INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Ethambutol (EMB) is the most frequently used "fourth drug" for the empiric treatment of tuberculosis (3). The standard short-course treatment of tuberculosis consists of isoniazid (INH), rifampin (RIF), and pyrazinamide (PZA), plus either EMB or streptomycin until susceptibility data are available (3). Also, EMB is frequently used for the treatment of infections caused by Mycobacterium avium complex (2). Limited information exists regarding the pharmacokinetics of EMB in healthy or infected individuals or regarding the effect of food or antacids on the gastrointestinal absorption of the drug (1, 12-15, 17-19, 21, 22). We examined the pharmacokinetics of EMB in healthy volunteers under fasting conditions (two replicates), with food, and with an aluminum-magnesium hydroxide antacid. This study describes the concentrations in serum and the pharmacokinetic behavior under optimal conditions, and the results can be used as benchmarks for comparison with those for samples obtained in other clinical settings.

(Part of this study was presented at the 37th Interscience Conference on Antimicrobial Agents and Chemotherapy, Toronto, Ontario, Canada, 28 September to 1 October 1997 [20].)

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

We conducted a four-period, randomized crossover study of EMB. The study protocol followed the guidelines of the Helsinki Declaration of 1975 and its amendments and was approved by the institutional review board at Millard Fillmore Hospital, Buffalo, N.Y. Written informed consent was obtained from each subject before the study. Sixteen healthy female and male volunteers were scheduled to participate. Subjects were eligible to participate if they were >18 years of age and were determined to be in good health as assessed by history, physical examination, and laboratory studies including serum chemistries, complete blood count with differential, and 12-lead electrocardiogram. Individuals were excluded if they had histories of a major disease of the kidneys (estimated creatinine clearance [CL_CR], 50 ml/min), the liver (transaminases, alkaline phosphatase, or bilirubin 2 times normal), or the cardiovascular system (New York class I to IV heart failure) or a hematocrit 36% at screening. They were also excluded if they had known gastrointestinal diseases that might affect the absorption of the drugs; known positive human immunodeficiency virus serology; AIDS; or histories of adverse reactions to INH, RIF, PZA, EMB, or related drugs. They were also excluded if they weighed >130% of ideal body weight, were pregnant or nursing, or donated blood within 30 days prior to the study (4). The subjects agreed to refrain from the use of prescription or nonprescription drugs (including vitamins) and alcohol during the entire study period. Women who were taking oral contraceptives at the start of the study were allowed to continue these during the study. They were required to agree to use additional contraceptive methods during the study period and for a week after the last dose of RIF. At the conclusion of the study, each subject underwent a brief physical examination and had blood drawn for serum chemistry and hematology, and female subjects had a repeat pregnancy test.

Experimental design. Sixteen subjects were randomized in four blocks of four subjects. The four treatments were fasting conditions (twice, to determine intrasubject variability), a high-fat meal, and aluminum-magnesium antacid. The subjects were housed at the study center from 10 h before to 24 h after dosing and returned for the 36- and 48-h collections. After eating a light snack prior to 2300, they fasted overnight. For three of the treatments, they continued to fast for 4 h after the dose. On one of these three fasting occasions, they also took 30 ml of aluminum-magnesium hydroxide (Mylanta) 9 h before dosing, at the time of the dose, after meals, and at bedtime postdose. For the fourth treatment, they consumed the standard Food and Drug Administration high-fat breakfast beginning 0.25 h before dosing. This meal consisted of 8 oz of whole milk, two scrambled eggs, two strips of bacon, two slices of toast with two butter pads, and one hash brown potato patty. The meal provided an estimated 53 g of carbohydrate, 33 g of protein, and 51 g of fat, for 792 kcal, 57% as fat. For all four treatments, subjects received single oral doses of 25 mg of EMB (median, 1,950 mg; dosed to the nearest 100 mg, with scored 500-mg tablets [Wyeth-Lederle, Philadelphia, Pa.) per kg of body weight with 240 ml of tap water. They also received 300 mg of INH, 600 mg of RIF, and 30 mg of PZA (median, 2,386 mg) per kg. Doses for all treatment periods were based on the subjects' prestudy weights. The subjects were allowed to ingest water ad libitum after the doses were given, and identical, nutritionally balanced meals were provided to all subjects during the remainder of the study period. There was a 14-day washout between each study period.

