Introduction
Highly active antiretroviral therapy (HAART) allows patients infected with human immunodeficiency virus type 1 (HIV-1) to live productive and relatively disease-free lives for long periods. However, even patients who respond well initially may suffer disease progression when the effectiveness of the regimen wanes. Treatment may fail for a number of reasons, including poor adherence to drug regimens, insufficient drug potency, pharmacokinetic factors, and the emergence of drug-resistant virus; poor adherence has been cited as the leading cause of treatment failure [1-5].
Although many antiretroviral agents are now available, these drugs are prescribed in combinations of three or more agents; hence, the number of effective combinations is still limited. Moreover, the durability of virological response tends to decrease progressively with each new treatment, until the patient is left with no therapeutic options other than highly complex and experimental combinations. A study from The Netherlands revealed that more than one-half of treatment-naive patients switched from their first regimen within 1 year [6], highlighting just how tenuous control of this infection can be. Optimizing available options is therefore crucial to long-term viral suppression.
Monitoring of HIV-1 infection is critical to helping physicians understand why treatment fails and to thereby optimize individualized treatment [7]. CD4 cell count helps track the immunological status of patients, and viral load assays are used to monitor the effectiveness of treatment and assess the need to consider regimen changes. Resistance assays are gradually becoming the standard of care, and current guidelines recommend resistance testing in specific settings, although conclusive evidence of their clinical utility is still lacking, and expert interpretation of results is clearly needed [8,9]. Emerging evidence of the link between drug exposure and antiviral efficacy [10,11], as well as toxicity [12,13], has begun to focus attention on the role of monitoring plasma drug levels in patients receiving HAART. In an analysis of pharmacokinetic data from the Viradapt study, suboptimal drug concentrations (less than twice the concentration inhibiting 95% of virus [IC95]) correlated significantly with the risk of virological failure [14].
If inadequate drug levels arising from pharmacokinetic or adherence factors are a major cause of treatment failure then monitoring these levels would seem a logical approach to improve the success rate of therapy. Monitoring of in vivo drug concentrations and adjusting the dosage regimen on the basis of these concentrations is a well-known therapeutic intervention defined as therapeutic drug monitoring (TDM).
TDM is used in a variety of clinical settings to adjust drug levels within a therapeutic window. Common examples include anticonvulsants and aminoglycoside antibiotics. In the setting of HIV-1 management, however, the role of TDM remains to be established. Although its potential utility is intuitive, a number of issues need to be resolved to justify the application of TDM in routine clinical practice. Experts who are sceptical regarding the introduction of TDM in the routine management of the HIV-1-infected patients cite a number of reasons that merit examination [15-17]: (i) plasma levels of antiretroviral agents may show high intra-patient variability, meaning that a single sample might not be representative; (ii) because adherence is difficult to assess, a low plasma drug level may reflect that the patient has not been taking the drug according to schedule, rather than a need for increased dosage; (iii) the effect of serum protein binding is poorly understood; (iv) the therapeutic indexes of most anti-HIV drugs are poorly defined; (v) logistical issues (difficulty getting patients to the clinic for sampling, and long turnaround times for assays); (vi) requirement for expert interpretation; (vii) poor quality control of assays; and (viii) lack of randomized, controlled clinical trials indicating that dose adjustments based on TDM are of clinical benefit.
Despite these considerations, TDM for antiretroviral drugs is already available in a number of centres, primarily in the western United States and in Europe. In such centres, physician requests for plasma concentration monitoring have been increasing in recent years; thousands of plasma concentrations have been generated in an attempt to use TDM as an additional tool to guide treatment in specific clinical settings. The French guidelines for HIV treatment already recommend TDM in certain situations [18].
TDM is available in the United Kingdom, and usage patterns suggest that it is often requested in situations where drug levels might be expected to be influenced or when there is reason to suspect inappropriate drug concentrations. In order of frequency, the most commonly cited reasons for requesting TDM were: (i) non-recommended dose; (ii) suspected treatment failure; (iii) pharmaco-enhancement; (iv) possible drug interaction; (v) suspected toxicity; (vi) change in therapy; (vii) clinical indications; and (viii) paediatric patients [19].
Similarly, the national guidelines for Germany and Austria suggest that TDM has potential benefit in patients who initiate a new therapy, who are changing treatment, or who are experiencing treatment failure, toxic effects, or adherence problems [20]. A study of TDM in clinical practice in Germany revealed that, in an unselected patient population, more than 50% of the patients had suboptimal drug exposure, and that this frequently resulted in treatment modification [21].
Hence, it is evident that we are already witnessing the gradual introduction of TDM in clinical practice, at least in large reference centres. The situation resembles, on a smaller scale, the introduction of resistance testing: TDM is being used cautiously, in the absence of specific guidelines, with the intention of evaluating its practical utility and the 'rules' for its optimal application while collecting valuable experience in an ongoing remodelling and refinement process of the test procedures and interpretations.
The broad range of opinions on the utility of TDM in HIV management calls for a detailed review of the issues. In October 2000, a group of pharmacologists and virologists with expertise in TDM met in Perugia, Italy, to discuss the present status and future uses of TDM in the management of HIV-1 infection, and to reach a consensus where possible. This paper reviews the issues discussed and the conclusions drawn. It has been updated prior to submission to include material published or presented at conferences up to the end of August 2001.
Which drug characteristics suggest TDM utility?
Drugs with the following characteristics are ideal candidates for TDM: (i) available pharmacokinetic data; (ii) high inter-patient variability in plasma concentrations; (iii) low intra-patient variability in plasma concentrations; (iv) established (and narrow) therapeutic range; (v) close correlation between plasma concentration and concentration at the site of action; (vi) pharmacological effect that persists for a long time and is dependent on the plasma concentration; (vii) long-term patient outcome compromised by lack of effect; and (viii) availability of cost-effective, accurate drug assays with a rapid turnaround time and a small blood volume requirement [22]. Additionally, there should be evidence that drug levels can be successfully manipulated to target levels, and that such manipulation may improve response to therapy.
There is a wealth of pharmacokinetic data on most of the available antiretroviral agents that point to a need to carefully monitor drug levels. Plasma levels of antiretroviral agents vary markedly among individuals given the same dose, often by more than 10-fold [19,23-26](Fig. 1). Preliminary studies from centres that offer TDM have shown that a substantial proportion of patients may have subtherapeutic protease inhibitor (PI) levels [21,27,28].
;)
Fig. 1: High inter-individual variability in protease inhibitor trough concentrations found at the therapeutic drug monitoring service in Liverpool, UK
[19]. HGC, hard gel capsule; SGC, soft gel capsule.
One of the main assumptions in pharmacology is that therapeutic agents exert both their efficacy and their toxicity in a concentration-dependent manner. The branch of clinical pharmacology that studies the concentration-response relationship is called pharmacodynamics. When a pharmacodynamic relationship exists between plasma concentration and therapeutic response, the concentrations can be manipulated through dosage adjustment or other interventions to achieve concentrations within a desired therapeutic window. The therapeutic window is defined as the range of concentrations associated with the optimal efficacy/toxicity ratio (Fig. 2). A correlation between drug concentrations and both toxicity and efficacy parameters has been shown for antiretroviral agents. However, given the broad range of potentially confounding variables, it is not surprising that 'some but not all' seems to be the major theme in studies assessing concentration-response correlations for antiretroviral drugs. Also, the studies reporting a significant concentration-response relationship do not generally show a very strong correlation, suggesting the role of confounding factors.
Fig. 2: Efficacy versus toxicity: the therapeutic window. Analysis of the exposure-response curves for both efficacy and toxicity allows identification of the therapeutic window. By maintaining plasma drug concentrations within this window, patients should be able to achieve the optimal balance between efficacy and safety. EC50, concentration eliciting 50% of the maximal response; TC50, concentration eliciting toxicity in 50% of patients.
One of the major confounders may be incomplete adherence. While patients who do not take their anti-HIV drugs on schedule or do not comply with dietary requirements would be expected to have low plasma levels and poor outcomes, these patients may have 'normal' plasma levels if they take their doses soon before a scheduled visit. Thus, patients who fail treatment because of low drug levels due to poor adherence may actually appear to have normal drug levels in outcome analyses. Hence, optimal design of trials aimed at the evaluation of concentration-response relationships should always include a thorough assessment of patient adherence to treatment. Monitoring of drug concentrations may be a reliable tool for assessing adherence, particularly if used in conjunction with other methods [29].
Another confounding factor may be the development of resistance. The concentrations required to inhibit the replication of a wild-type virus may be lower than those necessary for a strain with decreased susceptibility. For this reason, measures of drug exposure normalized to measures of viral susceptibility may correlate better with treatment response (see 'Target levels and influence of covariables').
Combination therapy also obscures the relationship between drug level and outcome: most first-line regimens contain three potent antiretrovirals, so a patient may still respond favourably to treatment, especially in the short term, even though one of the drugs is at suboptimal levels. Other covariables that may obscure a significant concentration-response relationship include protein binding and intracellular kinetics (see 'Target levels and influence of covariables').
Despite the numerous obstacles to observing a clear relationship between concentration and effect, numerous studies have demonstrated such correlations. A brief review of such studies follows.
Nucleoside reverse transcriptase inhibitors
Nucleoside reverse transcriptase inhibitors (NRTIs), which are converted to their active triphosphate forms within cells, may be at low or undetectable levels in plasma but at suppressive concentrations within cells. Because plasma levels may not be a good indicator of intracellular triphosphate concentrations, TDM is considered of limited value for this drug class. Although a significant relationship between the intracellular triphosphate moiety levels and outcome parameters for zidovudine and lamivudine has been shown [30], it is not feasible to measure the concentrations of intracellular triphosphate anabolites in routine clinical practice because of the complexity and cost.
In spite of the low correlation of plasma drug levels with response for NRTIs, plasma levels of zidovudine have been used as a basis for a concentration-controlled trial that yielded compelling results. The concept of concentration-controlled trials has been introduced to limit the influence of plasma concentration variability on the response variability [31]. As opposed to standard clinical trials in which the patients receive a fixed dose, patients in a concentration-controlled trial may receive a modified dosage regimen if their plasma concentrations are outside a pre-determined range. In other words, patients in such trials are randomized to a concentration range rather than to a particular dose. In a randomized, 24-week crossover study, concentration-controlled zidovudine therapy with a target plasma level of 0.7 μmol/l (versus the standard 500 mg/day dose) yielded better CD4 cell count responses than the standard of care (22% increase versus 7% decrease), a more rapid viral load decrease, and higher systemic [0.76 μmol/l (coefficient of variation, 12%) versus 0.62 μmol/l (coefficient of variation, 32%)] and intracellular (160 fmol versus 92 fmol/106 peripheral blood mononuclear cells) levels of zidovudine [32]. Concentration-controlled dosing also decreased inter-patient variability.
In a study in paediatric patients receiving triple-combination therapy with indinavir, didanosine and stavudine, the viral load decrease at 24 weeks of treatment was significantly correlated with didanosine under the plasma concentration versus time curve [33].
Even though a relationship between plasma concentrations of NRTI and outcome has been found in some studies, as already reported, the usefulness of NRTI plasma concentrations in predicting treatment response remains uncertain. Whether NRTI concentration monitoring may be useful for assessing patient adherence or for toxicity management should be thoroughly evaluated.
Non-nucleoside reverse transcriptase inhibitors
A common argument against TDM of non-nucleoside reverse transcriptase inhibitors (NNRTIs) is that the plasma concentrations of these drugs vary little during a dosing interval due to their long half-lives, and that these concentrations are adequate in most patients. However, studies have shown high inter-individual variability in plasma drug concentrations and a correlation between plasma levels and response; two characteristics that, taken together, strongly suggest a role for TDM for this drug class.
Nevirapine
In a study of nevirapine monotherapy, a high dose level (400 mg/day) yielded a more prolonged antiviral effect than did lower doses, as well as an increased rate of adverse events such as rash, thus suggesting a relationship between plasma levels and efficacy and toxicity [34]. Initial data indicate that a 400 mg once-daily regimen and a 200 mg twice-daily regimen yield equivalent daily exposure to nevirapine [24-h area under the curve (AUC)], although the once-daily regimen produces a higher maximum concentration achieved during a dosing interval (Cmax) and a lower minimum concentration during a dosing interval (Cmin), which usually occurs just before the next dose or shortly thereafter [35]. While the impact of these differences on clinical effectiveness and adverse reactions was not examined, the wide range between Cmax and Cmin suggests that TDM may be of particular value in once-daily regimens, which may, at least theoretically, be more prone to selection for nevirapine-resistant quasi-species.
Results of the INCAS trial suggest the usefulness of TDM for nevirapine. In this study, higher values of nevirapine exposure were associated with greater initial clearance of HIV-1 RNA from plasma, a higher probability of achieving undetectable levels of plasma HIV-1 RNA, and a more sustained suppression of HIV-1 replication [36,37].
Efavirenz
Because of its long half-life, efavirenz maintains a relatively stable plasma concentration between doses. Although this characteristic might limit the value of TDM for efavirenz, two pieces of evidence suggest that TDM could still be of benefit. A retrospective analysis of efavirenz treatment response in phase II clinical trials indicated a significant relationship between suboptimal plasma concentrations and treatment failure: failure was three times as probable (63% versus 21%) in patients with plasma Cmin levels below 3.5 μmol/l [38].
In a recent study [26], plasma concentrations of efavirenz varied broadly among patients, with a concentration range of over 1 log for each time point of the concentration versus time curve (Fig. 3). Plasma concentrations were also associated with the probability of viral suppression and central nervous system adverse effects. These findings allowed the authors to construct both a concentration-virological efficacy curve and a concentration-toxicity curve to define a possible therapeutic range (Fig. 4). The authors estimate that approximately 20% of patients will fall outside the suggested therapeutic window of 1-4 μg/ml. This raises the question of whether TDM should be performed in all patients in order to capture the patient subgroup that may have a probable benefit from TDM.
Fig. 3: Inter-individual variability of efavirenz concentrations. This graph represents the mean trend (with 95% confidence intervals) calculated from 171 determinations in 99 patients. Although the mean plasma concentration remains fairly stable during the dosing period, concentrations may vary 10-fold or more between patients. Reproduced with permission from
[26].
Predictive value of efavirenz concentrations for viral suppression (―) and central nervous system (CNS) adverse effects (―――): the therapeutic window. It is possible to define a range of plasma concentrations associated with a higher probability of efficacy and a lower probability of CNS adverse effects: the therapeutic window (indicated by the shaded area). Reproduced with permission from
[26].
Delavirdine
In a concentration-controlled study of delavirdine monotherapy, trough concentrations of delavirdine were adjusted weekly in order to achieve levels within three pre-defined ranges: low, middle and high [39]. Unfortunately, delavirdine alone showed a potent anti-HIV activity only at 2 weeks of treatment, while viral load reduction at 8 weeks was only of 0.10 log copies/ml on pooling the three study arms. Although the initial antiviral activity of delavirdine was concentration dependent, the virological response was transient even at higher concentrations in plasma. Antiviral activity could not be correlated with concentration target groups because the time to reach the targeted concentration range exceeded the duration of antiviral activity.
Protease inhibitors
Indinavir
In many studies, indinavir has shown a clear relationship between concentration and effect, in terms of virological response [40-42] as well as adverse events [43](Table 1). Most of these studies were carried out using indinavir in an 800 mg, three-times-daily dosage regimen with two NRTIs. It should be recognized, however, that some studies have failed to find such correlations [44,45].
Table 1: Pharmacodynamic studies of indinavir
[40,42,43,168-172].
Indinavir concentrations can vary markedly among patients receiving the same dosage. In a study of 43 HIV-1-infected adults receiving therapy with indinavir + NRTIs, the AUC0-8 ranged more than 12-fold, from 5.4 to 68.0 μM Ă— h. Conversely, the intra-individual variability in plasma concentrations was low [46,47](Fig. 5). In another study of 23 PI-naive patients, individuals with an undetectable plasma HIV-1 RNA level had significantly higher AUC values of indinavir (30.7 μM Ă— h versus 22.4 μM Ă— h, P = 0.035) than those with detectable HIV-1 RNA, indicating that higher plasma indinavir levels correlate with better antiviral response [23]. Similarly, in a study of 65 patients receiving indinavir as part of a triple-drug regimen, those with indinavir levels below 0.75 of the population average had a 3.5-fold higher risk of treatment failure [48]. In another study in 35 patients, an AUC higher than 25 mg Ă— h/l was significantly associated with a better virological outcome (reduction from baseline plasma HIV-1 RNA of 2.5 log10 versus 1.1 log10 copies/ml at 6 months) [42](Fig. 6).
Fig. 5: Intra-patient variability: indinavir. Serial indinavir concentrations were measured at week 2 (▪), week 30 (✦) and week 56 (▴) after administration of an observed dose to an HIV-infected patient. Intra-individual variability in plasma concentrations of protease inhibitors appears to be low. This has been shown for indinavir, amprenavir and nelfinavir. Reproduced with permission from
[47].
Relationship between the plasma concentration of indinavir and the antiretroviral effect. The relationship between virological efficacy [decrease in viral load at 24 weeks of treatment compared with viral load at time 0 (i.e., at the beginning of treatment) of indinavir (800 mg three times daily plus two nucleoside reverse transcriptase inhibitors)] and its 24-h area under the curve (AUC) were determined with the Bayesian approach in 35 protease inhibitor-naive, nucleoside reverse transcriptase inhibitor-experienced patients
[42].
Data from a small-scale study of indinavir pharmacokinetics suggest that children with a relatively small body surface area have higher AUC and lower Cmin values compared with adults, and may therefore benefit from more frequent dosing of indinavir in order to avoid renal effects and improve efficacy [49]. A report published around the same time used a concentration-controlled dosage to maintain trough indinavir concentrations at 0.1 mg/l or above in children receiving indinavir, didanosine and stavudine. Although the starting dosage for indinavir was 500 mg/m2 every 8 h, nine of the 18 children in the study needed 6-h dosing to maintain target drug levels [33]. In the nine children who completed this 24-week study, the plasma trough level of indinavir and the AUC of didanosine (already mentioned in 'Nucleoside reverse transcriptase inhibitors') correlated with the reduction in plasma HIV-1 RNA levels. The authors concluded that these findings demonstrate the potential of using pharmacokinetic information obtained during routine clinic visits to avoid under-dosing among paediatric patients.
