ABSTRACT
Therapeutic control of human immunodeficiency virus type 1 (HIV-1) in peripheral compartments does not assure control in the central nervous system. Inadequate drug penetration may provide a sanctuary from which resistant virus can emerge or allow development of psychomotor abnormalities. To characterize the effect of ritonavir on indinavir disposition into cerebrospinal fluid, seven HIV-infected adults underwent intensive sampling at steady-state while receiving twice-daily indinavir (800 mg) and ritonavir (100 mg). Serial cerebrospinal fluid and plasma samples were obtained at 10 time points from each subject. Free indinavir accounted for 98.6% of drug in cerebrospinal fluid and 55.9% in plasma. Mean cerebrospinal fluid Cmax, Cmin, and area under the concentration-time curve from 0 to 12 h (AUC0-12) values for free indinavir were 735 nM, 280 nM, and 6,502 nM h−1, respectively, and the free levels exceeded 100 nM in every sample. The cerebrospinal fluid/plasma AUC0-12 ratio for free indinavir was 17.5% ± 6.4%. This ratio was remarkably similar to results obtained in a previous study in which subjects received indinavir without ritonavir, indicating that ritonavir did not have a substantial direct effect on the barrier to indinavir penetration into cerebrospinal fluid. Low-dose ritonavir increases cerebrospinal fluid indinavir concentrations substantially more than 800 mg of indinavir given thrice daily without concomitant ritonavir, despite a lower total daily indinavir dose.
The central nervous system is involved by human immunodeficiency virus type 1 (HIV-1) during all stages of HIV disease (12, 14, 19, 23). The widespread use of potent antiretroviral agents, including HIV-1 protease inhibitors, has been associated with declines in HIV-related dementia and other overt neurological manifestations of infection (1-3, 24). Unfortunately, many patients receiving potent antiretroviral therapy do not achieve sustained control of viral replication (5). It is possible that the blood-brain barrier to entry of some HIV-1 protease inhibitors into the central nervous system contributes to systemic virologic failure (21). Ongoing viral replication in the brain may also cause neurologic sequelae despite the apparent control of virus in peripheral compartments (9).
Indinavir is a potent HIV-1 protease inhibitor that provides considerable virologic, immunologic, and clinical benefits (15, 18). It penetrates relatively well into cerebrospinal fluid (CSF) when given at 800 mg every 8 h (17, 30). In a previous study involving eight subjects who underwent intensive CSF sampling over entire dosing intervals while receiving indinavir (800 mg every 8 h) without ritonavir, the total indinavir levels in CSF exceeded 100 nM throughout the 8-h dosing interval in four subjects and during at least 85% of the dosing interval in seven subjects (17). In the present study, intensive CSF sampling was used to characterize steady-state indinavir pharmacokinetics in CSF and plasma among seven adults receiving indinavir (800 mg every 12 h) and ritonavir (100 mg every 12 h) twice daily.
MATERIALS AND METHODS
Study subjects.The present study enrolled seven subjects who were also participating in Merck Protocol 094, a single-arm study of four-drug therapy with indinavir, ritonavir, stavudine, and lamivudine in antiretroviral-naive adults. Inclusion criteria included greater than 75 CD4+ T cells/mm3, an HIV-1 RNA level in plasma of >5,000 copies/ml, and acceptable screening electrocardiogram, chest radiograph, urinary drug screen, hematology, chemistry, and coagulation studies. Potential enrollees were excluded for concomitant use of medications known or predicted to interact with cytochrome P450 3A4. The study was approved by the Vanderbilt University Institutional Review Board, and all subjects provided written informed consent. Only subjects judged likely to be compliant with study therapy and procedures were enrolled in this CSF sampling study.
Drug administration.Doses of indinavir of 800 mg (four 200-mg capsules) and ritonavir of 100 mg (one 100-mg capsule) were administered orally every 12 h. Pharmacokinetic sampling was conducted after a morning dose of indinavir, and subjects were required to fast from midnight the night before until 2 h postdose on that day. Both the indinavir dose at onset of pharmacokinetic sampling and the previous dose were directly administered by study personnel. Concomitant stavudine (40 mg every 12 h) and lamivudine (150 mg every 12 h) were continued throughout the study.
