Share this article on:

Therapeutic Drug Monitoring of Efavirenz: Trough Levels Cannot Be Estimated on the Basis of Earlier Plasma Determinations

López-Cortés, Luis F MD, PhD*; Ruiz-Valderas, Rosa MD, PhD*; Marín-Niebla, Ana MD†; Pascual-Carrasco, Rosario RN*; Rodríguez-Díez, Magdalena RN*; Lucero-Muñoz, María J PharmD, PhD‡

JAIDS Journal of Acquired Immune Deficiency Syndromes: 15 August 2005 - Volume 39 - Issue 5 - pp 551-556
Clinical Science

Objective: The few studies evaluating the relation between efavirenz (EFV) plasma levels and efficacy have only shown a weak association. They are based on population pharmacokinetics and/or estimation of trough levels from samples obtained at different time points, because this drug is usually taken at bedtime. We explore whether a limited sampling strategy could provide a good estimation of EFV trough levels and area under the curve (AUC).

Methods: Fifty-nine 24-hour profiles from patients on 600 mg of EFV administered once daily were included for the analysis. Plasma drug concentrations were determined by a validated high-performance liquid chromatography assay. The relation between EFV plasma concentrations at C8, C12, and C16 and the corresponding C24 and AUC0-24 values were assessed in different index sets, and their predictive performances were evaluated in the validation sets.

Results: EFV plasma levels at C12 and C16 are precise predictors of AUC0-24 (≤10% mean relative prediction errors). The estimation of C24 levels from EFV plasma concentrations at C8, C12, or C16 is not precise enough (18%-62% mean relative prediction errors), however, and the use of C12 and C16 in combination does not improve the accuracy of C24 prediction.

Conclusion: No adequate substitute for collecting blood 24 hours after dosing emerged from this analysis, which might explain the weak association observed between EFV plasma levels and treatment outcomes in the studies available.

From the *Infectious Diseases Service, Hospitales Universitarios Virgen del Rocío, Seville, Spain; †Hematology Service, Hospitales Universitarios Virgen del Rocío, Seville, Spain; and ‡Faculty of Pharmacy, University of Seville, Seville, Spain.

Received for publication January 16, 2005; accepted May 13, 2005.

Partially supported by grant 12/02 from the Consejería de Salud, Junta de Andalucía, and Red de Investigación en SIDA (ISCIII G03/173).

Reprints: Luis F. López-Cortés, Servicio de Enfermedades Infecciosas, Hospital Universitario Virgen del Rocío, Avda. Manuel Siurot, s/n. 41013, Seville, Spain (e-mail:

Although current antiretroviral therapy has significantly modified the natural history of HIV infection, a high incidence of therapeutic failures is still observed with the regimens currently used, being more frequent in antiretroviral treatment-experienced patients than in naive patients. Potential factors, such as the high interpatient variability in plasma levels of most antiretroviral drugs, the clear relation between trough levels and efficacy, and the lower virologic failure rates when optimal concentrations are achieved, have already been demonstrated in different studies.1-11 Although therapeutic drug monitoring (TDM) for antiretroviral drugs is not a routine practice, guidelines proposing specific indications and therapeutic ranges are already available.12 Not all the cutoff values given in these guidelines are fully supported by clinical data, however.

Regarding efavirenz (EFV), a nonnucleoside inhibitor of the HIV-1 reverse transcriptase mainly metabolized by CYP3A4 and CYP2B6 isoenzymes,13 the scarce published reports studying the relation between plasma levels and efficacy are based on a population pharmacokinetic approach, with blood samples taken between 8 and 20 hours after dosing, because this drug is usually taken at bedtime to minimize its psychiatric and central nervous system adverse effects. It is believed that the variability in EFV levels during a dose interval is minimal and that samples can be taken without regard to the time elapsed between intake and sampling because of its prolonged elimination half-life.12,14-17 Because there are no prospective studies validating the minimum plasma level recommended, we have explored whether plasma levels obtained at different time points could provide a good estimation of EFV trough plasma levels and/or area under the curve (AUC) so as to facilitate future studies on this issue.

