High Variability of Plasma Drug Concentrations in Dual Protease Inhibitor Regimens

Ritonavir (RTV) strongly increases the concentrations of proteaseinhibitors (PIs) in plasma in patients given a combination ofRTV and another PI. This pharmacological interaction is complexand poorly characterized and shows marked inter- and intraindividualvariations. In addition, RTV interacts differently with saquinavir(SQV), indinavir (IDV), amprenavir (APV), and lopinavir (LPV).In this retrospective study on 542 human immunodeficiency virus-infectedpatients, we compared inter- and intraindividual variabilityof plasma PI concentrations and correlations between the C_min(minimum concentration of drug in plasma) values for RTV andthe coadministered PI C_min values. Mean RTV C_mins are significantlylower in patients receiving combinations containing APV or LPVthan in combinations with SQV or IDV. With the most common PIdose regimens (600 mg of IDV twice a day [BID], 800 mg of SQVBID, and 400 mg of LPV BID), the interindividual C_min variabilityof patients treated with a PI and RTV seemed to be lower withAPV and LPV than with IDV and SQV. As regards intraindividualvariability, APV also differed from the other PIs, exhibitinglower C_min variability than with the other combinations. Significantpositive correlations between RTV C_min and boosted PI C_min wereobserved with IDV, SQV, and LPV, but not with APV. Individualdose adjustments must take into account the specificity thepharmacological interaction of each RTV/PI combination and thelarge inter- and intraindividual variability of plasma PI levelsto avoid suboptimal plasma drug concentrations which may leadto treatment failure and too high concentrations which may inducetoxicity and therefore reduce patient compliance.

INTRODUCTION

Top

Introduction

Ritonavir (RTV), a human immunodeficiency virus (HIV) proteaseinhibitor (PI), is relatively poorly tolerated at the full dose(600 mg twice a day [BID]) and has a potent inhibitory effecton the cytochrome P450 enzyme CYP3A, the major enzyme responsiblefor the metabolism of the PIs and many other drugs, in the intestinalwall and liver. RTV also inhibits P-glycoprotein transport,which may increase the absorption of PIs by this pathway. Evenat a reduced dose, RTV strongly increases the plasma drug concentrationsof other PIs administered concomitantly (10).

This property of RTV in which it strongly increases the concentrationsof other PIs is exploited therapeutically: low-dose RTV is increasinglyused to improve the pharmacokinetic profiles of other PIs byensuring higher and more stable plasma drug concentrations,while at the same reducing the number of daily doses and thenumber of tablets or capsules in each dose and sometimes avoidingthe effect of food on drug absorption (11).

These pharmacological interactions are complex and poorly characterizedand show marked interindividual variations (1). In addition,RTV interacts differently with saquinavir (SQV), indinavir (IDV),amprenavir (APV), and lopinavir (LPV).

Precise individual tailoring of the dose regimens of these PIcombinations is crucial to avoid subtherapeutic plasma drugconcentrations (which are predictive of virologic breakthrough)and excessively high concentrations (which may induce toxicityand therefore reduce patient compliance) (1-3, 11, 13).

In this study of patients receiving various PIs in combinationwith RTV, we compared the pharmacological interactions of thedifferent dual-PI regimens.

MATERIALS AND METHODS

Top

Materials and Methods

This retrospective study involved HIV-infected patients managedin the infectious diseases department of St. Antoine and Tenonhospitals (Paris, France) from September 1999 to August 2002.They were treated with RTV (100 mg BID) ("baby dose") in combinationwith SQV (600 to 800 mg BID), IDV (400 to 800 mg BID), APV (600mg BID), or LPV (400 mg BID). Patients receiving combinationsof more than two PIs (salvage therapy) were excluded from thestudy, as were those receiving nonnucleoside reverse transcriptaseinhibitors (owing to their interactions with PIs). If patientshad received successive PI/RTV dual combinations and undergonesuccessive drug determinations during the study period, onlythe pharmacological data collected for the first PI/RTV dualcombination were used in this study in order to preserve statisticalindependence between the groups in the interindividual variabilityanalysis.

When several drug assays were done in a given patient receivingthe same PI combination at the same doses, only the initialvalues were used for the analysis of interindividual variability.The analysis of intraindividual variability was exclusivelybased on values obtained in a given patient receiving the samePI combination at the same doses at each assay.

