Nelfinavir mesylate (Viracept, AG 1343) is a widely used HIV protease inhibitor. Because of its short plasma elimination half-life of 3.5–5 h, it has to be taken on a twice or thrice daily basis to allow sufficiently high plasma levels for antiviral activity. For sufficient absorption, nelfinavir should be taken with a meal. This results in a two- to threefold higher maximal plasma concentration (C max) and area under the concentration–time curve (AUC). Most of an oral dose of nelfinavir is excreted in faeces, either as unchanged drug or as metabolite [1,2]. Nelfinavir is metabolized by at least five different pathways in the human body . In vitro studies showed that cytochrome P450 (CYP) 3A4, 2C19 and 2D6 are the primary enzymes involved in the metabolism of nelfinavir. CYP3A4 is the major enzyme for three of the five routes, while CYP2D6 and CYP2C19 each catalyse one other route. The metabolism route catalysed by CYP2C19 leads to the formation of a pharmacologically active metabolite, M8 (hydroxy-t-butylamidenelfinavir; AG1402). The activity of CYP2C19, and thus the formation of M8, is genetically controlled. Poor metabolizers with CYP 2C19 possess two defective alleles resulting in inactive CYP2C19 protein and are expected to have lower or no M8 concentrations. After formation, M8 is further metabolized by CYP3A4 . M8 is reported to be as pharmacologically active as the parent compound. Zhang et al. showed in in vitro studies with HIV-1-infected cells that the median effective dose is 30 nmol/l (17.0 μg/l) for nelfinavir and 34 nmol/l (19.8 μg/l) for M8 . Several publications have demonstrated that M8 is present in plasma at sufficiently high levels after oral dosing of nelfinavir that a contribution of M8 to the antiviral activity must be considered [4,5].
Pharmacokinetic data on M8 published so far were obtained in rather small controlled studies. Here, we describe the pharmacokinetic characteristics of nelfinavir and M8 in a large, outpatient population, based on plasma samples collected for therapeutic drug monitoring. These have been used to characterize sources of variability in plasma levels and to evaluate the usefulness of therapeutic drug monitoring for nelfinavir treatment.
Materials and methods
Nelfinavir and M8 plasma concentrations were measured in blood samples taken from HIV-positive patients treated with nelfinavir-containing combination therapy. Blood samples were taken by standardized procedures from patients in various hospitals in the Netherlands between March and December 1999 for plasma concentration monitoring. Concentrations of nelfinavir were determined by a validated high performance liquid chromatographic (HPLC) method at the department of Clinical Pharmacy in the University Medical Centre Nijmegen . M8 was obtained from Agouron (San Diego, California, USA). M8 concentrations were determined simultaneously with nelfinavir using the same HPLC method without modifications. The retention time of M8 was 12.7 min. The lower and upper limits of quantification were 0.04 and 15.2 mg/l, respectively. Mean recovery was 92%. Based on quality control samples, intraday accuracy ranged from 93 to 106% of nominal concentrations. Intraday and interday precision varied between 2.8 and 4.3% and between 2.0 and 3.0%, respectively. Stability data obtained after 8 days in plasma at room temperature, three freeze–thaw cycles, 4 weeks in eluens at room temperature, and 1.5 years at −20°C demonstrated that M8 is stable under these conditions.
For each blood sample the following information was included in the database: patient number, gender, age, body weight, race, outpatient clinic, dosing scheme, reason for sampling, time of sampling, time of last drug intake before sampling and comedication. No information was available on metabolic genotype or phenotype for CYP2C19. Plasma concentrations of nelfinavir and M8 below the lower limit of quantification (0.04 mg/l) were considered to be 0.00 mg/l. If nelfinavir was undetectable, the ratio between the concentrations of M8 and nelfinavir was set as a missing value. Values for patients with an unknown dosing scheme or unknown time between intake of medication and blood sampling were excluded from the database. Plasma concentrations were assumed to be at steady state as generally no blood samples are taken within 2 weeks of starting therapy. For each sample, the ratio between the plasma concentrations of M8 and nelfinavir was calculated. Differences in molecular weight and any differences in protein binding, distribution characteristics and in vitro pharmacological activity were considered to have a negligible influence on the ratio. Information on a patient's gender, age, race or body weight that could not be obtained from the sampling forms or from the treating physician was regarded as missing value. Comedication was classified as follows: (i) compounds that induce CYP3A4, which included efavirenz, nevirapine, carbamazepine and rifabutin, (ii) compounds that inhibit CYP3A4, which included ritonavir, indinavir, itraconazole, and fluconazole, (iii) compounds that inhibit CYP2C19, which included fluoxetine, fluvoxamine, and omeprazole, (iv) antidiarrhetics, which included loperamide. Comedication was included only if there were demonstrated pharmacokinetic interactions in in vivo studies [7–9].
