The antiviral efficacy of HIV protease inhibitors (PI) in the central nervous system (CNS) is controversial. The biochemical characteristics of most HIV PI drugs suggest that they may not penetrate the blood–brain barrier well. Most of the drugs in this class are large hydrophobic molecules that are highly bound to plasma proteins and are substrates for P-glycoprotein, a multidrug resistance pump expressed on brain endothelial cells. As a result, most PI drugs achieve low concentrations in the CNS [1,2]. For example, nelfinavir was not detectable in multiple samples of cerebrospinal fluid (CSF) from six subjects . One study estimated that the CSF-to-plasma area-under-the-curve ratio of indinavir was only 6% even though 40% was unbound to plasma proteins . Kravcik and colleagues  studied the effects of ritonavir and saquinavir on HIV in plasma and CSF and found that, at the time of therapy failure, HIV RNA levels in the CSF exceeded those in plasma, reversing the ratio seen in untreated subjects. This suggested that the CNS remained untreated or escaped control earlier than the plasma/lymphatic compartment, supporting the conclusion that PI drugs may inadequately treat the CNS. Resolving this controversy is important because HIV replication in the CNS in the presence of subtherapeutic drug concentrations may lead to the evolution of resistant mutants.
Similar to other currently available PIs, a large proportion (97–99%) of lopinavir (LPV) is bound to plasma proteins, such as albumin, α1-acid glycoprotein . Since only unbound drug is available to distribute across the blood–brain barrier, LPV may attain low concentrations in the brain and CSF. In addition, several active transport systems extrude drugs from the CNS, limiting their concentrations. However, even low LPV penetration of the CSF may be sufficient to inhibit HIV replication in the CNS for at least three reasons. First, LPV is administered with small doses of ritonavir, which greatly increases the LPV concentrations in plasma. Second, ritonavir may reduce activity for some of the transporters that remove drug from the CNS. This seems to be true for indinavir when it is coadministered with high-dose ritonavir, although this effect may be dose dependent [7,8]. Third, LPV is very potent, plasma LPV concentrations seen clinically markedly exceed the drug's 50% inhibitory concentration (IC50) for wild-type viruses. When administered as a commercially available formulation that includes low-dose ritonavir, relatively low LPV concentrations in the CSF may still exceed the wild-type virus IC50 by several-fold. Therefore, LPV may suppress HIV replication in the brain despite the high degree of binding to plasma proteins.
Three publications have reported LPV concentrations in CSF below detection. Lafeuillade et al.  reported that LPV was ‘consistently undetectable’ in CSF. In a review, Cvetkovic and Goa  reported that LPV penetrated ‘poorly’ into CSF. Finally, in a recent cross-sectional study, Solas et al.  were unable to detect LPV in CSF specimens from 12 subjects using a high performance liquid chromatography method with a quantification limit of 50 μg/l. Our objective was to measure LPV in CSF using a more sensitive assay and to compare the results to a measure of drug susceptibility.
Subjects were enrolled in prospective, observational, cohort studies at the University of California San Diego HIV Neurobehavioral Research Center. The analysis described here involved 26 HIV-infected patients receiving highly active antiretroviral therapy (HAART) containing Kaletra (LPV and ritonavir) who were participating in a separate trial of HIV that required collected timed plasma and CSF specimens for assessment of HIV RNA. Specimens were stored at −70°C until analysis. All subjects reported taking 400 mg LPV in combination with 100 mg ritonavir orally twice daily as part of HAART. Subjects were in their early forties (median, 41 years), Caucasian [17/26 or (65%)] and predominantly male [24/26 (92%)]. The median blood CD4 cell count was 176 × 106 cells/l and HIV RNA was suppressed in 58% of CSF (< 50 copies/ml) and 35% of plasma (< 400 copies/ml) specimens. The University of California San Diego Human Research Protections Programs approved all studies and all subjects consented to the use of their specimens and data.