Sample collection. A 20-gauge angiocatheter was inserted into a forearm vein for the collection of blood samples and was maintained patent with a dilute heparin solution (10 to 15 U/ml). Two milliliters of blood was withdrawn and discarded prior to collection of each blood sample (12 ml) into plain red-top vacuum tubes. Serial blood samples for serum drug concentration analyses were collected at 0, 0.25, 0.5, 0.75, 1, 1.5, 2, 2.5, 3, 4, 6, 8, 10, 12, 16, 24, 36, and 48 h after the doses. Samples were allowed to clot for 30 min and then centrifuged at 2,500 to 3,000 × g for 10 min. Serum samples were then harvested and frozen at 70°C until assay.

Sample analysis. All assays were performed with a validated assay on a Hewlett-Packard (Wilmington, Del.) model 5890 Series II gas chromatograph with a Hewlett-Packard model 5971A mass selective detector. The serum standard curves for EMB ranged from 0.05 to 10 µg/ml. The absolute recovery of EMB from serum was 95.8%. The within-day precision (percent coefficient of variation [% CV]) of validation quality control (QC) samples was 2.2 to 4.1%, and the overall validation precision was 2.8 to 3.3%. The urine standard curves for EMB ranged from 0.05 to 50 µg/ml, with similar reproducibility. EMB urine samples were diluted 1:10 prior to extraction. The assay error pattern was determined from QC samples assayed over the course of the study. A line was fitted to the plot of the QC standard deviations (y) versus their means (x) at the three QC ranges (low, medium, and high). The assay error pattern used for the subsequent nonlinear regression was variance (standard deviation squared [SD²]) = (0.024 + 0.024x)² (11).

Pharmacokinetic analysis. Concentrations in serum below the quantification lower limit were treated as zeros in averaging the concentrations at a given collection time. The observed maximal serum concentration (C_max) and the time at which it occurred (T_max) were determined for each subject by inspection of the serum concentration-versus-time graphs. The area under the serum concentration-versus-time curve (AUC) from time zero to the time of the last quantifiable concentration (AUC_0-t^*) was determined by the linear trapezoidal rule. The last quantifiable concentration was designated C*. The AUC from time zero to infinity (AUC_0-) was determined as AUC_0-t^* + C*/, with  determined by ADAPT II (see below). The potential for accumulation of these drugs with multiple doses was evaluated by the principle of superposition (8). The accumulation of EMB with eight daily doses was simulated with the median serum concentration data from 0 to 24 h (first fasting treatment) and extrapolated from 24 h to day 8 with the median .

Compartmental analysis. ADAPT II software was used to construct candidate pharmacokinetic models, with nonlinear least-squares regression, weighted by the inverse of the assay variance, as described above (6). The Akaike information criteria were used to discriminate among candidate models, and a two-compartment model with apparent zero-order absorption was selected. The model included the zero-order absorption time (T_abs [milligrams per hour]), the absorption lag time (T_lag [hours]), the volume in the central compartment (V₁ [liters per kilogram]), the intercompartmental transfer rate constant (K₂₁ [1/h]), and the  and  elimination rate constants (1/h). The rate constant K₁₀ was calculated as [( × )/K₂₁], K₁₂ was calculated as [( + )  K₂₁  K₁₀], and total body clearance (CL [liters per hour]) was calculated as (V₁ × K₁₀). The steady-state volume of distribution (V_SS [liters per kilogram]) was calculated as [V₁ × (1 + K₂₁/K₁₂)]. The terminal elimination half-life (t_1/2) was calculated as ln(2)/.

