cmax in CSF was 1.3 ± 1.2 µmol/l (range, 0.4-3.6 µmol/l). Cmax in CSF were 17% ± 9% of the corresponding Cmax in plasma. ZDV was cleared from CSF with a mean t½ of 187.6 ± 69.3 minutes (Fig. 1). Mean AUCd in CSF was 257 ± 162 µmol ⋅ minutes/l and 57% ± 23% of that of plasma. Extrapolating the AUC to infinity (AUC∼) increased the AUCd on average by 29% ± 13%. Thus, mean AUC∼ in CSF was 358 ± 200 µmol ⋅ minutes/l and 75% ± 26% of that of plasma, indicating a high transfer of drug across the blood-CSF barrier. The detailed results of the pharmacokinetic analysis are presented in Table 1.

There was a significant correlation between Cmax in plasma and CSF (r = 0.88, p = .009), between AUCd in plasma and CSF (r = 0.86, p = .009), and between AUC∼ in plasma and CSF (r = 0.89, p = .014)(Fig. 2), indicating a concentration-dependent transport of ZDV into the CSF. The CSF-plasma ratio for albumin or IgG did not correlate with either Cmax, AUCd, or AUC∼.

Assuming a 12-hour dosing interval, trough levels in plasma and CSF were calculated by extrapolation to be 0.006 ± 0.005 µmol/l and 0.090 ± 0.065 µmol/l, respectively. The mean time interval from the end of the infusion until the drug concentration fell under the IC50 of susceptible HIV strains (0.05 µmol/l)(17) was 463 ± 82 minutes in plasma but 823 ± 303 minutes in CSF.

Because of the longer half-life of ZDV in CSF compared with plasma, ZDV concentrations in CSF exceeded those in plasma on average after 180 ± 45 minutes. Thus, the CSF-plasma ratio of ZDV was strongly time-dependent. Mean CSF-plasma ratio at the end of the infusion was 0.1 ± 0.1 and increased to 2.1 ± 0.4 after 6 hours (Fig. 3).

FIG. 1.
FIG. 1.:
Time concentration curve of zidovudine (ZDV) in plasma(•) and CSF (▾) in six HIV-infected patients after intravenous drug administration. Mean ± standard deviation is shown. The rapid clearance from plasma and the slow penetration into and elimination from CSF is clearly visible.
Pharmacokinetic parameters of zidovudine in cerebrospinal fluid and plasma

Time concentration curves of GZDV in plasma paralleled those of ZDV, with values ∼20% to 30% higher than the corresponding concentrations of the parent drug. In CSF, no GZDV was detected (data not shown).


Until this study, pharmacokinetic data on ZDV in CSF in humans had been established only by extrapolating single-point measurements in paired CSF-plasma samples(5,12-16). In our study, repeated CSF samples were taken by means of an intraspinal catheter. Thus, we were able to determine the pharmacokinetics of ZDV in CSF by sequential measurements of drug concentrations over a period of 6 hours.

Lumbar puncture and insertion of the catheter was well tolerated in all patients. Moreover, no adverse events other than mild postpunctural headache were noted in the patients investigated. Thus, this study also demonstrates the feasibility and safety of an intraspinal catheter for sequential CSF collection over at least 6 hours.

FIG. 2.
FIG. 2.:
Correlation between mean areas under the concentration time curve with extrapolation to infinity (AUC∼) in plasma and CSF in the six patients studied (r = 0.89, p =.014). Slope of the regression line is 0.98 and that of the x-intercept is 149.5µmol ⋅ min/l.

The blood-CSF barrier was intact in four patients and only slightly impaired in the remaining two. No influence of the integrity of the blood-CSF barrier on pharmacokinetic parameters of ZDV in CSF was observed. This is in accordance with Burger et al.(16), who could not demonstrate any influence of the CSF-plasma ratio of albumin or total protein on ZDV concentration in CSF or on the CSF-plasma ratio of ZDV, even in case of more severe disturbances. In addition, the degree of inflammatory involvement of the CNS had no measurable influence on the pharmacokinetics of ZDV in CSF or plasma.

The pharmacokinetics of ZDV in plasma after IV infusion were in good agreement with previous studies(1,13,20). ZDV uptake in CSF was slow, reaching Cmax 2 hours after the start of the infusion. The elimination of ZDV from CSF was also slow, with a t½ three times longer than that in plasma. Usually, the passage of drugs into or from the CSF is restricted by the blood-CSF barrier. Endothelial cells of brain capillaries are characterized by the absence of intercellular pores and pinocytic vesicles(21). Together with pericapillary glial cells, they restrict the aqueous flow across the membrane, limiting the transport of more hydrophilic drugs to the CSF. Only highly lipid-soluble compounds are able to cross the blood-CSF barrier freely(21). ZDV is a small, lipophilic compound that is uncharged at physiologic pH(13); therefore, it is expected to cross biological membranes by passive diffusion and to reach partition equilibrium quickly. For other compartments of the body, a rapid transport from and equilibrium to plasma could be shown. For example. ZDV enters erythrocytes by simple passive diffusion(22). Moreover, it is rapidly absorbed from the gastrointestinal tract, and concentrations of ZDV in saliva parallel those in plasma with nearly identical time to reach Cmax and elimination times in both compartments(23). However, one can speculate that the unique structure of the blood-CSF barrier restricts the nonfacilitated diffusion, leading to the slow penetration to and elimination from CSF of ZDV.