Collection of CSF and plasma samples.Sampling of CSF and plasma for pharmacokinetic analyses was performed after at least 14 days of continuous study drug therapy. CSF was collected through an 18-gauge indwelling lumbar intrathecal catheter. Catheters were inserted under 1% xylocaine local anesthesia within 1 h prior to the 8-a.m. dose of indinavir. After an initial CSF sample was obtained through the catheter for glucose, protein, protein binding, and cell count determinations, the catheter was secured and the end was capped. Two milliliters of CSF was collected directly into polypropylene tubes on ice at 0 h (i.e., predose) and at 0.5, 1, 1.5, 2, 3, 4, 6, 8, and 12 h postdose. Flow of CSF through the catheter allowed completion of each sampling within 10 min. The first 0.5 ml of each sample was discarded to account for the 0.2-ml tubing void volume. Samples were promptly placed on dry ice and then stored at −70°C until assayed. Whole blood was collected into EDTA tubes at the same time points. Plasma was separated by centrifugation at 4°C and stored at −70°C until assayed. All intensive CSF sampling procedures were performed at the Vanderbilt Clinical Research Center. For safety reasons, samples obtained immediately prior to catheter removal were also analyzed for glucose, protein, cell counts, and bacterial and fungal stains and cultures. Each sampling procedure required hospitalization for 36 h, including two overnight stays.
Quantification of total and free indinavir.Indinavir concentrations in CSF and plasma were determined by a validated liquid chromatography-mass spectroscopy-mass spectroscopy (LC/MS/MS) method (indinavir limits of quantitation, 50 ng/ml for plasma and 5 ng/ml for CSF) by BAS Analytics, West Lafayette, Ind. Indinavir concentrations were converted to a molar basis by using a molecular weight of 613.81. Indinavir and the internal standard, hexa-deuterated indinavir analog, were isolated from plasma samples by RapidTrace solid-phase extraction workstation. Indinavir and the internal standard were isolated from CSF samples liquid-liquid extraction at alkaline pH. For both plasma and CSF, the analytes were separated and detected by an LC/MS/MS system utilizing a 2.1-by-50-mm SB-CN column with an ammonium acetate buffer-methanol mobile phase. The standard curves were linear from 50 to 10,000 ng/ml for plasma and from 5 to 1,000 ng/ml for CSF. Daily standard curves were constructed from the peak area ratios of indinavir and internal standard against the concentrations of the standards. Unknown sample concentrations were calculated from the equation y = mx + b, as determined by the weighted (1/x) linear regression analysis of the standard curve. Studies of nonspecific binding to sampling tubes suggest small losses of drug with CSF samples (mean = 10% ± 3%). Assay accuracy ranged from 94.0% to 105.2% for CSF and from 93.5% to 104.2% for plasma.
The unbound fraction of indinavir in CSF and plasma was determined by ultrafiltration as described elsewhere (22). This assay has been validated by using radiolabeled indinavir, and binding of the drug to the ultrafiltration device is negligible (<1%). Briefly, 3H-labeled indinavir was added to plasma or CSF to yield a final 3H concentration of 10 μg/ml equivalent. After a 15-min incubation at 37°C, samples were centrifuged at 1,500 × g for 15 min at 37°C. The unbound fraction was estimated from the ratio of 3H concentration in the ultrafiltrate to that in the original sample. Protein binding was determined for each subject by analyzing a single CSF and plasma sample from each subject. Because protein binding of indinavir is linear over a wide range of concentrations, including those in excess of clinically observed concentrations (22), the unbound fraction in CSF or plasma determined for each individual was used as a constant to determine the concentration of free indinavir at each time point.
Pharmacokinetic analysis.Peak (Cmax) and trough (Cmin) drug concentrations and time to Cmax (Tmax) were obtained by inspection. Consistent with prior publications of indinavir pharmacokinetic data, the area under the concentration-time curve from 0 to 12 h (AUC0-12) was calculated by the modified trapezoidal method by using stable piecewise cubic polynomials to interpolate between the measured datum points (29). This approach produces stable and monotone interpolations that may be more reliable and less biased than other interpolation methods used to determine the AUC. Statistical analyses were performed by using SPSS software (version 9.0; SPSS, Inc., Chicago, Ill.). The results were compared across categorical variables by using Wilcoxon rank sum test. Continuous variables were compared by determining the Pearson correlation. Error values are presented as standard deviations unless otherwise indicated.
