Combination of protease inhibitors for the treatment of HIV-1-infected patients: a review of pharmacokinetics and clinical experience

Transcription

1 Antiviral Therapy 6: Review Combination of protease inhibitors for the treatment of HIV-1-infected patients: a review of pharmacokinetics and clinical experience RPG van Heeswijk 1 *, AI Veldkamp 1, JW Mulder 2, PL Meenhorst 2, JMA Lange 3, JH Beijnen 1 and RMW Hoetelmans 1 1 Department of Pharmacy & Pharmacology, Slotervaart Hospital, Amsterdam, The Netherlands 2 Department of Internal Medicine, Slotervaart Hospital, Amsterdam, The Netherlands 3 National AIDS Therapy Evaluation Centre and Department of Internal Medicine, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands *Corresponding author: Tel: ; Fax: ; rvanheeswijk@ottawahospital.on.ca Present address: The Ottawa Hospital, Division of Infectious Diseases, 501 Smyth Road, Ottawa, Ontario, K1H 8L6 Canada The use of highly active antiretroviral therapy, the combination of at least three different antiretroviral drugs for the treatment of HIV-1 infection, has greatly improved the prognosis for HIV-1-infected patients. The efficacy of a combination of a protease inhibitor (PI) plus two nucleoside analogue reverse transcriptase inhibitors has been well established over a period of up to 3 years. However, virological treatment failure has been reported in 40 60% of unselected patients within 1 year after initiation of a PI-containing regimen. This observation may, at least in part, be attributed to the poor pharmacokinetic characteristics of the PIs. Given as a single agent the PIs have several pharmacokinetic limitations; relatively short plasma-elimination half-lives and a modest and variable oral bioavailability, which is, for some of the PIs, influenced by food. To overcome these suboptimal pharmacokinetics, high doses (requiring large numbers of pills) must be ingested, often with food restrictions, which complicates patient adherence to the prescribed regimen. Positive drug drug interactions increase the exposure to the PIs, allowing administration of lower doses at reduced dosing frequencies with less dietary restrictions. In addition to increasing the potency of an antiretroviral regimen, combinations of PIs may enhance patient adherence, both of which will contribute to a more durable suppression of viral replication. The favourable pharmacokinetics of PIs in combination are a result of interactions through cytochrome P450 3A4 (CYP3A4) isoenzymes and, possibly, the multi-drug transporting P-glycoprotein (P-gp). Antiretroviral synergy between PIs and non-overlapping primary resistance patterns in the HIV-1 protease genome may further enhance the antiretroviral potency and durability of combinations of PIs. Many combinations contain ritonavir because this PI has the most pronounced inhibiting effects on CYP3A4. The combination of saquinavir and ritonavir, both in a dose of 400 mg twice-a-day, is the most studied double PI combination, with clinical experience extending over 3 years. Combination of a PI with a low dose of ritonavir ( 400 mg/day), only to boost its pharmacokinetic properties, seems an attractive option for patients who cannot tolerate higher doses of ritonavir. A recently introduced PI, lopinavir, has been co-formulated with low-dose ritonavir, which allows for a convenient three-capsules, twice-a-day dosing regimen. In an attempt to prolong suppression of viral replication combinations of PIs are becoming increasingly popular. However, further clinical studies are needed to identify the optimal combinations for treatment of antiretroviral naive and experienced HIV- 1-infected patients. This review covers combinations of saquinavir, indinavir, nelfinavir, amprenavir and lopinavir with different doses of ritonavir, as well as the combinations of saquinavir and indinavir with nelfinavir. Introduction Since multi-drug therapy was demonstrated to be more effective than mono- or dual-therapy for the treatment of HIV-1 infection, combinations of at least three different antiretroviral drugs have been used, and are generally referred to as highly active antiretroviral therapy (HAART). According to the current guidelines, 2002 International Medical Press /01/$

2 RPG van Heeswijk et al. Table 1. Characteristics of the protease inhibitors amprenavir, indinavir, nelfinavir, ritonavir and saquinavir Plasma Licensed No. of AUC (mean elimination IC 50 * (mean Cmin (mean Threshold Protease inhibitor dosage Food restrictions pills/day Bioavailability ±SD µg/ml h) half-life (h) ±SD ng/ml) ±SD ng/ml) conc. (ng/ml) Reference Amprenavir mg No % 18.9 ± ± ± [5,6,36,38] 280 Indinavir mg Without food, 6 70% 20.2 ± ± ± [7,8,34,35] extra fluid intake required ( 1.5 l/day) Nelfinavir mg With food 9 (tid), 10 (bid) 70 80% 13.3 (750 tid) ± [9,10] mg Ritonavir mg No % ± [11 13] Saquinavir (HGC) mg With food 9 4% 1.0 ± ± ±30 50 [14,15,33] Saquinavir (SGC) mg With food % relative 7.2 ± ± [14,16,33,37] to saquinavir HGC *IC 50 (ng/ml) for wild-type virus in the presence of 50% human serum (mean ±SD) [100]. Steady-state plasma trough concentration (ng/ml) in HIV-infected adults (mean ±SD given if cited in reference). Proposed plasma threshold concentration (ng/ml) to be maintained in vivo. Concentration calculated to yield 90% of the maximum antiviral effect in vivo (EC 90 ). AUC, area under the plasma concentration versus time curve over one dosing interval; HGC, hard-gelatin capsules; SGC, soft-gelatin capsules; tid, three-times-a-day; bid, twice-a-day International Medical Press

3 Combination of protease inhibitors HAART should consist of two nucleoside reverse transcriptase inhibitors (NRTIs), plus a non-nucleoside reverse transcriptase inhibitor (NNRTI), or plus one or two protease inhibitors (PIs) [1]. Six PIs are currently commercially available for the treatment of HIV-1-infection: amprenavir, indinavir, lopinavir (co-formulated with low-dose ritonavir), nelfinavir, ritonavir and saquinavir (formulated as hard- or softgelatin capsules). These drugs inhibit the HIV-1 protease enzyme, which is responsible for the posttranslational processing of gag and gag pol polyprotein precursors. As a result, non-infectious virions, which have the morphological features of immature particles, are produced. Thus, the main antiretroviral action of the PIs is to prevent infection of uninfected cells. PIs are active against both HIV-1 and -2 [2]. The clinical efficacy of triple drug combinations, including a PI, has been convincingly demonstrated during the recent years with long periods of follow-up [3,4]. As single agents, however, the PIs have