Back to Top | Article Outline



The data were obtained from 97 adult HIV-infected patients included in several EFV pharmacokinetic studies at the Infectious Diseases Service of our hospital. They were included only when the EFV dose administered was 600 mg daily, no acute diseases were present, and a complete EFV 24-hour pharmacokinetic profile was available, with blood samples collected before dosing and thereafter at 1, 2, 3, 4, 5, 6, 8, 12, 16, and 24 hours. The details of one of these clinical protocols have been previously reported.18 All these studies were approved by the local ethics committee. A total of 59 24-hour profiles from 44 patients were selected. Fifteen patients underwent 2 pharmacokinetic profiles, which were also included in the analysis because a change in antiretroviral treatment that was expected to alter EFV exposure had occurred between both profiles corresponding to each of these patients.

All the patients had been taking 600 mg of EFV once daily for at least 3 months before sampling and any concurrent drugs were allowed. In 15 cases, the regimen consisted of EFV plus 2 nucleoside analogue reverse transcriptase inhibitors (NRTIs; group 1); in the rest of the patients, the regimen included EFV plus rifampin (n = 11, group 2) or EFV in combination with protease inhibitors (PIs; n = 33, group 3). During the 2 weeks before the pharmacokinetic study, no other drugs with a potential effect on the EFV pharmacokinetic profile had been administered to any of the patients.

Back to Top | Article Outline


In a first step, the data from half of the patients in each treatment group, randomly selected, were used as the index set and the data from the remaining patients in the same treatment group were used as the validation set. In a second step, the data from the 15 patients in group 1 were used as the index set and the remaining 44 pharmacokinetic profiles were used as the validation set. Finally, the data of 29 patients, randomly selected among the 3 groups, were used as the index set, and the remaining 30 pharmacokinetic profiles were used as the validation set.

Back to Top | Article Outline

Determination of Efavirenz Concentrations

Plasma EFV levels were determined by reverse-phase high-performance liquid chromatography (HPLC) according to a validated method.19 The assay was found to be linear and was validated over a concentration range of 0.2 to 30 μg/mL. The lower limit of quantification was 0.1 μg/mL. Recovery of EFV from human plasma was 100.1% ± 1.23%. The mean intra- and interassay coefficients of variation were 3.2% (range: 1.5%-5%) and 4.7% (range: 2%-8%), respectively. Our laboratory is linked to the KKGT-International Interlaboratory Quality Control Program for TDM in HIV Infection (Department of Clinical Pharmacy, University Medical Center, Nijmegen, The Netherlands).

Back to Top | Article Outline

Pharmacokinetic Analysis

EFV pharmacokinetic parameters were determined by using the WinNonLin program (version 3.1; Pharsight Corporation, CA) according to a noncompartmental method from plasma drug concentration-time data. The terminal log-linear period (β) was defined by the data points from 12 to 24 hours after taking the drug (C12-C24).

Back to Top | Article Outline

Statistical Analysis

Statistical calculations were performed with the Statistical Product and Service Solutions (SPSS) for Windows (version 12.0; SPSS, Chicago, IL). The relation between the plasma EFV concentrations at C8, C12, and C16 and the corresponding trough levels (C24) and AUC0-24 values were assessed by univariate linear regression analysis. Their predictive performances were evaluated by using the root mean squared relative prediction errors (%RMSE) and standard errors (SE%) as a measure of precision and the mean relative predictor errors (%MPE) and SE% as a measure of bias. The %RMSE, the %MPE, and the corresponding SE% are given by the following equations:

in which N is the number of C24 and AUC0-24 pairs (ie, true with predicted values) and pei is the relative prediction error for each pair. Ideally, a predictive model should be precise (ie, the variation in the results, %RMSE, should be <15%) and unbiased (ie, the % deviation of the predicted values from the actual values, %MPE, should not be significantly different from 0).20,21 Comparisons of demographic and pharmacokinetic parameters among the different treatment groups were carried out by the Kruskal-Wallis test and analysis of variance. P ≤ 0.05 was considered statistically significant for all tests.

Back to Top | Article Outline


Data were obtained from 44 HIV-infected patients (34 male, 10 female) with a median age of 38 years (range: 22-66 years), a median weight of 66.1 kg (range: 44-104 kg), and a median CD4 cell count of 462 cells/μL (range: 7-926 cells/μL). All but 2 patients had an RNA HIV viral load less than 50 copies/mL when the pharmacokinetic profiles were obtained.

