Pharmacokinetics of rifabutin in HIV-infected patients with or without wasting syndrome

Abstract

Aims

The purpose of the study was to compare the pharmacokinetic parameters of rifabutin obtained in a group of patients without wasting syndrome (NWS) with those obtained in a group with wasting syndrome (WS).

Methods

A single dose of 300 mg rifabutin was administered in the fasting state to the patients in both study groups and blood samples were scheduled to be collected at the following times: 0 (predose), 0.5, 1, 2, 3, 4, 6, 8, 24, 48, 72 and 96 h following administration. Data were analysed using noncompartmental methods. The pharmacokinetic parameters of rifabutin in patients with and without wasting syndrome were compared using the Mann–Whitney U-test.

Results

C_max was 0.34±0.14 mg l⁻¹ in NWS patients and 0.55±0.16 mg l⁻¹ (P = 0.01) in patients with WS. t_max was 4.2±1.5 and 3.3±2.3 h (P = 0.17) in NWS and WS patients, respectively. The AUCs were similar in the two study groups. V/F was 2905±1646 l in NWS patients and 1701±492 l (P = 0.07) for the WS group. These differences are less pronounced following normalization of V/F to patients body weight (43.7±20.1 vs 35.4±10.3 l kg⁻¹). t_1/2,λz tended to be shorter in patients with WS (31.4±12.9 vs 46.0±23.5 h, P = 0.12).

Conclusions

Our study suggests that the pharmacokinetics of rifabutin in patients with wasting syndrome are not altered to a degree that is clinically important.

Introduction

HIV infection may significantly influence drug absorption and disposition. Such an effect appears unpredictable [1–9], although the majority of drugs investigated to date show a decrease in the parameters of drug exposure (C_max, C_min, AUC) [4, 5, 7], possibly due to a decrease in absorption as a consequence of gastrointestinal dysfunction caused by HIV, opportunistic pathogens or malignancies [10–13]. In the case of drugs requiring an acidic gastric pH for dissolution, such as ketoconazole and delavirdine [2–5], a decrease in exposure may result from lack of drug solubility in the stomach of HIV positive patients with hypochlorhydria. One study showed that the bioavailability of trimethoprim in addition to other pharmacokinetic variables did not differ in critically ill vs noncritically ill AIDS patients [8]. In contrast with the findings of a previous report [4], a recent study [9] could not demonstrate an association between concentrations of antimycobacterial drugs and factors such as HIV infection, diarrhoea or CD4⁺ lymphocyte count [4]. Paradoxically, the bioavailability was increased for clindamycin in AIDS patients compared with healthy volunteers [3] and for saquinavir in patients with advanced HIV disease and signs of gastrointestinal malabsorption compared with asymptomatic HIV positive patients [6]. This may be due to increased gastrointestinal permeability, decreased first-pass metabolism in the gastrointestinal tract or, in the case of saquinavir, decreased P-glycoprotein activity [14]. Pharmacokinetic modifications in HIV positive patients based on mechanisms not directly related to gastrointestinal dysfunction, such as changes in protein binding, volume of distribution or clearance cannot be excluded a priori [1–3].

To date, a formal study of the effect of HIV infection on rifabutin pharmacokinetics has not been reported. A recent review [15] showed that parameters of rifabutin exposure were comparable between healthy volunteers and patients at early stages of HIV disease. Also, in a retrospective analysis performed in patients stratified according to their CD4⁺ lymphocyte count [16], i.e. less than 20, greater than 20 and greater than 50 cells μl⁻¹ the parameters of rifabutin exposure did not differ between groups. However, such an analysis may be compromised as the data were obtained from patients enrolled in a phase III study where patient compliance was an issue. The investigation was extended by stratifing the patients, previously enrolled in 5 phase I controlled studies, into six groups according to their CD4⁺ cell count [17]. Even in this case no significant difference was observed between groups in terms of C_max and AUC. It could be argued that both analyses were conducted with the assumption that CD4⁺ lymphocyte count is a reliable parameter of HIV disease progression. However, the validity of such an assumption may be impaired by previous or concomitant antiretroviral therapy. Furthermore, CD4⁺ count may not be a precise index of dysfunction of physiological processes capable of influencing pharmacokinetics.