several pharmacokinetic limitations; relatively short plasmaelimination half-lives and a modest and variable oral bioavailability, which is influenced by food for some of the PIs (Table 1) [5 16]. Since several studies have shown exposure response relationships for the PIs, maintaining high plasma concentrations may be important for sustained suppression of viral replication. Due to the suboptimal pharmacokinetics of the PIs, maintaining high plasma concentrations can only be realised by administration of high doses (requiring large numbers of pills), often with food restrictions in twice- or three-times-a-day dosing regimens, which negatively influences patient adherence to the prescribed regimen [17,18]. A recent multivariate analysis that included baseline plasma HIV-1 RNA concentration, CD4 cell count and drug class, showed that the daily pill burden was significantly and inversely correlated with the percentage of patients who reached an undetectable plasma HIV-1 RNA concentration (<50 copies/ml) after 48 weeks of treatment [19]. Since adherence has been shown to be imperative for sustained suppression of viral replication [20], it may not be surprising that virological treatment failure has been reported in 40 60% of patients, within 1 year after initiation of a PIcontaining regimen [21 23]. Results of a recently presented meta-analysis of the effectiveness of triple combination therapy in 3115 patients in 22 clinical trials showed that the average proportion of patients with a plasma HIV-1 RNA concentration below 50 copies/ml after 48 weeks of therapy was 46% [19]. The use of a triple drug regimen has been shown to benefit only a minority of patients, particularly those with high baseline viral loads [24]. In an attempt to prolong the therapeutic advantage of antiretroviral therapy, drug regimens that include two PIs are becoming increasingly popular also as salvage therapy for patients who have failed treatment of HIV-1 infection [1]. Since it has been shown that the suppression of viral replication observed with a three-drug regimen can be improved by using an alternative multi-drug regimen, addition of a second PI is expected to enhance the antiviral potency of the therapeutic regimen [25]. Furthermore, combinations of PIs allow a reduction in both the dose and the dosing frequency because of the favourable pharmacokinetics of the PIs when used in combination [26]. This strategy may improve patient adherence with the therapeutic regimen [27]. The currently available PIs allow for several different combinations, each with distinct pharmacokinetic properties. In this review important clinical pharmacological characteristics of these combinations of PIs will be discussed. Rationale for the combined use of protease inhibitors Several studies have shown relationships between the exposure to PIs and the virological and immunological response to therapy. During monotherapy with ritonavir in doses ranging from 400 to 600 mg twice-a-day in 13 patients, the in vivo selection rate of resistance mutations was inversely correlated with the plasma trough concentration or the total drug exposure [area under the plasma concentration versus time curve (AUC)] [28]. It was also shown in this study that phenotypic resistance requires multiple mutations in the HIV-1 protease genome, which emerged in an ordered, stepwise fashion [28]. Indications for a similar relationship between the HIV-1-protease mutational rate and the exposure to a PI were observed by Schapiro et al. in a study with 40 patients treated with saquinavir monotherapy (3600 or 7200 mg/day) [29]. Two studies reported positive relationships between the exposure to saquinavir and both the decrease in plasma HIV-1 RNA concentration and the increase in CD4 lymphocyte count [29,30]. Furthermore, the initial rate of decline of HIV-1 RNA in plasma after initiation of a quadruple antiretroviral regimen has been shown to be positively correlated to the exposure to nelfinavir in a multivariate analysis in a study of 29 patients [31]. Since it has been hypothesised that the initial viral decay rate reflects the potency of an antiretroviral regimen, this finding suggests that the potency of a regimen can be improved by increasing drug exposure [32]. In the long-term (24 48 weeks), the exposure to saquinavir and indinavir has been shown to be an independent predictor of virological response in several retrospective studies [33 35]. In conclusion, these findings suggest that maintaining Antiviral Therapy 6:4 203

4 RPG van Heeswijk et al. Figure 1. Plasma drug concentrations and virus resistance Drug level Periodic inadequate drugs level Mutant selected with reduced susceptibility Rebound with highly resistant virus Periodic inadequate plasma drug concentrations (indicated by the dark grey bar) may enhance development of resistant viral strains. Consequently, the window of sub-therapeutic drug exposure increases, and eventually the virus becomes highly resistant. high plasma concentrations of the PIs throughout the dosing interval is crucial to maximally suppress viral replication and delay the development of resistant viral strains. This may be difficult to achieve when PIs are used in combination with two NRTIs only. Due to the suboptimal pharmacokinetic properties of the PIs their plasma trough concentrations rapidly fall to levels very close to or below the in vitro concentrations shown to be 50% effective (IC 50 ) in the presence of 50% human serum as well as the proposed in vivo threshold concentrations, shown in Table 1. Because the PIs show a marked inter-patient variation in the pharmacokinetic profiles and hence the exposure to the drug, a considerable proportion of the patients on PIcontaining regimens are at an increased risk for selection of resistant viral strains and subsequent treatment failure because of, at least in part, subtherapeutic plasma PI concentrations (Figure 1). The apparent differences between effective concentrations in vitro and in vivo illustrate that the clinical relevance of comparing trough concentrations with in vitro effective concentrations is uncertain. Several factors contribute to the in vivo efficacy of antiretroviral drugs (for example, plasma protein binding, the ability to penetrate different tissues, the formation of active metabolites), which cannot be easily mimicked in vitro, thus hampering extrapolation of in vitro efficacy to in vivo efficacy. Drug drug interactions between certain combination of PIs result in dramatically increased plasma concentrations, and are exploited in an attempt to overcome the pharmacokinetic shortcomings of the PIs