Single and multiple dose pharmacokinetics of nelfinavir and CYP2C19 activity in human immunodeficiency virus‐infected patients with chronic liver disease

Abstract

Aims

To evaluate the single‐dose and multiple‐dose pharmacokinetics of nelfinavir and its active M8 metabolite in eight HIV‐seropositive patients with liver disease, and to examine the relationship between CYP2C19 activity (genotype and plasma M8/nelfinavir metabolic ratio) and the severity of liver disease in these patients.

Methods

Nelfinavir was given as a single dose (500 or 750 mg) to patients beginning therapy and twice (500, 750 or 1000 mg) or three times (250 or 750 mg) daily during chronic therapy. Single‐dose pharmacokinetic values were used to predict multiple‐dose regimens. Peak and total plasma exposures between 2–4 µg ml⁻¹ and 45–75 µg ml⁻¹ h, respectively, and predose levels > 0.7 µg ml⁻¹ were targeted for multidose nelfinavir. Genotype was determined by analysis for CYP2C19*1, CYP2C19*2, and CYP2C19*3. Individuals were grouped according to their genotype, molar M8/nelfinavir AUC ratio (low: < 0.1, intermediate: 0.1–0.3, high > 0.3), and Child‐Pugh classification for severity of liver disease.

Results

Nelfinavir pharmacokinetics were characterized by wide interindividual variability, low clearance (181–496 ml min⁻¹ 70 kg⁻¹, n = 7), and prolonged half-life (5–20 h, n = 7). M8/nelfinavir AUC ratio increased 58% (n = 4) and α₁‐acid glycoprotein levels decreased up to 39% (n = 5) from single to multiple dosing. CYP2C19 activity was low (metabolic AUC ratio < 0.1) in four patients with moderate to severe liver disease even though they were genetically extensive CYP2C19 metabolizers (*1/*1 or *1/*2). Three patients required lower daily doses than the standard regimen of 750 mg every 8 h to achieve target concentrations and maintain virologic suppression at < 50 RNA copies ml⁻¹ (up to 20 months).

Conclusions

Acquired CYP2C19 deficiency from moderate or severe liver disease resulted in decreased M8 formation. Long‐term HIV suppression is possible using low nelfinavir doses in patients with liver disease.

Introduction

Drug disposition may be altered, leading to unpredictable and potentially harmful outcomes, if the liver is impaired by hepatitis, drug‐induced hepatotoxicity, cirrhosis, or other causes [1]. There is limited pharmacokinetic information regarding the elimination of nelfinavir in HIV‐seropositive persons with hepatic impairment following single‐dose and multiple‐dose regimens [2, 3].

Nelfinavir and its major metabolite in plasma (AG1402 or M8) have comparable in vitro activity against HIV [4]. The drug is > 98% bound to plasma proteins, mainly to albumin and α₁‐acid glycoprotein, has a large apparent volume of distribution, and has reduced absorption when it is administered without food [5]. Nelfinavir displays dose‐dependent and time‐dependent pharmacokinetics. There are indications that clearance decreases and half‐life increases when single doses are increased from 250 to 750 mg [5]. Nelfinavir may induce its own metabolism in healthy volunteers and patients, as suggested by decreases in plasma exposures during the first 6 or 9 days of chronic therapy [6, 7]. The formation of M8 is exclusively mediated by the polymorphic cytochrome P450 (CYP) 2C19 isoenyzme [8], and about 50% of nelfinavir is eliminated by this metabolic pathway in normal metabolizers [4], whereas the elimination of M8 is predominately controlled by CYP3A4 [8]. The activity of CYP2C19 may be impaired in patients with liver disease or reduced liver function, such that even those patients with genotypes associated with high levels of CYP2C19 activity can become functionally poor metabolizers [9, 10].

This paper presents data on the pharmacokinetics of nelfinavir and its active M8 metabolite in eight HIV‐infected patients with liver disease. The purpose of this observational case series was to assess single‐dose and/or multiple‐dose pharmacokinetic profiles of nelfinavir in HIV‐infected patients with chronic liver disease. Secondary objectives were 1) to develop dosing strategies for patients starting long‐term therapy of nelfinavir by targeting plasma exposures to those reported for the standard nelfinavir regimen of 750 mg every 8 h in individuals without liver disease, 2) to evaluate genetic and expressed CYP2C19 activity by determining genotype and plasma metabolic ratio of M8/nelfinavir, respectively and 3) to examine the relationship between CYP2C19 activity and severity of liver disease in these patients.

