The Early Bactericidal Activities of Rifampin and Rifapentine in Pulmonary Tuberculosis
http://www.100md.com
美国呼吸和危急护理医学 2005年第7期
Lead Programme for TB Research, Medical Research Council
Department of Pediatrics and Child Health, Tygerberg
Lead Programme for TB Research, Medical Research Council, Pretoria
King George V Hospital and Unit for Clinical and Biomedical TB Research, Medical Research Council, Durban
Unit for Biostatistics, Medical Research Council, Pretoria
Division of Pharmacology, University of Cape Town, Cape Town, South Africa
FINDdiagnostics, Geneva, Switzerland
St. George's Hospital Medical School, London, United Kingdom
ABSTRACT
Rationale: Comparison of the early bactericidal activity (EBA) of rifapentine and its pharmacokinetics with those of rifampin to determine the cause of poor clinical response and regrowth between doses, leading to rifamycin monoresistance at relapse. Objectives: Determination of the dose size of rifapentine that gives sufficient drug exposure to prevent regrowth. Methods: EBA study over initial 5 days of treatment of 123 patients, half at Durban and half at Cape Town, who received single rifapentine doses of 300, 600, 900, or 1,200 mg rifapentine or five daily doses of 150, 300, or 600 mg rifampin, with a pharmacokinetic study on 58 patients measuring standard parameters for each dose size of rifamycin and their desacetyl metabolites. Results: The EBAs for both rifamycins were similar, with a linear relationship to log dose at lower doses and a curvilinear response at higher doses giving a plateau at 1,136 mg rifapentine. The area under the concentrationeCtime curve (AUC) divided by the minimal inhibitory concentration (MIC) agreed well for both rifamycins on the assumption that the only free 2% of free rifapentine and the 14% of free rifampin after plasma binding were active in the lesions. Conclusions: Only the free proportions of the rifamycins were active in lesions. From consideration of the pulse size and the duration of the postantibiotic lag, a 1,200-mg dose of rifapentine seemed necessary to improve response and to prevent regrowth between doses, and hence rifamycin monoresistance.
Key Words: early bactericidal activity pharmacokinetics plasma binding postantibiotic effect rifapentine
Rifapentine (RPE) is a long-acting rifamycin with a half-life of approximately 12 hours in patients (1, 2). It has been shown to be effective when given once weekly in experimental murine tuberculosis, but its bactericidal efficacy was much influenced by the dose size (3, 4). In the published results of two large-scale studies of its use in the continuation phase of pulmonary tuberculosis, a dose of 600 mg given with 15 mg/kg isoniazid (INH) once weekly was found to be followed by higher relapse rates than with thrice- or twice-weekly rifampin (RMP) and INH (5, 6). Furthermore, when given to patients with seropositive HIV, four of the five patients who relapsed after treatment had strains resistant only to rifamycins, a phenomenon not encountered in those who were HIV seronegative (7). The dose size of RPE was originally chosen on the empiric ground that it was the same as the dose size of RMP and yielded high plasma concentrations. However, the disappointing results in clinical trials with RPE might have been caused by plasma binding of 98% of the drug (8). In addition to differences in plasma binding of RMP (9) and RPE and corresponding pharmacokinetics, there are also differences in the minimal inhibitory concentrations (MICs) for Mycobacterium tuberculosis (10), in the proportions converted to the microbiologically active desacetyl metabolite, and in their penetrations into macrophages (11). The effects of these differences on antibacterial activity during the first days of therapy could be explored in a study of early bactericidal activity (EBA). Although such studies are not efficient in measuring longer term sterilizing activity, there is no evidence to suggest that RMP and RPE differ in sterilizing activity, particularly because there is complete cross-resistance between them and they have similar bactericidal and postantibiotic effects (12). In this context, the EBA study was planned to establish the effects of the large differences in pharmacokinetic effects with RPE and RMP, but not to measure sterilizing activity. The EBA study was performed with a range of dose sizes of RMP and RPE during 5 days of treatment, with estimates of the numbers of colony-forming units (cfu) in sputum before the start of treatment, after 2 days and after 5 days of treatment. This design would give information on the bactericidal effect over the first 5 days of administration of each drug. Subsidiary information might be provided by comparing the bactericidal effects during the first 2 days and those during the succeeding 3 days. The study was performed to the same protocol in Durban and in Cape Town. Pharmacokinetic studies, measuring the plasma concentration of the rifamycins and their 25-desacetyl metabolites after a study dose, were done on all of the Cape Town patients. The results of these studies are reported here.