as single agents. The improved pharmacokinetics of the PIs when used in combination provide the potential for a reduction of the dose and the dosing frequency. Furthermore, simultaneous treatment with two PIs may provide additive or synergistic antiretroviral effects, and distinct, non-overlapping resistance patterns may theoretically create a higher genetic barrier to resistance, and thus hamper the development of resistant viral strains (Figure 2). Interactions involving cytochrome P450 The favourable pharmacokinetics of the PIs when used in combination (higher AUCs, longer half-lives) are probably predominantly caused by interactions through overlapping cytochrome P450 (CYP450) mediated metabolic pathways. The CYP450 mixed function oxidases are a family of enzymes that account for the majority of oxidative metabolic conversions of xenobiotics and endogenous biochemicals [39]. To date, over 30 different human CYP450 enzymes have been identified, of which CYP450 3A4 (CYP3A4) isoenzymes appear to be one of the most important representatives for the metabolism of many different classes of drugs [39]. The currently available PIs are Figure 2. Schematic presentation of strategies to improve the efficacy and durability of protease inhibitor therapy Drug level Incomplete suppression leads to resistance Increase potency Increase plasma levels Increase potency and increase trough levels International Medical Press

5 Combination of protease inhibitors metabolized for at least 80 95% by the CYP450 isoenzymes in the liver and small intestine [5]. Renal excretion plays only a minor role in the elimination of PIs [40]. For ritonavir and saquinavir less than 5% is excreted unchanged in the urine, for indinavir this percentage ranges from 10 15% [40]. In vitro the PIs are primarily biotransformed by CYP3A4 isoenzymes, and to a lesser extent by CYP2C9 and CYP2D6 [5]. The CYP3A4 content is highest in the liver, but this subfamily is also present in the small intestine, with the highest concentrations in the duodenum and with decreasing concentrations descending towards the colon. The CYP2C and CYP2D families have also been detected in human small intestine, although at substantially lower concentrations [41]. Extensive first-pass metabolism of PIs by CYP450 isoenzymes in the gut wall and the liver may contribute to the poor and variable bioavailability of this class of drugs (ranging from <5% for saquinavir to approximately 70% for ritonavir) [41]. In vitro studies revealed different capacities of the PIs for CYP450 inhibition [41 47]. Of the currently available PIs, ritonavir is the most potent inhibitor of CYP3A4 activity, followed by (in order of decreasing inhibiting capacity) indinavir, nelfinavir, amprenavir and saquinavir [44,47 49]. Amprenavir and saquinavir are weaker inhibitors of CYP3A4 metabolism [44,47,49]. The concentration of ritonavir that inhibits 50% of the metabolism (IC 50 ) of amprenavir, indinavir, nelfinavir and saquinavir in human liver microsomes has been estimated to be 0.10, 2.2, 0.62 and 0.25 µm, respectively [46]. Since PIs are primarily eliminated presystemically by CYP3A4, co-administration of a CYP3A4-inhibiting PI results in a positive drug drug interaction, which, depending on the combination, increases the exposure to either one or both of the PIs involved [50]. The mutual effects of PIs on CYP3A4 metabolism appear to be a result of competitive inhibition by rapid, reversible binding of the drug or its metabolite to CYP3A4 [39]. The inactivation of CYP3A4 by ritonavir occurs via the formation of a reactive metabolite that irreversibly inactivates cytochrome activity [42]. For nelfinavir, the inhibitory effect has been associated with the formation of reactive intermediates that reversibly inhibit CYP3A in a time- and concentrationdependent manner [51]. Inactivation of intestinal and hepatic CYP3A during absorption of an oral dose of nelfinavir or ritonavir increases drug bioavailability and plasma elimination half-life, resulting in a more favourable pharmacokinetic pattern compared with the other PIs (allowing for twice-a-day dosing). Of note, nelfinavir and ritonavir have wide variable affects on CYP3A metabolism, including both induction and inhibition, which complicates prediction of interactions in vivo, based on in vitro data [43]. Since the clearance of both PIs increases approximately twofold with multiple dosing during the first 2 weeks of therapy, thereby partly counteracting inhibiting effects [52 54], interactions between nelfinavir or ritonavir and other drugs are complex and cannot easily be predicted from in vitro experiments or from single-dose drug drug interaction studies [43,55]. Interactions involving P-glycoprotein Inhibition of CYP3A by PIs largely explains the improved pharmacokinetics of a co-administered PI. However, modulation of P-glycoprotein (P-gp) may also play a role in this respect, since many substrates metabolized by CYP3A4 are also substrates for P-gp [39,56]. P-gp is a transmembrane glycoprotein that functions as an energy-dependent efflux pump for a wide variety of structurally unrelated compounds, amongst which are the PIs [57 63]. P-gp is expressed in the epithelial cells of the gastrointestinal tract, liver, kidney, blood brain barrier and CD4 lymphocytes, and may therefore affect the absorption, distribution and elimination of PIs [63,64]. P-gp in the gastrointestinal tract transports substrates from the epithelial cells back into the lumen, and thus may contribute to the limited oral bioavailability of some of the PIs. Furthermore, it may function to secrete drugs, present in the systemic circulation, into the intestinal lumen, and thereby contribute to the elimination of the drug [65,66]. Hence, inhibition of intestinal P-gp may increase the oral bioavailability and decrease the clearance of PIs. This concept is already being used successfully in oncology where the combination of cyclosporin A (a potent inhibitor of P-gp) and the poorly bioavailable cytostatic agent, paclitaxel, resulted in an eightfold increase in the oral bioavailability of paclitaxel, thus enabling oral treatment with this drug [67,68]. After oral administration of indinavir, nelfinavir or saquinavir to P-gp-deficient mice (mdr1a/1b knockout mice) plasma concentrations were 2 5 fold higher than wild-type mice, suggesting a role of P-gp in the limited oral bioavailability of PIs [57]. Results from several in vitro studies also suggest that ritonavir, nelfinavir, indinavir and saquinavir are substrates/inhibitors of P-gp, with ritonavir being the most potent [58 61]. Ritonavir has been reported to have P-gp modulating properties that are at least as good as cyclosporin A in vitro [61]. Chemical modulation of P-gp (for example, by a PI) may therefore contribute to the improved bioavailability of co-administered PIs [69]. However, findings from a recent study using P-gp-deficient mice concluded that ritonavir was a poor P-gp inhibitor, even at high doses [70]. It was therefore concluded that the improved bioavailability of PIs in combination Antiviral Therapy 6:4 205