Systemic exposure levels are associated with virological response as well as adverse effects. In a study of 18 patients experiencing urological complications while taking 800 mg indinavir three times a day, indinavir concentrations from plasma samples (obtained at 1.5-8 h after the last indinavir dose) were compared with those from 14 patients on the same dosage who did not have urological complaints [43]. Remarkably, all but one (14/15; 93%) of the evaluable patients with urological complaints had plasma indinavir levels above the mean of the control group. When the indinavir dose was reduced to 600 mg three times daily for six patients, plasma levels dropped to within the 95% confidence limits of the control group levels. Urological complaints resolved, while antiviral efficacy was preserved (plasma HIV-1 RNA level remained < 500 copies/ml during follow-up). This study is an example of dose adjustments based on plasma concentrations. If the plasma levels had not been measured, the regimen components might have been needlessly changed.
Saquinavir
Some, but not all, retrospective studies have demonstrated a relationship between saquinavir pharmacokinetics and HIV-1 RNA response. In patients treated with high-dose (7200 mg/day) or low-dose (3600 mg/day) saquinavir monotherapy, plasma levels of the drug correlated well with virological decline [50]. This relationship has been also shown with saquinavir when used in combination with zidovudine and zalcitabine [51]. Data from a dose-ranging study indicate that the optimal dosage of saquinavir (soft gel) is 1200 mg three times daily [52]. However, at a dosage of 600 mg three times daily, 30.3% of 66 patients receiving saquinavir hard gel had trough concentrations below the IC95 (estimated to be 25 ng/ml) [53]. While this finding helps build a case for TDM of saquinavir, it should be noted that the hard gel formulation of saquinavir has lower bioavailability than the soft gel formulation. Still, at approved dosages, trough levels of saquinavir (soft gel) used as a sole PI were found to fall below the wild-type IC50 in a high percentage of patients [54]. While the addition of nelfinavir yielded higher saquinavir trough concentrations on average, trough concentrations below the IC50 were detected at least once during the 12-month study in more than 60% of the patients on this regimen; almost one-third of patients receiving ritonavir had trough saquinavir levels below the IC50 during this period. These low saquinavir levels, regardless of whether they are due to poor adherence or other factors, highlight the importance of monitoring drug levels during therapy with saquinavir-containing regimens.
Higher saquinavir exposure may increase the risk of gastrointestinal complaints. In the ADAM study, which assessed the efficacy of stavudine combined with lamivudine, nelfinavir and saquinavir, abdominal pain was associated with higher levels of saquinavir and nelfinavir [55]. The antiviral treatment was effective, despite relatively low plasma levels of these drugs. However, because these patients were treatment naive and the results were reported at 26 weeks, separating the effects of the PIs from those of the NRTIs would be difficult.
Nelfinavir
Nelfinavir has a large inter-patient variability and a low intra-patient variability with twice-daily and three-times-daily dosing [56], suggesting that TDM could help individualize dosages. This may be especially true for patients with chronic liver disease, in whom inter-individual variations are particularly high [57]. Of 18 heavily pre-treated patients receiving rescue therapy with efavirenz, nelfinavir and stavudine, 22% had plasma trough levels of nelfinavir below a pre-defined minimum effective concentration [58], similar to the rate reported in the ATHENA study (26%) [27].
A number of studies have investigated pharmacokinetic-pharmacodynamic relationships for nelfinavir. The ATHENA study showed that the median nelfinavir concentration ratio (i.e., the ratio between the observed concentration for an individual and the concentration in a reference population, obtained at the same time following administration) was the only factor significantly associated with treatment failure in patients receiving a nelfinavir-containing regimen (n = 45) [59]. Similarly, in another study, exposure levels to nelfinavir were significantly correlated with the rate of viral load decline following treatment initiation [10]. In a study of treatment-naive patients receiving a nelfinavir-based combination as their first regimen, 55 patients (31%) had viral rebound during a median follow-up of 19 months [60]. The relative risk of viral rebound was significantly higher among those with plasma concentrations below the lower limit of high-performance liquid chromatography quantification. Nelfinavir concentrations were also positively correlated with an increased risk of gastrointestinal adverse events [55], although which pharmacokinetic parameters are most closely associated with these adverse effects remains unclear. Additionally, nelfinavir has an active metabolite, M8, which may confound the interpretation of TDM. Although M8 concentrations have been determined in some studies [59-61], it is under debate whether and how they should be used for TDM.
Amprenavir
A pharmacodynamic analysis was conducted using data from a preliminary study in which amprenavir was administered in a dose-escalation manner, using doses of 300 mg twice daily, 300 mg three times daily, and 900, 1050 and 1200 mg twice daily. The decrease in viral load at 4 weeks versus amprenavir concentration showed a clear concentration-dependent pattern that was fitted to a typical sigmoid Emax model with a concentration eliciting 50% of the maximal response (EC50) value of 0.085 mg/l [62]. The Cmax value of amprenavir was positively correlated with the frequency of headache and oral numbness. However, these adverse events are rare and are usually mild to moderate. Therefore, it is debatable whether the application of TDM for reducing the incidence of such adverse events will be cost-effective.
In a study in HIV-1-infected children treated with amprenavir, the virological response was assessed in terms of time-averaged AUC minus baseline for HIV-1 RNA [63]. A multivariate logistical regression analysis revealed that the protein binding-corrected Cmin/IC50 ratio was the only factor associated with a time-averaged AUC minus baseline value > 1 log10. The Cmin/IC50 median ratio for the patients with a decline in HIV-1 RNA < 1 log10 versus > 1 log10 was 0.54 versus 1.66. The geometric means were 0.58 (range, 0.03-4.24) and 2.49 (range, 0.64-30), respectively (P = 0.003).
Plasma concentrations of amprenavir may predict the genotype of viral mutants in patients who fail treatment [64]. In particular, the higher the plasma concentration, the higher the probability of selecting highly resistant but significantly less-fit viral strains. The I50V mutation, which confers a high level of resistance to amprenavir but is associated with low viral fitness, was predominant in patients treated at higher doses of amprenavir (1200 and 1050 mg twice daily), while the I54L/M mutation was predominantly found in patients treated at a lower dosage (900 mg twice daily). Correspondingly, the median trough concentration of amprenavir was significantly higher in patients with the I50V mutation (421 ng/ml, n = 13) than in those with the I54L/M mutation (182 ng/ml, n = 14). This study suggests an intriguing interplay between drug exposure, resistance selection and viral fitness, which may have significant clinical implications. Overall, data support the need to maintain high plasma concentrations of PIs.
A study in 10 patients with >95% adherence, as assessed by medication diary and patient interview, revealed a low inter-individual variability of plasma concentrations of amprenavir determined at 30, 20 and 10 min before a 1200 mg (twice-daily) dose at steady state. Two hours after dosing, however, the inter-individual variability was wide (Fig. 7)[65]. This study also found low intra-individual variability of trough concentrations determined on three consecutive days.
Fig. 7: Inter-individual variations in amprenavir concentrations. This graph depicts the mean and standard deviations of the plasma concentrations obtained 30, 20 and 10 min before a 1200 mg amprenavir twice-daily dose and at 2 h after administration, indicating that amprenavir shows low inter-individual variability in plasma concentrations
[65].
Ritonavir
Dosage and plasma levels of ritonavir correlate with its antiviral efficacy [66] and risk of adverse effects [67]. Resistance-associated mutations may emerge more slowly at higher plasma concentrations (Cmin and AUC0-12) (Fig. 8)[68]. One study found that higher ritonavir concentrations increased the risk of neurological and gastrointestinal adverse events, and suggested that TDM may be useful in titrating the dose to avoid adverse effects while maintaining therapeutic concentrations [67].
Fig. 8: Effect of plasma concentrations of ritonavir on rate of accumulation of mutations. The rate of point mutation appearance in viral isolates from patients treated with ritonavir monotherapy correlates with the plasma trough concentration
[68], highlighting the importance of maintaining optimal drug levels to avoid the emergence of resistant viral strains.
C min, minimum concentration achieved during a dosing interval.
A recent study by Dumon et al.[69] showed the potential value of TDM for ritonavir. In this dose-ranging study, ritonavir (300-450 mg/m2, twice daily) was given as part of a triple-therapy regimen to 31 children with advanced HIV-1 infection. Plasma drug levels were determined twice during the observation period: after at least 4 weeks, and after 3 months of combined treatment. During follow-up (median, 19 months), the inter-subject variability of ritonavir concentrations was wide. The trough concentration, which increased between the first and second observations, correlated with virological response, and was significantly higher in patients with complete or partial response (> 1 log decrease in plasma HIV-1 RNA levels) than in non-responders. Plasma drug concentrations 2 h after drug ingestion did not, however, correlate with long-term virological response.
Inter-individual and intra-individual variability
As already outlined, plasma levels of antiretroviral agents vary greatly among patients receiving standard doses. HIV-1-infected patients constitute a highly heterogeneous population, in which covariables such as weight, gastrointestinal absorptive function, hepatic and renal function, adherence to drug intake schedule and food requirements, drug-drug interactions and many other factors may contribute to widen the range of plasma concentrations. The stage of HIV progression may play an important role in the inter-individual variability in pharmacokinetics. Several studies have shown significant differences in pharmacokinetic values between HIV-negative and HIV-infected patients, as well as among HIV-infected patients at different stages of disease progression [70-72].
Of note, the intra-individual variability in pharmacokinetics appears to be relatively low for some agents (indinavir, nelfinavir, amprenavir, nevirapine and efavirenz); a pre-requisite for the feasibility of TDM in clinical practice. Relatively few studies have investigated the intra-individual variability of antiretroviral drugs, however, and more data are needed to draw definite conclusions. In most studies, the intra-individual variability is lower than the inter-individual variability [46,56,65]. For example, the authors of a study specifically designed to address this issue for nelfinavir trough concentrations report a large inter-patient variability (coefficient of variation, 153%) and a modest intra-patient variability (coefficient of variation, 45%) [56]. However, how low does the intra-individual variability need to be in order to reliably use plasma concentrations for TDM? The low intra-patient variability found in controlled conditions (e.g., drug administered at the same time of the day, control of food intake, etc.) may not apply to settings in which the dose intake is not witnessed. In uncontrolled settings, a number of logistic factors may contribute to increased intra-individual variability in plasma drug levels: these include patients not reporting the precise time of the last dose intake, variable adherence to food requirements, and lack of full adherence to the last doses (i.e., patient not at steady state). Furthermore, drugs such as ritonavir and nelfinavir have been shown to exhibit a significant circadian effect, so that the concentrations after a morning dose are different from those after an evening dose [73,74], and indinavir and zidovudine pharmacokinetics may vary throughout the menstrual cycle [15].
Pharmacokinetic enhancement with ritonavir: implications for TDM
The use of low ritonavir dosages (100 or 200 mg in most cases) to enhance the pharmacokinetics of PIs has gained increasing attention. The latest PI approved in the United States is in fact a combination of a PI, lopinavir, co-formulated with low-dose ritonavir (33 mg ritonavir for each 133 mg lopinavir; recommended total daily dose, 400 mg lopinavir/100 mg ritonavir twice daily) [75]. Ritonavir potently inhibits the cytochrome P450-CYP3A4 metabolic system in the liver and, possibly, in the intestinal mucosa epithelium. The results are profound modifications in the shape of the concentration versus time curve of the other PIs and, therefore, of the parameters of systemic exposure: Cmin, Cmax and AUC.
The extent and type of the modifications caused by ritonavir varies among PIs, depending on the relative contribution of two pharmacokinetic processes inhibited by ritonavir: first-pass metabolism, and hepatic clearance. For the PIs primarily affected by the first-pass effect of ritonavir (saquinavir and lopinavir) [76-78], the bioavailability enhancement results in higher AUC, Cmax and Cmin values while the effect on the half-life is relatively minor. When the effect of ritonavir is primarily on the hepatic metabolism (indinavir and amprenavir) [79,80] (i.e., on hepatic clearance), the bioavailability enhancement is relatively small, the peak concentration increase is small, and the AUC, Cmin and half-life values are substantially enhanced. Pharmacokinetic enhancement of nelfinavir by ritonavir is less appreciable, because the P450 isoenzyme that is inhibited by ritonavir (CYP3A) is less involved in nelfinavir metabolism than for other PIs; however, the enhancement in concentration of its major metabolite, M8, is appreciable [81,82]. Interestingly, accumulating evidence suggests that ritonavir enhancement may be sufficient to allow a reduction in the frequency of dosing for some PIs. This is particularly appealing because it may facilitate once-daily dosing of PIs that are otherwise administered in a twice-daily dosage regimen [83,84].
Because ritonavir increases the plasma concentrations of other PIs, the future utility of TDM in HIV management has been questioned; with boosting, PI levels are less likely to fall below therapeutic ranges except in cases of poor adherence [16]. Several pieces of evidence, however, suggest that the widespread introduction of ritonavir pharmacokinetic enhancement may not reduce the utility of TDM: (i) the inter-individual variability of plasma concentrations remains high despite ritonavir pharmacokinetic enhancement; (ii) even with ritonavir boosting, PI concentrations may be sufficient to be active against wild-type virus but not against viral strains with multiple mutations; and (iii) increased PI concentrations due to ritonavir boosting may result in increased toxicity. Moreover, in the case of once-daily dosing, some patients may not maintain inhibitory concentrations throughout the 24 h following administration, although this may not be true for all the PIs.
Saquinavir
Saquinavir was the first PI studied in combination with ritonavir. A pilot study with 600 mg saquinavir/600 mg ritonavir twice daily in treatment-naive patients with contraindications for NRTIs showed a poor response rate (6/16) at week 13, although responders did have higher levels of each drug than did non-responders [85]. Importantly, saquinavir levels may decrease over time in patients on stable regimens, even if boosted with ritonavir [86]. In this study, the plasma concentrations of saquinavir decreased markedly from the first observation (at 4-12 months) to the second (at 9-15 months): Cmin by 30%, Cmax by 40%, and AUC0-8 by 33%.
Langmann et al. prospectively studied drug concentrations in patients receiving saquinavir (hard or soft gel) in addition to nelfinavir or ritonavir (therapeutic doses) [54]. Although both nelfinavir and ritonavir elevated saquinavir levels in plasma, 60% of patients receiving nelfinavir and 31% of those receiving ritonavir still had saquinavir trough levels below the IC50. In a dose-ranging study in 20 healthy men, Kurowski et al. found that the addition of 200 mg ritonavir increased saquinavir levels to an extent that 1600 mg saquinavir could be considered a once-daily dose [87]. Although patients received doses after a standardized breakfast, the high variability in drug concentrations led the authors to recommend monitoring of trough levels. In a similar study, once-daily dosing with 1600 mg saquinavir (soft gel) + 200 mg ritonavir (liquid) often yielded therapeutic plasma concentrations of saquinavir, but large inter-individual variations in exposure suggested the need to monitor saquinavir concentrations [88]. Together, these findings support the potential of TDM to individualize and optimize saquinavir-based regimens even when enhanced by subtherapeutic levels of ritonavir.
Indinavir
Indinavir has been studied at different dose combinations with ritonavir. The first study in healthy volunteers showed that administration of 400 mg ritonavir twice daily with 400 mg indinavir twice daily resulted in lower Cmax and higher Cmin values for indinavir, allowing a twice-daily dosage regimen for indinavir [79]. More common combinations are the 100 mg ritonavir/800 mg indinavir and 200 mg ritonavir/800 mg indinavir twice-daily dosages.
Resistance to antiretroviral agents is not absolute, as illustrated in a recent report by Condra et al.[89]. This study of isolates from patients failing therapy with 800 mg indinavir three times daily showed that the addition of 200 mg ritonavir twice daily results in average trough concentrations of indinavir far above the IC95 for all 20 virus strains, which had up to a 40-fold increase in IC95 with respect to wild-type virus [89](Fig. 9). These findings indicate that HIV resistance to PIs is a relative concept. In fact, viral strains considered resistant due to genotypic mutations resulting in increased IC95 might actually be clinically susceptible in patients with drug concentrations above this IC95. The role of TDM is suggested, however, by the high inter-individual variability in plasma concentration observed with indinavir regimens boosted with ritonavir: large ranges in trough levels have been reported both with 800 mg indinavir/100 mg ritonavir (mean, 0.38 μg/ml; range, 0.06-1.3 μg/ml) and with 800 mg indinavir/200 mg ritonavir (mean, 3.3 μg/ml; range, 0.3-15.2 μg/ml) [90]. Monitoring drug levels in this setting may be necessary in order to ensure optimal drug concentrations.
Fig. 9: The interplay between viral resistance and pharmacokinetics is suggested. Viral resistance may to some extent be a relative concept. Moderate increases in the concentration inhibiting 95% of virus (IC
95) to a certain drug (in this case, indinavir) due to genomic mutations may be overcome if plasma concentrations [shown here as steady-state minimum concentration achieved during a dosing interval (
C min) (―――)] can be increased above this IC
95, as in the case of exposure intensification with ritonavir. Reproduced with permission from
[89].