METHODS
Patients
Newly diagnosed, previously untreated patients with strongly smear-positive pulmonary tuberculosis were admitted to the study, organized by the South African Medical Research Council, between October 2000 and September 2002. Of the 123 patients admitted, 63 were in King George V Hospital, Durban, and 60 in Brooklyn Hospital for Chest Diseases, Cape Town. They were aged 18 to 60 years, weighed 40 to 60 kg, and were estimated to produce at least 15 ml of sputum during an overnight collection; female patients were not pregnant. Patients were excluded if they had serious concomitant disease or abnormal renal or hepatic function. Ethical approval for the study was given by the Medical Research Council of South Africa, and by ethics committees of the Universities of Natal and Stellenbosch. Patients gave written, informed consent. Their HIV status was tested by the double ELISA method with pre- and posttest counseling.
Treatment
Treatment was with RPE (Priftin) as 150-mg film-coated tablets or with 150-mg tablets of RMP (Rifadin). The RPE was given as a single dose, at most 15 minutes after taking an "English breakfast" shown to be capable of promoting absorption (13). RMP was given daily, 1 hour before starting the English breakfast. Urine specimens were examined before and during the study for isonicotinic acid (14) and for RMP (15) to detect unprescribed taking of INH or RMP. After 5 days of this treatment, the patients started on standard four-drug combined therapy with INH, pyrazinamide, RMP, and ethambutol.
Sputum Collection and Culture
Sputum was collected for 16 hours, from 4:00 P.M. to 8:00 A.M. the next morning, before the first drug dose (S0 collection), and after 2 (S2 collection) and 5 days (S5 collection). RMP or RPE was given soon after termination of the collection. The sputum was homogenized with a magnetic follower for 30 minutes. To one part (usually 10 ml) of homogenized sputum was added two parts of 1:10 dithiothreitol solution (Sputasol, code SR089A, Oxoid; Basingstoke, UK). After vortex mixing for 20 seconds, it was left at room temperature for 15 to 20 minutes. Serial dilutions were prepared in distilled water and were plated on 7H11 medium made selective by the addition of polymyxin B (200 U/ml), ticarcillin (100 mg/L), trimethoprim (10 mg/L), and amphotericin B (10 mg/L) (16). Plates were incubated with a Mycobacterium phlei plate to provide CO2, and colonies were counted after incubation for 3 weeks.
Pharmokinetic Study
Plasma samples were taken from eight to nine of the Cape Town patients in each treatment group at 0, 1, 2, 4, 6, 8, 12, 24, 48, 72, and 96 hours after the dose of RPE and at 0, 1, 2, 4, 6, 8, 12, and 24 hours after the first and the fifth doses of RMP. They were stored at eC80°C and protected from light, until analyzed for their content of the rifamycin and its 25-desacetyl metabolite (D-RPE and D-RMP) using high-performance liquid chromatography, with ultraviolet detection, measuring down to 0.3 e/ml RMP, 0.2 e/ml D-RMP, and 0.6 e/ml RPE and D-RPE. Reference standards for RPE, D-RPE, and D-RMP were donated by Aventis Pharma (Antony, France). RMP was purchased from Sigma (St. Louis, MO).
Statistical Handling
The Stata package, release 7 (Stata Corp., College Station, TX), was used to obtain summary statistics and for statistical modeling, and the WinNonLin package, version 3.3 (Pharsight Corp., Mountain View, CA), was used to generate pharmacokinetic parameters.
RESULTS
Summary Statistics
The patients were allocated at random and equally to the following treatment groups: (1) RPE, 300 mg as a single dose; (2) RPE, 600 mg as a single dose; (3) RPE, 900 mg as a single dose; (4) RPE, 1,200 mg as a single dose; (5) RMP, 150 mg daily; (6) RMP, 300 mg daily; or (7) RMP, 600 mg daily. The EBA was calculated as the fall in log10 cfu/ml of sputum/day, but for simplicity, values are given without units. Three EBAs were calculated as follows: EBA 0eC5 = (log10 S0 cfu/ml sputum eC log10 S5 cfu/ml sputum)/5, EBA 0eC2 = (log10 S0 cfu/ml sputum eC log10 S2 cfu/ml sputum)/2, and EBA 2eC5 = (log10 S2 cfu/ml sputum eC log10 S5 cfu/ml sputum)/3.