6 RPG van Heeswijk et al. with ritonavir may primarily result from CYP3A4 inhibition [70]. These contrasting results warrant further studies to unravel the relative contribution of CYP3A4 and P-gp to the improved bioavailability of PIs when used in combination. Furthermore, P-gp may play a role in the penetration of PIs into sanctuary sites for HIV-1 replication, such as the central nervous system (CNS) and the testes [63]. Poor penetration of antiretroviral drugs into these sites may explain the development of resistant HIV-1 isolates due to suboptimal suppression of HIV-1 replication [71,72]. Since only the unbound fraction of the PIs is available to cross membranes and penetrate sanctuary sites, the high plasma protein binding (for example, 60% for indinavir and >95% for the other PIs) limits disposition of these drugs [71]. Drug concentrations in cerebrospinal fluid (CSF) are often used as a measure of antiviral potency in the CNS. It is important to note, however, that CSF concentrations do not necessarily represent brain concentrations and that protein binding within, for example, CSF may be very different from (and much lower than) plasma, which makes the drug concentrations in various sanctuary sites difficult to interpret [73]. Studies with P-gp-deficient mice show a three- to sevenfold increased exposure to saquinavir in the brain tissue [57,70,74], and a 10- and 36-fold increase in the exposure to indinavir and nelfinavir, respectively [57]. The concentration of amprenavir in blood, brain and testes increased 1.3-, 27-, and fourfold in P-gp-deficient mice compared with control animals [62]. Pretreatment of wild-type mice with ritonavir had little affect on the concentrations of orally administered amprenavir and saquinavir in brain tissue [62,70]. However, the amprenavir concentration in blood and the testes increased 1.8- and 2.7-fold after pretreatment of wild-type mice with ritonavir [62]. Several studies reported low or undetectable PI concentrations in CSF and semen of patients treated with nelfinavir, saquinavir, ritonavir or a combination of two of these PIs [75 79]. In the Prometheus Study, the combination of saquinavir and ritonavir could not suppress HIV-1 RNA in CSF after 12 weeks of treatment [75]. Indinavir, on the other hand, reaches effective concentrations in the CSF. Interestingly, pharmacokinetic modelling suggested that indinavir is actively transported out of the brain [80]. Furthermore, co-administration of ketoconazole (an inhibitor of both CYP3A and P-gp) [56] in nine HIV-1-infected patients treated with saquinavir/ritonavir 400/400 mg twice-a-day resulted in a significantly larger increase in the CSF levels of the PIs relative to the increase of the total or unbound plasma concentration, which may be explained by P-gp inhibition [81]. After co-administration of ketoconazole the AUC of ritonavir and saquinavir in plasma increased by 30% and 36%, respectively. Ritonavir concentrations in CSF increased by 153%, and saquinavir was quantifiable (>0.2 ng/ml) in four out of nine patients when ketoconazole was co-administered, but CSF concentrations of both PIs remained below the protein-free IC 50 for wild-type virus [81]. The expression of P-gp in lymphocytes may indicate that it plays a role in the intracellular concentration of PIs, which is mediated by cellular influx, metabolism and efflux. The intracellular concentration of a PI (that is, at the site of HIV-1 replication) has been shown to be directly related to the antiviral activity [82]. Several studies have shown that indinavir, nelfinavir, ritonavir and saquinavir are less effective in P-gp expressing cells in vitro, at least at low dosages, and that inhibition of P-gp restored the antiviral activity [60,83,84]. Marked differences in the intracellular pharmacokinetics of the PIs in vitro have been reported, especially with regards to the cellular efflux, which was consistently slower for saquinavir and nelfinavir compared with amprenavir, indinavir and ritonavir [85 87]. These in vitro observations were recently confirmed in HIV-infected patients [88]. Differences in intracellular pharmacokinetics result in striking differences in the kinetics of restoration of HIV-1 infectivity after removal of extracellular drug from the cell cultures (post-antiviral effect) [85]. Furthermore, the durability of inhibition of HIV-1 replication in vitro is positively correlated to the intracellular PI concentration [85]. Modulation of cellular influx-efflux of PIs may thus be a way to optimize the intracellular concentrations and hence antiviral activity. Ritonavir has been shown to inhibit the cellular efflux of amprenavir and indinavir in vitro, allowing a 10- to 15-fold increase of the intracellular concentrations of these PIs [87]. Although ritonavir did not significantly increase the intracellular accumulation of saquinavir in HIV-infected patients [88], in vivo studies investigating the concept of modulating intracellular pharmacokinetics of PIs are warranted. However, the clinical implications of such treatment strategy are uncertain since the clinical efficacy of indinavir is well established despite the lack of intracellular accumulation in vivo [88], suggesting that high intracellular exposure to PIs may not be of paramount importance for long-term suppression of viral replication. In conclusion, the specific affects of the different PIs on P-gp function and vice versa are currently unclear and more research is necessary in order to further rationalise the combination of PIs in HAART. The relative contribution of P-gp modulation to the favourable pharmacokinetics of currently used PI combinations compared with inhibition of CYP3A4, International Medical Press