Blood-brain barrier integrity and intrathecal IgG production.Albumin and immunoglobulin G (IgG) were quantified by routine methods. The CSF-to-plasma albumin quotient was calculated as (albuminCSF ÷ albuminplasma). The blood-brain barrier was considered intact if the CSF-to-plasma albumin quotient was <0.0074 (8). The CSF IgG index was calculated as (IgGCSF ÷ IgGplasma) ÷ (albuminCSF ÷ albuminplasma). Intracerebral IgG production was considered present if the CSF IgG index was >0.66 (8). A CSF β2-microglobulin concentration of >2 mg/liter indicated immune activation.
Prior study of indinavir without concomitant ritonavir.Results of the present study were compared to our previous study of eight subjects receiving indinavir (800 mg every 8 h) without ritonavir, as described elsewhere (17). In that study, inclusion criteria included stable therapy with indinavir plus any two nucleoside analogues for at least 30 days and levels of HIV-1 RNA in plasma below the limits of detection within the prior 30 days and greater than 200 CD4+ T cells/mm3. In contrast to the present study, the previous study included CSF sampling at 5 and 7 h postdose but not at 1.5 and 12 h postdose. Similarly, the previous study included plasma sampling at 0.5, 5, and 7 h postdose but not at 12 h postdose. Sample collection methods were otherwise identical in these studies. Indinavir in CSF and plasma was quantitated by using a similar LC/MS/MS assay with some modification in the extraction method, chromatographic column, and mobile phase (17). The limit of quantitation was unchanged, and the assay accuracy was similar between the two studies. In the previous study assay, the accuracy ranged from 97.8 to 102.3% for CSF and from 98.6 to 101.3% for plasma. In the present study assay, accuracy ranged from 94.0 to 105.2% for CSF and from 93.5 to 104.2% for plasma.
RESULTS
Patient characteristics.The seven study subjects ranged in age from 31 to 52 years; four were Caucasian, and three were African-American; two of the seven were female. Patient characteristics are presented in Table 1. At the time of CSF sampling, all subjects had been receiving study drugs continuously for at least 14 days (range, 14 to 21 days).
Patient characteristics at baseline
Blood-brain barrier and intrathecal immune activation.Four of seven subjects had intact blood-brain barriers without intrathecal immune activation, as evidenced by normal CSF-to-plasma albumin quotients, CSF IgG indices, and CSF β2-microglobulin levels (Table 1). Subject B had an increased CSF-to-plasma albumin quotient and CSF total protein, β2-microglobulin, and IgG index levels indicating impairment of the blood-CSF barrier and increased intrathecal immune activation. Subjects E and G also had a slightly elevated CSF IgG indices.
Pharmacokinetic analysis.Figure 1 presents the concentration profiles of total indinavir in both CSF and plasma for each subject (Fig. 1A to G). The mean concentration plots based on data from all subjects are also presented (Fig. 1H). (A CSF assay result of 1,474 nM from subject E at 2 h was considered spuriously elevated and excluded from all analyses. A 12-h CSF sample from subject D was unavailable, so the predose value was also used for the 12-h value.) Indinavir concentrations in plasma increased rapidly and reached Cmax within 1 h in all but one subject. After peak levels were reached, the indinavir concentration in plasma declined gradually such that the trough concentrations in plasma were, on average, 15-fold less than the peak concentrations. The concentration-time profile for total indinavir in CSF was flatter than the plasma profile, with levels in CSF varying 3.0-fold (range, 1.8- to 5.4-fold) during the 12-h dosing interval. The indinavir Cmax in CSF occurred, on average, 5.3 h after oral dosing. The 95% inhibitory concentration (IC95) of indinavir in vitro against wild-type HIV-1 strains ranges from 25 to 100 nM (6, 10, 27). In the present study, total indinavir levels in CSF exceeded 100 nM throughout the entire dosing interval in every subject. Selected pharmacokinetic parameters for total indinavir are summarized in Table 2.