Of the 123 patients admitted, 12 were excluded during the study: one patient had negative cultures for M. tuberculosis, five had no sputum collections, two had M. tuberculosis resistant to rifamycins, and four had RMP or INH in their urine. In four patients, individual collections were missing, still allowing calculation of some of the EBAs. There remained 109 EBA 0eC5 values, 109 EBA 0eC2 values, and 107 EBA 2eC5 values for analysis. Summary statistics for each of the three EBAs are set out in Table 1 and Figure 1. With RMP, there is a linear relationship between log-dose size and both the EBA 0eC2 and the EBA 2eC5 (Figure 1). The mean for the EBA 0eC2 (0.121) is slightly higher than the mean for the EBA 2eC5 (0.093), but the difference of 0.028 is not significant (p = 0.25). The results are similar to those obtained in an earlier study (Figure 2) (17). The results with RPE show a smaller response to dose size and a wider interval between the EBA 0eC2 and EBA 2eC5, especially at dose sizes of 600 and 900 mg (Figure 1). The means for EBA 0eC2 and EBA 2eC5 with RPE are 0.257 and 0.194, respectively, a difference of 0.063, which is significant (p = 0.01). The term "therapeutic margin" has been defined as the ratio between the usual therapeutic dose size (600 mg for RMP) and the dose that just fails to yield a measurable EBA (150 mg for RMP). Thus, the therapeutic margin for RMP is 600/150 = 4.0, as found previously (Figure 2) (8, 17). The therapeutic margin for RPE could not be determined because none of the dose sizes was sufficiently low to produce an EBA approaching 0.
Statistical Modeling
The first issues were to establish by analysis of variance whether there were any differences between the two centers (Durban and Cape Town) in either the overall means of their EBA values or in their possible differences in dosage effects and whether the other characteristics of the patients had any similar effect. No such differences in HIV status, age, or body weight were found for each EBA in either the center main effect or the center x dose interaction. However, women had lower values of EBA 0eC5 by 0.074 (p = 0.006), with no evidence that this affected the linear or quadratic treatment components. Thus, the dose response was similar for men and women and the optimal dose of RPE should not depend on sex. The mean of the SDs for the EBA 0eC5 values within each treatment group were similar for Durban (0.140) and for Cape Town (0.146). Each of the mean EBA values was also similar in the 26 patients who were seropositive for HIV infection (mean EBA 0eC5, 0.178) and in the 82 who were seronegative (mean EBA 0eC5, 0.176). The results in the two centers and in HIV status could therefore be combined with confidence. In each analysis of variance, a highly significant treatment effect was found (p < 0.0001 in each; Table 2). The 6 degrees of freedom for treatment (there were seven treatment groups) were then partitioned into (1) the center effect, (2) a log-dose effect, (3) a (log-dose)2 effect to look for curvature of the line relating log-dose size to EBA, (4) a (log-dose)3 effect to look for deviations from the quadratic fit, (5) the difference between RPE and RMP (shown as RMP) and the interactions between RMP and the log-dose effects. Only the log-dose effect was consistently significant with each of the three EBAs (p < 0.0001 in each). The quadratic (log-dose)2 effect was nearly significant (p = 0.052) in the EBA 0eC5, indicating the curvature evident in Figure 1, but it was not significant in the analysis of either of the other two EBAs. Of particular importance, the RMP terms distinguishing the effects of RPE and RMP were clearly nonsignificant (p = 0.46, 0.17, and 0.80 in the analyses of EBA 0eC5, EBA 0eC2, and EBA 2eC5, respectively). The precision of the modeling allowed detection of a difference of 0.1 in the EBAs of the two groups given the same doses (300 and 600 mg) of the two rifamycins. Furthermore, the two interactions between the rifamycin difference and either log dose or (log dose)2 were clearly nonsignificant, so that the response curves for RPE and RMP can be taken as parallel. After removal of the cubic (log-dose)3 term, further regression analyses gave closely similar results.