Methods

Drug standards

Nelfinavir was isolated from the tablet formulation by extraction with methyl t‐butyl ether and purified by recrystallization from a mixture of methanol and water. Molecular structure was confirmed by mass spectrometry and infrared spectroscopy, and purity was assessed as > 98% by high‐performance liquid chromatography (h.p.l.c.). The hydroxy‐t‐butylamide metabolite of nelfinavir (AG1402 or M8) was provided by Agouron Pharmaceuticals Inc. (San Diego, CA).

Patient selection

Eight Caucasian patients (six male, two female) infected with HIV (CD4 T lymphocytes 20–700 cells mm⁻³, viral RNA < 500–200 000 copies ml⁻¹) and with liver disease participated in the study. They were assigned a number from 1 to 8 for reference in this paper. Patients had at least two of the following three criteria present for diagnosis of liver disease: (1) positive for hepatitis B (surface antigen and antibody to core antigen) or C (antibody), (2) aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels elevated greater than three times the upper limit of normal within the previous 6 months, and (3) clinical diagnosis of liver disease supported by biopsy or signs and symptoms of liver disease (e.g. hepatomegaly, ascites). Individuals were graded by severity of liver disease according to the Child‐Pugh classification (Child class A, Pugh score 5–6: mild liver impairment; Child class B, Pugh score 7–9: moderate liver impairment; Child class C, Pugh score 10–15: severe liver impairment); prothrombin time was not measured and its Pugh score was assumed to be 1, the minimum score possible (1–4 s prolonged) [11]. All participants gave written informed consent.

Drug regimens

Two patients (numbers 7, 8) were already receiving nelfinavir (750 mg every 8 h) for more than 8 weeks. Of the six people beginning nelfinavir therapy, four were naive to protease inhibitors and two (numbers 2, 6) had had protease inhibitor medications discontinued at least 3 months before pharmacokinetic sampling because of intolerance to therapy. Patients 1, 2 and 3 were taking stavudine (40 mg) and lamivudine (150 mg) every 12 h for at least 12 weeks before starting nelfinavir dosing.

On pharmacokinetic study days, nelfinavir was administered at 08.00 h and within 10 min of completing a standardized breakfast (566 kcal: 42% carbohydrates, 41% fat, and 17% protein). For patients starting nelfinavir, the drug was administered as a single dose (500 or 750 mg) followed by twice or thrice daily multiple‐dose regimens that were devised from the single‐dose profiles. The multiple‐dose regimens were started within 3 weeks after the single dose, and were designed to give plasma exposures of nelfinavir similar to those in persons without hepatic impairment, as defined later in the section on dosing strategy. Pharmacokinetic measurements of nelfinavir occurred after the single dose and at 10–19 days after beginning or adjusting chronic therapy. Participants who were already receiving nelfinavir were evaluated once to assess their steady‐state nelfinavir pharmacokinetics.

During multiple dosing, all participants took nelfinavir in combination with lamivudine (150 mg every 12 h) and either stavudine (40 mg every 12 h) or zidovudine (300 mg every 12 h). Patient 5 also received saquinavir (1000 mg Fortovase every 12 h). Five patients were taking other medications concurrently with their antiretroviral agents as outlined in Table 1.

Table 1.

Patient characteristics.

Patient	Gender	Age (years)	Weight* (kg)	CYP 2C19 genotype†	Comedications
1	M	39	61.9	*1/*1	none
2	M	34	74.9	*1/*2	none
3	M	42	70.5	*1/*1	fluconazole 400 mg bid clarithromycin 500 mg bid cotrimoxazole‡ frusemide 20 mg qd ethambutol 800 mg bid spironolactone 50 mg qd
4	F	29	55.8	*1/*1	rifabutin§ azithromycin 500 mg qd cotrimoxazole‡ ethambutol 800 mg qd pyrazinamide 1000 mg qd acyclovir 200 mg bid
5	F	35	61.1	*1/*1	methadone 30 mg qd cotrimoxazole¶
6	M	35	54.6	*1/*1	none
7	M	36	83.4	*1/*1	cotrimoxazole‡
8	M	31	44.0	*1/*1	cotrimoxazole¶
					salbutamol/beclomethasone

bid = twice daily, qd = once daily.

^*Body weight averaged over the study days.