Patients who forgot to take at least one drug dose during thetwo days prior to the pharmacological assay were not includedin the analysis. The nurse who performed the blood test askedpatients about their compliance, and information was based onpatients' answers. Only plasma PI concentrations measured 12± 2 h after the last administration (minimum concentrationof drug in plasma [C_min]) were taken into account in patientsreceiving a stable antiretroviral drug regimen for at least2 weeks before blood sampling.

These RTV and PI concentrations in plasma were determined byhigh-pressure liquid chromatography with quantification cutoffsof 5 ng/ml for APV and IDV and 10 ng/ml for LPV, SQV, and RTV(15).

Data were analyzed using nonparametric tests (Yates' and Fisher'schi-squared tests and the Mann-Whitney test); only series ofassays performed on samples from at least 10 subjects were usedfor the analysis of interindividual variability. Correlationsbetween the RTV C_mins and the PI C_mins were determined by linearregression curves.

RESULTS

Top

Results

A total of 955 plasma PI determinations were done during the36-month study period in 542 patients. A total of 413 patientsreceived low-dose RTV (100 mg BID) combined with another PI.The other 129 patients received only one PI (APV, IDV, or nelfinavir[NFV]) and were studied for comparison.

Interindividual variability of plasma RTV concentrations. RTV C_mins in the patients who were also receiving another PIshowed large interindividual variability, regardless of thecombination. The calculated coefficients of variation of RTVC_mins observed with each RTV/PI combination were generally high(Table 1).

View this table:
[in this window]
[in a new window]
TABLE 1. Interindividual variability of RTV and PI C_mins

Using a nonparametric Kruskal-Wallis test, distributions ofRTV C_mins are globally different between the groups (P <0.0001). Distributions are not statistically different, however,between the three IDV/RTV groups (P = 0.45 by the Kruskal-Wallistest) and between the two SQV/RTV groups (P = 0.55 by the Mann-Whitneytest) (Table 2).

View this table:
[in this window]
[in a new window]
TABLE 2. Comparison of mean RTV C_mins of RTV/PI combinations by the Mann-Whitney test

Using the Mann-Whitney test, mean RTV C_mins in the patientsreceiving SQV or IDV were not statistically different (P = 0.697).In contrast, mean RTV C_mins were significantly lower in patientsreceiving APV or LPV than in those receiving SQV or IDV (Tables1 and 2). The coefficients of variation of mean RTV C_mins weregenerally large, close to 60% in patients receiving APV or LPVand from 71 to 100% in those receiving SQV or IDV (Table 1).Mean RTV C_mins were statistically lower in the patients whowere receiving APV or LPV than in the other RTV/PI combinations(P = 0.009 by the Mann-Whitney test) (Table 2).

Therefore, SQV and IDV behaved differently from APV and LPVin their influence on RTV C_mins following a regimen of 100 mgof RTV BID.

Interindividual variability of plasma drug concentrations of the PI combined with RTV. The interindividual variability of C_mins observed in patientsreceiving IDV, NFV, and APV without RTV was shown for comparisonin Table 1 as the C_min coefficients of variation. The variabilityof the three drugs was established in groups of different sizes(29, 10, and 90 patients, respectively), but the values arein keeping with those reported elsewhere (4, 5, 16). The interindividualvariability of C_min was higher following administration of IDV(coefficient of variation, 182.6%) than with NFV or APV (coefficientof variation, 69.3 and 61.6%, respectively).

As shown by the comparison of mean C_mins of IDV and APV administeredwith and without RTV, the use of RTV as a pharmacological boosterresulted in a significant increase in the plasma PI concentrations.Combining APV with RTV multiplies the mean APV C_min by a factorof 5.5 while allowing the APV dose to be halved. The mean IDVC_min was increased by a factor of 2.5 in combination with RTV,despite a decrease in the IDV dose from 2,400 mg/day (threetimes a day) to 1,200 mg/day (BID). The booster effect of RTVon NFV was weaker (<2 with the same NFV doses) (Table 1).

However, the interindividual variability of the C_mins of eachPI remained large despite concomitant RTV administration (coefficientof variation, 45.0 to 106.4%). There was no marked differencesin the interindividual variability of the C_mins between thedifferent dosage regimens when IDV was combined with RTV (coefficientof variation, 76.5 to 92.0%). When SQV was combined with RTV,the coefficient of variation of the C_mins ranged from 83.7%(600 mg BID regimen) to 107.6% (800 mg BID regimen) (Table 1).