FIG. 3.
FIG. 3.:
CSF-plasma ratio of zidovudine (ZDV) concentrations in relation to the time after drug administration. Mean ± standard deviation is shown.

GZDV was not detected in CSF. Because of its higher polar charge, it is not able to cross the blood-CSF barrier. Moreover, as expected, there is no evidence for glucuronidation metabolism within the CNS.

In previous studies, the CSF-plasma ratio of ZDV calculated from single-point measurements was used to estimate the amount of drug entering the CSF(13-15); however, high variability of this parameter was observed. Considering the time after drug administration, a time-dependency of the CSF-plasma ratio was suggested(5,16). Our study proves this dependency. Therefore, the CSF-plasma ratio seems unsuitable to estimate the availability of ZDV in CSF.

The CSF-plasma ratio of the AUC seems to be the most precise estimate for the availability of ZDV in CSF. However, for practical reasons it is not suitable for routine determination in patients. Thus, the ZDV concentration in CSF at the end of the dosing interval might be the most suitable sampling time to monitor patient treatment. However, ZDV concentrations in CSF at the end of a 12-hour dosing interval are generally low, usually within the quantitation limit of HPLC methods. Thus, the feasibility of trough level determinations in CSF must be taken into account. Moreover, as ZDV is slowly eliminated from the CSF, it will probably accumulate in this compartment, and the CSF-plasma ratio after repeated dosing remains to be established. After a single dose, we found the AUC in CSF to be 75% of that of plasma, indicating a high overall transfer of drug into the CSF. In contrast, under a steady-state situation (i.e., continuous IV infusion), Pizzo et al.(2) observed a mean CSF-plasma ratio of ZDV concentrations of 0.24, indicating a 24% availability of drug in the CSF. There is no obvious explanation for this difference at the moment. However, continuous IV infusion results in constant but low plasma levels. As the transfer of ZDV across the blood-CSF barrier seems to be concentration dependent, one might speculate that higher plasma concentrations and a high concentration gradient are required to ensure ZDV transfer. In addition, the study population consisted of children: thus, results might not be directly comparable to adults.

Burger et al.(16) could not delineate a relation between plasma and CSF levels of ZDV or between the ZDV dose and drug concentrations in CSF. However, the authors admit that this finding might be confounded by methodologic problems arising from very heterogenous conditions. In contrast, we found a highly significant correlation between Cmax or AUC in plasma and CSF, indicating that the transfer of ZDV across the blood-CSF barrier is concentration-dependent and that the amount of drug available in CSF is in relation to the amount of drug present in plasma. It has been shown that Cmax and AUC in plasma increase with the ZDV dose administered over a wide dosage range(13,21). Therefore, our data clearly indicate that ZDV levels in the CSF depend on the dosage regimen. This is in accordance with data from Balis et al.(12), who studied ZDV under continuous IV infusion at different dose levels.

In our investigation, mean ZDV concentrations in CSF were expected to be above the IC50 for susceptible HIV strains(17) for more than 12 hours. The IV dose we used(2.5 mg/kg) equals an oral dose of about 3.5 mg/kg, assuming a bioavailability of 60%(13). In adults, this corresponds roughly to the most widely used single oral dose of 250 mg. If ZDV is administered as a 250-mg twice-per-day regimen, drug concentrations in CSF should exceed the IC50 during the whole time of therapy. However, because of the high variability in pharmacokinetic parameters after oral dosage and individual differences among patients, this conclusion may not be valid. In addition, although the penetration of a drug to the CSF may be indicative of its penetration to the brain tissue, where its site of action is believed, this is not proven for ZDV in humans.

ZDV has a distinct pharmacokinetic profile in CSF compared with plasma or other compartments of the body. Although peak concentrations in CSF are much smaller than in plasma, its overall bioavailability as indicated by the AUC is much higher than previously believed.

Acknowledgments: The authors thank the patients participating in this study. A. Vielhauer and B. Röttger are gratefully acknowledged for expert technical assistance. This study was supported in part by grant FKZ BGA III-002-89/FVP from the Bundesministerium für Forschung und Technologie.