Concentration curves of total indinavir (free and protein-bound) at steady-state in plasma (○) and CSF (•) for each study subject (A to G). Mean total (H) and free (I) indinavir values for all subjects are also shown. Error bars indicate SDs. The horizontal line indicates 100 nM, which approximates the upper limit of the cell culture IC95 for indinavir against wild-type HIV-1 (range, 25 to 100 nM).
Selected steady-state pharmacokinetic parameters for total and free indinavir in CSF and plasma during therapy with indinavir (800 mg) and ritonavir (100 mg) given twice daily
Plasma and CSF samples from each subject were assayed for protein binding to determine the unbound fraction of indinavir (i.e., free indinavir). In CSF 98.6% ± 1.3% of the drug was free (range, 97.0 to 100.0%), whereas 55.9% ± 4.1% was free (range, 50.1 to 60.9%) in plasma. Figure 1I shows the mean concentration profiles of free indinavir in CSF and plasma. Free indinavir levels in plasma exceeded free levels in CSF throughout the dosing interval.
The CSF/plasma ratio of free indinavir at single time points decreased during the first 30 min following drug administration to 6.3% ± 2.7% (mean ± the standard deviation [SD]), reflecting rapid increases in levels in plasma. This ratio then steadily increased, and free indinavir levels in CSF were 63% ± 34% (mean ± the SD) of free indinavir levels in plasma at the end of the 12-h dosing interval.
Among the seven subjects, total indinavir levels in CSF immediately before indinavir dosing (359 ± 64 nM, mean ± the standard error of the mean [SEM]) were the same as the 12-h postdose levels (352 ± 79 nM), suggesting that steady state had been achieved. Similarly, the indinavir levels in plasma before dosing (2,222 ± 593 nM) did not differ significantly from 12-h postdose levels (1,243 ± 460 nM [P = 0.217]).
Correlates of indinavir penetration into CSF.Predictors of indinavir levels in CSF (AUC0-12) were examined. The total CSF indinavir AUC0-12 correlated significantly with the total plasma Cmin (r = 0.77, P = 0.044) and free plasma Cmin (r = 0.77, P = 0.043) and tended to correlate with total plasma indinavir AUC0-12 (r = 0.72, P = 0.068) and free plasma AUC0-12 (r = 0.74, P = 0.059), but there was no apparent correlation with either total plasma Cmax (r = 0.41, P = 0.366) or free plasma Cmax (r = 0.54, P = 0.209).
The CSF/plasma AUC0-12 ratio for free indinavir is an index of the blood-CSF barrier to indinavir penetration. Among the seven study patients, the CSF/plasma AUC0-12 ratio did not correlate with CSF-to-plasma albumin quotients, CSF β2-microglobulin levels, CSF IgG indices, or with plasma indinavir parameters (plasma free or total indinavir Cmax, C12, or AUC0-12). Similarly, there were no significant differences in the CSF/plasma AUC0-12 ratio when the four subjects with entirely normal CSF indices were compared to the three subjects with at least one abnormal index, nor was there a correlation between CSF indinavir parameters (free or total indinavir Cmax, C12, or AUC0-12) and indices of intrathecal inflammation or blood-brain barrier integrity. In addition, analysis of combined data from the present study (seven subjects) and previous study (eight subjects) did not identify a correlation between these indices and any measure of indinavir disposition into CSF, despite the increased power associated with a larger sample size.
Comparison to a study of indinavir administered without ritonavir.In a previous study we used a similar intensive sampling approach to characterize CSF and plasma pharmacokinetics of indinavir in eight adults receiving indinavir three times daily (800 mg every 8 h) without ritonavir (17). Subjects in the present study had significantly lower CD4+-T-cell counts, a result likely reflecting more recent initiation of antiretroviral therapy, and somewhat high CSF protein concentrations, but the two groups were otherwise comparable (Table 3). The combined data from these two studies allowed a comparison of indinavir pharmacokinetics with these different dosing schemes (Fig. 2 and Table 4). Although the total daily indinavir dose was 800 mg lower in the present study, CSF indinavir Cmax, Cmin, and AUC0-24 with concomitant ritonavir were ca. 250% of previous values. In contrast, ritonavir did not alter the indinavir Cmax in plasma, modestly increased the plasma AUC0-24, but markedly increased Cmin in plasma. The lowest indinavir levels in plasma in any subject in previous studies and the present study were 84 and 534 nM, respectively, whereas the lowest CSF indinavir levels in these studies were 44 and 153 nM, respectively. Concomitant ritonavir increased the CSF/plasma AUC0-24 ratio for total indinavir <2-fold and had no significant effect on this ratio for free drug. Concomitant ritonavir significantly delayed time to indinavir Cmax in CSF but not in plasma, although this may be due to the change in dosing interval rather than the concomitant ritonavir.