The analysis also suggests that the interindividual variabilityof C_mins of the PIs combined with RTV was lower in the caseof APV and LPV (45.3 and 50.8%, respectively) than with IDVand SQV (76.4 to 91.4% and 83.7 to 107.6%, respectively) (Table1).

Intraindividual variability. C_mins of RTV/PI combinations were sometimes determined againin order to control the stability of the pharmacological interactionin the different combinations. A median of three plasma PI measurements(range, two to six) were done in 132 patients during identicaldose regimens (Table 3).

View this table:
[in this window]
[in a new window]
TABLE 3. Intraindividual variability of PI and RTV C_mins

The intraindividual variability of the C_mins of RTV and PI wasexpressed in terms of coefficient of variation, calculated fromthe weighted means of drug determinations performed for eachpatient. Table 3 shows the mean coefficients of variation andstandard deviations for RTV and PI C_mins for patients givenRTV/PI combinations. This intraindividual variability of C_minswas relatively large, as the coefficients of variation of meanvalues observed ranged from 28.4 to 61.4% for the PI and 21.9to 75.8% for RTV. As shown by the comparison of coefficientsof variation of mean values, the intraindividual variabilitiesof RTV and PI C_mins are similar. Globally, the observed valueswere similar to those obtained in the 26 patients treated withNFV without RTV. For the IDV/RTV combination, we observed adifference in the coefficients of variation of means accordingto the IDV dose prescribed (>50% for 800-mg IDV BID and 600-mgIDV BID regimens and <30% for 400-mg IDV BID regimen).

Correlations between C_mins of RTV and PI in patients given different RTV/PI combinations. (i) IDV and RTV. In patients given the IDV/RTV combination, the C_mins of RTVcorrelated with those of IDV: the higher the RTV C_min, the higherthe IDV C_min (correlation coefficient [r] = 0.72 and P <0.0001 for the combination of RTV [100 mg] and IDV [600 mg],both drugs given BID) (Fig. 1). The same correlation was foundwith the other IDV dose regimens (800 and 400 mg BID).

View larger version (19K):
[in this window]
[in a new window]
FIG. 1. Correlation between the C_mins for IDV (600 mg BID regimen) and RTV (100 mg BID).

(ii) SQV and RTV. In patients given the SQV/RTV combination, the C_mins of SQVand RTV correlated in the same way as the IDV/RTV combination.The C_mins of RTV and SQV correlated with each other, despitea wider dispersion of the results (r = 0.30 and P < 0.001for the combination of RTV [100 mg] and SQV [600 mg], both givenBID) (Fig. 2).

View larger version (17K):
[in this window]
[in a new window]
FIG. 2. Correlation between C_mins for SQV (600 mg BID regimen) and RTV (100 mg BID).

(iii) LPV and RTV. C_mins of RTV and LPV (400 mg BID) also correlated with eachother (r = 0.51 and P < 0.0001) (Fig. 3), again despite awide dispersion of the data in this small subset of patients.

View larger version (18K):
[in this window]
[in a new window]
FIG. 3. Correlation between C_mins for LPV (400 mg BID regimen) and RTV (100 mg BID).

(iv) APV and RTV. In contrast, C_mins of RTV and APV appeared to be independentof each other, regardless of the APV dose regimen. At a doseof 600 mg of APV BID, the correlation coefficient r was 0.025(P > 0.1) (Fig. 4).

View larger version (18K):
[in this window]
[in a new window]
FIG. 4. No correlation between C_mins for APV (600 mg BID regimen) and RTV (100 mg BID). NS, not significant.

DISCUSSION

Top

Discussion

There has been a large increase in the use of low-dose RTV incombination with a variety of PIs over the last 2 years. SQV,IDV, and APV are now almost always prescribed in combinationwith RTV. RTV improves the pharmacokinetic profile of the differentcoadministered PIs by raising their concentrations in plasma,lengthening their half-lives of elimination, and reducing theinfluence of food on their gastrointestinal absorption. Therefore,an increase in drug exposure may improve, in some patients,virologic success. By reducing the number of pills and/or decreasingthe frequency of doses and removing food restrictions, theseRTV combinations may also improve patients' comfort and compliance(1, 3, 11).