The conclusion from these analyses is that there is a strong relationship between log dose and each of the EBAs. There is also evidence of a curved (quadratic) effect with the EBA 0eC5 values (Figure 3). There is no evidence that the responses of RMP and RPE in the model are different; they can therefore be considered as parallel. As the curve for EBA 0eC5 flattens out between the 900- and the 1,200-mg RPE doses, (the curvature term) with the turning point (peak value) at 1,136 mg, there seems to be a maximal bactericidal effect. The equations for the fitted curves are shown in Table 3 together with turning points (peak values) of the curves, which are 981, 1,431, and 1,136 mg for the EBA 0eC2, EBA 2eC5, and EBA 0eC5, respectively. These turning points are all between 900 and 1,200 mg, suggesting that dose sizes of 900 or 1,200 mg have optimal bactericidal effect and that there would not be any advantage in taking the RPE dose above 1,200 mg. However, it should be noted that there is a rather large difference between the mean EBA 0eC2 of 0.364 and the mean EBA 2eC5 of 0.176 at the dose of 900 mg RPE. The difference between these means of 0.188 is 4.5 times the approximate standard error of 0.042 for the difference between any two means (calculated from the residual), so that it should be taken into account in the interpretation of results. At the 1,200-mg dose of RPE, this difference had disappeared so that the EBA 0eC2 and the EBA 2eC5 were similar in size.
Pulses of Rifamycin Activity
The results of the measurements of RMP, D-RMP, RPE, and D-RPE show several expected differences between the two rifamycins and their metabolites (Table 4). Mean time to peak (Tmax) times are longer, whereas mean peak drug concentration (Cmax) and particularly area under the timeeCconcentration curve extrapolated to infinity (AUCI) values are larger for RPE than for RMP. RMP and D-RMP values of AUCI fall between Days 1 and 5, as a result of induction. D-RPE AUCI values are larger relative to those for RPE than are the corresponding values for RMP and D-RMP. The results of the assays of RPE and D-RPE after dose sizes of 600 and 1,200 mg RPE (Figure 4) show typical curves for RPE, with peaks of 12.4 and 22.5 e/ml, respectively, at 8 hours, low concentrations at 72 hours, and just detectable mean concentrations of 0.33 and 0.95 e/ml, respectively, at 96 hours. The curves for D-RPE peaked later at 24 hours and thereafter had concentrations similar to those of RPE. Because D-RPE is less active than RPE against M. tuberculosis, a total active concentration was calculated as the concentration of RPE at each time point plus the concentration of D-RPE divided by 2.972, the geometric mean of the ratio between the MICs of RPE and D-RPE found in 14 pairs of assays by BacTec and 7H11 (Becton-Dickinson, Sparks, MD) plate methods on seven strains of M. tuberculosis (10). These values are shown for dose sizes of 600, 900, and 1,200 mg RPE (Figure 5). With the assumption that 98% of rifapentine is bound to plasma protein (8), leaving only 2% (1/50th) available in its free form to be distributed into the lesions, and that the mean MIC of RPT and D-RPT is 0.12 e/ml (10), a line AB has been drawn on the graph (Figure 5) at a concentration of 6.0 e/ml (0.12 x 50), which represents the lowest concentration of RPE + D-RPE that would have activity in the lesions. It is evident that as the dose of RPE was increased, there was an increase in the duration of the pulse of RPE activity in the lesions from 36.5 to 39 to 64 hours as the dose size was increased from 600 to 900 to 1,200 mg. The size of the pulse also increased with dose size.
The relationship between the pulse sizes and the EBA 0eC5 in those given the same sized doses of RMP and RPE are explored, with and without the effects of plasma binding, in Table 5. The best estimate of the antibacterial effect of a pulse was taken as the AUCI/MIC from experience gained with mouse pharmacokinetics of RMP (18). Values of the single-dose AUCI are taken from Table 4, with the means of the values for RMP at Days 1 and 5. Total AUCI was calculated for RPE by summing the AUCIs for RPE and D-RPE, each divided by its MIC. For RMP, a corresponding procedure was performed on the average AUCIs at Days 1 and 5. When no further correction was made for plasma binding, the total AUCI/MIC for RPE is four to seven times higher than for RMP. However, when it was assumed that plasma binding leaves only free drug available for antibacterial activity, estimated as 14% for RMP (9) and 2% for RPE (8), there is now good agreement between the AUCI/MICs of RMP and RPE. Furthermore, these values associate well with the EBA 0eC5 in the four dosage groups.
DISCUSSION
What does a comparison of the EBAs on RMP and RPE measure The EBA 0eC5 measures killing effects on a mixture of rapidly growing and persisting bacilli (19), and because it does not measure a pure effect on persisters, it is not a good indicator of the long-term sterilizing activity of a drug. However, RMP and RPE differ in their binding to plasma protein, their pharmacokinetics, and their penetration into macrophages. EBAs measure the effect these factors have on bacterial killing within cavitary lesions. They are characteristics that remain more or less constant throughout therapy during the change from activity to persistence in the bacterial population, so that the ratios between EBAs for RMP and RPE during the first week should remain throughout, regardless of the rate of growth and therefore the persistence of the bacilli.