Therefore, we undertook the present study in order to compare the single dose pharmacokinetic parameters of rifabutin in two groups of HIV infected patients at an advanced stage of disease using wasting syndrome as a discriminant factor. Furthermore, we studied the correlation of rifabutin pharmacokinetic parameters with 5 h urinary excretion of d-xylose, a specific marker of active absorption, as well as with a number of other demographic and physiopathological parameters.

Methods

Patients

Subjects were 20 HIV-infected patients, 10 without wasting syndrome (age, 37±7.0 years; weight, 63.8±7.7) and 10 with wasting syndrome (age 35.3±6.0 years; weight, 47.7±7.4 kg).

The study protocol was approved by the institutional review board for human investigation. Written informed consent was obtained from each patient.

Patients were included in the wasting syndrome group if they experienced a weight loss greater than 10% of usual weight during the year preceeding the study. Body mass index (BMI) was calculated as weight (kg)/height (m)². A complete medical history, a physical examination, and a panel of laboratory tests consisting of a chemistry screen and a complete blood cell count with differential and platelet count were all carried out within a week before the study. Additionally, gastrointestinal absorptive function was assessed by determining the percentage of d-xylose excreted in the urine over a 5 h period following the oral administration of a 25 g dose. The concentrations of d-xylose in urine were determined by spectrophotometry following a chromogenic reaction with phloroglucin [18]. Chronic diarrhoea was defined as more than three watery stools per day for at least 1 month before the study. It was recorded if patients had diarrhoea on the study day or had persistent diarrhoea within 2 weeks before the study day.

Medications administered concomitantly or within 2 weeks of the first study day were recorded.

Patients were excluded if they had fever (−38.5° C) or other signs of acute infection or if they had received polichemotherapy for malignancies within the previous 4 weeks. Karnofsky performance score had to be ≥60 for inclusion in the study.

Drug administration

All subjects received a single 300 mg dose of rifabutin (Mycobutin, Pharmacia & Upjohn S.p.A., Milan, Italy). Patients fasted for 8 h before to 2 h after administration of rifabutin.

Blood samples (7 ml) were drawn at 0 (predose), and at 0.5, 1, 2, 3, 4, 6, 8, 24, 48, 72, and 96 h following drug administration. Blood was collected in preheparinized Vacutainer® tubes, placed on ice and centrifuged within 30 min after collection. Plasma was harvested and stored at −70° C until analysis.

Analytical methods

Rifabutin concentrations were determined by h.p.l.c. Briefly, rifabutin was extracted from plasma samples using liquid-liquid column extraction (Extrelut, Merck, Darmstadt, Germany). Chromatography was achieved by means of a 5 μm Symmetry Shield RP8 column (Waters, Milford, MA) using a mobile phase of KH₂PO₄ (0.05 m) buffer: acetonitrile (51.3:48.7, v/v) and detection was accomplished by a L-4500 diode array detector (Merck-Hitachi, Darmstadt, Germany). Precision of the assay was <10% over the range 0.005–0.800 mg l⁻¹ except at the limit of quantification (0.005 mg l⁻¹) where precision was 13.8%.

Pharmacokinetic analysis

Pharmacokinetic variables for rifabutin were obtained by non-compartmental analysis using the program SIPHAR (Simed, Inc., Creteil, France). The area under the plasma concentration-time curve (AUC) from time 0 to time t_last was calculated by the trapezoidal rule. The AUC was extrapolated from the last observed concentration (C_last) to infinity dividing C_last by the terminal elimination rate consant (λ_z) which was obtained by log linear regression of at least the last three points of the plasma concentration vs time curve. Clearance/bioavailability (CL/F) was calculated as the AUC divided by the dose. The volume of distribution in the postdistributional phase (V/F) was calculated dividing CL/F by λ_z. The terminal elimination half-life (t_1/2,λz) was calculated as 0.693/λ_z. Maximum concentration (C_max), time to maximum concentration (t_max) and concentration at 24 h following administration were noted directly.

Statistical analysis

Data were analysed on a MacIntosh computer using the program Staview II (Abacus Concepts Inc., Berkeley, CA). All demographic and pharmacokinetic data are reported as mean±s.d. A value of P<0.05 was considered statistically significant.

The Mann–Whitney U-test was used to compare demographic and pharmacokinetic parameters between patients with or without wasting syndrome.