7 Combination of protease inhibitors may be overestimated and remains to be established. Inhibition of P-gp, by a co-administered PI or a specific P-gp-inhibitor, may theoretically be an attractive option to further improve the efficacy of antiretroviral therapy. However, continuous blockade of P-gp may not be without risks, and further research on this subject is warranted. Furthermore, other multi-drug transporters that may affect the pharmacokinetics of PIs should be considered, since it has been shown that PIs are not only substrates for P-gp but also for the multi-drug resistance-associated proteins 1 and 2 (MRP1 and MRP2) [61,84]. Antiretroviral synergy between protease inhibitors Besides the pronounced improvement in the pharmacokinetics of PIs when used in combination, additive or synergistic antiretroviral activity may provide an additional rationale for the simultaneous administration of two (or more) PIs. For the combinations of ritonavir with either saquinavir or nelfinavir, statistically significant synergism has been reported in vitro using MT4-cells infected with HIV IIIB in the presence of 50% human serum [89,90]. This synergism contrasts with the findings of Patick et al., who reported only additive effects for these combinations of PIs [91]. Furthermore, combination of ritonavir and indinavir or amprenavir resulted in an additive or slightly synergistic antiviral effect [89,90]. Amprenavir exhibits synergistic anti- HIV-1 activity with saquinavir, and additive activity with indinavir and nelfinavir in vitro [6]. The mechanism of in vitro synergy is currently unclear. Proposed mechanisms include modulation of P-gp and/or protein binding interactions [90]. Although indinavir has been shown to increase the exposure to saquinavir by five- to eightfold in a singledose pharmacokinetic study [92], this combination has not been evaluated in HIV-1-infected patients, since in vitro antagonism has been reported [93]. The interaction between nelfinavir and indinavir also appears to be slightly antagonistic [91]. Importantly, in vitro antagonism does not preclude a combination of two PIs in vivo. However, the potency of the combination will be less than expected in case their affects were additive [93]. Development of resistance for protease inhibitors Failure to maintain suppression of viral replication is associated with the appearance of drug-resistant viral strains in many patients [28]. Virological failure is, however, complex and not exclusively mediated by viral resistance but also by patient adherence and potency of the therapeutic regimen [94]. Several studies have shown that viral rebound occurred more often in patients who were assigned to less potent maintenance therapy after successful suppression of viral replication with triple- or quadruple-drug induction therapy [95 97], without readily proven resistance to antiretroviral drugs in the regimen [98,99]. Development of resistant viral strains severely compromises future treatment options due to cross resistance within each class of antiretroviral drugs [100]. Based on a daily virus production of virions and the inherent high mutational rate of HIV-1, it has been estimated that every base of the 10 4 nucleotidelong genome of HIV-1 may be prone to mutation every day [100,101]. The development of phenotypic resistance to PIs has been shown to require multiple mutations in the HIV-1 protease genome, which emerge in a stepwise fashion as a consequence of any residual virus replication in the presence of antiretroviral drugs [28,100,102]. Accordingly, the duration of suppression of viral replication by HAART has been shown to depend on the concentration of HIV-1 RNA at the nadir [103,104]. However, even in patients on HAART with less than 50 HIV-1 RNA copies/ml in plasma for up to 3 years, ongoing residual HIV-1 replication and, as a result, changes in the genetic sequence have been reported [105]. So far, 42 different PI resistance associated mutations have been reported from in vitro and in vivo studies with licensed and experimental drugs [100]. Based on their order of accumulation and their location on the HIV-1 protease genome, mutations are referred to as primary mutations (initially appearing, affecting the active site of the protease enzyme, except for mutations at residue 90), or secondary mutations (occurring outside the active site of the enzyme). Each PI has distinct primary mutations that confer to a twoto 10-fold decrease in in vitro susceptibility [106]. The primary mutations selected in vivo are located at residues 50 for amprenavir, 46 and/or 82 for indinavir, 30 for nelfinavir, 82 and/or 84 for ritonavir (acquisition of both primary mutations can result in a 10- to 20-fold decreased susceptibility) and 48 and/or 90 for saquinavir (acquisition of both mutations can result in >100-fold decreased susceptibility) [102,106]. Acquisition of (multiple) secondary mutations results in a dramatically decreased susceptibility and varying degrees of cross resistance between the PIs [100,102, ]. The rate of resistance development depends on the genetic barrier, the number of mutations required to render the virus insensitive for the drugs used and the potency of the treatment regimen. Thus, combinations of PIs may be beneficial, since the potency as well as the genetic barrier of the regimen are increased. Except for combinations with a low dose of ritonavir, therapeutic benefit may be expected from two PIs, thereby increasing the potency of the regimen. In addition, the favourable pharmacokinetics result in increased Antiviral Therapy 6:4 207

8 RPG van Heeswijk et al. plasma concentrations (increased potency) throughout the dosing interval, which is important since the rate of accumulation of mutations has been shown to be inversely correlated with plasma ritonavir trough concentration during monotherapy (Figures 1 and 2) [28]. High plasma concentrations may counteract moderate phenotypic resistance, and hence prolong suppression of viral replication. Furthermore, combinations of PIs increase the genetic barrier because of the distinct primary mutations associated with each of the PIs [102,106]. Theoretically, this finding may not be applicable to the combination of indinavir and ritonavir because of the similarities in their respective primary resistance patterns (that is, mutation at position 82 of the protease genome) [106,109]. Any residual replication of HIV-1 may therefore lead to failure of both drugs simultaneously. Bioanalysis and therapeutic drug monitoring of protease inhibitors The reported relation between the exposure to PIs and their activity and toxicity [28 31,33 35,110,111], in combination with the wide inter-individual variability in the pharmacokinetics of the PIs, has resulted in a growing interest in the therapeutic drug monitoring (TDM) of antiretroviral drugs as an additional tool in the management of HIV-1-infected patients [ ]. TDM of antiretroviral drugs might prove to be of benefit by facilitating timely dosage adjustments based on individual plasma concentrations, thus