The statistical modeling of the EBA results suggests that two smooth curves (Figure 3) can represent the doseeCresponse curve of the EBA 0eC5 for RPE and RMP. These curves are parallel, although the curve for RPE lies a little above the curve for RMP, possibly indicating slightly greater potency. The curve for RPE reaches a maximum at about a dose size of 900 to 1,200 mg RPE, suggesting that these are optimal dose sizes, and that no further increase would be beneficial. A similar flattening out of the EBAeCdose size curve has been found with INH (20). The analysis also implies that results with a single weekly dose of 600 mg RPE should have been at least as good as those with daily 600 mg RMP. This conclusion does not, however, agree with the findings of the two large-scale clinical trials, in which relapse rates were higher with 600 mg RPE than with 600 mg RMP, implying that the bactericidal activity of RPE was less good during the continuation phase (5, 6). How can we account for this disparity
One reason is that the experimental period of the EBA study was 5 days, whereas there were 7 days between the doses of RPE in the clinical trials. With RMP given daily, the EBA 2eC5 was only slightly lower than the EBA 0eC2, a nonsignificant difference of 0.028, whereas the difference of 0.063 between the corresponding means for RPE was larger and significant. Because plasma concentrations after the single dose of RPE peak during the first day after ingestion and then decrease during the next 3 days (Figure 4), the falling concentrations in plasma, and presumably also in lesions, might well account for lower EBAs toward the end of the 5-day period.
A second and more important reason lies in the possible regrowth of persisting bacilli during the last few days of the week, between RPE doses. Such regrowth would not be detected in EBA measurements. Regrowth of sensitive persisters would account for the higher relapse rates in the RPE arms of the two clinical trials. It could also account for the occurrence of rifamycin monoresistance in the relapse cultures of patients with seropositive HIV (7). Strains with rifamycin monoresistance must arise from rifamycin-resistant mutants. However, it is highly improbable that there would be any rifamycin-resistant mutants in the entire bacillary population at the end of the 8-week initial phase of treatment in the trials because of the very low frequency of mutations to resistance. The highest proportion of rifamycin mutants expected at the start of treatment has been calculated, using Luria and Delbruck fluctuation tests, as 1 in 3 x 10eC8 (21). Even if one assumed a very high estimate of 1010 viable bacilli in the entire initial bacterial population, the size of the population would be reduced to below 109 by the initial bactericidal action of INH and would soon be reduced substantially by the sterilizing action of pyrazinamide and INH (19) to bring it well below the number required for a mutation to be present. Further evidence that sufficient mutants would not be available without regrowth is shown by the rarity of the emergence of resistance to INH when continuation phase treatment, preceded by a 2-month, four-drug initial phase, was given with INH alone in a Tanzanian study (22), despite the higher proportion of mutant bacilli resistant to INH pretreatment (3 x 10eC6) than to RMP in a population of sensitive organisms. Without regrowth, no mutants would be available so that a companion drug would not be necessary at all to prevent RMR, except to guard against the occurrence of gross irregularity in drug taking. However, with regrowth, mutations would occur and mutants would multiply, particularly in immunocompromised patients; they might then develop rifamycin monoresistance (7). Failure of the companion drug INH could also lead to relapse, as occurred in a recent clinical trial (23).
However, regrowth would be unlikely to occur if the RPE concentrations estimated in the lesions were increased. The postantibiotic lag has been found to be 5 days after a pulse of 1 e/ml RMP or RPE lasting 96 hours and approximately 1 day after a pulse lasting 24 hours (12). Although a 6-hour exposure to 1 e/ml RPE was not followed by any lag (12), a recent study (24) indicates that lags increase with the higher RPE concentrations found when dose size was increased (Figure 4). The total length of the period of inhibition after a dose of RPE would then be the duration of the pulse for 3 to 4 days (Figure 4), with a peak concentration well above 1 e/ml, and a postantibiotic lag of at least 5 days (Table 6). Thus, bacterial multiplication would be inhibited throughout the week between doses.