Effect of ritonavir on mean steady-state concentration curves of total indinavir (free and protein-bound) in CSF and plasma. Concentrations in plasma (A) and CSF (B) are shown. Data are from seven subjects receiving indinavir (800 mg every 12 h) and ritonavir (100 mg every 12 h) in the present study (•) and from eight subjects receiving indinavir (800 mg every 8 h) without ritonavir (○) as described elsewhere (17). Error bars indicate the SDs.
Comparison of baseline patient characteristics in the present and prior studies
Percent change in steady-state total indinavir pharmacokinetic parameters associated with concomitant ritonavir administrationa
DISCUSSION
The present study demonstrates that twice-daily administration of indinavir (800 mg) with ritonavir (100 mg) achieves indinavir levels in CSF that exceed 100 nM throughout the entire dosing interval. The cell culture IC95 of indinavir for HIV-1 ranges from 25 to 100 nM (27), even in the presence of human serum (6). These drug levels in CSF should therefore provide substantial control of HIV-1 replication in this central nervous system compartment. Drug levels in CSF were relatively constant, varying on average only threefold during the 12-h dosing interval.
Administering indinavir at 800 mg every 8 h in multidrug regimens provides sustained control of HIV-1 replication with associated clinical benefits (18). Although ritonavir provides short-term clinical benefit when prescribed at 600 mg twice daily to patients with advanced AIDS (7), this dose is often not well tolerated and causes substantial hyperlipidemia. It is more commonly used as a “pharmacokinetic enhancer.” Ritonavir increases plasma AUCs, half-lives, and trough concentrations of indinavir and other protease inhibitors by inhibiting the 3A4 isoform of cytochrome P450 (13, 20, 25) and overcomes the negative effect of food on indinavir bioavailability. Comparing detailed pharmacokinetic data from our previous study (17) with data from the present study allowed the effect of ritonavir on indinavir disposition into CSF to be thoroughly characterized. Despite a lower total daily dose of indinavir, ritonavir provided CSF indinavir Cmax, Cmin, and AUC0-24 values ca. 250% of those achieved without ritonavir and delayed the time to CSF Cmax by ca. 2 h. Ritonavir more dramatically altered the shape of the indinavir concentration in plasma curve, greatly increasing Cmin but not affecting Cmax or time to Cmax. The effect of ritonavir on the indinavir concentration profiles in plasma is generally consistent with a recent study that showed that low-dose ritonavir increased the geometric mean indinavir Cmax by <2-fold, the AUC0-24 by <3-fold, and the Ctrough by >10-fold in HIV-negative volunteers (25). Since Cmin may best predict antiviral effect for HIV-1 protease inhibitors, concomitant ritonavir should enhance indinavir's antiviral effect in both peripheral and central nervous system compartments.
Two variables that differed between the present and previous studies were the coadministration of ritonavir with indinavir and the increased dosing interval from 8 to 12 h. Administering both drugs every 8 h, but at the doses used in the present study, would almost certainly have further increased both CSF and plasma Cmax, Cmin, and AUC0-24 values. However, such dosing would likely have caused intolerance and is not relevant to clinical practice.
The CSF AUC0-12 for indinavir correlated most strongly with plasma Cmin and AUC0-12. This result differs from what has been reported for indinavir administered without ritonavir (17), in which case only plasma Cmax and plasma AUC0-8 correlated with CSF/plasma AUC0-8 h. The correlation between plasma and CSF AUC values in both studies is consistent with the slowly equilibrating nature of indinavir transfer between plasma and CSF, since concentrations in the slowly equilibrating compartment (i.e., CSF) should be most influenced by the average concentrations in plasma over time rather than a minimum or maximum concentration in plasma that is sustained for only a short time. The differing shapes of the indinavir plasma profiles with or without coadministered ritonavir likely explain the differing extent to which plasma Cmax or Cmin correlated with CSF AUC rather than any direct effect of plasma Cmax or Cmin on drug transfer into CSF.