Table 1 and Figure 1 show the pharmacokinetic profile of EFV when administered alone and when associated with rifampin or PIs. No significant differences in demographic (sex, age, CD4 count, and HIV viral load) and pharmacokinetic parameters were found among the different treatment groups, except for the median weight of the patients who were concomitantly given rifampin, which was lower than that of patients from the other 2 groups (P = 0.04 and P = 0.01, respectively). Despite this fact, EFV plasma levels were lower in the rifampin group than in the other treatment groups, although the only statistically significant differences were observed with the group on PIs (P ≤ 0.01). Neither rifampin nor PIs had any influence on EFV elimination half-life (t½β), being similar among the 3 groups.

Predictability of EFV systemic exposure and trough levels was based on plasma determinations at 8, 12, and 16 hours.

The results of the univariate linear regression analysis of EFV levels at C8, C12, and C16, and the predictive performance of the equations for predicting C24 or AUC0-24 in the validation sets are shown in Table 2.

Although a close linear relation exists between EFV plasma concentrations at 8, 12, and 16 hours and C24 in the index sets (Fig. 2), EFV plasma levels at these time points do not allow a precise estimation of C24 levels in the validation sets (see Fig. 2), with high mean relative prediction errors (see Table 2). The use of both time points (C12 and C16) in combination does not improve the accuracy of the model either, and the results from the samples obtained within C2 to C6 yield even worse results (data not shown). The high interpatient variability in EFV t½β causes a wide variability in C24 in those patients with similar concentrations at a given time point (ie, 12 hours; Table 3). Therefore, no adequate substitute for collecting blood 24 hours after dosing emerged from this analysis. The data from these time points are not useful to estimate maximum concentration (Cmax; data not shown).

A similar situation occurs when estimating the systemic exposure to EFV (AUC0-24) based on plasma levels at C8. EFV plasma levels at 12 and 16 hours are precise predictors of AUC0-24, however, with mean relative prediction errors ≤10% (95% confidence interval [CI95] ≤15%) and biases not significantly different from 0. Figures 2 and 3 show the plot of the observed values for the C24 and AUC0-24 versus the estimated values based on plasma levels at C12 and C16.

Back to Top | Article Outline


The EFV pharmacokinetic data observed in our patients are similar to other previously reported data,14-16,21-25 with slight variations that might be attributable to the concomitant use of drugs such as rifampin or ritonavir. It is worth mentioning that neither rifampin nor PIs have any effect on EFV t½β. As in previous studies, our results show a high interpatient variability in EFV plasma levels. This variability and the low genetic barrier of EFV for the occurrence of high-level HIV phenotypic resistances are clear reasons for TDM of this drug. Moreover, neurologic adverse effects have been correlated with higher EFV plasma levels.24

Although with some antiretroviral drugs, a relation has been observed between efficacy and AUCτ, efficacy has usually been related to trough concentrations, and a consensus exists for using trough samples for therapeutic drug monitoring.1,3,12,25 Obtaining EFV through plasma levels carries certain difficulties in the clinical setting, however, because EFV is usually taken at bedtime. As a result, most of the studies evaluating EFV plasma concentrations in relation to its virologic efficacy are based on population pharmacokinetics and/or estimation of trough levels from plasma samples obtained at different time points after taking EFV but not on true trough levels.14-16

In the first study reported,24 treatment failure was approximately 3 times more frequent at a minimum concentration (Cmin) <1.1 mg/L (63%) than at a Cmin ≥1.1 mg/L (20.6%). The patients included in this study had participated in several heterogeneous clinical trials, however, and the efficacy of the different therapeutic regimens was probably affected by factors such as different EFV initial dosages (200-600 mg once daily) and the fact that some patients were treatment experienced and others were naive. Moreover, there was a considerable overlap in the range of plasma concentrations between the 2 groups.