Additionally, the two study groups were pooled and the Pearson correlation coefficient obtained by unweighted linear regression was used to evaluate the correlation between C_max, AUC, V/F and demographic parameters. This was done for weight, height, BMI, BSA, CD4⁺ lymphocytes, albumin, cholesterol, and haemoglobin. Stepwise linear regression was used subsequently for selection of interrelated covariates for their capability to predict pharmacokinetic variables.

The following demographic and biochemical parameters were dichotomized using the upper or lower limit of normal titre, depending on parameter, and analysed as categorical variables: d-xylose <50% the lower limit of normal, i.e. 8% of administered dose, triglycerides >170 mg dl⁻¹, cholesterol <130 mg dl⁻¹, GGT (gamma-glutamyltranspeptidase) >50 IU l⁻¹, ALT (alanine amino transferase) >40 IU l⁻¹, LDH (lactic dehydroganase) >460 IU l⁻, haemoglobin <14 g dl⁻ for males and <12 g dl⁻¹ for females. Categorical variables were compared using Fisher’s exact test (GWBasic, Round Rock, Texas).

Results

Patient demographics are shown in Table 1. The following parameters were not statistically different in the patients without wasting syndrome compared with patients with wasting syndrome: age, height (1.65±0.07 vs 1.65± 0.08 m), triglycerides (151±57 vs 140±66 mg dl⁻¹), proteins (7.5±0.8 vs 7.7±1.0 g dl⁻¹), albumin (3.97± 0.37 vs 3.65±0.87 g dl⁻¹), creatinine (0.93±0.13 vs 0.85±0.31 mg dl⁻¹), haemoglobin (11.9±2.7 vs 10.5±1.7 g dl⁻¹), ALT (43±24 vs 56±49 IU l⁻¹), LDH (347±77 vs 469±170 IU l⁻¹), white blood cells (4.07±1.7 vs 3.85±2.7 cells ×10³ μl⁻¹), CD4⁺ lymphocyte count, and d-xylose urinary test below normal (3 patients out of 10 vs 6 patients out of 9). The following parameters were significantly different: weight (P = 0.0009), body mass index (P = 0.0009), body surface area (P = 0.013), cholesterol analysed both as a continuous (P = 0.02) and as a categorical variable (P = 0.01), GGT (P = 0.004), and red blood cells (P = 0.02).

Table 1.

Patient demographics.

There were three female patients in the group without wasting syndrome and one in the group with wasting syndrome. There were two patients with diarrhoea or history of diarrhoea in the immediate time before the study in each of the study groups.

The pharmacokinetic variables obtained in patients with or without wasting syndrome are reported in Table 2 and the concentration vs time curves are shown Figure 1. C_max and C_{24 h} were significantly higher in patients with wasting syndrome (P = 0.01 and P = 0.03, respectively). Only nine patients for each study group were available for comparison of the other pharmacokinetic parameters of rifabutin since one patient in each study group complied only with the first 24 h of blood sampling. The portion of the total AUC extrapolated from the last observed concentration to infinity was less than 10% for all the patients where this parameter was estimated. AUC and CL/F normalized to body weight were similar between groups. There was a trend to V/F being smaller, and t_1/2,λz shorter, in patients with wasting syndrome (P = 0.07 and P = 0.12, respectively). The mean difference in volume of distribution was reduced from a factor of 1.71 to 1.23 by normalizing volume to patient body weight. C_max, C_{24 h} and V/F in the two study groups are shown in Figure 2.

Table 2.

Mean values for pharmacokinetic parameters of rifabutin in patients with (WS) and without (NWS) wasting syndrome.

Figure 1.

Concentration vs time profiles for patients without (a) or with (b) wasting syndrome.

Figure 2.

C_max, C_{24 h} (a), and V/F (b) in patients without (NWS) or with wasting syndrome (WS) with respective mean values.

Analysis of pooled data from the two groups resulted in a significant correlation of the following parameters with rifabutin C_max: weight (r = 0.55, P = 0.01), BMI (r = 0.54, P = 0.01), BSA (r = 0.50, P = 0.02), and cholesterol (r = 0.65, P = 0.002). Stepwise regression indicated cholesterol as a better predictor of C_max while the other covariates were excluded from the model. If cholesterol was not included in the analysis, patient weight was retained in the model while BSA and BMI were excluded. Among variables analysed as categorical, GGT (P = 0.01) and cholesterol (P = 0.007) were significant predictors of C_max. AUC was significantly different in patients with low GGT (P = 0.04) and ALT (P = 0.008). V/F correlated with weight (r = 0.67, P = 0.002), BSA (r = 0.63, P = 0.005), BMI (r = 0.15, P = 0.02), cholesterol (r = 0.62, P = 0.006), and haemoglobin (r = 0.54, P = 0.02). Stepwise regression showed that only weight remained significant when all the covariates were added to the model. Categorical variables correlating with V/F were GGT (P = 0.02) and cholesterol (P = 0.005). The covariates selected by the stepwise regression model for C_max and V/F are shown in Figure 3. Also, the relationship between AUC and ALT is shown in Figure 3.