Also, although the increase in indinavir peak concentration with ritonavir enhancement is relatively low, the overall exposure enhancement appears to increase the risk of renal toxicity significantly in boosted regimens. In the BEST study, 10% of patients receiving the 800 mg indinavir/100 mg ritonavir twice-daily combination experienced nephrolithiasis, versus 4% in the arm treated with indinavir standard dose [91]. Also, an analysis of plasma concentrations (Cmax and Cmin) of indinavir in 197 patients was performed in France [12]. The patients were on treatment with the combination indinavir/ritonavir in four different twice-daily dosing schemes: 400 mg/100 mg, 400 mg/400 mg, 600 mg/100 mg, and 800 mg/100 mg. The incidence of adverse events increased with indinavir Cmin and Cmax, and with indinavir dosage between 400 and 800 mg, regardless of ritonavir dosage. In a study in 59 HIV-infected patients receiving indinavir in combination with ritonavir (800 mg indinavir/100 mg ritonavir twice daily), nephrolithiasis, haematuria, or flank pain were observed in 22% of the patients, leading to withdrawal in 10% [13]. Indinavir peak levels were found to closely correlate with virological response and renal toxicity. Thus, TDM has potential for selecting the subset of patients with excessive exposure who may be eligible for a dose reduction aimed at limiting the risk of nephrolithiasis.
Amprenavir
Similar to other PIs, amprenavir is susceptible to a boosting effect with ritonavir. In a case-control study of HIV-1-infected patients, the addition of 100 mg ritonavir twice daily to 600 mg amprenavir twice daily resulted in an approximately fivefold increase in amprenavir trough concentrations and a 30% higher amprenavir AUC, compared with the standard dose of 1200 mg amprenavir twice daily [83](Fig. 10). This pharmacokinetic enhancement resulted in average concentrations well above the protein binding-corrected IC50 for strains resistant to other PIs, as well as for strains isolated from patients failing amprenavir at the dosage of 1200 mg amprenavir twice daily [80](Table 2).
Fig. 10: Effect of pharmacokinetic enhancement with ritonavir on median amprenavir (APV) concentrations
[83]. The trough concentrations of amprenavir/ritonavir, administered both as 600 mg/100 mg twice-daily (â–ª,
n = 12) and 1200 mg/200 mg once-daily regimens (â–´,
n = 15), are above the median concentration inhibiting 50% of virus (IC
50) found in patients with multiple failures on protease inhibitor (PI)-containing regimens (see
Table 2). Trough concentrations of unboosted 1200 mg amprenavir twice daily are also shown (♦,
n = 61). ART, antiretroviral therapy.
Amprenavir concentrations inhibiting 50% of virus (IC
50) for wild-type virus, viral strains isolated from patients failing amprenavir used as first protease inhibitor (PI), and viral strains isolated from patients with multiple PI failure
[80].
Data collected from routine clinical practice also confirm the advantages of boosting amprenavir levels with small doses of ritonavir in terms of pharmacokinetic enhancement and virological efficacy [92,93]. A recent study found that amprenavir, boosted or unboosted, showed the lowest inter-individual variability in plasma concentrations among all the PIs investigated [94]. This observation from everyday clinical practice confirms a previous report in a well-controlled setting with full-dose amprenavir [65].
The combination of 1200 mg amprenavir/200 mg ritonavir twice daily has been evaluated in healthy volunteers [95]. This boosted regimen increased the AUC and Cmin values of amprenavir by 40% and 600%, respectively, compared with the standard 1200 mg amprenavir regimen. This combination induced a sustained virological response in 20 of 30 patients with multiple PI failures; one-half of the responding patients had viral loads < 400 HIV-1 RNA copies/ml [96].
Amprenavir, boosted or unboosted, appears to be a suitable drug for TDM; furthermore, the strength of each pill (150 mg) allows for fine-tuning of dosage at the individual level. Interestingly, one study [83] showed that 1200 mg amprenavir administered once daily with 200 mg ritonavir results in a mean trough concentration at 24 h that is comparable with that observed at 12 h with the dosage of 600 mg amprenavir/100 mg ritonavir twice daily. The inter-individual variability was slightly lower with the 1200 mg amprenavir/200 mg ritonavir once-daily regimen than with the 600 mg amprenavir/100 mg ritonavir twice-daily regimen (Fig. 10). These data were observed in two parallel groups that were not matched for sex. However, if this observation is confirmed, it would suggest that TDM might not necessarily be more strongly indicated for once-daily than twice-daily dosage regimes (see 'Once-daily regimens').
Lopinavir
Alone, lopinavir has a particularly short half-life (< 1 h) that would make it unsuitable for clinical use. However, the addition of low-dose ritonavir potently enhances the pharmacokinetics of lopinavir, resulting in average lopinavir trough concentrations several-fold higher than the IC50 for wild-type virus [75]. On the contrary, the inter-individual variability in trough concentrations has been reported to be quite high (Table 3). The mean (± SD) trough concentrations in study M97-720 at weeks 4 and 24 were 3.63 ± 3.44 mg/l and 3.7 ± 2.58 mg/l, respectively. In study M97-765, these values were 2.35 ± 1.64 mg/l and 1.91 ± 1.41 mg/l, respectively [97]. In a recent report, lopinavir trough concentrations varied 11-fold (849-9298 ng/ml), indicating that TDM may be of value in detecting low concentrations in patients [98]. Therefore, it seems reasonable to hypothesize that a certain percentage of patients treated with lopinavir may have subinhibitory concentrations; particularly PI-experienced patients who harbour virus with reduced susceptibility to this agent. Indeed, the proportion of PI-experienced patients who responded to lopinavir/ritonavir was found to decrease in line with the magnitude of the fold-change in baseline susceptibility of isolates. While 93% of patients with fold-change < 10 achieved undetectable viral load, this proportion dropped to 78%, 67% and 50% when the fold-change was 10-20, 20-40 and > 40, respectively [99](Fig. 11).
Table 3: Inter-individual variability in pharmacokinetic parameters for lopinavir
[97].
Virological response at week 24 with respect to baseline phenotype among multi-protease inhibitor (PI)-experienced patients receiving lopinavir/ritonavir. The probability of response to treatment with lopinavir/ritonavir decreases with increasing numbers of point mutations in the viral genome. This observation, common to all PIs, together with the high inter-individual variability in plasma concentrations (see
Table 3), suggests that therapeutic drug monitoring may be of benefit even in patients treated with PIs in combination with small doses of ritonavir used as a pharmacokinetic enhancer
[99].
Taken together, these data suggest that, as with other PI combinations, the response to lopinavir/ritonavir is highly sensitive to the ratio of drug exposure to viral susceptibility, and that TDM may improve outcome in patients with suboptimal plasma drug concentrations.
Adherence
People living with HIV-1 infection face the challenge of adhering to long-term therapy that may be complex in terms of pill count, dosing frequency and dietary requirements for each of the three or more drugs constituting the treatment combination. Poor adherence to HAART not only reduces the effectiveness of treatment [100,101], but may also decrease the statistical power of clinical trials to detect differences in primary intent-to-treat analyses of outcomes among treatment arms [102].
Adherence to dietary requirements is an important consideration in the design of appropriate regimens. Adherence to requirements for intake of food in conjunction with drug intake may be difficult in the workplace and other social situations, and it may be problematic to adhere to water intake requirements. For drugs whose bioavailability is strongly influenced by the presence or absence of food, poor adherence to food requirements is likely to have the same effect on virological response as poor adherence to drug intake (Table 4)[75,103-105]. Thus, drugs with few or no dietary restrictions [106,107] may: (i) be more suitable to diverse patient lifestyles; (ii) reduce treatment complexity; and (iii) lower the risk of treatment failure due to insufficient exposure (Fig. 12).
Table 4: Food effect on protease inhibitor pharmacokinetics
[24,66,106,168,173-178].
Effects of food on bioavailability (24-h area under the curve) of protease inhibitors compared with Food and Drug Administration (FDA)-approved dosing. The effect of food on bioavailability adds a higher level of complexity to patient adherence: lack of adherence to food requirements may lead to suboptimal concentrations, as may be the case for nelfinavir (NFV) (ATHENA study
[164]). APV, amprenavir; IDV, indinavir; SQV, saquinavir; RTV, ritonavir; AE, adverse events.
The clinical use of TDM requires some level of adherence assessment to aid in the interpretation of parameters of drug exposure and their predictive value of efficacy and toxicity. On the contrary, TDM may be of help in assessing adherence.
The main argument against using TDM to monitor adherence is that the results may not reflect true behaviour, because patients may be more likely to take drugs as prescribed shortly before clinic visits if they know that their drug levels will be monitored. On the contrary, patients who normally adhere closely to their regimens may miss a dose shortly before a visit. In spite of this, low levels of monitored drugs may signal problems with adherence or pharmacokinetics, or both, and may provide advance notice of treatment failure if the underlying causes are not addressed.
Hugen et al. reported that adherence is best measured by a combination of methods, including pill count, electronic monitoring, self-reports/diaries, and TDM [29]. In their study of 28 patients (21 treatment experienced, and seven treatment naive), plasma levels of antiretrovirals were more variable in patients with lower adherence. In a subsequent report, Hugen et al. described the use of TDM to identify patients with poor adherence [108]. First, 8-h concentration curves were constructed on the basis of data from 12 patients after directly observed doses [108]. Subsequently, in patients with directly observed administration, very few plasma samples fell outside the 95% confidence limits around the median population values (3% for indinavir, 8% for nelfinavir, 5% for ritonavir, and 4% for saquinavir), but the proportions outside this range were markedly greater in random samples from people suspected of poor adherence (31% for indinavir, 55% for nelfinavir, and 46% for ritonavir and saquinavir).
An alternative adherence monitoring strategy that integrates pharmacokinetics and adherence measures has been proposed [109]. With this method, concentration values falling more than 40% above or below the predicted value (based on the AUC measured during the first 2 weeks of the study) are considered abnormal. In the initial evaluation of this approach, the authors collected up to 12 samples during 1 year from each of 50 children receiving a triple-drug regimen. This technique appeared to predict virological failure, in that viral rebound was significantly more common in those children with plasma drug levels below the 33rd percentile than in children with higher drug levels (53% versus 23%, P = 0.037).
There is, as yet, no well-established measure of adherence. Therefore, directly observed therapy may be the only way to distinguish adherence issues from pharmacokinetic issues, such as the patient's ability to absorb and metabolize the drug. If the trough levels of drug are particularly low in relation to a population average, but the witnessed dose yields a higher level, this may be a clue to poor adherence. Conversely, if low drug levels are achieved with witnessed dosing then the problem is more probably due to pharmacokinetic factors.
Patient selection and evidence of clinical utility
To exploit the full potential of TDM, the identification of patient categories that may benefit from this intervention should be one of the major objectives of future clinical studies evaluating TDM. Who would benefit from routine TDM? Every patient at the initiation of a new treatment regimen, regardless of prior treatment status? Only selected subpopulations who are at particular risk of subinhibitory or excessive concentrations? The following sections present an overview of the issues surrounding patient selection for TDM.
Initiating a new treatment
With so many uncertainties that have yet to be resolved, the current opinion shared by most pro-TDM authors is that, at least for now, TDM should be performed for research purposes in selected patient populations who have a hypothetical risk factor for excessively low or high concentrations. On the contrary, unpredictable factors may result in unfavourable concentrations, which may be identified (and resolved) with the use of TDM at the initiation of each new treatment. The obvious advantage of recognizing low concentrations at the beginning of treatment would be early intervention to correct suboptimal exposure and thereby avoid selection of resistant mutants.
The high inter-individual variability in plasma concentrations observed for all antiretroviral drugs may be due to causes that are often unpredictable, such as low absorption or rapid metabolism. A recent study in Caucasian patients receiving PI or NNRTI (efavirenz) treatment showed a significant relationship between plasma drug concentrations and allelic variations of genes encoding CYP enzymes and P-glycoprotein (PGP). Homozygosity for CC at PGP codon 1145 (exon 26) was present in 44% of 27 patients with high drug levels, but in only 3% of 36 patients with low levels. For CYP2D6, 37% of 27 patients with high drug levels were homozygous or heterozygous for at least one allele defining the poor metabolizer phenotype, as compared with 8.3% of 36 patients with low levels [110]. These data need to be confirmed in other trials, particularly because CYP2D6 is not considered an important metabolic pathway for the currently available antiretroviral drugs. Although rapid metabolism may explain part of the inter-individual variability of plasma concentrations, there is a lack of rapid assays for determining allelic patterns of genes encoding for P450 enzymes. Thus, TDM may be the only way to identify patients with low concentrations due to rapid metabolism.
An example of an unpredictable factor that may significantly alter drug concentrations is represented by non-antiviral medications that the patient may acquire over the counter. Moreover, with the explosive and largely unregulated growth of the herbal supplement industry, unforeseen interactions should also be considered. If TDM had been in routine clinical use, the important discovery that St John's Wort dramatically decreases indinavir exposure [111] might have come sooner, sparing many treatment failures.
Drug interactions
Because PIs are processed via P450 isoenzymes, particularly CYP3A, drugs that affect the activity of this enzyme system may raise or lower the plasma levels of these antiretrovirals, and vice versa(Tables 5-7). While all of the PIs inhibit CYP3A to some extent, ritonavir is the most potent inhibitor and saquinavir the least. In human liver microsomes, amprenavir metabolism is inhibited to various extents by different PIs, in the following order: ritonavir > indinavir > nelfinavir > saquinavir; amprenavir inhibits CYP more strongly than saquinavir, but much less strongly than ritonavir [112]. Many drugs used in the treatment of infections associated with HIV-1 disease (e.g., ketoconazole, fluconazole, terfenadine, astemizole, rifampin, methadone and rifabutin, as well as the herbal supplement St John's Wort) interact with P450 enzymes, and co-administration with antiretroviral agents can lead to serious toxicity-related complications or insufficient concentrations. A detailed list of such drug interactions is available (http://hivinsite.ucsf.edu/InSite.jsp?page=kb-03-01-07).
Table 5: Effects of concomitant medications on protease inhibitors.
Table 6: Effects of protease inhibitors on concomitant drugs
[179,180] (H Khanlou, personal communication, 2001).
Interactions between protease inhibitors.
NNRTIs also interact with CYP enzymes. Delavirdine interacts with CYP3A4, and can increase plasma concentrations of amprenavir, indinavir, nelfinavir and saquinavir. Delavirdine does not appear to affect plasma concentrations of ritonavir. Conversely, nevirapine and efavirenz induce PI-metabolizing enzymes, and may therefore decrease PI concentrations in plasma. Efavirenz reduces plasma levels of amprenavir, indinavir, saquinavir and lopinavir, and increases ritonavir levels, but produces no appreciable effect on nelfinavir levels (concentrations of nelfinavir are slightly elevated, but concentrations of the metabolite are slightly reduced) (reviewed in [113]). Nevirapine can lower plasma levels of saquinavir, lopinavir and indinavir, and has the potential to decrease amprenavir levels; it does not greatly affect plasma concentrations of nelfinavir or ritonavir. Detailed descriptions of interactions among antiretroviral agents are available (http://www.hiv-druginteractions.org).
Although knowledge of these interactions should prevent inappropriate combinations from being prescribed, or help define the appropriate dosage regimen to be used, data are lacking as to optimal dosages when more than two drugs affecting the P450 metabolic system are administered concomitantly. This treatment scenario is becoming increasingly common, especially in salvage patients who may be treated with a ritonavir-enhanced PI in combination with a NNRTI.
An increasing number of studies are aimed at careful definition of this complex issue. In one such study, pharmaco-enhancement of amprenavir (1200 mg twice daily) with ritonavir (200 mg twice daily) abolished the effect of efavirenz (600 mg once daily) on amprenavir levels [95]. Of note, efavirenz decreases the AUC of amprenavir (1200 mg twice daily) by approximately 40%. However, anecdotal reports suggest that a dose of 100 mg ritonavir in combination with 600 mg amprenavir twice daily may not be sufficient to completely counterbalance the effect of efavirenz [114,115]. In fact, amprenavir concentrations with the 600 mg amprenavir/100 mg ritonavir twice-daily combination + 600 mg efavirenz once daily appear to be lower than those observed without efavirenz, even though they remain higher than the concentrations obtained with the standard 1200 mg amprenavir twice-daily regimen without ritonavir.
The concentration-dependent effect of ritonavir in counterbalancing the effect of efavirenz extends to other PIs. As with amprenavir, 200 mg ritonavir twice daily can abolish the effect of efavirenz on indinavir. However, Aarnoutse et al. found that a smaller dose of ritonavir (100 mg twice daily) does not completely counteract the effect of 600 mg efavirenz once daily on plasma concentrations of indinavir [116]. Following 14 days of efavirenz administration, the steady-state trough concentration and the AUC of indinavir were lower than those without efavirenz (0.33 μg/ml versus 0.67 μg/ml, and 9.6 μg Ă— h/ml versus 16.5 μg Ă— h/ml, respectively). The authors concluded that dosage adjustment of indinavir may be indicated for PI-experienced patients. Interestingly, 100 mg ritonavir (the amount co-formulated with lopinavir) is also inadequate to balance the effect of efavirenz metabolic induction on lopinavir [117].
From the aforementioned examples, it appears that the number of studies necessary to define clearly the optimal dosing for each drug in each combination is extremely high, due to the number of possible drug combinations and dose-dependent effects. Therefore, drug-drug interactions, especially when more than two P450-interacting agents are co-administered, may be an indication for TDM. The potential advantages of TDM in this particular situation should be thoroughly investigated.
Hepatic impairment
Patients with drug addiction as a risk factor for HIV-1 infection are often co-infected with hepatitis C virus or hepatitis B virus [118], and may present with some degree of hepatic cirrhosis and liver dysfunction. While there is a general lack of data on dosage adjustments in patients with severe hepatic impairment, recommendations for dosage adjustment in patients with mild-to-moderate hepatic impairment are available for some (indinavir, amprenavir), but not all (efavirenz), of the antiretroviral agents eliminated primarily by the liver. However, determining the extent of hepatic impairment may require invasive procedures that patients are reluctant to undergo.
The inter-individual variability in plasma concentration is particularly high in patients with liver dysfunction [57]. Hepatic impairment may dramatically decrease systemic drug clearance. For instance, the dosage of amprenavir must be radically decreased from the standard regimen of 1200 mg amprenavir twice daily for patients with hepatic impairment, in order to achieve levels comparable with those of patients without hepatic impairment. The recommended amprenavir dosage is therefore 450 mg amprenavir twice daily for patients with moderate hepatic impairment (Child-Pugh score, 5-8) and 300 mg amprenavir twice daily for those with severe impairment (Child-Pugh score, 9-15) [119].