The most referenced study on EFV pharmacokinetics14 has set the current recommendation that the optimal range is between 1 and 4 μg/mL in samples withdrawn between 8 and 20 hours after drug intake. In this study and in the third report,15 which is an extension of the former in which a larger sample of patients is analyzed, the treatment regimens used were heterogeneous (EFV plus 2 NRTIs or in combination with a PI with or without NRTIs), no previous therapies or resistances were referred, and the relation between EFV plasma levels and virologic status of the patients was only punctually analyzed because of a lack of longitudinal data. Indeed, virologic failure was observed in 22% and 18% of patients with EFV levels of 1 to 4 μg/mL and >4 μg/mL, respectively, with the inverse correlations between EFV concentrations and the viral load values being practically negligible in both studies (r = 0.34 and r = 0.14, respectively). As the authors explained later, the association between EFV plasma levels and treatment outcomes was weak, and the therapeutic range of 1 to 4 μg/mL was proposed for validation from an intervention trial.26,27 Our results could explain this weak association because they demonstrate that the estimation of C24 levels from EFV plasma concentrations between 8 and 16 hours is not precise enough to be considered reliable.

Our results could also explain the lack of relation between EFV plasma levels and efficacy in a study performed in heavily pretreated patients,28 although other additional factors possibly influencing the results of this study were also present, such as the heterogeneity of the population studied and concomitant antiretroviral drugs. Moreover, the low plasma levels of EFV (<1 μg/mL) detected in some patients in the samples withdrawn in the first hours of the morning strongly suggest nonadherence in these cases.

Despite the high correlation coefficients existing between EFV plasma levels at different points and C24, the estimation of C24 cannot be made precisely enough. This is not unexpected if we take into account that coefficient of correlation can only estimate the degree of association between 2 variables along the best line relating them but that it cannot predict the actual value of one of the variables based on the other variable's value (ie, if we systematically overpredict a given concentration by 200%-300%, the correlation between the actual concentration and the predicted one would be perfect but completely wrong).20

Conversely, our data suggest that EFV AUC0-24 can be accurately estimated from single plasma samples obtained at 12 or 16 hours. The relation between EFV AUC0-24 and efficacy has not been evaluated yet, however.

Therefore, prospective studies based on plasma samples obtained at 24 hours after drug intake should be carried out to clarify if there is any relation between EFV trough plasma levels and efficacy and which are the effective EFV plasma concentrations to be recommended in TDM programs.