LPV was measured in 31 CSF–plasma pairs from these 26 individuals. Lopinavir was measured by validated liquid chromatography mass spectrometry (MS) and high performance liquid chromatography for CSF and plasma samples, respectively. Briefly, CSF samples were mixed with acetonitrile that included an internal standard, A-86093, kindly donated by Abbott Laboratories. Samples remained at room temperature for 5–10 min and were then centrifuged at room temperature for 10 min at 14 000 × g. The supernatant was transferred to an autosampler vial with glass insert for duplicate 10 μl injections. LPV was separated using an ACE 5 C18 150 mm × 2.1 mm column, using acidified water (with formic acid) and acetonitrile as mobile phases. LPV was analyzed using atmospheric pressure chemical ionization in positive ion mode (APCI+) mass spectrometry (Thermo Electron, Finnigan LCQ duo, San Jose, California, USA). Quantification was performed by MS/MS using LPV parent compound/fragment ion (m/z) 629/447. The assay had a lower limit of quantification of 3.7 μg/l for CSF samples, with interassay variability < 13%. The plasma samples were processed in a similar manner with a lower limit of quantification of 100 μg/l and interassay variability < 7.5%
The median time between CSF and plasma sampling was 24 min. Concentrations were compared with the LPV IC50 range derived from the control virus for the ViroLogic PhenoSense assay [median, 3.0 nmol/l (1.9 μg/l) and 99th percentile 8.1 nmol/l] . Other routine measures and methods included blood CD4 lymphocytes, enumerated by flow cytometry, and HIV RNA measured by Roche Amplicor assay (standard in plasma, ultrasensitive in CSF; Roche Molecular Systems, Branchburg, New Jersey, USA).
LPV concentrations are summarized in Table 1. Subjects had been receiving LPV therapy for a median of 88 days [interquartile range (IQR), 41–230] at the time of sampling. The median post-dose sampling interval was 4.3 h (IQR, 2.5–6.0) for both CSF and plasma specimens.
Seven samples had very low LPV concentrations in plasma, suggesting poor adherence. Based on sample collection times, these seven samples fell below the 5th percentile for expected LPV plasma concentrations derived from published pharmacokinetic studies and were all lower than 1700 μg/l. Five of these CSF samples and three of the plasma samples had LPV concentrations below the detection limit of the assay. Three of these subjects had a second set of CSF–plasma samples collected on a separate visit and LPV evaluations in each of these had typical plasma and measurable CSF concentrations, consistent with other subjects. When these seven pairs with suspected poor adherence were excluded, all of the remaining CSF samples had measurable LPV concentrations. These samples had a median LPV concentration in plasma of 5889 μg/l (IQR, 4805–9620). The median LPV concentration in the matched CSF specimens was 17.0 μg/l (IQR, 12.1–22.7) and the CSF–plasma ratios ranged from 0.12 to 0.75% (median, 0.23%). Figure 1 depicts LPV concentrations in plasma and CSF by post-dose time. Figure 1a shows the median and 10th and 90th percentile fitted curves for plasma with the seven excluded specimens indicated by open triangles. Figure 1b shows the fitted curve for CSF specimens as well as the median IC50 for LPV using the ViroLogic PhenoSense assay . Measured CSF concentrations exceeded this level by a median 5.3-fold (IQR, 3.8–7.2).
These results suggest that LPV penetrates sufficiently into the CNS, despite being more than 98% bound to plasma proteins, to have significant antiviral activity in this potential reservoir. Using a sensitive assay, we were able to detect LPV in all CSF samples from patients with typical plasma LPV concentrations. Lopinavir CSF concentrations exceeded the 99th percentile for HIV wild-type susceptibility and surpassed the median IC50 by approximately five-fold. Compared with indinavir, the LPV CSF–plasma ratio is in closer agreement with the reported free fraction in plasma, suggesting that CSF penetration of LPV is less impaired by active transport out of the CSF.
In contrast to the current study, prior studies have not detected LPV in CSF of patients receiving LPV/ritonavir. This difference can be entirely attributed to differences in assay sensitivity. If prior assay limits of quantification are applied to our results, few of our CSF samples would have yielded quantitative results.