The reduced and delayed concentration peaks in CSF relative to plasma and the flatter profile shape in CSF result from a drug transfer rate across the blood-brain barrier that is slow relative to the rate of change of drug concentrations in plasma. If passive diffusion were the only mechanism controlling indinavir disposition in the central nervous system, the AUC0-12 of free indinavir in CSF and plasma would be equal under steady-state conditions. The observed CSF/plasma AUC0-12 ratio for free indinavir of 17.5% ± 6.4% indicates that additional distribution or clearance mechanisms act upon indinavir. The locations of such additional distribution or clearance within the central nervous system cannot be inferred directly from this comparison. However, indinavir is a substrate for P-glycoprotein, a multidrug efflux transport protein found at the blood-brain barrier, suggesting that this transporter accounts at least in part for the additional clearance (21).
In the present study, free and total indinavir concentrations in CSF and plasma were quantified for all seven subjects. The proportion of free indinavir observed in the present (56%) study is somewhat greater than the ca. 40% value from previous reports, including studies that used the same methodology (10, 17, 26). The reason for this difference between studies is uncertain. Displacement of indinavir from plasma proteins by ritonavir is unlikely given the very low concentrations in plasma associated with the ritonavir dose used in the present study.
Although indinavir penetrated relatively well into CSF, strategies to enhance the delivery of HIV-1 protease inhibitors into the central nervous system are warranted. One potential strategy is targeted inhibition of P-glycoprotein (21). In animal studies, systemic inhibition of P-glycoprotein disproportionately increases concentrations of HIV-1 protease inhibitors in the brain (4). In addition, ritonavir inhibits P-glycoprotein in vitro (11). We were therefore interested in assessing whether ritonavir enhanced penetration of indinavir into CSF at least in part by inhibiting P-glycoprotein in the blood-brain barrier, in which case the CSF/plasma AUC0-24 ratio would likely increase. The remarkably similar CSF/plasma AUC0-24 ratios for free indinavir in the two studies (17.5 and 17.0%, Table 4) suggest that the contribution of efflux transport was not altered by ritonavir and that the primary mechanism of enhanced central nervous system penetration of indinavir with coadministered ritonavir was the increase in the concentrations of indinavir in plasma due to inhibition of hepatic cytochrome P450 by ritonavir rather than P-glycoprotein inhibition at the blood-brain barrier.
Serial CSF sampling provides information not readily addressed by single-point measurements. A prior study using single-time-point CSF samples also showed increased indinavir concentrations in CSF in a patient receiving concomitant ritonavir, but the authors of that study acknowledged that although a “AUCCSF-to-AUCplasma” ratio is considered the gold standard for drug entry…it is not feasible to perform several lumbar punctures on one day to calculate the AUC in CSF” (28). Accurate characterization of the pharmacokinetics of drug transfer to the CSF in individual patients (i.e., CSF/plasma AUC0-12 ratio), determination of interindividual variability in the blood-brain barrier, and accurate characterization of total drug exposure in CSF (i.e., AUC0-12, Cmax, and Cmin) are facilitated by intensive CSF sampling. In addition, defining whether indices of blood-brain barrier integrity and central nervous system immune activation influence the blood-CSF barrier to drug penetration may best be achieved by this intensive approach (16).
The present study provides strong evidence that administering indinavir with concomitant low-dose ritonavir will enhance control of HIV-1 replication in the central nervous system. This effect may be of particular benefit to patients with AIDS dementia or other central nervous system complications of ongoing viral replication in the central nervous system.
ACKNOWLEDGMENTS
We are grateful to the persons with HIV infection who volunteered for this study and to Jeanne Schroeder for regulatory support. Indinavir protein-binding assays were performed by I-Wu Chen and Janice Rowe.
This study was supported in part by NIH grant RR 00095 (GCRC).
FOOTNOTES
- Received 10 December 2002.
- Returned for modification 17 March 2003.
- Accepted 31 March 2003.
- American Society for Microbiology