Pharmacokinetic population curves were constructed for the period 0–12 h after the morning dose, using MW/Pharm software (version 3.30, Medi/Ware & University Center for Pharmacy, University of Groningen, Groningen, the Netherlands). The curves were fitted on the median values of the plasma concentrations grouped for each 30 min. The median concentration of the morning trough values was used as an additional data point at zero time. Knowledge of plasma concentration–time curves constructed previously by a rich sampling strategy was used during the fitting procedure [10,11]. These curves suggest that nelfinavir pharmacokinetics are characterized by a lag time in absorption and by a circadian rhythm, the morning trough value being about 2.5-fold higher than the evening one.
After constructing the pharmacokinetic population curves, the ratio of each observed plasma concentration to the population median curve at that time point was calculated. Influence of demographic factors and comedication on the nelfinavir and M8 plasma concentrations was investigated by comparing the ratio between groups. Age and body weight were examined by grouping concentrations for each 10 years or 10 kg. Statistics were performed with SPSS for Windows, version 9.0 (SPSS Inc., Chicago, Illinois, USA). Statistical significance of differences between two patient groups was assessed using the Mann–Whitney test.
The patients included in the database were taking different dosages of nelfinavir. As almost 80% of the blood samples were taken from patients on a 1250 mg twice daily dosing scheme, statistical analysis focused on this dosage. Samples taken during a twice daily scheme but at a different dose to 1250 mg (13% of the samples) were analysed only for the investigation of changing dose. A thrice daily dosing scheme was used in 8% of the samples, which was considered insufficient for construction of a pharmacokinetic population curve for this dosage.
For the 1250 mg twice daily dosage, 618 plasma samples from 355 patients were available. The characteristics for these samples are shown in Table 1. There were 1 to 10 samples available for each patient; however, in the database, all random samples were included as individual samples because the method used to investigate differences between patient groups was not able to discriminate between intra- and interindividual variability. Plasma concentrations for the same patient were compared only for the investigation of changing nelfinavir dosage. Collection of blood samples was either in the morning, 9 to 15 h after the evening dose, or during the day, 0.15 to 12 h after the morning dose. Influence of comedication inhibiting CYP3A4 was not studied further because the number of people taking these drugs concomitantly was too low for powered assessment.
Nelfinavir and M8 population pharmacokinetics
Pharmacokinetic population curves for a 1250 mg twice daily dosage were constructed for nelfinavir and M8. The best fit for the curves was obtained with a one-compartment model with a lag-time of 1.3 h for both nelfinavir and M8. Constructed population curves for nelfinavir and M8 are shown in Fig. 1. Corresponding pharmacokinetic parameters for nelfinavir and M8, respectively, are C max3.4 and 1.1 mg/l, time to maximum plasma concentration 3.1 and 2.9 h, AUC0−−12 h 24.0 and 7.6 mg/l.h and half-life for elimination 4.1 and 4.3 h. The medians for the whole population comparing observed plasma concentrations with the constructed population curves were 1.07 for nelfinavir and 0.98 for M8, showing a good fit.
The ratio of M8 to nelfinavir appeared to be independent of the time after administration. Figure 2 shows the distribution of this ratio. The median ratio of the whole population was 0.29. M8 concentrations were below the limit of quantification in 31 samples with quantifiable nelfinavir concentrations. M8 concentrations were unquantifiable in all of the 16 patients with a nelfinavir concentration below the limit of quantification.