Figure 3.

Association of C_max (a), AUC (b) and V/F (c) with demographic parameters. The demographic parameters resulted significant in the stepwise regression model for C_max and V/F are shown.

Discussion

The values obtained in this study for AUC were higher than the values of 5.32 and 6.19 mg l⁻¹ h reported in symptomatic HIV infected patients and healthy volunteers [15], respectively, and lower than the values of 8.16 and 8.84 mg l⁻¹ h found in patients with hepatic dysfunction and non-HIV elderly subjects [15]. The values of V/F and t_1/2,λz in the group without wasting syndrome were similar to the ones reported in the literature for normal healthy volunteers [15]. On average, t_1/2,λz was 1.5 times higher in patients without wasting syndrome compared with patients with wasting syndrome. It should be pointed out that two patients in the NWS group had an estimated value for half-life which was only slightly less than sampling time. Therefore, the confidence of such estimates as well as V/F estimates for such patients is low. Furthermore, the power of the study to detect a 40% difference at a 5% alpha level was only 60% for V/F and 70% for t_1/2,λz while it was >90% for AUC and C_max. It is likely that the higher C_max in patients with wasting syndrome reflects the lower volume of distribution in this patient population, even though a faster absorption rate constant or a higher bioavailability may have a partial role in explaining this finding.

Wasting syndrome represents a pathological condition reflecting organ and metabolic dysfunction at multiple levels. Even though a primary cause of wasting syndrome is malabsorption due to abnormalities of structure and function of small intestinal mucosa [11–13], there is evidence that poor dietary intake [19] and hypermetabolism at concomitantly reduced substrate disposal may play an important role [20].

We used d-xylose urine test as a specific marker of active absorption, and studied the correlation of such parameter with wasting syndrome as well as rifabutin pharmacokinetic parameters. While not achieving statistical difference (P = 0.13) there was a higher number of patients with abnormal d-xylose tests in the wasting syndrome group (30%) compared with the group without wasting syndrome (67%). The association of d-xylose with body mass index was close to significance (P = 0.07). Our data are in agreement with the literaure [21–23]. Even though d-xylose absorption is, on average, lower in patients with advanced disease stage, such a parameter has a broad interpatient variability and may be altered in patients who are well while it may be within normal values in patients with advanced stage of disease, wasting syndrome and/or diarrhoea. It is also interesting to note that in some patients [21–24], the extent of malabsorption evidenced by common tests such as d-xylose is often far greater than expected on the basis of jejunal and ileal morphological alteration. Therefore, mechanisms other that intestinal mucosal inflammation and dysfunction may also play a role in altered absorption. For example, it is proposed [21] that the intestinal autonomic neuropathy which is often seen in AIDS patients may lead to a rapid intestinal transit similar to that seen in diabetic autonomic neuropathy. This mechanism may explain both d-xylose malabsorption and wasting syndrome as well as drug malabsorption in absence of major alterations of intestinal mucosa.

d-xylose did not correlate with any of the pharmacokinetic parameters of rifabutin. Our study confirms that d-xylose is a poor predictor of drug absorption in HIV infected patients. The bioavailability of zidovudine, for instance, did not correlate with d-xylose absorption [7]. Absorption of saquinavir correlated negatively with d-xylose [6], while it correlated positively with the ratio lactose/mannitol, a parameter of passive diffusion, suggesting that for drugs with low bioavailability, the increased permeability due to mucosal inflammation may actually result in an increased extent of absorption. Two patients in each study group had diarrhoea. Therefore, this variable may be excluded as a confounding factor.

In a previous study, C_max, AUC, and V/F were higher in women compared with men [25]. However, such differences disappeared when data were normalized for body fat [26]. Our data set was slightly unbalanced regarding gender but we suggest that this factor is not a confounder, since pharmacokinetic parameters in the male and female patients did not differ significantly.