^†*1/*1 = homozygous for wild‐type alleles in exons 2, 4 and 5; *1/*2 = heterozygous for *2 mutation without*3 mutation.

^‡Trimethoprim 160 mg and sulphamethoxazole 800 mg once daily.

^§Rifabutin 150 mg once daily during nelfinavir single dose and twice weekly during nelfinavir multiple doses

^¶Trimethoprim 80 mg and sulphamethoxazole 400 mg three times per week.

Blood sampling

Serial blood samples (5 ml) were collected via indwelling catheter or venepuncture from the antecubital vein into Vacutainer tubes containing heparin, immediately before drug administration and at 0.5, 1, 2, 3, 3.5, 4, 4.5, 5, 6, 8, 12, 24, and 30 h after single‐dose administration. Sampling during multiple‐dose therapy followed the same schedule until the end of the 8 or 12 h dosing interval. Samples were equilibrated for 15 min at 4 °C before centrifugation (1500 g, 4 °C, 10 min). The isolated plasma was stored at −80 °C until analysis.

Drug analysis

Concentrations of nelfinavir and M8 in plasma were measured by h.p.l.c. Internal standard (ritonavir) in 50% aqueous methanol (0.2 ml) and water (1 ml) were added to plasma samples (1 ml). The mixture was basified with 2 m tripotassium phosphate (0.1 ml), washed with hexane (3 ml), and extracted twice with methyl t‐butyl ether (2 × 5 ml). The combined organic extracts were evaporated to dryness and the residue was dissolved in 0.02 m potassium dihydrogen phosphate in 50% methanol (0.2 ml). The solution was washed with hexane (3 ml) and analysed (0.075 ml) by h.p.l.c. The h.p.l.c. conditions were: column: Supelcosil ABZ‐plus, 3 µm particle size, 4.6 × 250 mm; column temperature: 45 °C; mobile phase composition: 0.02 m potassium dihydrogen phosphate + 0.01 m octylsulphonic acid adjusted to pH 4.0 (45%): methanol (30%): acetonitrile (25%); flow rate: 1 ml min⁻¹; u.v. detector wavelength: 240 nm. The retention time of M8, ritonavir and nelfinavir was 7.2, 12.7 and 16.7 min, respectively. The lower limit of quantification for the analytes was 30 ng ml⁻¹ with associated intrabatch precision of < 13% for 14 sets of duplicate samples per analyte. The 90% confidence limits around the mean assay biases of nelfinavir and M8 quality‐control samples pooled over 14 batches ranged from −2.5% to −10.3% for low, medium, and high concentrations (n = 31–32 samples/analyte).

Biochemical measurements

Concentrations of serum albumin, α₁‐acid glycoprotein, AST, ALT, and total bilirubin were measured by standard procedures in a single, accredited laboratory.

CYP2C19 genotyping

All patients were genotyped for the wild‐type allele (CYP2C19*1) and the two principal defective alleles associated with deficient enzyme activity (CYP2C19*2 variants in exons 2 and 5 and CYP2C19*3 in exon 4) [12]. Venous blood samples were collected in Vacutainer tubes containing acid citrate dextrose (ACD) anticoagulant and dried on untreated filter paper (Schleicher & Schuell, Keene, NH). The genomic deoxyribonucleic acid was extracted and amplified by polymerase chain reaction followed by restriction fragment length analysis according to the procedure of de Morais et al. [13]. This procedure does not separate the *2 variant alleles on exon 5 (*2 A) and exon 2 (*2B) [12]. Alleles that had no *2 or *3 mutations were classified as *1 (wild‐type) alleles. Those who were homozygous (*1/*1) or heterozygous (*1/*2 or *1/*3) for *1 were considered genetically extensive metabolizers of CYP2C19, whereas those who were homozygous for *2 (*2/*2) or heterozygous for the two mutations (*2/*3) were considered poor metabolizers of CYP2C19 [14].