Statistical analysis. Data analysis was performed with JMP version 3.1.6 (SAS Institute, Cary, N.C.), with supplemental analyses done with Excel version 4.0 (Microsoft, Seattle, Wash.). Frequency distributions (JMP) included plots of the data, distribution curves to test for normality, parametric and nonparametric measures of central tendency and dispersion, and the Shapiro-Wilk W test for normality. Means are reported ± the SD. The percent CV was calculated as SD/mean multiplied by 100%. Differences among the treatment groups were determined with an analysis of variance model that tested differences based on period, treatment, sequence, and subject (sequence). Pairwise differences across the four treatments were evaluated with individual linear contrasts. Bioequivalence criteria were tested according to the 1992 Food and Drug Administration guidelines (7). C_max and AUC_0- were log transformed and were analyzed with the analysis of variance model described above. Mean estimates and standard errors were obtained from the linear contrasts, and these were used to calculate the geometric means and the lower and upper 90% confidence limits. Comparison treatments were considered bioequivalent to the reference treatment (fasting treatment 2) if the comparison parameter 90% lower limit was 80% and the upper limit was 125%.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Fourteen subjects completed all four treatments: six white females, three black males, and five white males. The remaining subjects dropped out for personal reasons. The mean age was 39.1 ± 7.4 years, and the mean weight was 79.3 ± 13.2 kg. The subjects received a mean EMB dose of 1,936 ± 343 mg (24.9 ± 0.4 mg/kg). All subjects denied the use of any nonprotocol medications during the study period. CL_CR estimates were a mean of 103 ± 25 ml/min.

The absorption characteristics for EMB with the four treatments are described in Table 1, and the corresponding mean EMB serum concentration-versus-time profiles across the 14 subjects are shown in Fig. 1. Under fasting conditions, variability in absorption of EMB was small (Table 1) and the individual results were quite reproducible (Fig. 2). The mean EMB C_max was significantly reduced by antacids (29%) and, to a lesser extent, by food (17%) (P = 0.0003). The mean EMB T_max was increased by antacids (+17%) and, to a greater extent, by food (+29%) (P = 0.0787). The mean EMB AUC_0- was modestly decreased by antacids (10%) and minimally by food (4%) (P = 0.1625). With the bioequivalence criteria, food did not significantly affect the C_max (90% confidence interval [CI], 88.3 to 100.0%) or the AUC_0- (90% CI, 96.4 to 103.6%). Antacids did not significantly affect the C_max (90% CI, 83.6 to 94.6) or AUC_0- (90% CI, 93.2 to 100.2).

View larger version (13K): [in this window] [in a new window]   FIG. 1.   Mean EMB serum concentrations for 14 subjects across the four treatments.

View larger version (12K): [in this window] [in a new window]   FIG. 2.   Pairings of individual 2.5-h serum concentrations for 14 subjects between the two fasting treatments.

Simulated multiple daily doses of EMB with the median  value showed that, on day 4, after 6 to 7 EMB half-lives, the 2.5-h EMB concentration (C_max) was 7.7% higher than that on day 1 but only 1.8% higher than that on day 2 and 0.4% higher than that on day 3. Subsequent simulated C_max values remained constant. Given the day 1 C_max range of 2.0 to 6.0 µg/ml (first fasting treatment [Table 1]), the calculated steady-state C_max range for EMB (~25-mg/kg dose) was 2.1 to 6.4 µg/ml.

Table 2 shows the parameter estimates for EMB following the 25-mg/kg dose as calculated with ADAPT II (first fasting treatment). The various parameter estimates were not significantly different across the four treatments. T_abs was somewhat longer in the feeding treatment, but this did not reach statistical significance (P = 0.0856).