Drug abuse may be a cause of poor nutritional intake. There were three patients in each study group with drug abuse as a risk factor for acquiring HIV disease. Therefore, drug abuse may be ruled out as a confounding factor in our analysis.

The other variables which resulted significantly different between patient groups were weight, BSA, BMI, GGT, cholesterol and RBC.

We thought that the analysis of such variables as well as of other patient variables by pooling the rifabutin parameters obtained in the two patient groups could provide further insights regarding possible factors capable of predicting the pharmacokinetics of rifabutin.

GGT was higher in patients with wasting syndrome. Only patient 10 in the group without wasting syndrome had a value higher than the upper limit of normal while there were eight patients in the group with wasting syndrome. GGT was significantly associated with C_max, AUC, and V/F. Due to the aspecificity of GGT as a marker of organ dysfunction, the value of GGT as a predictor of rifabutin pharmacokinetic parameters is debatable. The fact that GGT was correlated with C_max and V/F may simply reflect the higher values of GGT in patients with wasting syndrome who had a smaller volume of distribution. On the other hand, the association with AUC may suggests that patients with higher GGT had lower clearance or higher bioavailability.

ALT, a specific parameter of liver function which did not differ between study groups, was significantly associated with rifabutin AUC. This may reflect either an actual lower clearance or higher bioavailability in patients with values of ALT higher than the upper limit of normal or, more likely, it may be a spurious observation. In fact, although rifabutin is extensively metabolized by the liver, in patients with alcoholic liver disease only a slight increase of rifabutin AUC has been observed [27]. None of the patients enrolled in the study had overt cirrhosis as evidenced by liver biopsy or CT scan. ALT values were within 2.5 times the upper value of normal except in patient 7 who had a higher value than normal. This patient did not have an acute infective episode at the time of the study. He was positive for HCV infection and his ALT value had been stable for several months.

Cholesterol concentrations were, on average, within the normal range in the group without wasting syndrome but relatively lower in patients with wasting syndrome.

Cholesterol may be considered an aspecific parameter of malabsorption. In fact, hypocholesterolaemia is significantly associated with advanced stages of disease and weight loss, and is considered a prognostic parameter of poor clinical outcome [28]. Low cholesterol concentrations may also be caused by liver dysfunction. In this data set, cholesterol tended to be higher at lower levels of ALT, LDH and GGT but was not significantly correlated to these variables by linear regression. It was associated with C_max, and V/F, when analysed both as a continuous and a categorical variable, and was even a better predictor of C_max than patient weight.

Albumin may also be considered an aspecific marker of malabsorption. Recently, albumin was shown to decrease significantly in patients at an advanced stage of disease although, in agreement with our results, it was not different in patients without wasting syndrome compared with patients with wasting syndrome [29]. None of the pharmacokinetic parameters of rifabutin correlated with serum albumin concentrations.

Haemoglobin was below the lower limit of normal in both groups and had lower values in patients with wasting syndrome, although the difference was not statistically significant, while RBC were significantly lower in patients with wasting syndrome. Both haemoglobin and RBC were significantly correlated with V/F. Such a correlation is likely to reflect the lower weight in patients with lower haemoglobin and RBC. We did not include RBC in the stepwise regression for V/F because the correlation appeared mostly due to the fact that the patient with the highest RBC had also the highest value for V/F.

CD4⁺ lymphocyte count did not correlate with rifabutin pharmacokinetic parameters, according to previous studies [16, 17]. While a measure of viral load was not available at the time of the study, this parameter would have added little to our analysis. Viral load is a good predictor of the rate of disease progression but is a poor predictor of disease stage.

We chose to carry out a single dose study in order to avoid confounding factors such as patient compliance and interpatient variability in rifabutin metabolism autoinduction [15]. Many of the comparative prospective studies of the effect of HIV infection on drug pharmacokinetics have been carried out following administration of single doses [1, 3, 5, 7]. Our findings should be confirmed by conducting a multiple dose study.

In conclusion, our single-dose study suggests that HIV infected patients with wasting syndrome may have a slightly higher peak and 24-h concentration, and a trend for shorter terminal half-life and smaller volume of distribution, compared with patients without wasting syndrome. Multivariate analysis did not reveal any significant association between rifabutin and demographic or physiopathological parameters other than those correlated with wasting syndrome.