Pharmacokinetic analysis

The plasma concentration (C) vs time (t) data of nelfinavir and M8 were analysed by noncompartmental methods. The highest concentration (C_max), hour 0 concentration (predose C₀), hour 8 concentration (postdose C₈), hour 12 concentration (postdose C₁₂), and the time to reach C_max (t_max) were obtained directly from the observed data. The apparent terminal disposition half‐life (t_½,z) was calculated from the final slope (– λ_z) of the log‐linear concentration‐time curve (ln C, t) by least‐squares linear regression. The slope was estimated from the data set (n ≥ 3 points) with the smallest 90% confidence interval around the slope. Area under the plasma concentration‐time curve (AUC) from time zero to 30 h (AUC(0,30h)) or over the dosing interval (τ) (AUC(τ)) was calculated by the linear trapezoidal method. After single dose, AUC(0,∞) was estimated by adding (C₃₀)λ_z⁻¹ to AUC(0,30h), where C₃₀ is the predicted plasma concentration at 30 h. Extrapolated tail segments (30 h to ∞) of total AUC were < 18%. Apparent oral plasma clearance of nelfinavir (CL/F where F is the bioavailability) was calculated by dividing the dose by AUC(0,∞) or AUC(τ). CL/F values were normalized to 70 kg body weight. Individuals were grouped according to their molar M8/nelfinavir AUC ratio (low: < 0.1, intermediate: 0.1–0.3, high > 0.3) [8].

Dosing strategy

To develop multiple‐dose regimens for patients with liver disease, we compared nelfinavir pharmacokinetic profiles to reference ranges defined by us for patients without liver disease and receiving 750 mg every 8 h. Because a therapeutic range has not been previously defined for nelfinavir, we developed target reference ranges for steady‐state nelfinavir concentrations from average C_max, predose, and AUC(0,8h) data determined in‐house and reported in the literature [4, 5, 15, 16]. For example, in 30 adult patients taking 750 mg every 8 h, the AUC, C_max, and predose values as mean (± s.d.) were 18.5 ± 7.6 µg ml⁻¹ h, 3.16 ± 1.21 µg ml⁻¹, and 1.50 ± 0.82 µg ml⁻¹, respectively [5]. We therefore defined our target reference ranges – AUC normalized over 24 h (AUC(0,24h): 45–75 µg ml⁻¹ h; C_max: 2–4 µg ml⁻¹; C₀, C₈ and C₁₂: > 0.7 µg ml⁻¹. The predose critical value of 0.7 µg ml⁻¹ was selected to be greater than the estimated minimum in vitro concentration to inhibit replication in 50% of wild‐type HIV isolates in 50% human serum (IC₅₀ = 520 ng ml⁻¹ [17]), and accounts for the intraindividual pharmacokinetic variability of nelfinavir trough levels (∼ 40%, in‐house data) and the assay bias and within‐assay variability at these concentrations (∼ 5%).

The frequency and dose of multiple‐dose regimens were devised from single‐dose data to meet the reference ranges. The selection of a twice rather than a three times daily regimen was given priority. The degree of nelfinavir accumulation from single to multiple dose was estimated from single‐dose AUC data by the ratio of AUC(0,∞)/AUC(τ)_(s.d.), where AUC(τ)_(s.d.) is the single‐dose AUC over 8 h (thrice daily dosing) or 12 h (twice daily dosing). The minimum daily dose was estimated by multiplying the single‐dose clearance value by the lower limit of the target AUC(0,24h) range. Single‐dose C_max, C₈ and C₁₂ values were multiplied by the accumulation factor for a particular dosing frequency to predict the respective values at steady‐state. These predictions assume that any potential nonlinear pharmacokinetic characteristics of nelfinavir, such as auto‐induction of metabolism, will have minimal effects on oral clearance during multiple dosing.

Results

Individual patient characteristics and comedications are tabulated in Table 1. The mean (± s.d.) age and weight were 35 ± 4 years (range, 29–42 years) and 63 ± 13 kg (range, 44–83 kg), respectively. Except for patients 4, 5 and 6 who were not receiving any antiretroviral agents during single dose, all had viral RNA measurements < 500 copies ml⁻¹ on pharmacokinetic study days. Seven individuals were homozygous for the CYP2C19 wild‐type allele in exons 2, 4 and 5 (*1/*1) and patient 2 was heterozygous for *2 mutation without *3 mutation (*1/*2). All were therefore genotypically extensive metabolizers of this isoenyzme.