Influence of patient characteristics
Gender, age and body weight did not influence the nelfinavir or the M8 concentrations for a 1250 mg twice daily dosage. Median M8 concentrations were 39% and 49% lower in Black and Asian patients, respectively, than in the Caucasian population (P < 0.05). The ratio of M8 to nelfinavir in Black and in Asian people was significantly lower (P < 0.01) than in Caucasians (0.20 and 0.19 compared with 0.30) (Fig. 3b,c). Although Black and Asian patients tended to have higher nelfinavir concentrations than Caucasians, differences were not statistically significant. The median ratios of nelfinavir concentration compared with the population value were 1.23, 1.17 and 1.07 for the Asian, Black and Caucasian patients, respectively (Fig. 3a). Patients taking comedication inducing enzymes or inhibiting CYP2C19 were excluded from this analysis.
Influence of CYP3A4-inducing drugs
M8 concentrations were lower in patients using inducers of CYP3A4 as comedication than in patients not taking any comedication that influenced drug-metabolizing enzymes. The ratio of M8 concentration compared with the population curve was 0.53 (range, 0.00–4.41) for samples with inducing comedication and 1.05 (range, 0.00–5.85) without (P < 0.01). As nelfinavir concentrations are not influenced by concomitant intake of enzyme inducers, the ratio between M8 and nelfinavir was also half of the expected value: 0.14 (range, 0.00–1.01) with inducing comedication and 0.31 (range, 0.00–1.22) without (P < 0.001). The lower M8 concentrations were not associated with lower concentrations of total active compounds: total median nelfinavir plus M8 concentration was only 4% lower.
Influence of CYP2C19 inhibitors
Nine patients were reported to use omeprazole, one fluoxetine and one fluvoxamine. The median M8 concentration ratio in these 11 patients was 0.34 (range, 0.05–7.17) compared with the population median of 1.07 (range, 0.00–6.09). The median ratio of M8 to nelfinavir in these patients was 0.15 (range, 0.02–0.75) compared with 0.29 (range, 0.00–1.22) in the population. The variability in the ratio is large and, therefore, the effect of the CYP2C19 inhibition is not evident in all patients. Although the differences in median values for M8 are clear, the group of patients taking CYP2C19-inhibiting comedication is too small to demonstrate any statistical significance for the difference. CYP2C19 inhibition did not result in higher nelfinavir levels: the median nelfinavir concentration was 13% lower and the nelfinavir plus M8 concentration was 21% lower in the group taking CYP2C19-inhibiting comedication than in patients not taking enzyme influencing comediation.
Influence of diarrhoea
To study the nelfinavir concentrations in patients treated for diarrhoea, we selected patients from our database taking loperamide as comedication. The median plasma concentration of nelfinavir in these patients was 35% lower than the median population value (P < 0.05). Median M8 concentration in patients taking loperamide was also lower, although the difference was not statistically significantly. As a consequence, the ratio of M8 to nelfinavir was not altered by loperamide, but the summed concentration of active compounds was 28% lower (P < 0.05).
Rationale for therapeutic drug monitoring
Plasma concentrations of drugs can be checked for routine purposes to prevent virological failure but also when the physician suspects subtherapy, a compliance problem or intoxication. Plasma concentrations were available from five patients in whom intoxication occurred that was not caused by comedication. In three of them, nelfinavir plasma levels were indeed two- to threefold higher than normal. In the other two patients, nelfinavir plasma levels were very low, 15–20% of the median value, which may be explained by non-compliance owing to side-effects or poor absorption owing to diarrhoea.
In the groups that were sampled because of therapy failure or suspicion of non-compliance, the median plasma concentrations were 0.83 and 0.92, respectively, of the concentration in the group sampled for routine purposes. In these groups, the proportion of patients having nelfinavir concentrations less than 75% of the population median, which is regarded as subtherapeutic, was larger than for samples sent in for routine purposes (38% and 43%, respectively, versus 25%).
Change in nelfinavir dose
Eighteen patients taking 1250 mg twice daily were advised to increase their dosage to 1500 mg twice daily because their nelfinavir concentrations were too low (generally less than 75% of the population median). After the dose increase, plasma concentrations were monitored again. Eight patients were then advised to decrease the dosage from 1250 to 1000 mg twice daily because of high plasma levels (more than two times the population median). Patients changing comedication at the same time were excluded from the analyses. The effect of the dose adjustments was not consistent, as can be seen in Fig. 4. In a number of patients, the dose adjustment resulted in the desired decrease or increase in plasma concentration, while in other patients no effect or the opposite effect was observed.