Malavaud et al.[120] found a significant increase of indinavir-related nephrolithiasis in patients with hepatitis B or hepatitis C compared with those without. Although plasma concentrations were not determined, these data suggest a strong link between liver impairment, elevated concentrations of indinavir, and nephrolithiasis.
In another observation, TDM was performed for 18 patients with hepatitis C virus-HIV-1 co-infection receiving a nelfinavir-containing regimen (1250 mg nelfinavir twice daily) [61]. The objective of the study was to maintain nelfinavir trough concentrations within a range of 0.3-1 μg/ml. At day 60 of the study, only two patients had maintained their initial nelfinavir regimen, while the other patients had all undergone a dose reduction, down to 1000 or 750 mg nelfinavir twice daily. Although nelfinavir is generally well tolerated and patients did not experience additional toxicity due to the high initial concentration, TDM in this case had the obvious advantage of reducing the number of unnecessary pills, cutting the drug costs, and preventing long-term toxicity, while maintaining antiviral and immunological response.
Patients with liver impairment appear to be ideal candidates for TDM, in light of the wide uncertainty of concentration predictions based on common hepatic impairment classification algorithms, the prospect of severe toxicity due to overexposure, and the potential for exacerbation of pharmacokinetic changes in the event of drug-drug interactions. Examples of the application of TDM for antiretroviral drugs in patients with liver impairment have been published [57,61].
Gender and body weight
Because of the large number of covariables that may affect the pharmacokinetics of antiretroviral drugs, the influence of body weight and gender on pharmacokinetics is often overlooked. Only three of the currently available antiretroviral drugs include a recommendation for dosage adjustment based on patient weight (nevirapine, didanosine and stavudine). In a study of 93 patients (40% men and 60% women), women (66%) had significantly higher rates of ritonavir-related neurological and gastrointestinal adverse effects than did men (34%), possibly reflecting higher plasma concentrations due to lower body weight [121]. Thus, TDM may be useful in patients with very low or very high body weight.
Pregnant women
Recent US guidelines [122] state that antiretroviral treatment should not be withheld from pregnant women, although this patient group warrants some unique considerations (including potential changes in dosage requirements). Drug pharmacokinetics in pregnant women may be profoundly affected by physiological changes that affect absorption, distribution, metabolism, and elimination. The intestinal transit time is prolonged; body fat, body water, blood flow, and ventilation all increase; plasma protein concentration decreases; hepatic metabolism pathways change; drugs can be transported through the placenta; and the placenta and foetus can contribute to the drug volume of distribution and play a part in its elimination.
In spite of all the potential changes during pregnancy, there is a general lack of pharmacokinetic data in pregnant women. Available information is generally limited to compounds with a potential or proven use in decreasing the risk of HIV-1 transmission to the foetus. While zidovudine, didanosine and lamivudine pharmacokinetics do not change to an extent necessitating dose modification [123-125], the nevirapine half-life is significantly prolonged in pregnant women [126]. A report of two women who were receiving 800 mg indinavir three times daily, and in whom indinavir concentrations were determined during their second and third trimesters and for up to 3 months postpartum, revealed that plasma indinavir AUC0-8 values in the third trimester were 63-86% lower than in the postpartum period [127]. This was despite the fact that the indinavir dosage remained unchanged. The researchers concluded that 'pregnant women may be exposed to subtherapeutic drug levels during the later stages of pregnancy', and suggested that such patients may therefore require higher than standard PI dosages or the concomitant use of a second PI.
Given the absence of data, particularly for PIs, and the possibility of significant differences in this patient category, TDM seems to have a potential benefit in pregnant women.
Children
HIV-1-infected children are an extremely heterogeneous category. All pharmacokinetic processes undergo profound changes as children grow, reflecting changes in gastric pH, gastric and intestinal motility, bile acids, extracellular water/total body water ratio, percent body fat, plasma protein concentrations, liver enzyme capacity, and glomerular and tubular function. Also, there are many unique considerations in the evaluation of treatment efficacy in children: viral load is less predictive of outcome, especially in younger children; the CD4 cell percentage is better than the CD4 cell count as a marker of immunological depletion (in children aged < 6 years); treatment efficacy is predicted by a combination of viral load and CD4 cell percentage growth and development are also markers of disease progression; and adolescents infected perinatally or early in life through blood products may have a unique clinical course that differs from that of adolescents infected later in life [128].
A striking aspect emerging from the available pharmacokinetic data in HIV-1-infected children is the poor performance of attempts to apply pharmacokinetic data from adults to children, and from one stage of the developmental continuum to the other. Brundage et al.[129] examined plasma levels of efavirenz and nelfinavir in young children (10 children aged < 2 years, and 18 children aged > 2 years). The doses used were 720 mg efavirenz/day, allometrically scaled to body size using a standard formula (weight Ă— 700.7), and 25-45 mg/kg nelfinavir three times daily. Plasma levels of both drugs were tested at weeks 2 and 6 to adjust AUC values to between 190 and 380 μmol Ă— h for efavirenz and to above 10 mg Ă— h/l for nelfinavir. Among the 18 patients aged > 2 years, 11 had AUC values below target levels for efavirenz and five below target levels for nelfinavir. Among the 10 younger children, AUC values were below the target level in seven cases for efavirenz and in five cases for nelfinavir. Because of the unacceptably high percentage of patients with low AUC values, the study was stopped and then restarted with a higher dose of both drugs. The authors concluded that dose definition based on pharmacokinetic data in older children may not be applicable to younger children.
A study of nelfinavir pharmacokinetics in neonates showed that exposure decreased after 7 days, possibly because of hepatic enzyme maturation and/or autoinduction [130]. The higher dose tested (45 mg/kg twice daily) resulted in trough concentrations above the target level of 1 mg/l in only 48% of the newborns.
As previously mentioned, indinavir dosages may need to be adjusted or carefully controlled in children, with higher dosages or shorter dosing intervals than in adults [33,49]. Parameters of drug exposure have been found to correlate significantly with the antiviral efficacy of indinavir in children. In a study in children aged > 3 months, five of 11 children with AUC < 20 mg Ă— h/l had detectable viraemia (> 500 HIV-1 RNA copies/ml) after 6 months of treatment, whereas none of the 11 children with AUC > 20 mg Ă— h/l had detectable viral load at this time point [131]. In this study, patients started with an indinavir dose of 400 mg/m2; however, the dose was raised to 600 mg/m2 three times daily in some patients because the AUC was below the lower limit in the target range (10 mg Ă— h/l). Another paediatric study found a significant relationship between indinavir Cmin values and viral load reduction at 24 weeks of treatment [33].
In one study [132], the exposure to saquinavir-soft gel capsule (SGC) (dose, 33 mg/kg twice daily) was lower than expected and the regimen was increased to 50 mg/kg twice daily. In another arm of the same study, children receiving saquinavir-SGC in combination with nelfinavir had higher exposure to saquinavir compared with children not receiving nelfinavir.
The newly introduced approach of potentiating PI exposure with low-dose ritonavir may find an important application in the treatment of HIV-1-infected children [133]. However, although some data are already available [134], further studies are needed to define the appropriate dosage regimens for each combination.
Most researchers agree that HIV-1-infected children may be a target category for TDM due to the unpredictability of plasma concentrations based on the administered dose.
Once-daily regimens
TDM may be of particular benefit in once-daily regimens of ritonavir-enhanced PIs [16]; these regimens are still investigational and the pharmacokinetic/pharmacodynamic data are somewhat sparse. Monitoring of concentrations in once-daily regimens may be particularly appropriate for PIs with short half-lives and/or low plasma concentrations when not boosted with ritonavir [87,88,135]. Metabolism inhibition by ritonavir is a competitive process, and so concentrations of the boosted PI may not be sustained after low-dose ritonavir is completely removed from the body. Since the half-life of ritonavir ranges from 3 to 5 h, this may occur before the 24th hour of a once-daily dosage [107]. Drugs with a longer half-life (Table 8) may show less variability at 24 h in once-daily regimens; preliminary data suggest that this may be the case for amprenavir [83] (see 'Pharmacokinetic enhancement with ritonavir: implications for TDM'). In general, however, once-daily dosage regimens currently appear to be a suitable indication for TDM, although not necessarily for all drug combinations.
Table 8: Pharmacokinetic comparison of protease inhibitors
[174,176,178].
Patients with toxicity
It is often difficult in a multi-drug regimen to understand which drug is causing toxicity, even with the help of TDM. Some of the adverse effects of antiretroviral drugs have been shown to correlate well with plasma concentrations. Perhaps the most notable of these associations is between indinavir plasma levels and their renal side effects, and TDM has already been used successfully to guide dosages of indinavir in patients with renal toxicity [43,136]. The initial side effects of ritonavir, particularly perioral paresthesia and gastrointestinal side effects, also appear to correlate with Cmax values, and dose reduction guided by TDM may reduce the risk of such adverse effects while allowing patients to continue therapy [67].
Patients with viral rebound or with an unsatisfactory response
TDM has potential benefit in patients with a slow rate of viral decay, those experiencing viral rebound after an initial response, and those who are responding but have frequent viral 'blips', which may reflect selection of mutant virus and ultimately result in treatment failure. The ability to recognize that treatment failure is due to low plasma drug concentrations may help in selecting a new treatment, or may alert physicians to the fact that a particular patient may be at risk for low concentrations even with a new treatment.
TDM may also help guide dosage adjustment to avoid toxicity during exposure intensification, which can be used to improve the virological response in patients with intermediate resistance and low drug concentrations. Theoretically, treatment can be intensified by two routes: by increasing the dose of one or more drugs in the regimen, or by adding low-dose ritonavir to a failing PI-containing regimen. This latter approach has been successfully applied to saquinavir-based regimens [137]. Patients failing a saquinavir (hard gel)-containing regimen (HIV-1 RNA levels, 957-8750 copies/ml at baseline) were switched from 600 mg saquinavir three times daily to 1600 mg saquinavir/200 mg ritonavir once daily; the backbone NRTI was maintained until at least week 3 of the study. In most patients (14/18), including four of 10 patients with the L90M mutation at baseline, HIV-1 RNA levels had dropped to < 200 copies/ml at week 3, and remained there through week 12 in 12 of these 14 patients.
In another study, Havlir et al. added 400 mg ritonavir twice daily to indinavir-containing regimens in patients with viral loads between 50 and 50 000 copies/ml; the indinavir dosage was reduced concomitantly to 400 mg indinavir twice daily [138]. Although this strategy increased indinavir trough levels, it did not affect peak levels. At 24 weeks, intensification with ritonavir had resulted in a decline in viral load to < 400 copies/ml in 52% of patients and to < 50 copies/ml in 44% of patients (on-treatment analysis). As expected, viral load reductions were greatest in patients with higher indinavir trough levels and fewer baseline PI mutations.
These studies suggest that intensification of exposure to the current regimen may salvage the efficacy of regimens, particularly in the setting of intermediate resistance and low plasma drug concentrations. TDM could help guide plasma concentrations to the appropriate target. TDM in this setting might have a higher probability of being useful if concentrations are adjusted based on each patient's viral IC50 (see next section). In case treatment failure is primarily due to the lack of adherence to drug regimen or dietary requirements, TDM may be of some help in the differential diagnosis of poor adherence versus poor pharmacokinetics (see 'Adherence') and may suggest interventions aimed at improving patient adherence.
Target levels and influence of covariables
While TDM may be a promising means of individualizing and optimizing existing treatment options for HIV management, one of the major caveats to its application is the lack of broadly accepted target concentrations. A relatively straightforward application of TDM is dose reduction to overcome toxic effects clearly known to correlate with drug concentrations, such as renal toxicity for indinavir and oral paresthesia for ritonavir. Although target levels have not been defined, when either high Cmax values or high drug concentrations in random samples are detected by TDM (as compared with a reference population or literature values) in a patient with signs of toxicity, a dosage reduction is likely to be of benefit. Dosage reduction may also be advisable to prevent short-term or long-term toxicity, even in patients without signs of toxicity at the time of TDM. This approach has been reported in studies describing the successful application of TDM in preventing, reducing, or resolving toxicity while maintaining the patient on successful therapy at lower dosages [43,136].
The concept of inhibitory quotient
Applying TDM to enhance treatment efficacy appears to be a more complex proposition than its application to manage drug toxicities. One of the objectives of pharmacokinetic/pharmacodynamic studies is to evaluate whether the three measures of systemic exposure (AUC, Cmax, or Cmin) correlate with a certain treatment response. Because the AUC, Cmax and Cmin values are strongly correlated with one another, drug response parameters (e.g., decrease in viral load) that correlate with one of them will most probably also correlate with the other two. No study has directly compared Cmax, AUC and Cmin as predictors of treatment efficacy. Studies of virological efficacy have found a significant correlation both with Cmin and AUC. The ratio of Cmin to IC50, defined as the 'inhibitory quotient' (IQ), is used by many researchers and is believed to be the parameter most likely to predict efficacy. Although this is still under debate, the trough concentration has correlated with outcome in a number of studies and it is the easiest to obtain in patients from a logistical standpoint. Additional data supporting the value of the trough concentration for efficacy prediction have been generated using in vivo modelling approaches [139]. The AUC, a measure of systemic exposure, may also be an important target since it is more likely to reflect exposure in body compartments. Also, whether PI and NNRTI activity is concentration dependent or time dependent is unknown. A concentration-dependent activity would favour AUC or Cmax as predictors of efficacy. On the contrary, even though methods for predicting AUC based on one or two plasma concentrations are available [140], determinations of AUC based on a single patient sample are not yet feasible in the TDM setting.
Analogous with the approach used for antibiotics that have time-dependent activity, dosage adjustment based on the Cmin/IC50 ratio is widely thought to be the most suitable target for TDM in HIV-1 infection. In the absence of inhibitory concentrations, viral mutants may be selected rapidly. The IC50 is almost invariably used as the denominator of the IQ, although the fold-increase in IC50 with respect to wild-type virus could be used as an alternative measure for adjusting dosages. IC90 or IC95 would be more logical parameters for a correct assessment of the IQ, due to their closer association with the concept of viral escape from therapy; nevertheless, they are not commonly used because their calculation is associated with a higher degree of error. This is because the viral inhibition versus drug concentration curve in virological assays has a sigmoidal shape, and thus concentration values calculated at its plateau (where IC90/IC95 values usually lie) are subject to a high standard deviation around the mean estimate.
Keeping this in mind, the optimal target may be a value of Cmin/IC50 > 1. An argument for targeting the trough concentration to a multiple of the IC50 may be found in the attempt to achieve inhibitory concentrations in certain body sites, such as the brain and testes. This may be the case for drugs such as nevirapine, indinavir and amprenavir, which are understood to penetrate these body sites, albeit to a possibly lower concentration than in plasma. It may not be appropriate for drugs that achieve only negligible concentrations in body tissues, due to high plasma protein binding. It may be argued that the optimal target does not necessarily have to be a multiple of the IC50. Since drugs may have different intracellular accumulation properties [141,142], plasma concentrations required to fully inhibit the virus at the site of action may be lower than those predicted by the IQ deduced from plasma concentrations.
An additional level of complexity in IQ interpretation is that the IQ value for a treatment-naive patient may be different from that needed for a highly treatment-experienced patient in order to achieve the same outcome. It is possible that patients who have failed a greater number of treatment regimens will harbour a higher number of viral subpopulations with a lower genetic barrier to resistance, which would need fewer mutations in order to become fully resistant. This may be true in spite of an apparently adequate Cmin/IC50, using the IC50 of the predominant viral species. Even treatment-naive patients may harbour a large number of mutant virus subpopulations in addition to the predominant viral species. Ideally, therefore, the IQ of all mutant viral variants that are present should be considered when predicting efficacy based on IQ.
Protein-binding correction and concentration-dependent binding
When the Cmin/IC50 ratio is used to predict efficacy, the effect of protein binding on the free, active drug concentrations should always be taken into account. This is generally the case in most present-day studies. Examples of calculations that can be used to derive the protein binding-corrected Cmin/IC50 of PIs are presented in Tables 9 and 10. There is currently, however, a lack of standardization of the assays used for this purpose.
Table 9: Effects of human plasma proteins on the antiviral effects of protease inhibitors
[77,177,178,181-184].
Estimated concentration inhibiting 50% of virus (IC
50) shifts for protease inhibitors (reported for HIV-1 IIIB strain compared with population steady-state trough levels reported in the literature)
[181,183].
A more conservative approach is to correct the Cmin/IC50 ratio for the free drug fraction, calculated from in vitro or ex vivo studies using ultrafiltration, ultracentrifugation, or equilibrium dialysis. The ratio is corrected by simply multiplying the numerator by the free drug fraction, which ranges in value from 0 to 1. In vivo, however, the free drug is not only in equilibrium with the plasma protein-bound fraction, but also with the fraction bound to HIV protease. Thus, the amount of free drug available for anti-HIV activity in vivo depends on the ratio of the affinity constants of the drug to these proteins.
An alternative approach to calculate the effect of protein binding on the Cmin/IC50 ratio is to determine the IC50 in the presence of human serum, albumin or alpha-1-acid glycoprotein (AAG), alone or in combination. In general, the correction factor using this method is lower than that derived using the previous method. A drawback in the interpretation of this assay is that cells cannot grow in the presence of 100% serum; the addition of buffer that is necessary for cell growth therefore dilutes serum proteins.