We propose that the pulse of RPE must be much smaller than would be estimated from the measured plasma concentrations, presumably because of 98% plasma binding. This has been shown by the close correlation between AUCI/MIC and EBA 0eC5 for both RMP and RPE (Table 5). That only the 2% free portion of RPE is available in lesions has also been indicated by experience with RMP (8). The finding in the present and in a previous study (17) that the therapeutic margin for RMP is 4, whereas it is approximately 20 for INH (see Figure 2) (20), with similar ratios between the peak plasma concentrations and the MICs of both drugs, has suggested that only 20% (4/20) of the plasma concentration is available in lesions. This estimate is in reasonable agreement with the proposition that the proportion of RMP that is active in the lesions is only the unbound 14% of the dose (8). If only the 2% of free RPE is available in tuberculous cavities, the size and duration of the RPE pulse reaching the bacilli would be smaller than what appear from plasma estimates, and the lag after the pulse would also be shorter (Tables 5 and 6, with duration of pulse from Figure 4). The lag after a pulse is difficult to estimate from the data on only three pulse sizes (12), even when supplemented by more recent data on much higher RPE concentrations (24). The estimates suggest that the pulse after a 600-mg dose and the lag after it would last 3 days so that there would be 4 days available for regrowth to occur. Using our pharmacokinetic and postantibiotic lag data, an increase in the dose of RPE to 1,200 mg would increase the duration of the pulse to 2.5 days and also the lag period after it, leaving only 1.5 days for regrowth. However, the size and height of the pulse are substantially increased as the dose size is raised from 600 to 1,200 mg with a probable increase in the lag period, which is then estimated to be at least 3.5 days (Table 6). There would then be a maximum of only 1 day, allowing only one bacterial division each week, insufficient to create rifamycin monoresistance. Furthermore, the EBA 2eC5 increased substantially when the RPE dose was increased from 900 to 1,200 mg, suggesting better bactericidal activity toward the end of the week. We suggest that the balance between the bactericidal effect of a pulse from a 600 mg dose and subsequent regrowth at the end of the treatment cycle may be insufficient to create an overall bactericidal effect in all patients, but the balance might move to a greater overall effect, without appreciable regrowth, when a larger dose is given. Further estimates of the effect of pulse size and height on postantibiotic lag would be helpful.
In conclusion, it seems reasonable to assume that the most effective dose size of RPE would be 1,200 mg and that this dose size is likely to inhibit growth in lesions for approximately 6 of the 7 days between doses. The increase in dose size should be sufficient to prevent the development of rifamycin monoresistance, whether or not an effective companion drug is given. Thus, there seems no necessity to find a better companion drug than INH if the dose size is increased. It does not seem from this study that any benefit in terms of preventing relapse would be likely to accrue from raising the dose size above 1,200 mg, because our study indicates that there would be no further increase in bactericidal action.
Rifapentine EBA Collaborative Study Group:
D. P. Parkin, A. Venter, M. Bester, Stellenbosch University; J. Allen, T. C. P. Mthiyane, Unit for Clinical and Biomedical Research, Medical Research Council, Durban; A. S. Pala, A. Ramjee, I. H. Master, King George V TB Hospital, Durban; J. L. S. Pillay, Umlazi Chest Clinic, Prince Mshiyeni Hospital, Durban; A. Olowolagba, T. Chinappa, Ethekweni Municipality, Durban: S. Siwendu, Brooklyn Hospital for Chest Diseases, Cape Town; P. Smith, University of Cape Town, Cape Town, South Africa.
Acknowledgments
The authors acknowledge the excellent assistance of the staff of the King George V Hospital, Durban, and the staff of Brooklyn Hospital for Chest Diseases, Cape Town.
Members of the Rifapentine EBA Collaborative Study Group are listed at the end of the article.
REFERENCES
Arioli V, Berti M, Carniti G, Randisi E, Rossi E, Scotti R. Antibacterial activity of DL473, a new semisynthetic rifampicin derivative. J Antibiot 1981;34:1026eC1032.
Keung A, Eller MG, McKenzie KA, Weir SJ. Single and multiple dose pharmacokinetics of rifapentine in man: part II. Int J Tuberc Lung Dis 1999;3:437eC444.
Truffot-Pernot C, Grosset J, Bismuth R, Lecoeur H. Activitee de la rifampicine adminutesistreee de manieere intermittente et de la cyclopentyl rifamycine (our DL473) sur la tuberculeuse experimental de la souris. Rev Fr Maladies Respir 1983;11:875eC882.
Dhillon J, Dickinson JM, Guy JA, Ng TK, Mitchison DA. Activity of two long-acting rifamycins, rifapentine and FCE 22807, in experimental murine tuberculosis. Tuber Lung Dis 1992;73:116eC123.