In the Netherlands, plasma levels of nelfinavir are measured routinely for therapeutic drug monitoring to prevent virological failure. Plasma levels may also be measured to provide guidance in cases of suspected interactions, intoxication, non-compliance or drug intake without sufficient food. We presented data on 618 plasma concentrations of nelfinavir and its metabolite M8 from 355 HIV-patients, treated with nelfinvair containing antiretroviral treatment in various Dutch outpatients clinics. The M8 metabolite of nelfinavir is reported to have an equal pharmacological activity in vitro ; however, its contribution to the in vivo therapeutic activity of nelfinavir has not yet been established. In this study, it was presumed that they had an additive pharmacological effect as both compounds act on the same receptor.
The ratio of M8 to nelfinavir appeared to be independent of the time after nelfinavir ingestion and the M8 plasma concentration–time curve followed that of nelfinavir over time, without accumulation of the metabolite. Therefore, formation of M8 by CYP2C19 and not the subsequent metabolism by CYP3A4 is considered to be the rate-limiting pharmacokinetic process for M8. The median value of 0.29 for the ratio M8 to nelfinavir in plasma for patients on a 1250 mg twice daily dosage is in good agreement with previously reported data. Zhang et al. reported a ratio of 0.28 for AUC0−−8 h in 10 HIV-positive patients  and Lillibrigde et al. calculated a mean AUC ratio of 0.38 in 65 patients or healthy volunteers . Liver failure has been reported to influence M8 concentrations, probably through changes in CYP2C19 activity. In eight HIV-positive patients with mild-to-moderate liver failure, the ratio of AUC values for M8 and nelfinavir varied between 0.01 and 0.49 . In 18 patients with liver disease, the mean ratio of AUC values varied between 0.02 and 0.04 .
Possible sources of variation in nelfinavir and M8 plasma levels were studied by comparing the plasma concentrations for different groups of patients with the population concentration. Patients of Black or Asian descent generally had lower M8 concentrations than patients from the Caucasian population. This is in agreement with the incidence of poor metabolizers for CYP2C19 among different races and populations. CYP2C19 is the sole enzyme responsible for formation of M8 and no M8, or lower M8 concentrations, is expected in CYP2C19 poor metabolizers . Xie et al. [14,15] reported the frequencies of poor metabolism by CYP2C19 in the three major populations based on meta-analyses: Caucasians 2.1% (95% confidence interval (CI), 1.3–2.8), Black-Africans and Black-Americans 3.9% (95% CI, 2.7–5.2) and Oriental 14.3% (95% CI, 12.3–16.4) .
Coadministration of CYP2C19 inhibitors would be expected to result in lower M8 concentrations. In our database, omeprazole, fluoxetine and fluvoxamine were codrugs within this group [7–9]. In the small group of patients (n = 11) using CYP2C19 inhibitors, M8 concentrations were, on average, 68% lower than in the group of patients not taking comedication known to influence metabolism of other drugs. There was a large variation in the effect of the CYP2C19-inhibiting comedication. The elimination half-life of omeprazole is less than 1 h ; consequently, the time between intake of nelfinavir and omeprazole could be of crucial importance for the magnitude of the interaction. Nelfinavir concentrations did not appear to be affected by reduced metabolism by CYP2C19, suggesting that other metabolism routes for nelfinavir, catalysed by CYP3A4 or CYP2D6, can compensate for the reduced CYP2C19 activity.
The effect of CYP induction on nelfinavir metabolism has been studied in in vivo interaction studies after coadministration of the inducers efavirenz, nevirapine, rifampicin and rifabutin . The AUC and C max of nelfinavir was increased by 20% and 21%, respectively, when it was coadministered with efavirenz. No statistically significant changes in nelfinavir pharmacokinetic parameters were seen after addition of nevirapine in HIV-infected patients but rifampicin and rifabutin decreased the nelfinavir plasma AUC by 82% and 32%, respectively. Therefore, apart from the combination with rifampicin, nelfinavir pharmacokinetics were hardly influenced by CYP inducers, which is confirmed by our study. However, M8 concentrations were twofold lower in patients treated with efavirenz, carbamazepine, rifabutin or nevirapine. Apparently, metabolism of M8 is increased by induction of CYP3A4 while its formation by CYP2C19 is not affected by enzyme-inducing comedication.