For both approaches, standardization is urgently needed. With the first approach, results may vary widely depending on which methodology is used (e.g., ultrafiltration versus equilibrium dialysis). With the second approach, results reported by different authors vary substantially depending on experimental conditions, such as the percentage of human serum used and whether albumin and/or AAG were added. For example, in one study, protein binding was found to increase the IC50 of lopinavir 5.5-fold in cells grown in the presence of 10% fetal calf serum + 50% human serum + 40% buffer versus cells grown in 10% fetal calf serum + 90% buffer [143], whereas in another study the IC50 of lopinavir increased 28-fold in the presence of 1 mg/ml AAG [144]. These differences in protein binding correction of the IQ suggest that the IQ should not be used for drug comparison, until assay standardization has been achieved.
In general, there is a lack of data on the effect of varying concentrations of the binding protein on the free drug concentration and its antiviral activity. This is particularly true for those patient populations, such as paediatric patients, in which concentrations of albumin and/or AAG may vary substantially. Evaluation of the effects of varying concentrations of AAG appears to be of particular importance, since AAG is an acute-phase protein and shows increases in plasma levels of up to fourfold in pathological and physiological states such as stress, infection and trauma. All PIs are bound by plasma proteins, especially AAG (reviewed by Barry et al.[145]). In vitro, a fourfold increase in AAG concentration substantially decreases the activity of most PIs against partially resistant isolates: by 2% for indinavir, by 30% for saquinavir, by 37% for nelfinavir, by 37% for ritonavir, and by 42% for amprenavir [146].
Fluctuating AAG concentrations may influence the choice of target drug levels. Unlike albumin, AAG has a low molar concentration and few binding sites, and is therefore easily saturated by drugs at physiological concentrations. This results in concentration-dependent binding (i.e., the fraction of free drug increases with increasing total drug concentrations). Of note, the effect of AAG in vivo may be complex and vary broadly from drug to drug. A study of amprenavir showed that lower levels of AAG were associated with a lower AUC of total drug, due to an increase in drug clearance. On the contrary, lower total AAG concentrations should not correlate with lower efficacy, since the free drug concentrations are not changed (i.e., lower AAG levels result in higher free drug fractions) [147]. This implies that a drug-drug interaction based on competition for the binding site may be misinterpreted as a P450 induction effect, since it would result in lower total drug concentration; and, erroneously, suggests the need for a dosage increase. This is an example of the relevance of protein binding studies for the definition of adequate target levels.
A worldwide standardization of the assay for determining the viral IC50 and the effect of plasma proteins on the IC50, as is done for antibiotics by the National Committee for Clinical Laboratory Standards, would enhance the reliability of the protein binding-corrected Cmin/IC50 ratio as a comparative tool between drugs, and would exploit its full potential for establishing target levels for TDM.
Virtual IQ
Currently, when the IQ is used in TDM for antiretroviral drugs, only the drug concentration in plasma is determined; its value is defined as optimal or suboptimal with respect to an IC50 obtained from the literature (e.g., for wild-type virus). Alternatively, the optimal concentration can be defined in light of the confidence intervals of a standard population [148] (i.e., without considering viral susceptibility). However, viral susceptibility may be a major confounding factor in the use of TDM. Mutations tend to accumulate over time and with treatment failure, such that patients who have been on therapy for long periods of time or who have failed on numerous regimens are likely to harbour viral populations with many primary and secondary mutations and reduced susceptibility to a range of drugs. With the increasing prevalence of drug-resistant virus, primary HIV resistance is also increasingly common [149,150]. Adjusting drug dosage on the basis of plasma concentrations without regard to viral susceptibility, particularly in patients who have experienced failure on multiple treatment regimens, is therefore very much a shot in the dark.
A better approach would be to integrate pharmacokinetic and resistance data obtained in a particular patient and to adjust the dose on the basis of a parameter of exposure, such as the trough level, and a parameter of viral susceptibility, such as the IC50 of the virus isolated from that patient. Cost is a major drawback for the use of IC50 determination in routine clinical practice. An innovative and potentially successful approach may be based on the virtual IQ [151]. This is calculated by dividing the Cmin value by the virtual phenotype (the expected fold-change in viral susceptibility associated with a particular genotypic mutation pattern) multiplied by the serum-adjusted wild-type IC50 value. This approach, which may represent a lower cost solution for fine-tuning dosage regimens based on individual pharmacokinetic and resistance data, still requires validation in clinical practice. In a preliminary study, virtual IQ was a significant predictive factor of virological response in patients treated with ritonavir/indinavir [152].
Alternative definitions of IQ have been proposed. For example, IQ may be calculated as the ratio of Cmin/fold-change in resistance, since most commercial phenotype assays report the fold-change in IC50 with respect to wild type rather than the IC50. Recently, it has been proposed that the IQ for a patient could be normalized to the population trough concentration and to the phenotype assay cut-off for resistance [153]. The normalized IQ is proposed as the ratio of patient IQ/reference IQ, in which patient IQ = measured patient trough/fold-change in virtual phenotype, and reference IQ = population trough concentration/cut-off value defining resistance for virtual phenotype. The author suggests that the advantage of this approach is, with some limitations, the elimination of the need for complex protein binding corrections. A significant correlation (r = 0.89) was found between change in plasma viral load at 24 weeks of treatment and amprenavir normalized IQ in treatment-experienced patients receiving a complex therapeutic regimen including amprenavir.
Other methods may be advocated to define appropriate target levels. For example, the EC50 value (the level of drug exposure eliciting 50% of the maximal response in a population pharmacokinetic/pharmacodynamic clinical trial), or a multiple thereof, could be a suitable target. Use of this parameter would also obviate the need to correct for protein binding.
Antiretroviral drug combinations
The NRTIs used in a combination may be a factor capable of influencing target levels of a PI or a NNRTI. For example, if a patient's viral population is highly resistant to the backbone NRTI, then adjusting the levels of the PI or NNRTI in the regimen may be of little benefit. On the contrary, if two active drugs are administered and both drugs need dose adjustment, what is the target level for each of them? Is there an influence of an antiretroviral drug on the target concentration of another antiretroviral drug administered in combination? Studies are needed in order to address these issues.
P-Glycoprotein and intracellular concentrations
The aforementioned considerations reflect a basic assumption: plasma concentrations are in equilibrium with concentrations at the site of drug activity, and therefore adjusting the dose on the basis of plasma concentration will result in an appropriate concentration at the site of action (i.e., the intracellular environment) for both PIs and NNRTIs. Because several studies have found a correlation between plasma concentration and virological response, this assumption appears to be valid. On the contrary, there is an urgent need for more data on intracellular pharmacokinetics of antiretroviral drugs.
Considerable attention is focused on the effect of PGP activity on PI intracellular concentrations. PGP, the MDR1 multi-drug transporter [154], is a transmembrane efflux pump that exports lipophilic drugs from the intracellular environment; it has been implicated in resistance to anti-cancer agents and other drugs. PGP is found in tumours, the endothelial cells of the blood-brain barrier, the intestines, liver and kidneys, and in a small proportion of T lymphocytes. The activity of this pump may vary among individuals and with disease states, including HIV-1 infection [155] and lupus [156]. PGP activity may lead to reduced drug absorption from the gastrointestinal tract and enhanced drug elimination into bile and urine, and may represent an obstacle to achieving adequate concentrations in the central nervous system. Conversely, inhibiting or blocking production of PGP can increase plasma levels of PIs in sanctuary sites (brain and testes) of mice [157,158]. In an in vitro study [141], ritonavir significantly increased the intracellular accumulation of indinavir and amprenavir, possibly through inhibition of PGP activity.
A recent study suggested a new and intriguing mechanism by which PGP may affect treatment outcome in HIV-1-infected patients. PGP expression appeared to inhibit viral replication, possibly because of inhibition of HIV binding or entry into cells [159]. Therefore, while enhanced PGP activity may be expected to reduce intracellular levels of certain drugs, it may also prevent or slow infection of cells that express PGP.
A summary of IQ caveats
At the 2nd International Workshop on Clinical Pharmacology of HIV Therapy (Noordwijk, The Netherlands, 2-4 April 2001), caveats to using IQ as a predictor of treatment efficacy and as a target parameter in TDM were discussed. These are summarized in the following:
i) Caution is recommended in relating the IQ of the predominant species to outcome. The IQ of viral subpopulations other than the predominant one may be required to predict clinical outcome.
ii) The calculated IQ may vary, depending on:
• differences in resistance assays (replication based, enzyme based), isolates (laboratory strain, patient strain), and cell types (activated cells, resting cells);
• the reliability of drug concentration measurements;
• the patient population analysed (e.g., drug concentrations and AAG concentrations may differ between healthy volunteers and HIV-1-infected patients);
• body compartment: IQ in plasma may differ from IQ in other compartments due to differences in drug concentrations and genetically distinct viral subpopulations in sanctuary sites; and
• variability in the assays used to determine the influence of protein binding on activity.
iii) Because of the numerous factors contributing to variability in results provided by different investigators, IQ should not be used to compare drugs until assay standardization is achieved. At the current status, IQ values may be used for between-patient comparison for patients receiving a specific antiretroviral drug within a specific study or patient group.
iv) The IQ for TDM may work better in cases of intermediate antiretroviral resistance. The presence of highly resistant virus may have confounded the results of a recent study that failed to show an advantage of TDM over standard of care [160] (see 'Ongoing or planned trials'). The predictive value of the IQ may differ according to viral fold-changes in resistance. For example, IQ = 2 may predict a different percentage response in patients with wild-type virus to that in those harbouring virus with several resistance-associated mutations.
v) For a correct evaluation of the IQ, the contribution of other drugs in the regimen (e.g., NRTI) needs to be included.
vi) An evaluation of how best to adjust for in vivo changes in protein binding needs to be undertaken. AAG has a low molar concentration and few binding sites, and is therefore easily saturated. This results in concentration-dependent binding of drug. How do we adjust for this in clinical practice? Use an average value? On an individual basis, by measuring the concentration of the binding protein?
vii) How do intracellular concentrations come into play?
Quality control issues
TDM faces quality control issues similar to those encountered with genotypic resistance assays. Although few studies have addressed this issue, initial results are not encouraging. In a quality control study, plasma samples spiked with various concentrations of PIs (indinavir, nelfinavir, ritonavir, saquinavir) and NNRTIs (efavirenz, nevirapine) were sent to 13 laboratories in the United States, Canada, Europe and Australia [161]. Values within 20% of the actual level were considered accurate. Despite this fairly liberal definition, accuracy was relatively low: 72% for indinavir, 77% for nelfinavir, 83% for ritonavir, 92% for saquinavir, 74% for efavirenz, and 62% for nevirapine.
A similar study in France yielded equally poor results, for NNRTIs as well as PIs [162]. Accuracy, also as already defined, was lowest for the NNRTIs (28.6% for efavirenz and 50% for nevirapine), and was only slightly better for the PIs [53.3% for ritonavir, 75% for saquinavir, 73.3% for nelfinavir, 60.2% and 69.2% for indinavir (two samples), and 63.6% and 81.8% for amprenavir (two samples)]. Factors that could affect accuracy include the reference compound used (lyophilized versus sterile) and the compartment analysed (serum versus plasma).
While these results suggest the need for improvement, standardization should improve with increasing demand. At the time of writing, several commercial laboratories have announced an intention to offer drug concentration assays for routine clinical use.
Logistical aspects: blood sampling and dosage adjustment
One of three procedures has been used in most TDM studies to date: (i) determination of serial or peak and/or trough levels, with subsequent dosage adjustment based on investigational cut-off values; (ii) determination of one concentration using a blood sample obtained randomly within a dosing interval, with subsequent dosage adjustment if the concentration is outside cut-off values based on concentrations obtained in a reference population; and (iii) random concentrations evaluated with a population pharmacokinetics model and with dosage adjustment based on parameters of exposure predicted by the model with respect to investigational cut-off values.
Determination of peak and/or trough levels
TDM may be performed using peak and trough concentrations. However, the correct sampling time for peak determination is highly variable between and within patients; therefore, the confidence of having drawn a true peak by sampling at the mean time to maximum concentration (Tmax) reported in the literature is low. Obtaining true trough concentrations is also problematic. Because of delayed absorption, plasma concentrations of PIs and NNRTIs may, in some cases, continue to fall following dose administration. In these cases, the trough level obtained just before dosing does not represent the lowest concentration of a dosing interval. Also, practical considerations (e.g., patients arriving late for appointments and high patient volume) make true trough levels difficult to obtain; trough concentrations are generally obtained without directly observed dosing, meaning that the time since the last dose in a twice-daily regimen may not actually be 12 h.
The concentration ratio method
Random time samples can also be used, as with the concentration ratio method employed in the ATHENA study. To establish the population normal values, full 24-h curves are obtained for at least 20 patients for each drug. A median curve is then created, along with curves representing pre-determined percentiles (Fig. 13). The area included between the percentile curves is considered the therapeutic range. Patient samples can be obtained at any time; the patient is simply asked how long it has been since the last dose was taken. Once an abnormal drug level is detected, the dosage is modified accordingly. The reliability of this method may be low in the initial period of time following administration, due to the high inter-patient variability in drug absorption.
Fig. 13: The concentration ratio method for adjusting dosages of antiretroviral agents. With this method, a median curve (―) and confidence intervals [25% and 75% quartiles (―――)] are constructed using data from a reference population, and plasma concentration values obtained from patients undergoing therapeutic drug monitoring are then plotted on this graph. If the concentration falls outside predetermined values, poor adherence and pharmacokinetic factors responsible for the abnormal concentrations should be evaluated.
The Bayesian approach
Bayesian estimation, a mathematical process capable of predicting plasma concentrations on the basis of pharmacokinetic information obtained in a patient population, represents a more complex and sophisticated tool suitable for TDM application [163]. The following basic equation illustrates how Bayesian estimation works: EQUATION 1
where Ci ... Cm are measured concentrations of the individual, Ĉi ... Cm are predicted individual concentrations, SD2Ci is the variance of measured concentrations, Pj ... Pn are population pharmacokinetic parameters, P̀‚j ... Pn are individual pharmacokinetic parameters, and SD2Pj is the variance of population parameters.
The predicted individual concentrations, on the left-hand side of the equation, are balanced against the predicted population values, on the right-hand side of the equation. This process essentially shows how different from the population the individual is. One of the advantages of this approach is that the primary parameters of drug exposure for a patient (Cmax, Cmin and AUC) can be based on one or more random samples and calculated a posteriori, as the most probable values generated on the basis of the population model. This process requires minimal data, and balances uncertainty of data against uncertainty of the model parameters. Another advantage is that the model used may incorporate a wealth of information regarding the variability of the pharmacokinetic parameters and the influence of covariables on them, potentially allowing greater confidence in predicting the concentrations that will be achieved following a particular dosage adjustment. For example, the model may be built with information on the shape of clearance variability and on the effect of hepatic impairment or of a concomitantly administered drug on clearance. Also, the same model does not have to be used for all patients; it can be individualized so that it can take into account, for example, the possibility that a particular patient may have delayed absorption while most patients do not. Clearly, the model needs to be validated before using it in clinical practice. In particular, the precision with which a certain parameter of exposure (e.g., the trough concentration) is predicted by the model needs to be taken in account.
Fletcher and colleagues have used the Bayesian approach for concentration-targeted dosage [163]. First an intensive pharmacokinetic profile is obtained for each patient, followed by randomly timed samples at monthly intervals. In the simplest scenario, a one-compartment algorithm is used: FIGURE
where Ka is the absorption rate, Vc is the compartment volume, and Ke is the elimination rate. Additional compartments can be added as necessary to explain drug disposition. These values can then be incorporated into the Bayesian algorithm to guide dosage adjustments. For example, the equation for zidovudine would be: EQUATION 2
where CL/F represents oral clearance, which in this case is obtained by multiplying Vc by Ke, and 0.19 mg/l is the desired steady-state concentration.
Obviously, this approach needs a population pharmacokinetics expert to analyse the data and to help physicians interpret the results.
Ongoing or planned trials
In The Netherlands, the ongoing prospective ATHENA study is examining the clinical utility of TDM. In this study, plasma samples are taken at random times between two doses at steady state. The concentration obtained is then evaluated in light of the median concentration obtained in a reference population. If the ratio between a patient concentration and the median concentration in the population is below or above pre-defined cut-off values (e.g., for indinavir, 0.75 and 2), then the concentration is considered suboptimal or excessive, respectively. TDM is performed at two central laboratories, for patients from 22 sites. As of December 1999, 391 patients had been randomized and 1828 drug concentrations had been measured. At baseline, the rates of suboptimal exposure were 41% for saquinavir, 28% for indinavir, 27% for ritonavir, 26% for nelfinavir, and 10% for nevirapine [27]. Over-exposure was less common: 11.5% for saquinavir, 9.9% for ritonavir, 5.9% for indinavir, 5% for nelfinavir, and 3.5% for nevirapine.
A number of factors appear to limit the statistical power of this study and therefore may limit its ability to definitively answer questions surrounding TDM in HIV management. First, physicians are not required to comply with TDM-based dosage adjustments. In fact, early results suggest that advice is followed in fewer than 20% of cases. Second, TDM is widely available in The Netherlands, and physicians whose patients are randomized to the control (no TDM) arm may go outside the study protocol to have drug levels measured. Third, because many patients entered the ATHENA study with undetectable viral loads, achieving sufficient statistical power to identify a meaningful effect may be difficult. Also, the protocol does not incorporate resistance data in the dosage adjustment calculation.
Nevertheless, a preliminary analysis of two subsets of patients enrolled in this study appears to suggest the benefit of TDM in clinical practice. In the first subset (Fig. 14), 55 antiretroviral-naive patients were randomized to TDM or control groups [136]. Patients were treated with indinavir three times daily or 800 mg indinavir/100 mg ritonavir twice daily. In the TDM group, the indinavir dose was reduced in several patients because of excessive concentrations. There was a lower rate of discontinuation in the TDM group than in the control group (9.5% versus 40%, P = 0.03). Analysis of the proportion of patients with HIV-1 RNA levels < 500 copies/ml at 12 months indicated that patients receiving TDM were more likely to achieve a virological response than control patients (75% versus 48.1%, P = 0.04, intention-to-treat [ITT] analysis). The authors concluded that TDM improved outcome, primarily by allowing better management of toxicity.