Liver assessments (Table 2) indicate that AST or ALT levels were greater than or equal to three times the upper limit of normal during the study. All participants had had AST and ALT levels greater than three times the upper limit of normal (range, 4‐to 16‐fold) within the previous 6 months. Five patients were seropositive for hepatitis C, one was positive for hepatitis B, and one was positive for both hepatitis B and C. A liver biopsy revealed signs of cirrhosis in patients 1 and 2 and steatohepatitis in patient 4, and ultrasound images indicated mild to moderate hepatomegaly in patient 6. By the Child‐Pugh classification, patients 1 and 3 had moderate liver impairment (Child class B) and the others had mild liver impairment (Child class A). However, the biopsy or ultrasound results on two individuals with Child class A impairment (numbers 2, 6) indicated a more significant degree of liver disease. Patients 1 and 3 had below normal levels of both serum albumin (< 35 g l⁻¹) and α₁‐acid glycoprotein (< 0.4 g l⁻¹). Concentrations of α₁‐acid glycoprotein decreased up to 39% during the initial weeks of multiple dosing in all persons.

Table 2.

Functional,^* prognostic and diagnostic indicators of liver assessment.

	AST	ALT	Total bilirubin	Serum albumin	Child‐Pugh
Patient	(12–36 U l−1)†	(8–40 U l−1)†	(4–24 µmol l−1)†	(> 35 g l−1)†	Class‡	Available test results	History of liver disease
1	9 X	6 X	54	33	B8	Biopsy: chronic active hepatitis, cirrhosis	Hepatitis C, alcohol abuse
2	8 X	10 X	13	41	A5	Biopsy: moderate active hepatitis, severe fibrosis, nodules, cirrhosis	Hepatitis C
3	3 X	2 X	44	32	B8	None	Hepatitis B, cirrhosis, ascites Chronic liver disease
4	2 X	3 X	9	44	A6	Biopsy: steatohepatitis U/S: nonspecific and fatty changes, ascites	Hepatitis at age 10 Left upper quadrant pain
5	3 X	2 X	11	40	A5	U/S: no ascites or cirrhosis	Hepatitis C
6	8 X	4 X	34	37	A6	U/S: mild to moderate hepatomegaly	Hepatitis B/C
7	3 X	4 X	18	47	A5	None	Hepatitis C, alcohol abuse
8	3 X	3 X	12	42	A5	None	Hepatitis C, hepatomegaly

AST = aspartate aminotransferase, ALT = alanine aminotransferase, U/S = ultrasound.

^*Blood samples for liver function tests were collected within 1 week before the first pharmacokinetic visit. Measurements are expressed as the magnitude above the upper limit of normal.

^†Normal range of values is in brackets.

^‡Child class A, Pugh score 5–6: mild liver impairment; Child class B, Pugh score 7–9: moderate liver impairment; Child class C, Pugh score 10–15: severe liver impairment. The Pugh score for prothrombin time was assumed to be 1.

Individual single‐dose and multiple‐dose pharmacokinetic parameters of nelfinavir are presented in Table 3. A wide interindividual variability in nelfinavir pharmacokinetics was observed. Generally, the pharmacokinetics was characterized by decreased clearance (< 500 ml min⁻¹) and longer half‐life (> 5 h) values compared to those expected for patients taking 750 mg every 8 h and without liver disease. In seven individuals, clearance values normalized to 70 kg body weight ranged from 181 to 496 ml min⁻¹ at steady‐state and half‐lives ranged from 5 to 20 h; patient 3 had clearance values > 600 ml min⁻¹ and patient 5 had a half‐life of 3.3 h. The four patients with significant liver impairment by Child‐Pugh classification (numbers 1, 3), biopsy (numbers 1, 2) or other clinical tests (number 6) all had low levels of M8 (< 250 ng ml⁻¹), low metabolic AUC ratios (< 6%), and, except for patient 3, low clearance values. Within patients, the metabolic M8/nelfinavir AUC ratio increased from single to multiple dosing in all four participants for which data are available (mean increase 58%, s.d. 31%; median increase 60%, range 19% to 88%).

Table 3.

Observed and predicted nelfinavir pharmacokinetic parameters^*.