The CSF penetration, described by the CSF–plasma ratio, was variable among subjects. This may be the result of differences in plasma protein binding and sample collection times relatively to dose. The cross-sectional design and the lack of susceptibility testing of patient-specific CSF viral isolates prevented more conclusive determination of the contribution of LPV to suppressing viral replication in the CNS. While LPV achieved suppressive concentrations against wild-type virus in the CSF, these levels may not be adequate for virus in treatment-experienced patients and may not reflect the brain parenchyma LPV concentrations achieved. It is also possible that monocyte-derived macrophages (the type of cell that can be productively infected in the CNS) may have different LPV sensitivity than the lymphocytes used in reference assays.
Lopinavir as a common component of HAART therapy likely contributes to suppression of viral replication in the CNS. While LPV concentrations in CSF are much lower than plasma, they exceed levels that suppress wild-type HIV replication. They exceed the IC50 to a greater degree than has been reported for other PI drugs. While these results are reassuring, further work characterizing the determinants of LPV disposition and its antiviral effects in the CNS are needed.
Sponsorship: This project was supported by Abbott Laboratories (Investigator-initiated funding), the University-wide AIDS Research Program, and the National Institute of Mental Health (K23 MH01779, R01 MH58076, P30 MH 62512).
1. Beach JW. Chemotherapeutic agents for human immunodeficiency virus infection: mechanism of action, pharmacokinetics
, metabolism, and adverse reactions. Clin Ther 1998; 20:2–25.
2. Enting RH, Hoetelmans RM, Lange JM, Burger DM, Beijnen JH, Portegies P. Antiretroviral drugs and the central nervous system. AIDS 1998; 12:1941–1955.
3. Aweeka F, Jayewardene A, Staprans S, Bellibas SE, Kearney B, Lizak P, et al
. Failure to detect nelfinavir in the cerebrospinal fluid
-1-infected patients with and without AIDS dementia complex. J Acquir Immune Defic Syndr Hum Retrovirol 1999; 20:39–43.
4. Letendre SL, Capparelli EV, Ellis RJ, McCutchan JA. Indinavir population pharmacokinetics
and cerebrospinal fluid
. The HIV
Neurobehavioral Research Center Group. Antimicrob Agents Chemother 2000; 44:2173–2175.
5. Kravcik S, Gallicano K, Roth V, Cassol S, Hawley-Foss N, Badley A, et al
. Cerebrospinal fluid HIV
RNA and drug levels with combination ritonavir and saquinavir. J Acquir Immune Defic Syndr 1999; 21:371–375.
6. Anon. ABT-378/r Investigator's Manual
. Abbott Park, IL: Abbott Laboratories; 2 May 2000.
7. van Praag RM, Weverling GJ, Portegies P, Jurriaans S, Zhou XJ, Turner-Foisy ML, et al
. Enhanced penetration of indinavir in cerebrospinal fluid
and semen after the addition of low-dose ritonavir. AIDS 2000; 14:1187–1194.
8. Haas DW, Johnson B, Nicotera J, Bailey VL, Harris VL, Bowles FB, et al
. Effects of ritonavir on indinavir pharmacokinetics
in cerebrospinal fluid
. Antimicrob Agents Chemother 2003; 47:2131.
9. Lafeuillade A, Solas C, Halfon P, Chadapaud S, Hittinger G, Lacarelle B. Differences in the detection of three HIV
-1 protease inhibitors in non-blood compartments: clinical correlations. HIV
Clin Trials 2002; 3:27–35.
10. Cvetkovic RS, Goa KL. Lopinavir
/ritonavir: a review of its use in the management of HIV
infection. Drugs 2003; 63:769–802.
11. Solas C, Lafeuillade A, Halfon P, Chadapaud S, Hittinger G, Lacarelle B. Discrepancies between protease inhibitor concentrations and viral load in reservoirs and sanctuary sites in human immunodeficiency virus-infected patients. Antimicrob Agents Chemother 2003; 47:238–243.
12. Parkin NT, Hellmann NS, Whitcomb JM, Kiss L, Chappey C, Petropoulos CJ. Natural variation of drug susceptibility in wild-type human immunodeficiency virus type 1. Antimicrobial Agents Chemother 2004; 48:437–443.