Fig. 14: Effects of therapeutic drug monitoring (TDM) of indinavir on virological response: the ATHENA study. This study suggested that TDM significantly increased the likelihood of patients achieving HIV-RNA levels < 500 copies/ml at 12 months of treatment (intention-to-treat analysis), pooling results for both indinavir and nelfinavir versus control (a). (b) In the case of indinavir, this advantage was driven by the lower number of patients discontinuing indinavir because of toxicity; for nelfinavir, this effect was due to the lower percentage of patients discontinuing nelfinavir because of virological failure
[136,164].
In the second subset of the ATHENA study (Fig. 14b), 92 antiretroviral-naive patients treated with 1250 mg nelfinavir twice daily were randomized to TDM or control groups [164]. Again, patients in the TDM arm had a greater likelihood of achieving HIV-1 RNA levels < 500 copies/ml at 12 months (80.5% versus 58.8%, P = 0.03, ITT analysis). This result was driven by a lower percentage of patients in the TDM arm discontinuing nelfinavir because of virological failure. Indeed, 50% of patients in the TDM arm had a dose increase because of low drug levels. The authors indicated during their presentation of the results in Noordwijk that they felt the effect of food on nelfinavir levels to be very important and that TDM may help in the management of this issue. Thus, interventions arising from TDM may not be limited to dose modifications in patients with inappropriate drug concentrations, but may also include patient counselling aimed at guiding and correcting food intake behaviour.
A second study conducted in France (PharmAdapt) failed to find a significant benefit of TDM versus standard care [160]. At 3 months, there was no advantage in terms of viral load in the 96 patients enrolled in the TDM arm compared with the 84 patients in the control arm. Several factors may account for this lack of effect: (i) plasma samples were obtained at week 4 of treatment, and dose modification in patients with suboptimal concentrations was performed at week 8, so patients with low concentrations were treated for a long period of time with suboptimal doses, possibly allowing emergence of resistance; (ii) only 17% of the patients in the TDM arm had a dosage modification, posing doubts regarding the statistical power of this study; (iii) 12 weeks may be a relatively short observation time; and (iv) because the study enrolled treatment-experienced patients, some subjects may have been harbouring highly resistant virus; however, the trough cut-off values for dosage modification were based on the IC50 for wild-type virus.
A recent analysis suggests how the design of controlled studies aimed at evaluating the clinical benefit of TDM (e.g., ATHENA and PharmAdapt studies) could be improved [165]. To date, TDM trials have been designed with patient randomization to TDM or no TDM, an approach requiring a very high sample size in order to achieve an adequate statistical power. The suggested innovative approach is based on a preliminary screening for patients with low plasma concentrations or IQs with subsequent patient randomization to dose adjustment or no dose adjustment.
A number of other ongoing or planned prospective studies reflect the interest that is being given to TDM and its possible introduction in routine clinical practice.
An example of TDM use in the setting of a concentration-controlled clinical trial has been published [166]. In the Pediatric AIDS Clinical Trials Group 382 study, TDM was used to adjust the dosages of efavirenz and nelfinavir for many of the study participants. In week 2 of this study, in which children received efavirenz, nelfinavir and one or more NRTI, increased doses of efavirenz were recommended for 22 patients, and decreased doses for three patients. At week 6, increased doses were recommended for nine children, and decreased doses for six. Overall, the percentage of children with 24-h AUC values within the target range increased from 44% (22/50) at week 2 to 56% (28/50) at week 6. For nelfinavir, 74% of children had levels within the target range at week 2 and 80% at week 6 (nelfinavir doses were increased for two children at week 2).
Consensus
Conceptually, the panel agrees that TDM may represent a practical tool to improve the outcome of patients receiving HAART. However, at present, we cannot recommend its use in routine clinical practice for three reasons: (i) existing data are not adequate to confirm the utility of TDM in this setting; (ii) current assays are not sufficiently reliable or standardized; and (iii) interpretation of data is complex, and may require expert advice. To address these issues, we suggest that the following measures are necessary: (i) large randomized trials to assess the clinical utility of TDM in the management of HIV-1 infection; (ii) assay standardization; and (iii) education programmes for pharmacists, physicians and patients.
With the caveat that any application of TDM in HIV management should be validated in clinical trials before being incorporated into routine clinical practice, the panel has formulated the following list of position statements, which may help guide TDM trial objectives as well as provide points of reflection for centres in which TDM is already in use.
i) Levels of systemic exposure to PIs and NNRTIs correlate with their efficacy as well as some of their adverse effects (i.e., a concentration-effect relationship exists). Therefore, adjusting PI and NNRTI concentrations within a therapeutic window may be of benefit, especially in light of the high inter-individual and relatively low intra-individual variability of plasma levels.
ii) TDM has potential utility for patients who are initiating therapy with any PI or NNRTI or who are changing regimens, regardless of treatment history or pharmaco-enhancement with ritonavir. TDM may help to ensure that appropriate plasma drug concentrations are achieved and to identify absorption or metabolic problems, as well as unexpected interactions with over-the-counter or herbal medications that may lead to suboptimal drug levels.
iii) TDM is more likely to be of benefit in specific situations that may occur at the beginning of a new treatment protocol or during ongoing treatment:
• if a malabsorption syndrome is suspected;
• if drug interactions likely to cause clinically significant concentration changes are suspected;
• if more than two drugs with an influence on P450 activity are administered;
• in patients with hepatic impairment;
• in patients with particularly high or low body weight compared with the population average;
• during pregnancy;
• in children;
• if there is a change in clinical or physiological status that is suspected of causing abnormal drug levels;
• for evaluation of unsatisfactory virological response;
• for dose intensification of failing regimens;
• in once-daily regimens of PIs using ritonavir pharmacokinetic enhancement;
• in deep salvage therapy in order to expose patients to maximal tolerable levels while limiting the risk of toxicity (even in cases of ritonavir boosting);
• for preventing toxicity in patients with high plasma drug levels; and
• in patients who develop adverse reactions while taking PIs, to adjust the dosage downward while still maintaining therapeutic levels.
iv) Pharmaco-enhancement with ritonavir does not necessarily mitigate the utility of TDM. Because of the high inter-individual variability in plasma drug concentrations, and the possibility of higher than normal IC50 values in pre-treated patients, individual patients may have subinhibitory trough/IC50 ratios even though the average trough/wild-type IC50 ratios are high.
v) TDM may have a role in monitoring adherence. Adherence is probably best monitored with a combination of methods, including patient self-report, pill count, medication event monitoring systems, and possibly TDM. On the contrary, the evaluation of patient adherence to treatment is important for a correct interpretation of TDM results.
vi) Possible interventions resulting from TDM include adjusting dosage or switching drugs if concentrations are excessive or inadequate; or, if possible, addressing the factor causing the unsatisfactory drug levels (e.g., patient education or discontinuation of an interacting drug).
vii) The parameters of drug exposure that should be monitored, and their optimal target values, have not been adequately defined. These may differ depending on patient treatment history, concomitant antiretroviral drugs in the regimen, and other factors. For example, PI-experienced patients may harbour virus populations with multifold increases in drug IC50 values compared with wild-type virus; thus, target concentrations may be higher in this clinical setting. TDM trials and available TDM services should focus on the definition of such targets by accruing data that integrate pharmacokinetics and resistance in diverse patient populations.
viii) The future of TDM in HIV management may be an integration of pharmacokinetics and resistance. One of the major objectives of TDM studies should be to evaluate methods of integrating these data to optimize treatment. The innovative concept of the virtual IQ should be investigated, since it has potential to be a cost-effective approach and, as such, to be adaptable to routine clinical practice.
ix) Since the target drug level is usually defined as the ratio of a pharmacokinetic measure and a resistance measure, the variability of resistance testing assays, concentration determination and evaluations of the effect of protein binding on IC50 constitute a major obstacle. An effort should therefore be made towards assay standardization and quality control programmes.
x) The use of TDM to adjust dosages on the basis of plasma drug concentrations assumes that the free drug concentration at the site of action (the intracellular environment) correlates well with plasma concentrations. Potential confounding factors that may cause a lack of correlation, such as concentration-dependent protein binding, tissue penetration and intracellular accumulation, need to be thoroughly investigated.
xi) Procedures for sampling and dose adjustment used to date include protocols based on peak and trough levels, concentration ratios, and Bayesian methods. Further studies are needed to determine which method will yield the most useful results or whether alternative methods may be considered.
Conclusions
Because initial HAART regimens are highly effective in the short term, determining that TDM improves outcomes will be difficult in prospective studies. On the contrary, heavily pre-treated patients may harbour a highly resistant virus that will result in treatment failure regardless of the clinical intervention used, including TDM. The efficacy of TDM may be more easily demonstrated in patients with virological failure in the early stages of the therapeutic continuum, and this may be a better setting for a pivotal study. PIs represent the best candidates for TDM with current assay techniques, although NNRTIs should not be overlooked.
Large-scale studies of TDM should be designed to include substudies aimed at elucidating the role of key factors such as protein binding and the relationship between intracellular and plasma drug levels. In addition, the value of any TDM study design that does not include resistance testing, as well as an analysis of patient adherence to treatment, will be extremely limited. Physician and pharmacist education is needed, as the information derived from TDM will be of little value unless it is acted on in order to optimize treatment. Finally, studies will need to include pharmaco-economic evaluations, which will be essential for insurance coverage and reimbursement.
All antiretroviral drugs are expensive; their use should be optimized to avoid treatment with a drug that is not helping the patient and that will probably compromise future options for therapy. New agents offer hope in dealing with HIV-1 infection more effectively, but approval of many new additions to the anti-HIV armamentarium is not expected in the short term. In addition, while the unique resistance profiles of new agents will offer expanded options, HIV-1 has shown a remarkable ability to adapt to therapeutic challenge and there is no evidence that it cannot escape pressure from any agent we may include in a therapeutic protocol.
Treatment options are not limitless, and the available options must be used with particular care for optimization. As Gallant eloquently stated in a review of strategies for long-term success in the treatment of HIV-1 infection, "the best way to deal with resistance is to prevent it" [167]. TDM has the potential to help ensure that patients receive adequate drug levels from the outset of therapy, and thereby prolong the effectiveness of existing treatment options for HIV-1-infected individuals.
Acknowledgements
The authors wish to acknowledge the unrestricted support of Vertex Pharmaceuticals (Cambridge, MA, USA) for the realization of this project. They are also indebted to the following people for helpful discussion and/or critical review of the manuscript: Prof. Adriano Lazzarin of San Raffaele Hospital (Milan, Italy), Prof. Gioacchino Angarano of the University of Lecce (Italy), and Pravin Chaturvedi, Varun Garg and Lucas Beeler of Vertex Pharmaceuticals. In addition, the authors would like to thank Dr Helmut Liess and Dr Shahin Gharakhanian, of Vertex Pharmaceuticals (Europe) Ltd., for participation and discussion.
References
1. Descamps D, Flandre P, Calvez V,
et al.Mechanisms of virological failure in previously untreated HIV-infected patients from a trial of induction-maintenance therapy. Trilege (Agence Nationale de Recherches sur le SIDA 072 Study Team).JAMA 2000,
283:205-211.
2. Markowitz M.
Resistance, fitness, and potency. Mapping the paths to virological failure. [Editorial].JAMA 2000,
283:250-251.
3. Austin DJ, Ferguson NM, Fraser C,
et al.Understanding anti-retroviral treatment failure: the role of pharmacokinetic/dynamics and patient non-compliance in the emergence of resistance.7th European Conference on Clinical Aspects and Treatment of HIV-Infection. Lisbon, October 1999 [abstract 212].
4. Paris D, Ledergerber B, Weber R,
et al.Incidence and predictors of virological failure of antiretroviral triple-drug therapy in a community-based cohort.AIDS Res Hum Retroviruses 1999,
15:1631-1638.
5. Watson DC, Farley JJ.
Efficacy of and adherence to highly active antiretroviral therapy in children infected with human immunodeficiency virus type 1.Pediatr Infect Dis J 1999,
18:682-689.
6. Wit FW, van Leeuwen R, Weverling GJ,
et al.Outcome and predictors of failure of highly active anti-retroviral therapy: one-year follow-up of a cohort of human immunodeficiency type-1-infected persons.J Infect Dis 1999,
179:790-798.
7. Deeks SG.
Determinants of virological response to antiretroviral therapy: implications for long-term strategies.Clin Infect Dis 2000,
30(suppl 2):S177-S184.
8. The Panel on Clinical Practices for Treatment of HIV Infection convened by the Department of Health and Human Services (DHHS) and the Henry J. Kaiser Family Foundation.
Guidelines for the Use of Antiretroviral Agents in HIV-Infected Adults and Adolescents. Published at
http://www.hivatis.org, 28 January 2000. Available at
http://www.hivatis.org/guidelines/adult/pdf/A&ajani.pdf, accessed 9 February 2000.
9. Hirsch MS, Brun-Vezinet F, D'Aquila RT,
et al.Antiretroviral drug resistance testing in adult HIV-1 infection: recommendations of an International AIDS Society-USA Panel.JAMA 2000,
283:2417-2426.
10. Hoetelmans RM, Reijers MH, Weverling GJ,
et al.The effect of plasma drug concentrations on HIV-1 clearance rate during quadruple drug therapy.AIDS 1998,
12:F111-F115.
11. Murphy RL, Sommadossi JP, Lamson M, Hall DB, Myers M, Dusek A.
Antiviral effect and pharmacokinetic interaction between nevirapine and indinavir in persons infected with human immunodeficiency virus type 1.J Infect Dis 1999,
179:1116-1123.
12. Lamotte C, Peytavin G, Perre P,
et al.Increasing adverse events with indinavir dosages and plasma concentrations in four different ritonavir-indinavir containing regimens in HIV-infected patients.Eighth Conference on Retroviruses and Opportunistic Infections. Chicago, February 2001 [abstract 738].
13. Casado JL, Moreno A, Sabido R,
et al.A clinical study of the combination of 100 mg ritonavir plus 800 mg indinavir as salvage therapy: influence of increased plasma drug levels in the rate of response.HIV Clin Trials 2000,
1:13-19.
14. Durant J, Clevenbergh P, Garraffo R,
et al.Importance of protease inhibitor plasma levels in HIV-infected patients treated with genotypic-guided therapy: pharmacological data from the Viradapt Study.AIDS 2000,
14:1333-1339.
15. Piscitelli SC.
The limited value of therapeutic drug monitoring in HIV infection.Medscape HIV/AIDS 1999,
5.
16. Flexner C.
What's in a trough? Therapeutic drug monitoring and antiretroviral regimens.The Hopkins HIV Report 2000,
12:5-12.
17. AIDS Clinical Trials Group Pharmacology Committee.
AIDS Clinical Trials Group (ACTG) Pharmacology Committee position paper on therapeutic drug monitoring for antiretroviral drugs. Using drug concentrations to guide individual therapy.9th Annual Clinical Care Options for HIV Symposium. Laguna Niguel, June 1999.
18. Recommandations du Groupe d'Experts.
Rapport 2000 Sous la Direction du Pr JF Delfraissy. Prise en Charge Therapeutique des Personnes Infectees par le VIH. Paris: Medecine-Sciences, Flamarion; 2000.
19. Gibbons ES, Reynolds HE, Tjia JF,
et al.The Liverpool therapeutic drug monitoring service - a summary of the service and examples of use in clinical practice. 5th International Congress on Drug Therapy in HIV Infection [abstract P259 A].AIDS 2000,
14(suppl 4):S89.
20. Brockmeyer N, Salzberger B, Doerr H,
et al.Antiretrovirale Therapie der HIV-Infektion.Deutsches Aerzteblatt 2001,
4:175-181.
21. Aries SP, Schaaf BM, Wilmsen U,
et al.Routine therapeutic drug monitoring in patients on protease inhibitors: impact on therapeutic decisions? 5thInternational Congress on Drug Therapy in HIV Infection [abstract P264 A].AIDS 2000,
14(suppl 4):S91.
22. Acosta EP.
The promise of therapeutic drug monitoring in HIV infection [expert column].Medscape HIV/AIDS 1999,
5.
23. Acosta EP, Henry K, Baken L, Page LM, Fletcher CV.
Indinavir concentrations and antiviral effect.Pharmacotherapy 1999,
19: 708-712.
24. Barry MG, Merry C, Lloyd J,
et al.Variability in trough plasma saquinavir concentrations in HIV patients: a case for therapeutic drug monitoring.Br J Clin Pharmacol 1998,
45:501-502.
25. Gatti G, Di Biagio A, De Pascalis C,
et al.High variability of indinavir parameters of systemic exposure in HIV+ patients observed prospectively in everyday clinical practice.38th Interscience Conference on Antimicrobial Agents and Chemotherapy. San Diego, September 1998 [abstract A66].
26. Marzolini C, Telenti A, Decosterd LA,
et al.Efavirenz plasma levels can predict treatment failure and central nervous system side effects in HIV-1-infected patients.AIDS 2001,
15:71-75.
27. Burger D, Hoetelmans R, Hugen P,
et al.ATHENA: a randomized, controlled clinical trial to evaluate whether therapeutic drug monitoring (TDM) contributes to reduced HIV-related morbidity and mortality.1st International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, March 2000 [abstract 6.6].
28. Gibbons ES, Reynolds HE, TIJA JF,
et al.Therapeutic drug monitoring in the management of subjects on the protease inhibitors nelfinavir and saquinavir: results of the Roche UK TDM service.5th International Congress on Drug Therapy in HIV Infection. Glasgow, October 2000 [abstract P259].
29. Hugen P, Burger DM, Langebeek N,
et al.Compliance to HIV-protease inhibitors (PIs) is more accurately measured by combining various methods, including MEMS (electronic monitoring) and therapeutic drug monitoring.XIII International Conference on AIDS. Durban, July 2000 [abstract ThPeB5029].