Patient	Dose (mg)	AUC(0,∞)(s.d.), 24 (MD) observed (predicted) (µg ml−1 h)	Cmax observed (predicted) (µg ml−1)	C0, Cτ observed (predicted) (µg ml−1)	CL/F (ml min−1)	CL/F (ml min−1 70‐kg−1)	t½,z (h)	Molar AUCM8/AUCNFV (%)
1	750 s.d.	45.1	2.7	–	277	318	11.3	3.2
	250 bid	26.9 (30.1)	1.7 (2.1)	0.9, 0.9 (1.1)	309	348	10.1	4
	500 bid	58.6 (60.5)	2.9 (4.1)	1.8, 2.4 (2.2)	284	318	19.7	5.8
2	500 s.d.	42.1	3.6	–	198	183	5.1	2.5
	500 bid	87.2 (84.2)	5.3 (5.9)	3.3, 2.2 (2.3)	191	181	6.3	2.6
	250 tid	46.5 (63.2)	2.5 (4.1)	1.4, 1.3 (2.9)	269	251	3.8	3.1
3	500 s.d.	8.6	0.8	–	971	971	8.4	ND
	750 tid	61.5 (38.6)	3.5 (2.7)	3.2, 2.6 (1.5)	611	602	9.8	1.3
4	500 s.d.	12.7	1	–	655	822	5.6	14.1
	1000 bid	101 (48.7)	6.8 (3.8)	2.4, 3.5 (2.5)	330	414	4.8	19.6
5	500 s.d.	19.5	1.5	–	427	496	3.3	8
	750 bid	63.5 (57.5)	3.1 (3.3)	2.8, 2.5 (2.1)	394	445	ND	15
6	500 s.d.	27.8	1.7	–	300	384	8.3	3.6
7	750 tid	64.2	3.3	2.5, 1.9	584	490	5.1	49.2
8	750 tid	146	7.4	3.5, 6.1	257	409	ND	29.6
Reference †	750 tid	45–75	2–4	> 0.7	–	500–850	3.5–5	33 (EM) < 10 (PM)

NFV = nelfinavir; M8 = hydroxy‐t‐butylamide metabolite of nelfinavir; s.d. = single dose; bid = twice daily; tid = three times daily; ND = not determined; EM = patients in whom nelfinavir is extensively metabolized by cytochrome P450 2C19 enyzmes; PM = patients in whom nelfinavir is poorly metabolized by cytochrome P450 2C19 enyzmes; AUC = area under the concentration plotted against time curve from time zero to infinity (AUC(0,∞)) after single dose and over the 8 h or 12 h dosing interval normalized to 24 h (AUC(0,24 h)) after multiple doses; C_max = highest observed plasma drug concentration; C₀ = observed predose plasma drug concentration at the beginning of the dosing interval; C_τ = observed predose plasma drug concentration at the end of the 8 h or 12 h dosing interval; CL/F = apparent oral plasma clearance; t_½,z = apparent terminal disposition half‐life.

^*Predicted values of AUC, C_max, and C_τ are shown in round brackets. Multiple‐dose parameters were determined 10–19 days after beginning chronic therapy. Patient 6 did not start chronic therapy.

^†Normal values for patients receiving nelfinavir 750 mg every 8 h [in house data 4, 5, 15, 16].

To achieve target concentrations during chronic therapy, dosage adjustments from the standard regimen (750 mg every 8 h) were made for patients 1, 2, 4 and 5 on the basis of predictions from single‐dose data (Table 3). Patient 6 did not start chronic therapy. The daily doses necessary to attain target levels ranged from 750 to 2000 mg and were less than that required for the standard regimen. Two follow‐up dose modifications were necessary for patients 1 and 2. The former mistakenly took 250 mg instead of 500 mg every 12 h, whereas the latter was the first patient in the study and his predicted plasma exposures were targeted near the upper reference range to account for the expected decrease in concentrations caused by auto‐induction of nelfinavir metabolism.

The other three patients were maintained on nelfinavir 750 mg thrice daily. Target concentrations in patient 3 were achieved with the standard regimen despite moderate liver disease and fluconazole coadministration. Patients 7 and 8 appeared to have less severe liver impairment as judged clinically and by liver function tests. Although patient 8 had AUC and C_max values above the target range, his choice was to continue the regimen as he was not experiencing intolerable side‐effects.

For patients with dosage adjustments from the standard regimen, patient 1 has continued on 500 mg twice daily for 20 months with viral RNA < 50 copies ml⁻¹. Patient 2 maintained viral RNA < 50 copies ml⁻¹ for 3 months with a low dose of nelfinavir (250 mg every 8 h), but a liver biopsy at that time revealed chronic moderate hepatitis (not related to nelfinavir) and all medications were stopped. Patient 4 was naive to protease inhibitors and had a baseline viral load of 200 000 RNA copies ml⁻¹. Despite high plasma exposures of nelfinavir (C_max: 6.8 µg ml⁻¹, AUC: 101 µg ml⁻¹ h) she was still not suppressed after 5 months of combination therapy with twice daily nelfinavir (1000 mg), lamivudine and zidovudine, and all medications were discontinued. The viral load for patient 5 was 4468 copies ml⁻¹ during single dose, decreased to 106 copies ml⁻¹ after 2 weeks of twice daily nelfinavir (750 mg), saquinavir, lamivudine and stavudine, and was undetectable at < 50 RNA copies ml⁻¹ at 10 weeks. Her viral load has remained undetectable for 8 months. The other three patients taking 750 mg every 8 h have continued therapy and maintained viral suppression (< 50 RNA copies ml⁻¹) without adverse effect over a follow‐up period of 15–22 months.