The D-optimal sampling times for all subjects over the period 0.5 to 24.0 h were 0.6 and 11 h for the two-sample strategy and 0.5, 6.2, and 6.2 h (duplicate) for the three-sample strategy if the entire interval was available for sampling. D-optimal sampling times were not affected by changes in the initial sampling times chosen but did change based on the constraints placed on the sampling times. Restriction of the D-optimal sampling period from T_max to 24 h resulted in the selection of T_max as the first sampling time. Table 3 shows that the 2.5-h sample came closest to C_max for the greatest number of the 14 subjects, followed by the 2- and 3-h samples.

The recovery of EMB in the urine is shown in Table 4. Most of the urinary excretion of EMB occurred during the first 12 h postdose, with about 30% of the dose recovered unchanged in the urine over 24 h. Intersubject variability was small. The total recovery of EMB in the urine, the percentage of the dose recovered in the urine, and the CL_R were not different across the four treatments. Subjects receiving antacid treatment had the lowest recovery in the period 0 to 12 h (415 mg versus first fasting result of 498 mg) but the highest recovery in the period 12 to 24 h (123 mg versus first fasting result of 102 mg, P < 0.03 for each comparison), resulting in a total recovery comparable to those of the other treatments.

The EMB results were analyzed with JMP, and the nonparametric measures of association are reported (Spearman rho). C_max and T_max showed a modest negative correlation (r = 0.5192, P = 0.0571), with early absorbers showing the higher C_max values. AUC_0- was somewhat higher in those with higher serum creatinine levels (r = 0.7603, P = 0.0016), but it did not correlate with CL_CR (r = 0.1736, P = 0.5528). The EMB CL correlated with CL_CR (r = 0.5341, P = 0.0492) and with EMB CL_R (r = 0.5385, P = 0.0470). EMB pharmacokinetic parameters were not dependent upon age, gender, or race in this group of 14 subjects.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

Determinations of the absolute bioavailability of EMB from the tablets versus an intravenous dosage form were not performed. All parameters were estimated assuming F = 1.

EMB was not rapidly absorbed, with most T_max values near 2.5 h. Similar results were described by Lee et al., who studied six healthy volunteers (two females and four males) and determined T_max values of 2.8 ± 0.7 h (12). They also determined C_max values of 4.0 ± 0.8 µg/ml following smaller doses of 15 mg/kg, given as tablets. Under fasting conditions, variability across our 14 subjects was small, especially for the C_max and AUC_0- values, and was highly reproducible between the two fasting treatments. Concentrations in serum increased by 7.7% over 4 days of simulated dosing. However, sampling as early as the second day of treatment will reflect steady-state serum concentrations. EMB taken with food or antacids was bioequivalent to EMB taken in the fasting state.

Antacids reduced the EMB C_max by 29% and reduced the EMB AUC_0- by 10%. Therefore, antacids should be avoided near the time of EMB dosing. These findings are generally consistent with the work of Mattila et al., who measured EMB serum concentrations 2, 4, and 10 h postdose (15). In our study, food reduced the EMB C_max by 17% but reduced the EMB AUC_0- by only 4%. Ameer et al. previously showed a similar lack of effect by food on the EMB AUC, although the standardized meal given to their subjects was not described (1). Their paper does not describe the effect on C_max or T_max. Place and Thomas found slightly higher serum concentrations in the nonfasted state, again, without a detailed description of the study conditions (21). Therefore, it may be preferable to give EMB on an empty stomach whenever possible. However, when this is not possible, the absorption of EMB should still be adequate. This may facilitate the dosing of EMB with other drugs such as nelfinavir, ritonavir, saquinavir, or rifapentine that are preferentially given with food, assuming that there are no direct interactions with these drugs (16, 23).