preventing subtherapeutic or toxic-drug exposure. This benefit is illustrated by studies reporting suboptimal plasma PI concentrations in approximately 30% of the patients on PI-containing regimens, and successful plasma concentration-guided adjustments of indinavir doses in preventing indinavir-related urological toxicity, while maintaining suppression of viral replication [ ]. Prospective controlled trials evaluating the contribution of TDM in the optimisation of antiretroviral therapy are underway [115,118]. Several relatively simple techniques, which can readily be used in a hospital laboratory for the measurement of up to five PIs simultaneously, have been described [ ]. Sample pretreatment consists of either solid-phase extraction or liquid liquid extraction, before reversed-phase high performance liquid chromatography (HPLC) with isocratic or gradient elution to separate the PIs from interfering endogenous compounds. All assays use UV detection, at either one or two wavelengths. Of note, there is up to a 10-fold variation in the lower limit of quantification for the respective PIs between the assays. In particular, the assay by Marzolini et al. [123], which can measure five PIs plus the NNRTI efavirenz, may not be sensitive Figure 3. Steady-state plasma concentration versus time curves of saquinavir and ritonavir in HIV-1-infected patients (a) Plasma saquinavir concentration (µg/l) (b) Plasma ritonavir concentration (µg/ml) Time (h) 3x1200 mg 2x400 mg (+ 2x400 mg RTV) 1x1600 mg (+ 1x100 mg RTV) Time (h) 2x600 mg 2x400 mg (+ 2x400 mg SQV) 1x100 mg (+ 1x1600 mg SQV) (a) Saquinavir (SQV) hard-gelatin capsules 1200 mg three-times-a-day (n=20), SQV hard-gelatine capsules 400 mg twice-a-day in combination with ritonavir (RTV) 400 mg twice-a-day (n=12), and SQV soft-gelatine capsules 1600 mg once-a-day in combination with ritonavir 100 mg once-a-day (n=12). (b) RTV 600 mg twice-a-day (n=5), RTV 400 mg twice-a-day in combination with SQV hard-gelatine capsules 400 mg twice-a-day (n=12), and ritonavir 100 mg once-a-day in combination with SQV soft-gelatin capsules 1600 mg once-aday (n=12) in HIV-1-infected patients. enough for the quantification of trough concentrations in patients on single-pi containing regimens. Double protease inhibitor regimens Saquinavir plus ritonavir Pharmacokinetics. The combination of saquinavir and ritonavir is the most extensively studied double PI regimen, both as first-line therapy and as part of salvage regimens for patients failing other PIcontaining regimens. Saquinavir is the most potent PI in vitro, but its poor bioavailability (approximately 4% for the hard-gelatin capsules) and the corresponding low plasma concentrations advise against its use as a single PI [4,21,102,125,126]. To overcome these pharmacokinetic shortcomings, a soft-gelatin capsule formulation of saquinavir has been developed with a bioavailability of 331% relative to the hardgelatin capsules, and could be used as a single PI in HAART in a dosage of 1200 mg twice-a-day with food [1,14]. However, both formulations of saquinavir are International Medical Press

9 Combination of protease inhibitors Table 2. Mean steady-state plasma pharmacokinetics of double protease inhibitor combinations in HIV-1-infected patients Combination Dose Frequency AUC dosing interval (µg/ml h) Cmax (µg/ml) Cmin (µg/ml) No. of patients in study References Saquinavir (HGC) 400 mg bid [132] Ritonavir 400 mg Indinavir 400 mg bid [162,163] Ritonavir 400 mg NA Nelfinavir 750 mg bid [185] Indinavir 1000 mg Nelfinavir 1000 mg bid [186] Indinavir 1000 mg Nelfinavir 500 mg bid 28* NA NA 10 [13] Ritonavir 400 mg 60* NA NA Nelfinavir 1250 mg bid [170] Saquinavir (HGC) 1000 mg AUC, area under the plasma concentration versus time curve during a dosing interval; Cmax, maximal plasma concentration; Cmin, minimal plasma concentration; HGC, hard-gelatin capsules; NA, not available; bid, twice-a-day. *Median value. best used in combination with ritonavir to enhance the pharmacokinetic properties of this potent PI (Figure 3) [1]. The exposure to saquinavir (either as a hard-gelatin or soft-gelatin capsule) is markedly increased when coadministered with ritonavir, even at dose of a 100 mg, allowing for a convenient twice-a-day dosing regimen, while maintaining high plasma concentrations throughout the dosing interval [127,128,130,131]. Furthermore, ritonavir has been shown to reduce the inter-subject variability in the saquinavir AUC from 60 to 28% [55]. The effect of saquinavir on the exposure to ritonavir has been shown to be negligible (+6.4% in AUC) in single dose studies on healthy volunteers [55]. The most widely used combination consists of 400 mg saquinavir and 400 mg ritonavir in a twice-a-day regimen. However, the influence of food on the bioavailability of saquinavir in this combination with ritonavir is currently unknown. In this combination, the steady-state AUC of saquinavir in HIV-1-infected patients is increased by six- to 20-fold when compared with patients receiving saquinavir hard-gelatin capsule 600 mg three-times-a-day [14,130,132], while the AUC of ritonavir has decreased dose-proportionally (Table 2) [11,13,130,132]. Both the increased exposure to saquinavir and the decreased exposure to ritonavir (Figure 3) contribute to the popularity of this combination. Higher exposure to saquinavir produces a greater and more durable suppression of viral replication and increase in CD4 cell counts, without development of severe toxicity [29,33,133]. On the other hand, high exposure to ritonavir has been associated with an increased risk of neurological and gastrointestinal side-effects [111]. It has been shown that therapeutic regimens containing ritonavir as the only PI (in a dosage of 600 mg twice-a-day) are changed in 37 55% of patients because of ritonavirrelated toxicity (mainly nausea and vomiting), which occurs more often than in regimens that include indinavir (29%), saquinavir (22%), or ritonavir plus saquinavir (14 20%) [21, ]. These findings suggest that the lower exposure to ritonavir in the saquinavir/ritonavir 400/400 mg twice-a-day regimen results in a better tolerance than the ritonavir 600 mg twice-a-day regimen. Important within this respect is the recent finding that a similar proportion of PI-naive patients achieved HIV-1 RNA concentrations below 400 copies/ml after randomization to 24 weeks of treatment with a reduced dose of ritonavir (400 mg twice-a-day; 26/30, 88%) or the standard dose (600 mg twice-a-day; 17/24, 78%), in combination with stavudine and lamivudine [137]. These results, although based on small numbers, suggest similar potency for both doses of ritonavir in PI-naive patients. In a 2-week multiple dose pharmacokinetic study in healthy volunteers it was shown that doubling the dose of saquinavir from 400 to 800 mg, in combination with ritonavir 400 mg, did not significantly affect the exposure to saquinavir. The mean AUC 24h (area under plasma concentration versus time curve from 0 24 h) of saquinavir in a dose of 400 and 800 mg, in combination with 400 mg ritonavir, was 36 and 38 mg/ml h, respectively [138]. Increasing the dose of ritonavir from 400 to 600 mg, in combination with saquinavir 600 mg, resulted in a 1.8-fold and 1.5-fold increase in the AUC 24h of ritonavir and saquinavir, respectively [138]. These results suggest non-linear pharmacokinetics of ritonavir and saquinavir in this combination. Antiviral Therapy 6:4 209