Discussion

In this study we dosed nelfinavir in patients with various types and degrees of liver disease. Pharmacokinetic data were compared with historical data from patients without liver disease and taking the standard 750 mg thrice daily regimen, and were used with the clinical status of the individuals to guide multidose therapy. In three patients, low nelfinavir clearance allowed the use of reduced doses to maintain viral load at < 50 RNA copies ml⁻¹ (up to 20 months duration).

Intra‐ and interindividual variation in protease inhibitor drug exposure can influence the safety and effectiveness of anti‐HIV therapy. Within patients, estimates of multiple‐dose exposures from single‐dose exposures were within 25% of predicted values in three of the five patients administered both single and multiple doses of nelfinavir, because clearance and half‐life values were also within 25% of single‐dose values. Even if the observed levels were > 25% of predicted values, target concentrations were achieved with the selected dosage regimen because of the wide range of target values. Several factors could have affected the multiple‐dose predictions from single‐dose data. These include the amount of food ingested, levels of α₁‐acid glycoprotein, dose adjustments of interacting comedications, variability in HIV and liver disease, and nonlinearity in nelfinavir pharmacokinetics.

Dose‐ and time‐dependent pharmacokinetics of nelfinavir are present in individuals with normal liver function. We observed lower concentrations of α₁‐acid glycoprotein and an increased M8/nelfinavir AUC ratio at steady‐state in the five patients administered both single and multiple doses. However, these nonlinear characteristics appear to have minimally influenced the predicted steady‐state estimates from single‐dose pharmacokinetics in our patients with liver disease. Although the increased M8/nelfinavir AUC ratio is suggestive of CYP2C19 induction of M8 formation, we did not observe increased nelfinavir clearance during chronic therapy, as would be expected for auto‐induction or decreased α₁‐acid glycoprotein levels, except during the second dose adjustment for patient 2. Decreased concentrations of α₁‐acid glycoprotein at steady‐state rather than auto‐induction have explained the lower than expected multiple‐dose concentrations from single‐dose data of amprenavir [18]. CYP2C19 is a major pathway of nelfinavir elimination in patients with normal liver function, but the low metabolic ratios (< 20%) observed in the five patients suggest that the fraction of nelfinavir metabolized to M8 is less in patients with liver disease. Therefore, the metabolic ratio will be more sensitive to CYP2C19 induction than AUC or CL/F parameters of parent drug. Thus, liver disease may be partly responsible for the apparent lack of changes in nelfinavir oral clearance, which allowed estimations of multiple‐dose pharmacokinetics from single‐dose data in our patients.

The increase in clearance during the second dose adjustment in patient 2 is unlikely to be related to the 39% decrease in α₁‐acid glycoprotein levels because levels of this protein and albumin remained within normal limits. Similar to saquinavir [19], decreased binding to α₁‐acid glycoprotein is expected to be compensated by increased binding to albumin such that the fraction of unbound nelfinavir remains unchanged. Even if total binding decreased, no change, instead of the observed decrease, in the half‐life of nelfinavir is expected for such a drug with a large volume of distribution [5, 20]. Patient 3 consumed more breakfast on the multiple‐dose study day, which probably contributed to decreasing nelfinavir single‐dose clearance from 971 ml min⁻¹ to 611 ml min⁻¹ during chronic therapy. Nelfinavir bioavailability increases at least twofold in the fed compared to fasted state [5].

Concomitant medications likely to influence pharmaco-kinetic determinations were fluconazole and spironolactone in patient 3 and rifabutin in patient 4. Elimination of nelfinavir is predominantly controlled by CYP3A4 and CYP2C19 isoenzymes whereas formation and elimination of M8 are exclusively mediated by CYP2C19 and CYP3A4 pathways, respectively [8]. Fluconazole is a moderate metabolic inhibitor of these enzymes, rifabutin is a moderate metabolic inducer of CYP3A4, and spironolactone can both up‐regulate CYP3A activity and inactivate CYP expression.