1. Yarchoan R, Berg G, Brouwers P, et al. Response of human immunodeficiency virus-associated neurological disease to 3'-azido-3'-deoxythymidine. Lancet 1987;1:132-5.
2. Pizzo PA, Eddy J. Falloon J, et al. Effect of continuous intravenous infusion of zidovudine (AZT) in children with symptomatic HIV infection [see comments]. N Engl J Med 1988;319:889-96.
3. Schmitt FA, Bigley JW, McKinnis R. Logue PE, Evans RW, Drucker JL. Neuropsychological outcome of zidovudine (AZT) treatment of patients with AIDS and AIDS-related complex. N Engl J Med 1988;319:1573-8.
4. Portegies P, de Gans J, Lange JM, et al. Declining incidence of AIDS dementia complex after introduction of zidovudine treatment[see comments]. BMJ 1989;299:819-21, erratum BMJ 1989;299:1141.
5. Tartaglione TA, Collier AC, Coombs RW, et al. Acquired immunodeficiency syndrome: cerebrospinal fluid findings in patients before and during long-term oral zidovudine therapy. Arch Neurol 1991;48:695-9, erratum Arch Neurol 1991;48:1238.
6. Gray F, Geny C, Dournon E, Fenelon F, Lionnet F, Gherardi R. Neuropathological evidence that zidovudine reduces incidence of HIV infection of brain [letter] [see comments]. Lancet 1991;337:852-3.
7. Vago L, Castagna A, Lazzarin A, Trabattoni G, Cinque P, Costanzi G. Reduced frequency of HIV-induced brain lesions in AIDS patients treated with zidovudine. J Acquir Immune Defic Syndr 1993;6:42-5.
8. Sidtis JJ, Gatsonis C, Price RW, et al. Zidovudine treatment of the AIDS dementia complex: results of a placebo-controlled trial. AIDS Clinical Trials Group. Ann Neurol 1993;33:343-9.
9. Tozzi V, Narciso P, Galgani S, et al. Effects of zidovudine in 30 patients with mild to end-stage AIDS dementia complex. AIDS 1993;7:683-92.
10. Portegies P, Enting RH, de Gans J, et al. Presentation and course of AIDS dementia complex: 10 years of follow-up in Amsterdam, The Netherlands. AIDS 1993;7:669-75.
11. Fischl MA, Parker CB, Pettinelli C, et al. A randomized controlled trial of a reduced daily dose of zidovudine in patients with the acquired immunodeficiency syndrome. The AIDS Clinical Trials Group. N Engl J Med 1990;323:1009-14.
12. Balis FM, Pizzo PA, Murphy RF, et al. The pharmacokinetics of zidovudine administered by continuous infusion in children. Ann Intern Med 1989;110:279-85.
13. Klecker RW Jr, Collins JM, Yarchoan R, et al. Plasma and cerebrospinal fluid pharmacokinetics of 3'-azido-3'-deoxythymidine: a novel pyrimidine analog with potential application for the treatment of patients with AIDS and related diseases. Clin Pharmacol Ther 1987;41:407-12.
14. Surbone A, Yarchoan R, McAtee N, et al. Treatment of the acquired immunodeficiency syndrome (AIDS) and AIDS-related complex with a regimen of 3'-azido-2',3'-dideoxythymidine(azidothymidine or zidovudine) and acyclovir: a pilot study. Ann Intern Med 1988;108:534-40.
15. Yarchoan R, Thomas RV, Grafman J, et al. Long-term administration of 3'-azido-2',3'-dideoxythymidine to patients with AIDS-related neurological disease. Ann Neurol 1988;23(Suppl):S82-7.
16. Burger DM, Kraaijeveld CL, Meenhorst PL, et al. Penetration of zidovudine into the cerebrospinal fluid of patients infected with HIV. AIDS 1993;7:1581-7.
17. Larder BA, Darby G, Richman DD. HIV with reduced sensitivity to zidovudine (AZT) isolated during prolonged therapy. Science 1989;243:1731—.
18. Good SS, Reynoids DJ, De Miranda P. Simultaneous quantification of zidovudine and its glucuronide in serum by high-performance liquid chromatography. J Chromatogr 1988;431:123-33.
19. Heinzel G, Woloszczak R, Thomann P. Topfit 2.0: pharmacokinetic and pharmacodynamic data analysis system for the PC. Stuttgart, Germany: Gustav Fischer Verlag, 1993:25-66.
20. Blum MR, Liao SH, Good SS, de Miranda P. Pharmacokinetics and bioavailability of zidovudine in humans. Am J Med 1988;85:189-94.
21. Goodman LS, Gilman A, eds. The pharmacological basis of therapeutics. New York: Macmillan, 1990:10-11.
22. Zimmermann TP, Mahony WB, Prus KL. 3'-Azido-3'-deoxythymidine: an unusual nucleoside analogue that permeates the membrane of human erythrocytes and lymphocytes by nonfaciliated diffusion. J Biol Chem 1987;262:5748-54.
23. Rolinski B, Wintergerst U, Matuschke A, et al. Evaluation of saliva as a specimen for monitoring zidovudine therapy in HIV-infected patients. AIDS 1991;5:885-8.

Zidovudine (ZDV); Cerebrospinal fluid (CSF); Central nervous system (CNS); Analysis; Pharmacokinetics; Blood-CSF barrier

© Lippincott-Raven Publishers.