10 RPG van Heeswijk et al. Table 3. Virological and immunological response to double protease inhibitor combinations in protease inhibitor-naive HIV-1- infected patients Median baseline No. of patients Median change from baseline plasma log 10 Median baseline with plasma Plasma log 10 CD4 cell No. of HIV-1 RNA CD4 cell count Follow-up HIV-1RNA <400 HIV-1 RNA count Drug regimen patients (copies/ml) (cells/µl) (weeks) copies/ml* (%) (copies/ml) (cells/µl) Reference SQV/RTV 400/400 mg bid (65%) 266 [129] (±2 NRTIs after 48 weeks) SQV/RTV 400/400 mg bid /32 (70%) [148] + 1 NRTI versus IDV 800 tid + 2 NRTIs /36 (83%) SQV/RTV 400/400 mg bid (63%) 185 [147] versus SQV/RTV 400/400 mg + d4t (69%) 185 IDV/RTV 400/400 mg bid /24 (96%) [166] + d4t/3tc IDV/RTV 400/400 mg bid /92 (66.3%) [167] + 2 NRTIs NFV/SQV 1250/1200 bid /280 (41%) 225 [172] + 2 NRTIs NFV/SQV 750/600 tid 15 > (47%) [173] + d4t/ 3TC versus NFV/SQV1250/1000 mg 17 > (65%) bid + d4t/3tc NFV/RTV 500/400 mg bid [13,182] NFV/RTV 750/400 mg bid NFV/IDV 750/1000 mg bid 18 >4.47 > (61%) 133 [187] *Intent-to-treat analysis unless stated otherwise. No. of patients (%) with a HIV-1 RNA concentration below 200 copies/ml. On-treatment analysis. Mean value. No. of patients (%) with a HIV-1 RNA concentration below 80 copies/ml. No. of patients (%) with a HIV-1 RNA concentration below 50 copies/ml. SQV, saquinavir; RTV, ritonavir; IDV, indinavir; NFV, nelfinavir; NRTI, nucleoside analogue reverse transcriptase inhibitor; d4t, stavudine; 3TC, lamivudine; bid, twice-aday; tid, three-times-a-day. Based on a combined analysis of 120 subjects receiving different dose combinations of saquinavir ( mg) and ritonavir ( mg) it was concluded that ritonavir increases plasma saquinavir concentrations by inhibition of first-pass metabolism [resulting in an increased maximal plasma concentration (Cmax)], but with no subsequent effect on the plasma half-life [139]. The increase in Cmax was similar for ritonavir doses ranging from 100 to 400 mg twice-a-day [139]. Recently, Gisolf et al. reported a decrease in the exposure to saquinavir in patients who were on longterm therapy, including the combination of saquinavir/ritonavir 400/400 mg twice-a-day [140]. The steady-state plasma pharmacokinetics of saquinavir and ritonavir were assessed in six HIV-1- infected patients on two occasions, the second time was 9 15 months after the first. The median AUC of saquinavir was 33% lower on the second occasion than on the first, while the AUC of ritonavir was similar. Proposed explanations for this observation included induction of activity or expression of P-gp, or other transport molecules, or changes in CYP450 activity [140]. Induction of P-gp expression by PIs in cell culture was recently reported by Perloff et al. [141]. The combination of saquinavir and ritonavir enables simultaneous administration of PIs and rifampin for the treatment of HIV-1-infected patients co-infected with Mycobacterium tuberculosis [142]. M. tuberculosis infection remains an important problem in HIV-1-infected patients, and poses treatment difficulties for physicians because of drug drug interactions with rifamycins, which induce CYP3A4 metabolism, resulting in subtherapeutic plasma concentrations of co-administered PIs (for example, International Medical Press

11 Combination of protease inhibitors Table 4. Virological and immunological response to double protease inhibitor combinations in protease inhibitor-pretreated HIV-1-infected patients Median baseline No. of patients Median change from baseline plasma log 10 Median baseline with plasma HIV-1 Plasma log 10 CD4 cell Previous No. of HIV-1 RNA CD4 cell count Follow-up RNA <400 copies/ HIV-1 RNA count Drug regimen PIs patients (copies/ml) (cells/µl) (weeks) ml* (%) (copies/ml) (cells/µl) Reference SQV/RTV 400/ SQV (50%), (49%) 227 [153] 400 mg bid + RTV (14%), AZT/3TC or IDV (2%) d4t/3tc SQV/RTV 400/ RTV, IDV Only 4 patients [154] 400 mg bid + achieved >0.5 2 NRTIs log decrease SQV/RTV 400/ IDV 23 >3.80 > (57%) [152] 400 mg bid + 2 NRTIs SQV/RTV 400/ NFV (71%) 141 [155] 400 mg bid + d4t/3tc IDV/RTV 400/ IDV /25 (40%) 52 [162] 400 mg bid + 2 NRTIs NFV/SQV RTV, IDV (24%) [180] 1250/1200 mg bid +NRTIs/ NNRTIs *Intent-to-treat analysis unless stated otherwise. No. of patients (%) with a HIV-1 RNA concentration below 500 copies/ml. On-treatment analysis. No. of patients (%) with a HIV-1 RNA concentration below 200 copeis/ml. PI, protease inhibitor; SQV, saquinavir; RTV, ritonavir; IDV, indinavir; NFV, nelfinavir; AZT, zidovudine; d4t, stavudine; 3TC, lamivudine; NRTI, nucleoside analogue reverse transcriptse inhibitor; NNRTI non-nucleoside analogue reverse transcriptse inhibitor; bid, twice-a-day. rifampin decreases PI concentrations by 35 92%) [ ]. The results by Veldkamp et al., although based on only a limited number of patients, suggest that the strong inhibiting effect of ritonavir on saquinavir metabolism counteracts the enzymeinducing effect of co-administered rifampin, thereby maintaining therapeutic saquinavir concentrations [142]. This strategy may be of paramount importance for the treatment of tuberculosis in HIV-1-infected patients and should be investigated further. Similarly, it was shown that in combination with ritonavir, saquinavir concentrations were not affected by the CYP3A4 inducers rifabutin and efavirenz, which decrease saquinavir concentrations by 45% and 60%, respectively, in the absence of ritonavir [132,144,146]. Clinical experience. Several clinical trials have established the safety and efficacy of the combination of saquinavir/ritonavir 400/400 mg twice-a-day in antiretroviral-naive and -experienced patients (Tables 3 and 4, respectively). In a randomized trial comparing the safety and efficacy of four different dose combinations of saquinavir and ritonavir, 