However, despite the widespread use of such RTV/PI combinations,few data are available on the plasma PI concentrations achievedin clinical practice. Available data on the clinical pharmacologyof PIs were obtained in small groups of patients and under experimentalconditions that differ somewhat from the usual ambulatory follow-upof HIV-infected patients.

In France, therapeutic drug monitoring of antiretroviral treatmentsis increasingly used to determine the best dosage regimen adaptedto each individual patient in order to reduce the risk of virologicfailure (due to low plasma drug concentrations), to limit thetoxicity linked to high plasma drug concentrations, and alsoto facilitate treatment from the patient's point of view. Therapeuticdrug monitoring of these drugs could provide practical informationfor clinicians (3, 15).

In this retrospective study of a large cohort of ambulatorypatients managed in two specialized hospital units, we assessedthe interindividual and intraindividual variability of plasmaPI concentrations obtained during routine assays in patientstreated with RTV-boosted PI combinations. To avoid bias dueto retrospective data analysis, patients have been carefullyselected. They need to receive RTV (100 mg BID) in a first-linetreatment of dual-PI regimen. If they had received successivePI/RTV dual combinations, only pharmacological data collectedduring the first PI/RTV dual regimen were used in this study.Patients treated with drugs known to have potent pharmacologicalinteractions with PIs (rifampin, rifabutin, ketoconazole, itraconazole,and phenobarbital) were excluded from the analysis, as werepatients receiving nonnucleoside reverse transcriptase inhibitors.For patients who underwent several consecutive drug determinationsbefore and after a change in the dosage of the PI combined withRTV, only the initial values (collected during the first PIdosage) were used for the analysis of interindividual variability.For patients who had been on a stable antiretroviral combinationfor at least 2 weeks, only samples taken within the allocatedinterval (12 ± 2 h after the last dose) were used here.

The nurse in charge of taking blood samples for pharmacologicalassays systematically asked patients about compliance with thedrug treatment during the two previous days and noted the exacttime of the last drug dose taken. A patient who forgot one doseduring the two previous days was considered noncompliant andexcluded from the analysis.

Compliance of antiretroviral therapy is difficult to assessin clinical practice, and collecting information about drugintake regularity is always partially inaccurate. This inaccuracycould probably be limited by using the most recent recollectionof the patient, and that is why we usually focused questionsabout treatment compliance on the last 2 days. All patientswere informed about the pharmacological assays. One can thereforeassume that the degree of inaccuracy in the results due to complianceproblems was equally distributed among the groups.

Interindividual variability of plasma RTV concentrations. Large interindividual variability of mean RTV C_mins was observedamong the different groups of patients (coefficient of variation,57.4 to 99.9%). Variability of mean RTV C_mins was found to belower in patients receiving combinations containing APV or LPV(59.9 and 57.4%) than in those receiving combinations containingSQV or IDV (71.7, 76.1, and 95.4% for IDV and 70.8 and 77.9%for SQV) irrespective of the dose of the coadministered PI,as shown by coefficients of variation. Mean RTV C_mins were foundto be significantly lower in patients receiving combinationscontaining APV or LPV than in those receiving combinations containingSQV or IDV. These results could be explained by an inductioneffect of LPV and APV on the metabolism of RTV.

This observation should be taken into account when another PImust be added to a combination of LPV/RTV or APV/RTV in salvagetherapy. Indeed, if plasma RTV concentrations are reduced byAPV or LPV, the expected booster effect of RTV on the thirdPI contained in the combination may be correspondingly reduced(9).

In some patients receiving the SQV/RTV or IDV/RTV combination,we observed high RTV C_mins, so this PI may participate directlyto the overall antiretroviral activity. However, in patientsreceiving the APV/RTV or LPV/RTV combination, RTV C_min alwaysremained under its minimal effective trough level in plasma.