The lower M8 levels observed in Black and Asian patients, or during coadministration of CYP2C19 and CYP3A4 inducers, were associated with marginal change in the nelfinavir concentrations and in the total amount of active compounds. M8 constitutes 23% of the plasma levels of total active compounds. If nelfinavir concentrations are unaltered, changes in M8 will have only a small effect on the total amount of active compounds. These findings support the conclusions of Zhang et al. that low M8 concentrations had no obvious effect on tolerability or antiviral response to nelfinavir . Accordingly, Jackson et al. demonstrated that the M8 to nelfinavir ratio did not appear to influence clearance/bioavailability of nelfinavir in 110 patients .
Diarrhoea is the most frequently reported adverse event in patients receiving nelfinavir and can be treated with the antidiarrhetic agent loperamide . Plasma concentrations of nelfinavir and M8 in patients treated with loperamide were lower than in the overall population. The ratio between the two compounds was not altered, indicating that the relative contribution of CYP3A4 and CYP2C19 to metabolism of nelfinavir is not influenced by reduced absorption of nelfinavir.
Monitoring plasma levels may be helpful to identify the cause of intoxication or therapy failure . Samples suspected to have high concentrations indeed showed increased plasma levels in a number of patients. The group of samples that were sent in on suspicion of low plasma levels contained relatively more low plasma levels than the routine group. In a small but significant number of patients, adverse events or therapy failure could not be related to the nelfinavir plasma levels, and other causes must be considered. Patients with nelfinavir concentrations that are too low or too high can be advised to increase or decrease the dose, respectively. Measurements before and after dose adjustment showed no clear dependency of the plasma nelfinavir concentration on the dose in these patients. Adjustment resulted in the desired change in plasma concentration for only some of the patients, although in controlled trials nelfinavir clearance after multiple doses was shown to increase proportionally to the dose . The effectiveness of dose adjustment may depend on the source of variation. If nelfinavir concentrations are too low because the patient does not take the medication or takes it without food, a dose increase is not likely to be effective. If low concentrations are caused by, for instance, poor intestinal absorption or comedication, a dose increase could be effective.
Concentrations of nelfinavir and its active metabolite M8, measured in outpatients, show large variability. Possible sources of this variation and in the ratio of M8 to nelfinavir were studied. Black and Asian descent and coadministration of CYP2C19 inhibitors or CYP3A4 inducers were identified as factors reducing the concentration of M8. Nevertheless, nelfinavir and total nelfinavir plus M8 concentrations were only marginally affected in these patient groups. On the basis of these results, it would not be essential to measure M8 in addition to nelfinavir for therapeutic drug monitoring of nelfinavir in outpatients.
Agouron is acknowledged for the supply of M8. The authors would like to thank the technicians of the Department of Clinical Pharmacy for analysing the plasma samples. All Dutch HIV/AIDS physicians are thanked for sending in random nelfinavir samples.
1. Jarvis B, Faulds D. Nelfinavir, a review of its therapeutic efficacy in HIV infection. Drugs 1998, 56: 147 –167.
2. Pai VB, Nahata MC. Nelfinavir mesylate: a protease inhibitor. Ann Pharmacother 1999, 33: 325 –339.
3. Sandoval TM, Grettenberger HM, Zhang KE et al. Metabolism of nelfinavir mesylate, an HIV-1 protease inhibitor by human liver microsomes and recombinant human isoforms. XII Annual Meeting and Exposition of the American Association of Pharmaceutical Scientists.
San Francisco, November 1998 [abstract 1096].
4. Lillibridge JH, Lee CA, Pithavala YK et al. The role of polymorphic CYP2C19 in the metabolism of nelfinavir mesylate. XII Annual Meeting and Exposition of the American Association of Pharmaceutical Scientists.
San Francisco, November 1998 [abstract 3035].
5. Zhang K, Wu E, Patick A et al. Plasma metabolites of nelfinavir, a potent HIV protease inhibitor, in HIV positive patients: Quantitation by LC-MS/MS and antiviral activities. 6th European ISSX meeting.