Tam CM, Chan SL, Kam KM, Sim E, Staples D, Sole KM, Al-Ghusein H, Mitchison DA. Rifapentine and INH in the continuation phase of a 6-month regimen. Interim report: no activity of INH in the continuation phase. Int J Tuberc Lung Dis 2000;4:262eC267.
Tuberculosis Trials Consortium. Once-weekly rifapentine and INH versus twice-weekly rifampin and INH in the continuation phase of therapy for drug-susceptible pulmonary tuberculosis: a prospective, randomised clinical trial among HIV-negative persons. Lancet 2002;360:528eC534.
Vernon A, Burman W, Benator D, Khan A, Bozeman L, for the Tuberculosis Trials Consortium. Acquired rifamycin mono-resistance among patients with HIV-related tuberculosis treated with supervised once weekly rifapentine and INH. Lancet 1999;353:1843eC1847.
Mitchison DA. Development of rifapentine: the way ahead. Int J Tuberc Lung Dis 1998;2:612eC615.
Boman G, Ringberger V-A. Binding of rifampicin by human plasma proteins. Eur J Clin Pharmacol 1974;7:369eC373.
Rastogi N, Goh KS, Berchell M, Bryskier A. Activity of rifapentine and its metabolite 25-O-desacetylrifapentine compared with rifampicin and rifabutin against Mycobacterium tuberculosis, Mycobacterium africanum, Mycobacterium bovis and M. bovis BCG. J Antimicrob Chemother 2000;46:565eC570.
Dhillon J, Mitchison DA. Activity in vitro of rifabutin, FCE22807, rifapentine and rifampin against Mycobacterium microti and M. tuberculosis and their penetration into mouse peritoneal macrophages. Am Rev Respir Dis 1992;145:212eC214.
Dickinson JM, Mitchison DA. In vitro properties of rifapentine (MDL473) relevant to its use in intermittent chemotherapy of tuberculosis. Tubercle 1987;68:113eC118.
Chan SL, Yew WW, Porter JHD, McAdam KPWJ, Allen BW, Dickinson JM, Ellard GA, Mitchison DA. Comparison of Chinese and Western rifapentines and improvement of bioavailability by prior taking of various meals. Int J Antimicrob Agents 1994;3:267eC274.
Ellard GA, Greenfield C. A sensitive urine-test method for monitoring the ingestion of isoniazid. J Clin Pathol 1977;30:84eC87.
Mitchison DA, Allen BW, Miller AB. Detection of rifampicin in urine by a simple microbiological assay. Tubercle 1970;51:300eC304.
Mitchison DA, Allen BW, Carrol L, Dickinson JM, Aber BR. A selective oleic acid albuminutes agar medium for tubercle bacilli. J Med Microbiol 1972;5:165eC175.
Sirgel FA, Botha FJH, Parkin DP, Van de Wal BW, Donald PR, Clark PK, Mitchison DA. The early bactericidal activity of rifabutin in patients with pulmonary tuberculosis measured by sputum viable counts: a new method of drug assessment. J Antimicrob Chemother 1993;32:867eC875.
Jayaram R, Gaonkar S, Kaur P, Suresh BL, Mahesh BN, Jayashree R, Nandi V, Bharat S, Shandil RK, Kantharaj E, et al. Pharmacokinetics-pharmacodynamics of rifampin in an aerosol infection model of tuberculosis. Antimicrob Agents Chemother 2003;47:2118eC2124.
Jindani A, Doree CJ, Mitchison DA. The bactericidal and sterilising activities of antituberculosis drugs during the first 14 days. Am J Respir Crit Care Med 2003;167:1348eC1354.
Donald PR, Sirgel FA, Botha FJ, Seifart HI, Parkin DP, Vandenplas ML, Van de Wal BW, Maritz JS, Mitchison DA. The early bactericidal activity of isoniazid related to its dose size in pulmonary tuberculosis. Am J Respir Crit Care Med 1997;156:895eC900.
David HL. Probability distribution of drug-resistant mutants in unselected populations of Mycobacterium tuberculosis. Appl Microbiol 1970;20:810eC814.
Tanzania/British Medical Research Council. A controlled trial of a 4-weekly supplement of rifampicin, pyrazinamide and streptomycin in the continuation phase of a 7-month daily chemotherapy regimen for pulmonary tuberculosis. S Afr Med J 1996;86:960eC965.