30. Fletcher CV, Kawle SP, Kakuda TN,
et al.Zidovudine triphosphate and lamivudine triphosphate concentration-response relationships in HIV-infected persons.AIDS 2000,
14:2137-2144.
31. Sanathan LP, Peck CC.
The randomized concentration-controlled trial: an evaluation of its sample size efficiency.Control Clin Trials 1991,
12:780-794.
32. Fletcher CV, Acosta EP, Henry K,
et al.Concentration-controlled zidovudine therapy.Clin Pharmacol Ther 1998,
64:331-338.
33. Fletcher CV, Brundage RC, Remmel RP,
et al.Pharmacologic characteristics of indinavir, didanosine, and stavudine in human immunodeficiency virus-infected children receiving combination therapy.Antimicrob Agents Chemother 2000,
44:1029-1034.
34. Havlir D, Cheeseman SH, McLaughlin M,
et al.High-dose nevirapine: safety, pharmacokinetics, and antiviral effect in patients with human immunodeficiency virus infection.J Infect Dis 1995,
171:537-545.
35. van Heeswijk RP, Veldkamp AI, Mulder JW,
et al.The steady-state pharmacokinetics of nevirapine during once daily and twice daily dosing in HIV-1-infected individuals.AIDS 2000,
14:F77-F82.
36. Montaner J, Reiss P, Cooper D,
et al.A randomized, double-blind trial comparing combinations of nevirapine, didanosine, and zidovudine for HIV infected patients.JAMA 1998,
279:930-937.
37. Veldkamp AL, Weverling GJ, Lange JM,
et al.High exposure to nevirapine in plasma is associated with an improved virological response in HIV-1-infected individuals.AIDS 2001,
15:1089-1095.
38. Joshi AS, Barrett JS, Fiske WD,
et al.Population pharmacokinetics of efavirenz in phase II studies and relationship with efficacy.39th Interscience Conference on Antimicrobial Agents and Chemotherapy. San Francisco, September 1999 [abstract 1201].
39. Para MF, Meehan P, Holden-Wiltse J,
et al.ACTG 260: a randomized, phase I-II, dose-ranging trial of the anti-human immunodeficiency virus activity of delavirdine monotherapy. The AIDS Clinical Trials Group Protocol 260 Team.Antimicrob Agents Chemother 1999,
43:1373-1378.
40. Stein DS, Fish DG, Bilello JA, Preston SL, Martineau GL, Drusano GL.
A 24-week open-label phase I/II evaluation of the HIV protease inhibitor MK-639 (indinavir).AIDS 1996,
10:485-492.
41. Havlir DV, Hellmann NS, Petropoulos CJ,
et al.Low plasma concentrations of indinavir are related to virological treatment failure in HIV-1 infected patients on indinavir-containing triple therapy.Antivir Ther 1998,
3:215-220.
42. Gatti G, Di Biagio A, Mastropasqua M,
et al.The relationship between systemic exposure to indinavir and virological efficacy.1st International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, March 2000 [abstract 5.4].
43. Dieleman JP, Gyssens IC, van der Ende ME, de Marie S, Burger DM.
Urological complaints in relation to indinavir plasma concentrations in HIV-infected patients.AIDS 1999,
13:473-478.
44. Dalmau D, Ochoa De Euchaguen AOE, Martinez-Lacasa JML,
et al.Indinavir pharmacokinetics and their correlation with virological and immunologic parameters.XII International Conference on AIDS. Geneva, June 1998 [abstract 42276].
45. Perello L, Goujard C, Delfraissy JF, Taburet AM.
Therapeutic drug monitoring in HIV-infected patients receiving indinavir.XII International Conference on AIDS. Geneva, June 1998 [abstract 42272].
46. Gatti G, Di Biagio A, De Pascalis CR,
et al.Interoccasion variability of indinavir in HIV infected patients [abstract].Clin Pharmacol Ther 2000,
67:156.
47. Acosta EP, Kakuda TN, Brundage NC,
et al.Pharmacodynamics of human immunodeficiency virus type 1 protease inhibitor.Clin Infect Dis 2000,
30(suppl 2):S151-S159.
48. Burger DM, Hoetelmans RMW, Mulder JW,
et al.Low levels of indinavir (IDV) are highly predictive of virological treatment failure in patients using IDV-containing regimens.XII International Conference on AIDS. Geneva, June 1998 [abstract 42275].
49. Gatti G, Vigano' A, Sala N,
et al.Indinavir pharmacokinetics and pharmacodynamics in children with human immunodeficiency virus infection.Antimicrob Agents Chemother 2000,
44:752-755.
50. Schapiro JM, Winters MA, Stewart F,
et al.The effect of high-dose saquinavir on viral load and CD4+ T-cell counts in HIV-infected patients.Ann Intern Med 1996,
124:1039-1050.
51. Vanhove GF, Gries JM, Verotta D,
et al.Exposure-response relationships for saquinavir, zidovudine, and zalcitabine in combination therapy.Antimicrob Agents Chemother 1997,
41:2433-2438.
52. Gieschke R, Fotteler B, Buss N, Steiner J-L.
Relationship between exposure to saquinavir monotherapy and antiviral response in HIV-positive patients.Clin Pharmacokinet 1999,
37:75-86.
53. Barry MG, Merry C, Lloyd J
et al.Variability in trough concentrations in HIV patients - a case for therapeutic drug monitoring.Br J Clin Pharmacol 1998,
45:501-502.
54. Langmann P, Zilly M, Weissbrich B,
et al.Therapeutic drug monitoring of saquinavir in patients during protease inhibitor therapy with saquinavir alone or in combination with ritonavir or nelfinavir.Eur J Med Res 2000,
5:59-62.
55. Reijers MH, Weigel HM, Hart AA,
et al.Toxicity and drug exposure in a quadruple drug regimen in HIV-1 infected patients participating in the ADAM study.AIDS 2000,
14:59-67.
56. Marzolini C, Buclin T, Decosterd LA,
et al.Nelfinavir plasma levels under twice-daily and three-times-daily regimens: high interpatient and low intrapatient variability.Ther Drug Monit 2001,
23:394-398.
57. Khaliq Y, Gallicano K, Seguin I,
et al.Single and multiple dose pharmacokinetics of nelfinavir and CYP2C19 activity in human immunodeficiency virus-infected patients with chronic liver disease.Br J Clin Pharmacol 2000,
50:108-115.
58. Seminari E, Maggiolo F, Villani P,
et al.Efavirenz, nelfinavir, and stavudine rescue combination therapy in HIV-1-positive patients heavily pretreated with nucleoside analogues and protease inhibitors.J Acquir Immune Defic Syndr 1999,
22:453-460.
59. Burger DM, Hugen PWH, Aarnoutse RE,
et al.Treatment failure of nelfinavir-containing triple therapy can largely be explained by low nelfinavir plasma concentrations.5th International Congress on Drug Therapy in HIV Infection. Glasgow, October 2000 [abstract 258].
60. Peytavin G, Le Moing V, Ecobichon JL,
et al.Relationship between nelfinavir plasma concentrations and late failure in APROCO (ANRS EP 11) cohort study.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 5.13].
61. Landman R, Peytavin G, Lamotte C,
et al.Therapeutic drug monitoring of nelfinavir in a prospective study (LIVIR IMEA 014) in HIV-HCV co-infected patients with chronic liver disease.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 6.4].
62. Sadler BM, Gillotin C, Lou Y, Stein DS.
Pharmacokinetic and pharmacodynamic study of the human immunodeficiency virus protease inhibitor amprenavir after multiple oral dosing.Antimicrob Agents Chemother 2001,
45:30-37.
63. Stein DS, Lou Yu, Johnson M,
et al.PK/PD analysis of variables that influence antiviral response in children treated with amprenavir containing combination antiviral therapy (PROB2004).2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 5.6].
64. Elston R, Randall S, Xu F,
et al.High plasma trough levels favour selection of the 150V mutation pathway during development of APV resistance.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 5.1].
65. Luber AD, Gunawan S, Wolfe PR,
et al.Serum and plasma drug levels of amprenavir display limited inter- and intra-patient variability.5th International Congress on Drug Therapy in HIV Infection. Glasgow, October 2000 [abstract P267].
66. Danner SA, Carr A, Leonard JM,
et al.A short-term study of the safety, pharmacokinetics, and efficacy of ritonavir, an inhibitor of HIV-1 protease. European-Australian Collaborative Ritonavir Study Group.N Engl J Med 1995,
333:1528-1533.
67. Gatti G, Di Biagio A, Casazza R.
The relationship between ritonavir plasma levels and side-effects: implications for therapeutic drug monitoring.AIDS 1999,
13:2083-2089.
68. Molla A, Korneyeva M, Gao Q,
et al.Ordered accumulation of mutations in HIV protease confers resistance to ritonavir.Nat Med 1996,
2:760-766.
69. Dumon C, Solas C, Thuret I,
et al.Relationship between efficacy, tolerance, and plasma drug concentration of ritonavir in children with advanced HIV infection.Ther Drug Monit 2000,
22:402-408.
70. Gatti G, Guerra M, Miletich F,
et al.Comparative study of dapsone plasma levels in HIV+ patients and healthy volunteers using a population approach.36th Interscience Conference on Antimicrobial Agents and Chemotherapy. New Orleans, September 1996 [abstract A52].
71. Gatti F, Flaherty F, Bubp J,
et al.Comparative study of the bioavailabilities and pharmacokinetics of clindamycin in healthy volunteers and patients with AIDS.Antimicrob Agents Chemother 1993,
37:1137-1143.
72. Gatti G, Di Biagio A, De Pascalis CR, Guerra M, Bassetti M, Bassetti D.
Pharmacokinetics of rifabutin in HIV-infected patients with or without wasting syndrome.Br J Clin Pharmacol 1999,
48:704-711.
73. Hsu A, Granneman GR, Witt G,
et al.Multiple-dose pharmacokinetics of ritonavir in human immunodeficiency virus-infected subjects.Antimicrob Agents Chemother 1997,
41:898-905.
74. Hill A.
Reliability ofCmin:
inhibitory concentration ratios.J Infect Dis 2001,
183:992-994.
75.
Kaletra US Prescribing Information. Data. Abbott Laboratories, Abbott Park, IL, USA.
76. Piketty C, Race E, Catiel
et al.Efficacy of five drug combination of ritonavir, saquinavir, and efavirenz in patients who failed on conventional triple drug regimen: phenotypic resistance to protease inhibitors predicts outcome of therapy.AIDS 1999,
13:F71-F77.
77. Sham HL, Kempf DJ, Molla A,
et al.ABT378, a highly potent inhibitor of the human immunodeficiency virus protease.Antimicrob Agents Chemother 1998,
42:3218-3224.
78. Saag MS, Kilby M, Ehrensing E,
et al.Modulation of saquinavir steady-state pharmacokinetics with 'baby' doses of ritonavir in healthy volunteers.7th European Conference on Clinical Aspects and Treatment of HIV Infection. Lisbon, October 1999 [abstract 829].
79. Hsu A, Granneman K, Cao G,
et al.Pharmacokinetic interactions between ritonavir and indinavir in healthy volunteers.Antimicrob Agents Chemother 1998,
42:2784-2791.
80. Sadler BM, Piliero PJ, Preston SL, Yu L, Stein DS.
Pharmacokinetics and safety of amprenavir and ritonavir following multiple-dose, co-administration to healthy volunteers. drug-interaction between amprenavir and ritonavir in HIV-seronegative subjects after multiple, oral dosing.AIDS 2001,
15:1009-1018.
81. Aarnoutse R, Burger D, van Oosterhaut J,
et al.Multiple dose pharmacokinetics and tolerability of once daily nelfinavir and ritonavir combinations in healthy volunteers.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 1.3].
82. Raines CP, Flexner C, Sun E,
et al.Safety, tolerability and antiretroviral effects of ritonavir-nelfinavir combination therapy administered for 48 weeks.J AIDS, 2000,
25:322-328.
83. Wood R, Trepo C, Livrozet JM,
et al.Enhancement of pharmacokinetic parameters of amprenavir when combined with low dose ritonavir (APV 600/RTV 100 mg BID) and preliminary efficacy results.5th International Congress on Drug Therapy in HIV Infection. Glasgow, October 2000 [abstract P283].
84. Saag MS, Kilby M, Ehrensing E,
et al.Saquinavir systemic exposure and safety of once daily administration of Fortovase (saquinavir) soft gel capsule (FTV) in combination with low dose ritonavir (RTV).39th Interscience Conference on Antimicrobial Agents and Chemotherapy. San Francisco, September 1999 [poster 330].
85. Lorenzi P, Yerly S, Abderrakim K,
et al.Toxicity, efficacy, plasma drug concentrations and protease mutations in patients with advanced HIV infection treated with ritonavir plus saquinavir. Swiss HIV Cohort Study.AIDS 1997,
11:F95-F99.
86. Gisolf EH, van Heeswijk RP, Hoetelmans RW, Danner SA.
Decreased exposure to saquinavir in HIV-1-infected patients after long-term antiretroviral therapy including ritonavir and saquinavir.AIDS 2000,
14:801-805.
87. Kurowski M, Muller M, Donath F, Mrozikiewicz M, Mocklinghoff C.
Single daily doses of saquinavir achieve HIV-inhibitory concentrations when combined with baby-dose ritonavir.Eur J Med Res 1999,
4:101-104.
88. van Heeswijk RP, Veldkamp AI, Mulder JW,
et al.Once-daily dosing of saquinavir and low-dose ritonavir in HIV-1-infected individuals: a pharmacokinetic pilot study.AIDS 2000,
14:F103-F110.
89. Condra JH, Petropoulos CJ, Ziermann R,
et al.Drug resistance and predicted virological responses to human immunodeficiency virus type 1 protease inhibitor therapy.J Infect Dis 2000,
182:758-765.
90. O'Brien WA, Atkinson TA, HanX, Sova M, East J.
Combination therapy with indinavir and ritonavir in antiretroviral-experienced patients.39th Interscience Conference on Antimicrobial Agents and Chemotherapy. San Francisco, September 1999 [abstract 2209].
91. Gatell JM, Lange J, Arnaiz JA, Faetkenheuer G, Mallolas J, Clotet B.
A randomized study comparing continued indinavir (800 mg tid) vs switching to indinavir/ritonavir (800/100 mg bid) in HIV patients having achieved viral load suppression with indinavir plus 2 nucleoside analogues The BID efficacy and safety Trial (BEST).XIII International Conference on AIDS. Durban, July 2000 [abstract WeOrB484].
92. Romer K, Fatkenheuer G, Veldkamps AI,
et al.Pharmacokinetics and virological efficacy of the combination of 600mg amprenavir BID, 100mg ritonavir BID and efavirenz QD in multiply pretreated HIV positive patients.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 5.10].
93. Katlama C, Schneider L, Agher R,
et al.Ritonavir/amprenavir combination therapy in HIV infected patients who failed several protease inhibitors containing regimens.5th International Congress on Drug Therapy in HIV Infection. Glasgow, October 2000 [abstract P41].
94. Poirier JM, Meynard JL, Guiard-Schmid JB,
et al.Unexpected drug-drug interaction between amprenavir and ritonavir.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 1.6].
95. Piscitelli S, Bechtel C, Sadler B, Falloon J, and the Intramural AIDS Program.
The addition of a second protease inhibitor eliminates amprenavir-efavirenz drug interactions and increases plasma amprenavir concentrations.Seventh Conference on Retroviruses and Opportunistic Infections. San Francisco, January-February 2000 [abstract 78].
96. Hsu R, Bellman P.
The combination of amprenavir with low dose ritonavir in salvage regimens overcomes clinically multiple-drug resistant HIV.XIII International AIDS Conference. Durban, July 2000 [abstract WePeB4204].
97. Bertz R, Lam W, Brun S,
et al.Multiple dose pharmacokinetics of ABT378/ritonavir in HIV+ subjects.39th Interscience Conference on Antimicrobial Agents and Chemotherapy. San Francisco, September 1999 [abstract 327].
98. Poirier JM, Meynard JL, Guiard-Schmid JB,
et al.Lopinavir and ritonavir trough plasma concentrations in HIV-experienced patients treated with Kaletra.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 1.7].
99. Kempf D, Brun S, Rode R,
et al.Identification of clinically relevant phenotypic and genotypic breakpoints for ABT 378/r in multiple PI-experienced, NNRTI-naive patients [abstract].Antivir Ther 2000;
5(suppl 3):70.
100. Gifford AL, Bormann JE, Shively MJ, Wright BC, Richman DD, Bozzette SA.
Predictors of self-reported adherence and plasma HIV concentrations in patients on multidrug antiretroviral regimens.J Acquir Immune Defic Syndr 2000,
23:386-395.
101. Paterson DL, Swindells S, Mohr J,
et al.Adherence to protease inhibitor therapy and outcomes in patients with HIV infection.Ann Internal Med 2000,
133:21-30.
102. Kastrissios H, Suarez JR, Hammer S, Katzenstein D, Blaschke TF.
The extent of non-adherence in a large AIDS clinical trial using plasma dideoxynucleoside concentrations as a marker.AIDS 1998,
12:2305-2311.
103.
Fortovase (saquinavir) soft gel capsule.US Prescribing Information. Roche Laboratories Inc., Nutley, NJ, USA; 1997.
104.
Crixivan (indinavir sulfate) capsules.US Prescribing Information. Merck and Co. Inc., West Point, PA, USA; February 2000.
105.
Viracept (nelfinavir mesylate) tablets and oral powder.US Prescribing Information. Agouron Pharmaceuticals Inc., San Diego, CA, USA; January 1998.
106.
Agenerase (amprenavir) capsules.US Prescribing Information. GlaxoWellcome, Research Triangle Park, NC, USA; May 2000.
107.
Norvir (ritonavir capsules) soft gelatin.US Prescribing Information. Abbott Laboratories, Abbott Park, IL, USA; March 2000.
108. Hugen P, Burger DM, Aarnoutse RE,
et al.Concentration ratios of protease inhibitors can be applied to assess non-compliance.1st International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, March 2000 [abstract 3.1].