Gothenburg, Sweden, June 1997 [abstract 128].
6. Hugen PWH, Verweij-van Wissen CPWGM, Burger DM, Wuis EW, Koopmans PP, Hekster YA. Simultaneous determination of the HIV-protease inhibitors indinavir, nelfinavir, saquinavir, and ritonavir in human plasma by reversed-phase high performance liquid chromatography. J Chromatogr B, Biomed Sci Appl 1999, 727: 139 –149.
7. Jeppesen U, Rasmussen BB, Brosen K. Fluvoxamine inhibits the CYP2C19-catalyzed bioactivation of chloroguanide. Clin Pharmacol Ther 1997, 62: 279 –286.
8. Suri A, Bramer SL. Effect of omeprazole on the metabolism of cilostazol. Clin Pharmacokinet 1999, 37 (Suppl 2): 53 –59.
9. Jeppesen U, Gram LF, Vistisen K, Loft S, Poulsen HE, Brosen K. Dose-dependent inhibition of CYP1A2, CYP2C19 and CYP2D6 by citalopram, fluoxetine, fluvoxamine and paroxetine. Eur J Clin Pharmacol 1996, 51: 73 –78.
10. Petersen A, Johnson M, Nelson M, Peters B, Opravil M, Clendeninn N. Long-term comparison of BID and TID dosing of nelfinavir (NFV) in conbination with stavudine (d4T) and lamivudine (3TC) in HIV patients. XII International Conference on AIDS.
Geneva, June 1998 [abstract 12224].
11. Hugen PWH, Burger DM, Aarnoutse RE, Baede P, Nieuwkerk PT, Hekster YA. Concentration ratios of protease inhibitors (PIs) can be applied to assess non-compliance. First International Workshop on Clinical Pharmacology of HIV Therapy
. Noordwijk, the Netherlands, March 2000 [abstract 3.1].
12. Khaliq Y, Gallicano K, Sequin 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.
13. Hsyu PH, Lillibridge JH, Beeby S, Schultz MD, Heine PR, Kerr BM. Pharmacokinetics of nelfinavir and metabolite M8 in patients with liver impairment after a single oral 750 mg dose of Viracept®. First International Workshop on Clinical Pharmacology of HIV Therapy
. Noordwijk, the Netherlands, March 2000 [abstract 8.3].
14. Xie HG, Stein CM, Kim RB, Wilkinson GR, Flockhart DA, Wood AJ. Allelic, genotypic and phenotypic distributions of S-mephenytoin 4′-hydroxylase (CYP2C19) in healthy Caucasian populations of European descent throughout the world. Pharmacogenetics 1999, 9: 539 –549.
15. Xie HG, Kim RB, Stein CM, Wilkinson GR, Wood AJ. Genetic polymorphism of (S
)-mephenytoin 4′-hydroxylation in populations of African descent. Br J Clin Pharmacol 1999, 48: 402 –408.
16. Xie HG, Xu ZH, Luo X, Huang SL, Zeng FD, Zhou HH. Genetic polymorphisms of debrisoquine and S
-mephenytoin oxidation metabolism in Chinese populations: a meta-analysis. Pharmacogenetics 1996, 6: 235 –238.
17. Andersson T. Pharmacokinetics, metabolism and interactions of acid pump inhibitors. Focus on omeprazole, lansoprazole and pantoprazole.
Clin Pharmacokinet 1996, 31: 9 –28.
18. Malaty L, Kuper JJ. Drug interactions of HIV protease inhibitors. Drug Safety 1999, 20: 147 –169.
19. Zhang MH, Pithavala YK, Lee CA et al. Apparent genetic polymorphism in nelfinavir metabolism: evaluation of clinical relevance. 12th International Symposium on Microsomes and Drug Oxidation
. Montpelier, France, July 1998 [abstract 264].
20. Jackson KA, Rosenbaum SE, Kerr DM, Pithavala YK, Yuen G, Dudley MN. A population pharmacokinetic analysis of nelfinavir mesylate in human immunodeficiency virus-infected patients enrolled in a phase III clinical trial. Antimicrob Agents Chemother 2000, 44: 1832 –1837.
21. Roche Registration Ltd UK. Viracept ® (nelfinavir)
. Summary of Product Characteristics.