400 mg of each PI in a twice-a-day regimen was the best tolerated combination, and was equally effective as higher dose combinations in PI-naive patients (n=141) after 48 weeks follow-up [130]. The Prometheus Study investigated the effect of treatment with ritonavir/saquinavir 400/400 mg twice-a-day plus stavudine versus ritonavir/saquinavir alone, with treatment intensification if needed, in PI- and stavudine-naive HIV-1-infected patients (Table 3) [147]. Overall, no significant difference between the treatment arms was observed, with respect to the number of patients with less than 400 HIV-1 RNA copies/ml in plasma at 48 weeks of therapy. However, in patients with a high baseline HIV-1 RNA concentration (>5 log 10 copies/ml), the triple drug treatment was superior to the double PI regimen. Most frequently reported side-effects were diarrhoea (18%, median duration 7 weeks) and nausea (13%, median duration 1 week). Ten percent of the patients discontinued the study due to adverse events [147]. Antiviral Therapy 6:4 211

12 RPG van Heeswijk et al. In extensively PI-pretreated patients, baseline resistance to ritonavir, saquinavir or both has been associated with a poor antiviral response to treatment with saquinavir/ritonavir, 400/400 mg twice-a-day [149,150]. Patients naive to PIs were reported to be almost seven times more likely to suppress their plasma HIV-1 RNA concentrations to below 500 copies/ml, after starting treatment with saquinavir/ritonavir 400/400 mg twice-a-day, than those who were PI experienced [151]. Salvage therapy seems to be more effective when initiated early in failure at low plasma HIV-1 RNA concentrations [152]. Data on the efficacy of saquinavir/ritonavir 400/400 mg twice-a-day in PIexperienced patients are limited and often based on small numbers of patients (Table 4). Promising results were reported by Tebas et al. who studied the efficacy of saquinavir/ritonavir 400/400 mg twice-a-day in patients who had a detectable plasma viral load during nelfinavir treatment for a median of 48 weeks [155]. After 24 weeks 17/24 patients (71%) achieved a HIV-1 RNA concentration below 500 copies/ml, and 10 of these patients (59%) also achieved a plasma viral load of less than 50 copies/ml [155]. Patients failing a saquinavir ritonavir salvage regimen, after previous failure of an indinavir-containing salvage regimen, do not respond virologically to a subsequent PI-containing regimen. However, the CD4 cell count may continue to increase during treatment with saquinavir/ritonavir [156]. Nearly 70 80% of the viral isolates from 16 patients failing a rescue regimen with saquinavir/ritonavir were resistant to all available PIs (with a median number of 12 mutations associated with PI resistance), although nelfinavir had not been used before [156]. These results emphasise the need for new treatment options for the growing number of heavily pretreated patients. Indinavir plus ritonavir Pharmacokinetics. The recommended dosage of indinavir is 800 mg orally every 8 h. For optimal absorption, indinavir should be administered with water 1 h before or 2 h after a meal. Compared with ingestion on an empty stomach, ingestion of indinavir with a meal has been shown to reduce the AUC by 30 68% in HIV-1-infected patients, depending on the viscosity, and protein, carbohydrate and fat content of the meal [157]. Furthermore, patients taking indinavir must drink at least 1.5 l of fluid daily to ensure adequate hydration, in an attempt to reduce the risk of nephrolithiasis [7]. The safety and efficacy of indinavir-containing regimens has been well established with results showing durable suppression of viral replication for up to 3 years [3]. Patient adherence to an indinavir-containing regimen is, however, complicated by the three-times-a-day regimen and the food and drink restrictions. Co-administration of indinavir with (low dose) ritonavir has been shown to improve the overall pharmacokinetics of indinavir, allowing for a twice-aday administration of reduced dosages without food restrictions [ ]. With the widely used combination of indinavir 400 mg and ritonavir 400 mg in a twice-a-day dosing regimen, both drugs contribute to the antiviral potency of this combination. The AUC 24h of ritonavir, when used in this combination, was about 1.5-fold higher than when ritonavir was given alone (600 mg twice-a-day), in a multiple dose study in healthy volunteers [160], which may be explained by CYP3A4 inhibition by indinavir. The steady-state plasma AUC 24h for indinavir in this combination in HIV-1-infected patients is comparable to the AUC 24h observed with the licensed dosage (Table 2) [162]. However, the indinavir plasma trough concentration is increased by approximately fivefold, while the maximal concentration is approximately half, compared with the concentrations for the indinavir 800 mg three-times-a-day regimen [162]. The higher indinavir trough concentration is expected to enhance the potency of this drug, since several studies have shown relationships between indinavir trough concentrations and virological response [8,34,35]. The lower maximal concentration may attenuate the incidence of indinavir-related urological complications, which have been associated with elevated indinavir plasma concentrations [117]. The overall incidence of indinavir-induced nephrolithiasis is approximately 9% but a higher incidence has been reported in a warm climate [7,164]. A retrospective study of 89 patients, who had received indinavir/ritonavir 400/400 mg twice-a-day for a median period of 34 weeks, without additional hydration, revealed no cases of nephrolithiasis, haematuria, flank pain, or creatinine elevations >20% of baseline values [165]. These results suggest a significant reduction in the risk of urological complaints for the indinavir/ritonavir 400/400 mg twice-a-day regimen as compared with indinavir in the licensed dosage. In a multiple-dose study in healthy volunteers it was shown that for a constant indinavir dose ( mg), increasing the dose of ritonavir from mg twice-a-day does not significantly affect the pharmacokinetics [AUC, Cmax and minimal plasma concentration (Cmin)] of indinavir [159,161]. However, increasing the ritonavir dose from 100 mg to 200 mg does increase the exposure to indinavir, mainly due to a decreased post-absorptive clearance of indinavir resulting in an approximately 2.3-fold increase in the Cmin [161]. With a constant ritonavir dose, an increase in the indinavir dose resulted in a proportional increase in the AUC of indinavir, a less than proportional increase in the Cmax, and a more than International Medical Press