Back to Top | Article Outline


1. Van Heeswijk RP. Critical issues in therapeutic drug monitoring of antiretroviral drugs. Ther Drug Monit. 2002;24:323-331.
2. Acosta EP, King JR. Methods for integration of pharmacokinetic and phenotypic information in the treatment of infection with human immunodeficiency virus. Clin Infect Dis. 2003;36:373-377.
3. Back DJ, Khoo SH. The role of clinical pharmacology in optimizing antiretroviral therapy. Br J Clin Pharmacol. 2003;55:473-476.
4. Evans WE, McLeod HL. Pharmacogenomics: drug disposition, drug targets, and side effects. N Engl J Med. 2003;348:538-549.
5. 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.
6. Gieschke R, Fotteler B, Buss N, et al. Relationships between exposure to saquinavir monotherapy and antiviral response in HIV-positive patients. Clin Pharmacokinet. 1999;37:75-86.
7. Pellegrin I, Breilh D, Montestruc F, et al. Virologic response to nelfinavir-based regimens: pharmacokinetics and drug resistance mutations. AIDS. 2002;16:1331-1340.
8. 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. Antivir Ther. 1998;3:215-220.
9. 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.
10. Veldkamp AI, 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.
11. Fletcher CV, Anderson PL, Kakuda TN, et al. Concentration-controlled compared with conventional antiretroviral therapy for HIV infection. AIDS. 2002;16:551-560.
12. Back D, Blascke T, Boucher C, et al. A practical guide to performing therapeutic drug monitoring for antiretroviral agents. Version 1.0. April 2003. Available at:
13. Virology BMS. Efavirenz (Sustiva®): prescribing information. Princeton, NJ: Bristol-Myers Squibb; 2002.
14. 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.
15. Csajka C, Marzolini C, Fattinger K, et al. Population pharmacokinetics and effects of efavirenz in patients with human immunodeficiency virus infection. Clin Pharmacol Ther. 2003;73:20-30.
16. Barrett JS, Joshi AS, Chai M, et al. Population pharmacokinetic meta-analysis with efavirenz. Int J Clin Pharmacol Ther. 2002;40:507-519.
17. López-Cortés LF, Ruiz-Valderas R, Viciana P, et al. Once-daily saquinavir-sgc plus low-dose ritonavir (1200/100 mg) in combination with efavirenz: pharmacokinetics and efficacy in HIV-infected patients with prior antiretroviral therapy. J Acquir Immune Defic Syndr. 2003;32:240-242.
18. Sarasa-Nacenta M, López-Púa Y, López-Cortés LF, et al. Determination of efavirenz in human plasma by high-performance liquid chromatography with ultraviolet detection. J Chromatogr B Biomed Appl. 2001;763:53-59.
19. Sheiner LB, Beal SL. Some suggestions for measuring predictive performance. J Pharmacokinet Biopharm. 1981;9:503-512.
20. Veldkamp AI, van Heeswijk RP, Mulder JW, et al. Limited sampling strategies for the estimation of the systemic exposure to the HIV-1 nonnucleoside reverse transcriptase inhibitor nevirapine. Ther Drug Monit. 2001;23:606-611.
21. Veldkamp AI, Harris M, Montaner JS, et al. The steady-state pharmacokinetics of efavirenz and nevirapine when used in combination in human immunodeficiency virus type 1-infected persons. J Infect Dis. 2001;184:37-42.
22. Fiske W, Benedek IH, Joseph JL, et al. Pharmacokinetics of efavirenz and ritonavir after multiple oral doses in healthy volunteers [abstract 42269]. Int Conf AIDS. 1998;12:827.
23. Joshi AS, Barrett JS, Fiske WD, et al. Population pharmacokinetics of efavirenz in phase II studies and relationship with efficacy [abstract 1201]. Presented at: 39th Interscience Conference on Antimicrobial Agents and Chemotherapy; 1991; San Francisco.
24. Gallego L, Barreiro P, del Rio R, et al. Analyzing sleep abnormalities in HIV-infected patients treated with efavirenz. Clin Infect Dis. 2004;38:430-432.
25. Back D, Gatti G, Fletcher C, et al. Therapeutic drug monitoring in HIV infection: current status and future directions. AIDS. 2002;16(Suppl 1):S5-S37.
26. López-Cortés LF, Alarcón A, Viciana P. Efavirenz plasma concentrations and efficiency. AIDS. 2001;15:1192-1193.
27. Marzolini C, Telenti A, Decosterd L, et al. Efavirenz plasma levels can predict treatment failure and central nervous system side effects in HIV-1-infected patients. AIDS. 2001;15:1193-1194.
28. Langmann P, Weissbrich B, Desch S, et al. Efavirenz plasma levels for the prediction of treatment failure in heavily pretreated HIV-1 infected patients. Eur J Med Res. 2002;7:309-314.

Cited By:

This article has been cited 3 time(s).

Therapeutic monitoring is necessary for the association itraconazole and efavirenz in a patient with AIDS and disseminated histoplasmosis
Huet, E; Hadji, C; Hulin, A; Botterel, F; Bretagne, S; Lévy, Y
AIDS, 22(14): 1885-1886.
PDF (356) | CrossRef
JAIDS Journal of Acquired Immune Deficiency Syndromes
Efavirenz Induces CYP2B6-Mediated Hydroxylation of Bupropion in Healthy Subjects
Robertson, SM; Maldarelli, F; Natarajan, V; Formentini, E; Alfaro, RM; Penzak, SR
JAIDS Journal of Acquired Immune Deficiency Syndromes, 49(5): 513-519.
PDF (128) | CrossRef
Therapeutic Drug Monitoring
Limited-Sampling Strategy for the Prediction of Boosted Hard-Gel Saquinavir Exposure at a Dosage of 1000/100 mg Twice Daily in Human Immunodeficiency Virus-Infected Individuals
Dickinson, L; Back, D; Pozniak, A; Khoo, S; Boffito, M
Therapeutic Drug Monitoring, 29(3): 361-367.
PDF (138) | CrossRef
Back to Top | Article Outline

efavirenz; pharmacokinetics; drug monitoring; antiretroviral therapy; HIV

© 2005 Lippincott Williams & Wilkins, Inc.