Interindividual variability of plasma drug concentrations of the PI combined with RTV. Large interindividual C_min variability of PIs was observed duringtreatment without RTV among our patients (61.7, 69.3, and 182.5%for APV, NFV, and IDV, respectively), as in previous studies(4, 5, 12, 16). Interestingly, despite the pharmacokinetic enhancementof these PIs by the addition of low-dose RTV, the interindividualvariability of these concentrations remains high in our study(45.0, 74.6, and 78.8%, respectively). With the most commonPI dose regimens (800, 600, and 400 mg of IDV BID; 800 and 600mg of SQV BID; and 400 mg of LPV or APV BID), the interindividualC_min variability of the PI combined with RTV was lower for LPVand APV (51.6 and 45%) than for IDV (76.5 to 92.0%) or SQV (83.7to 106.3%) in terms of coefficients of variation observed ata given dose regimen.

The interindividual variability of PI concentrations probablydepends on many parameters, including intestinal and hepaticmetabolism, body weight (or the body mass index), age, gender,genetic factors, viral hepatitis or opportunistic infections,alcohol consumption, etc. Nonoptimal treatment compliance (drugnot taken or inadequate time between medications or betweendrug intake and plasma sampling) participate in this overallvariability. However, if the drug dose was taken later thanscheduled, the inadequate time between the last drug dose andplasma sample would introduce less variability for PIs withlonger half-lives. Therefore, with respect to their terminalhalf-life, the interpatient variability in C_mins of boostedLPV and APV could be lower than for the other PIs.

However, pharmacological interactions between RTV and the differentPIs are different for the different combinations, as shown here.

Intraindividual variability. Regarding intraindividual variability, all the assay valuesused here were obtained after at least 14 days of stable treatment,suggesting that our results were not influenced by the autoinductionof RTV metabolism observed during the first days of treatmentwith this PI. Treatment compliance may play a determining role(especially with respect to the dosing schedule).

The overall intraindividual RTV and PI C_min variability wassmaller than the corresponding interindividual variability.

The wide intraindividual variability must be taken into accountin order to avoid erroneous dosage adjustment based on a singletrough drug concentration measurement. For these reasons, weperform the first assay after 2 weeks of treatment, with atleast one further assay after 2 months if the dosage regimenremains unchanged. If the dosage was adjusted, another drugdetermination was made after 2 weeks.

Correlations between the C_mins of RTV and PI for different RTV/PI combinations. The study of the relationship between the C_mins of RTV and thePI in patients given RTV/PI combinations, expressed in termsof their correlation (linear regression), showed significantpositive correlations in the case of IDV, SQV, and LPV (P <0.0001). In contrast, no such correlation was found betweenthe RTV and APV C_mins (P > 0.1).

The increase in the concentration of RTV in plasma thereforeaccentuates its booster effect on IDV, SQV, and LPV, but notAPV. According to these results, the C_mins of IDV, SQV, andLPV can be raised by increasing the dose regimen of either RTVor PI. However, in the case of APV, only an increase in theAPV dosage will increase its C_min, in accordance with previousfindings in healthy volunteers (B. M. Sadler, P. J. Pileiro,S. L. Preston, L. Yu, and D. L. Stein, Abstr. 7th Conf. Retrovir.Opportun. Infect., abstr. 77, 2000).

According to the observed correlations between PI and RTV C_mins,inter- and intraindividual variability of PI C_mins could bepartially due to those of RTV C_mins, except for APV.

Many retrospective studies have reported concentration relationshipsbetween the exposure to PIs and antiretroviral efficacy and/ora related toxicity (2, 6-8, 14). The use of low-dose RTV incombination with various PIs improves the pharmacokinetic profileof the PI given in combination with RTV, thereby in theory reducingthe risk of suboptimal or toxic plasma PI concentrations. However,the inter- and intraindividual variability of plasma drug concentrationsof RTV and the coadministered PI remains high in patients thustreated. Consequently, therapeutic drug monitoring of PIs mayrepresent an additional tool for the management of HIV-infectedpatients either to improve antiretroviral response or to decreasetoxicity. Individual dosage adjustments must take into accountthe specificity of each pharmacological interaction, but theclinical relevance and benefit of therapeutic drug monitoringremain to be demonstrated.

FOOTNOTES

* Corresponding author. Mailing address: Department of Clinical Infectious and Tropical Diseases, Tenon Hospital, 4, rue de la Chine, 75970 Paris Cedex 20, France. Phone: 33 1 56 01 74 21. Fax: 33 1 56 01 74 23. E-mail: jean-baptiste.guiard-schmid{at}tnn.ap-hop-paris.fr.

REFERENCES

Top