In patient 4, rifabutin exposure was lowered by extending the dosing interval from once daily to twice weekly after multiple dosing of nelfinavir was started. The decrease in nelfinavir exposure from rifabutin inducing nelfinavir metabolism is expected to be less pronounced in a twice‐daily nelfinavir regimen than during single‐dose administration of nelfinavir [21] and when rifabutin exposures are reduced. This partly explains the decrease in nelfinavir clearance from 655 ml min⁻¹ to 330 ml min⁻¹ after adjustment of rifabutin therapy. This patient's antiviral regimen was maintained with levels above the target range because viral load was not decreasing.

Our patients with liver impairment showed wide interindividual variability in nelfinavir pharmacokinetics. Physiologic and environmental factors that appeared to influence the interindividual variability were intrinsic hepatocellular activity, protein binding, and concurrent medications. The magnitude of total (protein bound plus unbound) oral clearance for drugs that are extensively metabolized in the liver is proportional to the fraction of drug unbound to protein and the intrinsic hepatocellular activity. Both of these physiologic variables can be influenced by liver disease. Although the presence of liver disease may explain the reduced oral clearance and prolonged half‐life in our cohort, we did not observe these characteristics in all patients with significant liver disease by Child‐Pugh classification or clinical tests. The below normal serum albumin and α₁‐acid glycoprotein levels in patient 3 possibly led to decreased plasma protein binding and associated higher oral clearance values compared to those of the others. Additionally, concurrent spironolactone in patient 3 and rifabutin in patient 4 likely caused higher clearances than expected and influenced metabolic AUC ratios.

Despite classification as genetically extensive metabolizers of CYP2C19, all four patients with significant liver disease showed low metabolic AUC ratios, suggesting that they were functionally poor metabolizers of CYP2C19 with a reduced capacity to form M8. The other patients with less severe liver disease had intermediate or high metabolic ratios. The low ratio may be caused by heritable CYP2C19 deficiency, disease‐related hepatic dysfunction, or inhibition of CYP2C19 from coadministered drugs. Lillibridge et al. [8] demonstrated that low M8/nelfinavir AUC ratios in 18 of 19 HIV‐infected patients were explained by poor metabolizer genotype (*2/*2) (n = 6) or administration of putative CYP2C19 inhibitors in genetically extensive metabolizers (*1/*1 or *1/*2) (n = 12). The discordance between genotype and metabolic ratio in our four patients implicates liver impairment, although CYP2C19 activity could have been lowered in one patient by fluconazole and in another by the presence of only one *1 allele. Moreover, because CYP2C19*2 and CYP2C19*3 account for about 87% of poor metabolizer alleles in Caucasians, we cannot exclude the possibility that these patients have other defective alleles (*4, *5, *6, *7 and *8) known to reduce catalytic activity toward the classical CYP2C19 substrates S‐mephenytoin and tolbutamide [12, 22–24]. Also, there may be additional as yet unidentified mutations of CYP2C19 that could diminish enzyme activity.

Our results are consistent with other studies that have identified acquired CYP2C19 deficiency in subjects with liver disease. Adedoyin et al. [9] reported that CYP2C19 activity, as measured by S‐mephenytoin hydroxylation, was more affected in patients with moderate than mild liver disease. Rost et al. [10] used the omeprazole metabolic ratio as a measure of CYP2C19 activity to show that patients with liver disease displayed a metabolic state similar to that of genetically impaired poor metabolizers of this enzyme.

The contribution of M8 to antiviral response in vivo is unclear. Although M8 and nelfinavir have similar antiviral activity in vitro, viral load remained suppressed at < 50 RNA copies ml⁻¹ in the three patients with the lowest metabolic ratio and M8 levels. Viral suppression was also observed by Zhang et al. [4] in patients deficient in M8 but high in nelfinavir plasma exposures. Thus, reduced M8 formation may be of minor clinical importance if nelfinavir levels remain within normal ranges. In summary, nelfinavir pharmacokinetics were characterized by wide interindividual variability, low clearance, and prolonged half-life. Acquired CYP2C19 deficiency in M8 formation was evident, particularly in patients with moderate to severe liver disease. Administration of a single dose was helpful to guide dosing before starting chronic therapy. Impairment of clearance was compensated by dose reduction or extension of the dosing interval. We demonstrated that some patients with liver disease can maintain long‐term virologic suppression with a lower daily dose of nelfinavir.