While a two-compartment model with first-order absorption produced good results, the two-compartment model with apparent zero-order absorption was superior based on the Akaike information criteria. The small T_lag corresponds to the first sampling time. This is consistent with the assay's low limit of quantification, which produced measurable concentrations for nearly all subjects and treatments from 0.25 to 48 h. EMB displayed a large V₁ and V_SS, which in part reflects its binding to erythrocytes and its uptake by macrophages (12, 14, 17). Both Peets and Lee found erythrocyte/plasma concentration ratios of 1.1 to 1.8 in vitro, with higher ratios described by Peets for human subjects (12, 17). Entry into macrophages is particularly useful, since a portion of the total body burden of Mycobacterium tuberculosis and M. avium complex is found within macrophages. Previous estimates of the EMB V₁ and V_SS have been smaller (12, 13). Our CL estimates were quite similar across the 14 subjects, whether or not normalized to body weight. These also were larger than previously described (12, 13).

The EMB serum concentrations clearly displayed a biexponential decline, with a median t_1/2 of about 1.3 h and a t_1/2 of about 12.4 h. Visual inspection of the serum concentration-versus-time data shows an apparent decrease in the slope occurring about 12 h postdose. Similar findings were described previously by Lee et al. (12, 13). Following oral doses, they described a decline of concentrations in serum over the first 12 h postdose, with a least-squares regression t_1/2 of 4.0 ± 0.5 h. They showed a second phase from 12 to 24 h with a least-squares regression t_1/2 of 8.8 ± 2.2 h. Differences between our curve fitting techniques and theirs account for the discrepancies in the apparent t_1/2s. Reanalysis of our data by their techniques produced a t_1/2 over the first 12 h of 3.3 ± 0.8 h and a secondary t_1/2 over 12 to 48 h of 13.9 ± 2.1 h. EMB's long terminal t_1/2 renders it suitable for once-daily dosing, particularly since it is used against the slow-growing M. tuberculosis and M. avium, which have doubling times of 24 h.

The clearance of EMB has been described as occurring predominantly through renal mechanisms. We recovered only 30% of the single oral doses unchanged in the urine over 24 h. The serum AUC_0-24 represented a median 83% of the AUC_0-, so longer collection periods might have resulted in roughly 36% recovery, assuming that CL_R is constant over the range of concentrations in serum. The completeness of oral absorption, and other sources of elimination, including metabolites, were not determined in this study. In two studies, Lee et al. documented 24-h urinary recoveries of 53% after oral EMB doses and 73% after intravenous EMB doses (12). The reasons for lower recovery in our study are not apparent but may include incomplete absorption from the oral doses here versus the intravenous doses used by Lee. Also, there may be some differences in specificity between the two methods of detection used. Renal dysfunction has been shown previously to have a significant impact on the CL of EMB, and frequency of dosing should be reduced in patients with renal dysfunction (18, 24).

If the sampling times were not restricted to begin at T_max, the EMB D-optimal sampling times included a point during the absorptive phase and one point in the elimination phase, but not the true C_max. The three-point strategy produced identical sampling times for the second and third parameters. Samples collected at 2.5 h postdose captured most of the C_max values and would be preferred for that parameter.

EMB has modest activity against both M. tuberculosis and M. avium (9, 10). Using radiometric techniques, Heifets determined the MIC of EMB to be 1 to 4 µg/ml against M. tuberculosis and to be 4 to 8 µg/ml against M. avium (9). Against an isolate of M. tuberculosis having a MIC of 1 µg/ml, the EMB C_max/MIC ratio may range from 2:1 to 6:1, with a time above MIC from 5 to 12 h. However, EMB barely achieves inhibitory concentrations in serum against M. avium, even with good absorption. Since the absorption of EMB has been shown to be poor in patients with AIDS, EMB doses higher than 25 mg/kg may be required for some of these patients in order to inhibit the pathogens (19).

To conclude, the concentrations in serum found in this study were consistent with those previously described. In contrast, our V₁, V_SS, and CL estimates were larger than previously described. The kinetic behavior of EMB was consistent between the two fasting treatments. Food had a minimal effect on the absorption of EMB, while antacids should be avoided near the time of EMB dosing. Samples drawn between 2 and 3 h postdose approach C_max for most subjects, and samples drawn as early as day 2 of daily EMB therapy will produce concentrations in serum that approach steady-state values.