109. Brundage RC, Yong FH, Fenton T,
et al.Variability in efavirenz concentrations predicts virological outcome in HIV-infected children.1st International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, March 2000 [abstract 3.2].
110. Fellay J, Marzolini C, Greub G,
et al.Predictive power of P-glycoprotein and CYP2D6 polymorphisms for plasma levels of antiretroviral agents.Eighth Conference on Retroviruses and Opportunistic Infections. Chicago, February 2001 [abstract 260].
111. Piscitelli SC, Burstein AH, Chaitt D, Alfaro RM, Falloon J.
Indinavir concentrations and St John's Wort.Lancet 2000,
12(355):547-548.
112. Decker CJ, Laitinen LM, Bridson GW,
et al.Metabolism of amprenavir in liver microsomes: role of CYP3A4 inhibition for drug interactions.J Pharm Sci 1998,
87:803-807.
113. Barry M, Mulcahy F, Merry C,
et al.Pharmacokinetics and potential interactions amongst antiretroviral agents used to treat patients with HIV infection.Clin Pharmacokinet 1999,
36: 289-304.
114. Goujard C, Meynard JL, Choudet N,
et al.Study PROF1004: steady-state pharmacokinetics of amprenavir 600mg BID and ritonavir 100mg BID with or without NNRTI in HIV-1 infected patients.5th International Congress on Drug Therapy in HIV Infection. Glasgow, October 2000 [abstract P268].
115. Lamotte C, Duval X, Peytavin G,
et al.Amprenavir plasma concentrations are dramatically increased by association to ritonavir baby doses in HIV-infected patients: possible combination with efavirenz.1st International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, March 2000 [abstract 2.7].
116. Aarnoutse RE, Burger DM, Hugen PWH,
et al.A pharmacokinetic (PK) study to investigate the influence of efavirenz (EFV) on a BID IDV/RTV regimen (800/100) in healthy volunteers.40th Interscience Conference on Antimicrobial Agents and Chemotherapy. Toronto, September 2000 [abstract 423].
117. Clumeck N, Brun S, Sylte J,
et al.ABT378/r and efavirenz: 24 week safety/efficacy evaluation in multiple PI experienced patients.XIII International Conference on AIDS. Durban, July 2000 [abstract TuPeB3196].
118. Sulkowski MS, Thomas DL, Chaisson RE,
et al.Hepatotoxicity associated with antiretroviral therapy in adults infected with human Immunodeficiency virus and the role of hepatitis C or B virus infection.JAMA 2000,
283:74-80.
119. Veronese L, Rautaureau J, Sadler B,
et al.Single dose pharmacokinetics of amprenavir, a human immunodeficiency virus type 1 proteases inhibitor, in subjects with normal or impaired hepatic function.Antimicrobial Agents Chemother 2000,
44:821-826.
120. Malavaud B, Dinh B, Bonnet E,
et al.Increased incidence of indinavir nephrolitiasis in patients with hepatitis B or C virus infection.Antivir Ther,5:3-5.
121. Gatti G, Di Biagio A, Beltrame A,
et al.Gender as a risk factor for ritonavir intolerance.39th Interscience Conference on Antimicrobial Agents and Chemotherapy. San Francisco, September 1999 [abstract 22].
122. Perinatal HIV Guidelines Working Group Members.
Public Health Service Task Force recommendations for use of antiretroviral drugs in pregnant HIV-1-infected women for maternal health and interventions to reduce perinatal HIV-1 transmission in the United States.Living Document. Available at
http://www.hivatis.org/guidelines/perinatal/May03_01/Perinatal-May04_01.pdf, accessed 23 May 2001.
123. Moodley J, Moodley D, Pillay K,
et al.Pharmacokinetics and antiretroviral activity of lamivudine alone or when coadministered with zidovudine in human immunodeficiency virus type 1-infected pregnant women and their offspring.J Infect Dis 1998,
178:1327-1333.
124. Wang Y, Livingston E, Patil S,
et al.Pharmacokinetics of didanosine in antepartum and postpartum human immunodeficiency virus-infected pregnant women and their neonates: an AIDS Clinical Trials Group study.J Infect Dis 1999,
180:1536-1541.
125. Rodman JH, Flynn PM, Robbins B,
et al.Systemic pharmacokinetics and cellular pharmacology of zidovudine in human immunodeficiency virus type 1-infected women and newborn infants.J Infect Dis 1999,
180:1844-1850.
126. Mirochnick M, Fenton T, Gagnier P,
et al.Pharmacokinetics of nevirapine in human immunodeficiency virus type 1-infected pregnant women and their neonates. Pediatric AIDS Clinical Trials Group Protocol 250 Team.J Infect Dis 1998,
178:368-374.
127. Hayashi S, Beckerman K, Homma,
et al.Pharmacokinetics of indinavir in HIV-positive pregnant women.AIDS 2000,
14:1061-1062.
128. Grubman S, Gross E, Lerner-Weiss N,
et al.Older children and adolescents living with perinatally acquired human immunodeficiency virus infection.Pediatrics 1995,
95:657-663.
129. Brundage RC, Fletcher CV, Fenton T,
et al.Efavirenz and nelfinavir pharmacokinetics in HIV infected children under 2 years of age.Seventh Conference on Retroviruses and Opportunistic Infections. San Francisco, January-February 2000 [abstract 719].
130. Rongkavilit C, Van Heeswijk RPG, Risuwanna P,
et al.The safety and pharmacokinetics of nelfinavir in a dose-escalating study in HIV-1-exposed newborn infants: HIV-NAT 007.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 3.1].
131. Burger DM, Van Rossum AMC, Hugen PWH,
et al.The indinavir AUC in HIV-1 infected children should be above 20 mg/L/h to obtain virological response.1st International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, March 2000 [abstract 5.3].
132. Kline MW, Brundage RC, Fletcher CV,
et al.Combination therapy with saquinavir soft gelatin capsules in children with human immunodeficiency virus infection.Pediatr Infect Dis J 2001,
20:666-671.
133. Hoffman F, Notheis G, Wintergerst U,
et al.Comparison of ritonavir plus saquinavir- and nelfinavir plus saquinavir-containing regimens as salvage therapy in children with human immunodeficiency type 1 infection.Pediatr Infect Dis J 2000,
19:47-51.
134. Van Rossum AM, De Groot R, Hartwig NG,
et al.Pharmacokinetics of indinavir and low-dose ritonavir in children with HIV-1 infection.AIDS 2000,
14:2209-2210.
135. Mole L, Schmidgall D, Holodniy M.
A pilot trial of indinavir, ritonavir, didanosine, and lamivudine in a once-daily four-drug regimen for HIV infection.J Acquir Immune Defic Syndr 2001,
27:260-265.
136. Burger DM, Hugen PWH, Droste J,
et al.Therapeutic drug monitoring of indinavir in treatment-naive patients improves therapeutic outcome after 1 year: results from ATHENA.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 6.2a].
137. Blanco JL, Mallolas J, Sarasa M,
et al.Intensification therapy with saquinavir soft gel/ritonavir QD in patients failing on a saquinavir hard gel containing HAART.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 5.8].
138. Havlir D, Gallant J, Race E,
et al.Ritonavir intensification in indinavir recipients: tolerance, antiviral effect and pharmacokinetics.XIII International Conference on AIDS. Durban, July 2000 [abstract 4122].
139. Drusano GL, D'Argenio DZ, Preston SL,
et al.Use of drug effect interaction modeling with Monte Carlo simulation to examine the impact of dosing interval on the projected antiviral activity of the combination of abacavir and amprenavir.Antimcrob Agents Chemother 2000,
44:1655-1659.
140. Regazzi MB, Villani P, Maserati R,
et al.Pharmacokinetic variability and strategy for therapeutic drug monitoring of saquinavir (SQV) in HIV-1 infected individuals.Br J Clin Pharmacol 1999,
47:379-382.
141. Garaffo R, Durant J, Roptin C,
et al.Protease inhibitors kinetics in a human cell line expressing or not MDR1 gene and role of ritonavir to increase intracellular PI accumulation.1st International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, March 2000 [abstract 4.5].
142. Khoo S, Hennessy M, Mulcahy F,
et al.Differences in intracellular accumulation between protease inhibitors. Presented at
Eighth Conference on Retroviruses and Opportunistic Infections. Chicago, February 2001 [abstract 258].
143. Hsu A, Bertz R, Hickman D,
et al.Assessment of plasma protein binding of lopinavir and ritonavir between in vitro and ex vivo experiments and between HIV-infected patients and HIV-negative healthy volunteers.Eighth Conference on Retroviruses and Opportunistic Infections. Chicago, February 2001 [abstract 753].
144. Craig C, Rose J, Haywood J,
et al.Protein binding: effect on antiviral efficacy.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 3.7].
145. Barry M, Gibbons S, Back D, Mulcahy F.
Protease inhibitors in patients with HIV disease. Clinically important pharmacokinetic considerations.Clin Pharmacokinet 1997,
32:194-209.
146. Zhang XQ, Schooley RT, Gerber JG.
The effect of increasing alpha1-acid glycoprotein concentration on the antiviral efficacy of human immunodeficiency virus protease inhibitors.J Infect Dis 1999,
180:1833-1837.
147. Sadler BM, Gillotin C, Lou Y,
et al.In vivo effect of alpha-1 acid glycoprotein on pharmacokinetics of amprenavir, a human immunodeficiency virus protease inhibitor.Antimicrob Agents Chemother 2001,
45:852-856.
148. Burger D, Aarnoutse R, Hugen P.
Validation of the use of the concentration ratio to estimate the area under the curve of indinavir in HIV-1 infected patients.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 6.6].
149. Simon V, Vanderhoeven J, Hurley A,
et al.Prevalence of drug resistant HIV-1 variants in newly infected individuals during 1999-2000.Eighth Conference on Retroviruses and Opportunistic Infections. Chicago, February 2001 [abstract 423].
150. Hicks, Eron J, Lennox J,
et al.Resistance to antiretroviral agents in a cohort of patients with acute HIV infection acquired in non urban areas in southeastern USA.Eighth Conference on Retroviruses and Opportunistic Infections. Chicago, February 2001 [abstract 425].
151. Larder BA, Kempf SD, Hertogs K.
Quantitative prediction of phenotypic drug resistance from genotypes: the virtual phenotype (Virtual/
Phenotypeâ„¢) [abstract].Antiviral Ther 2000,
5(suppl 3):49.
152. Kempf D, Hsu A, Jiang P,
et al.Response to ritonavir intensification in indinavir recipients is highly correlated with virtual inhibitory quotient.Eighth Conference on Retroviruses and Opportunistic Infections. Chicago, February 2001 [abstract 523].
153. Piscitelli SC, Metcalf J, RMW Hoetelmans RMW,
et al.The normalized inhibitory quotient: a method for predicting response to amprenavir.Eighth European Conference on Clinical Aspects and Treatment of HIV. Athens, October 2001 [abstract 164].
154. Lee CGL, Gottesman MM, Cardarelli CO,
et al.HIV-1 protease inhibitors are substrates for the MDR-1 multidrug transporter.Biochemistry 1998,
37:3594-3601.
155. Andreana A, Aggarwal S, Gollapudi S, Wien D, Tsuruo T, Gupta S.
Abnormal expression of a 170-kilodalton P-glycoprotein encoded by MDR1 gene, a metabolically active efflux pump, in CD4+ and CD8+ T cells from patients with human immunodeficiency virus type 1 infection.AIDS Res Hum Retroviruses 1996,
12:1457-1462.
156. Diaz-Borjon A, Richaud-Patin Y, Alvarado de la Barrera C, Jakez-Ocampo J, Ruiz-Arguelles A, Llorente L.
Multidrug resistance-1 (MDR-1) in rheumatic autoimmune disorders. Part II: Increased P-glycoprotein activity in lymphocytes from systemic lupus erythematosus patients might affect steroid requirements for disease control.Joint Bone Spine 2000,
67:40-48.
157. Washington CB, Wiltshire HR, Man M.
The disposition of saquinavir in normal and P-glycoprotein deficient mice, rats, and in cultured cells.Drug Metab Dispos 2000,
28:1058-1062.
158. Choo EF, Leake B, Wandel C,
et al.Pharmacological inhibition of P-glycoprotein transport enhances the distribution of HIV-1 protease inhibitors into brain and testes.Drug Metab Dispos 2000,
28:655-660.
159. Flexner C.
Role of multidrug transporters in HIV pathogenesis.Eighth Conference on Retroviruses and Opportunistic Infections. Chicago, February 2001 [abstract S4].
160. Clevenbergh P, Garraffo R, Durant J,
et al.PharmAdapt: a prospective multicenter randomized controlled trial to evaluate the usefulness of protease inhibitor therapeutic drug monitoring: 12 weeks results.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 6.1].
161. Aarnoutse R, Burger DM, Verwij-Van Wissen C,
et al.An international interlaboratory quality control (QC) program for therapeutic drug monitoring (TDM) in HIV-infection: first results.Eighth Conference on Retroviruses and Opportunistic Infections. Chicago, February 2001 [abstract 734].
162. Palette C, Taburet AM, Marquet P,
et al.Evaluation of serum HIV drugs determinations in France: preliminary results of a national quality control survey.First International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, March 2000 [abstract 1.5].
163. NONMEM Project Group.
NONMEM Users Guides. Edited by Beal SL, Sheiner LB. San Francisco, CA: University of California; 1992.
164. Burger DM, Hugen PWH, Droste J,
et al.Therapeutic drug monitoring of nelfinavir 1250BID in treatment-naive patients improves therapeutic outcome after 1 year: results from ATHENA.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 6.2b].
165. Hill A.
Are clinical trials of TDM underpowered to detect effects on HIV-RNA?: Three stage modeling analysis of sample sizes.2nd International Workshop on Clinical Pharmacology of HIV Therapy. Noordwijk, April 2001 [abstract 6.9]
166. Starr SE, Fletcher CV, Spector SA,
et al.Combination therapy with efavirenz, nelfinavir, and nucleoside reverse-transcriptase inhibitors in children infected with human immunodeficiency virus type 1. Pediatric AIDS Clinical Trials Group 382 Team.N Engl J Med 1999,
341:1874-1881.
167. Gallant JE.
Strategies for long-term success in the treatment of HIV infection.JAMA 2000,
283:1329-1334.
168. Burger DM, Hoetelmans RM, Hugen PW,
et al.Low plasma concentrations of indinavir are related to virological treatment failure in HIV-1-infected patients on indinavir-containing triple therapy.Antiviral Therapy 1998,
3:215-220.
169. Harris M, Durakovic S, Rae S,
et al.A pilot study of nevirapine, indinavir, and lamivudine among patients with Advanced human immunodeficiency virus disease who have had failure of combination nucleoside therapy.J Infect Disease 1998,
177: 1514-1520.
170. Fletcher CV, Brundage RC, Remmel RP,
et al.Pharmacologic characteristics of indinavir, didanosine, and stavudine in human immunodeficiency virus-infected children receiving combination therapy.Antimicrob Agents Chemother 2000,
44:1029-1034.
171. Murphy RL, Sommadossi JP, Lamson M, Hall DB, Myers M, Dusek A.
Antiviral effect and pharmacokinetic interaction between nevirapine and indinavir in persons infected with human immunodeficiency virus type 1.J Infect Dis 1999,
179:1116-1123.
172. Acosta EP, Havlir DV, Richman DD,
et al.Pharmacodynamics (PD) of indinavir (IDV) in protease-naive HIV-infected patients receiving ZDV and 3TC.Seventh Conference on Retroviruses and Opportunistic Infections. San Francisco, January-February 2000 [abstract 455].
173. Perry CM, Noble S.
Saquinavir soft-gel capsule formulation. A review of its use in patients with HIV infection.Drugs 1998,
55:461-486.
174. Hsu A, Granneman GR, Bertz RJ.
Ritonavir. Clinical pharmacokinetics and interactions with other anti-HIV agents.Clin Pharmacokinet 1998,
35:275-291.
175. Yeh KC, Deutsch PJ, Haddix H,
et al.Single-dose pharmacokinetics of indinavir and the effect of food.Antimicrob Agents Chemother 1998,
42:332-338.
176. Pai VB, Nahata MC.
Nelfinavir mesylate: a protease inhibitor.Ann Pharmacother 1999,
33:325-339.
177. Jarvis B, Faulds D.
Nelfinavir. A review of its therapeutic efficacy in HIV infection.Drugs 1998,
56:147-167.
178. Sadler BM, Hanson CD, Chittick GE, Symonds WT, Roskell NS.
Safety and pharmacokinetics of amprenavir (141W94), a human immunodeficiency virus (HIV) type 1 protease inhibitor, following oral administration of single doses to HIV-infected adults.Antimicrob Agents Chemother 1999,
43:1686-1692.
179. Flexner C.
HIV-protease inhibitors.N Engl J Med 1998,
338: 1281-1292.
180.
Amprenavir Data On File. Glaxo Wellcome, Research Triangle PArk, NC, USA.
181. Livington DJ, Pazhanisamy S, Porter DJ, Partaledis JA, Tung RD, Painter GR.
Weak binding of VX-478 to human plasma proteins and implications for anti-human immunodeficiency virus therapy.J Infect Dis 1995,
172:1238-1245.
182. Markowitz M, Conant M, Hurley A,
et al.A preliminary evaluation of nelfinavir mesylate, an inhibitor of human immunodeficiency virus (HIV)-1 protease, to treat HIV infection.J Infect Dis 1998,
177:1533-1540.
183. Molla A, Granneman GR, Sun E, Kempf DJ.
Recent developments in HIV protease inhibitor therapy.Antiviral Res 1998,
39:1-23.
184. Taylor S, Back DJ, Workman J,
et al.Poor penetration of the male genital tract by HIV-1 protease inhibitors.AIDS 1999,
13: 859-860.
Section Description
Supported by an unrestricted educational grant from Vertex Pharamceuticals