Weiner M, Burman W, Vernon A, Benator D, Peloquin CA, Khan A, Weis S, King B, Shah N, Hodge T, and the Tuberculosis Trials Consortium. Low isoniazid concentrations and outcome of tuberculosis treatment with once-weekly isoniazid and rifapentine. Am J Respir Crit Care Med 2003;167:1341eC1347.
Chan C-Y, Au-Yeang C, Yew W-W, Leung C-C, Cheng AFB. In vitro postantibiotic effects of rifapentine, isoniazid, and moxifloxacin against Mycobacterium tuberculosis. Antimicrob Agents Chemother 2004;48:340eC343.(Frik A. Sirgel, P. Bernar)
Department of Pediatrics and Child Health, Tygerberg
Lead Programme for TB Research, Medical Research Council, Pretoria
King George V Hospital and Unit for Clinical and Biomedical TB Research, Medical Research Council, Durban
Unit for Biostatistics, Medical Research Council, Pretoria
Division of Pharmacology, University of Cape Town, Cape Town, South Africa
FINDdiagnostics, Geneva, Switzerland
St. George's Hospital Medical School, London, United Kingdom
ABSTRACT
Rationale: Comparison of the early bactericidal activity (EBA) of rifapentine and its pharmacokinetics with those of rifampin to determine the cause of poor clinical response and regrowth between doses, leading to rifamycin monoresistance at relapse. Objectives: Determination of the dose size of rifapentine that gives sufficient drug exposure to prevent regrowth. Methods: EBA study over initial 5 days of treatment of 123 patients, half at Durban and half at Cape Town, who received single rifapentine doses of 300, 600, 900, or 1,200 mg rifapentine or five daily doses of 150, 300, or 600 mg rifampin, with a pharmacokinetic study on 58 patients measuring standard parameters for each dose size of rifamycin and their desacetyl metabolites. Results: The EBAs for both rifamycins were similar, with a linear relationship to log dose at lower doses and a curvilinear response at higher doses giving a plateau at 1,136 mg rifapentine. The area under the concentrationeCtime curve (AUC) divided by the minimal inhibitory concentration (MIC) agreed well for both rifamycins on the assumption that the only free 2% of free rifapentine and the 14% of free rifampin after plasma binding were active in the lesions. Conclusions: Only the free proportions of the rifamycins were active in lesions. From consideration of the pulse size and the duration of the postantibiotic lag, a 1,200-mg dose of rifapentine seemed necessary to improve response and to prevent regrowth between doses, and hence rifamycin monoresistance.
Key Words: early bactericidal activity pharmacokinetics plasma binding postantibiotic effect rifapentine
Rifapentine (RPE) is a long-acting rifamycin with a half-life of approximately 12 hours in patients (1, 2). It has been shown to be effective when given once weekly in experimental murine tuberculosis, but its bactericidal efficacy was much influenced by the dose size (3, 4). In the published results of two large-scale studies of its use in the continuation phase of pulmonary tuberculosis, a dose of 600 mg given with 15 mg/kg isoniazid (INH) once weekly was found to be followed by higher relapse rates than with thrice- or twice-weekly rifampin (RMP) and INH (5, 6). Furthermore, when given to patients with seropositive HIV, four of the five patients who relapsed after treatment had strains resistant only to rifamycins, a phenomenon not encountered in those who were HIV seronegative (7). The dose size of RPE was originally chosen on the empiric ground that it was the same as the dose size of RMP and yielded high plasma concentrations. However, the disappointing results in clinical trials with RPE might have been caused by plasma binding of 98% of the drug (8). In addition to differences in plasma binding of RMP (9) and RPE and corresponding pharmacokinetics, there are also differences in the minimal inhibitory concentrations (MICs) for Mycobacterium tuberculosis (10), in the proportions converted to the microbiologically active desacetyl metabolite, and in their penetrations into macrophages (11). The effects of these differences on antibacterial activity during the first days of therapy could be explored in a study of early bactericidal activity (EBA). Although such studies are not efficient in measuring longer term sterilizing activity, there is no evidence to suggest that RMP and RPE differ in sterilizing activity, particularly because there is complete cross-resistance between them and they have similar bactericidal and postantibiotic effects (12). In this context, the EBA study was planned to establish the effects of the large differences in pharmacokinetic effects with RPE and RMP, but not to measure sterilizing activity. The EBA study was performed with a range of dose sizes of RMP and RPE during 5 days of treatment, with estimates of the numbers of colony-forming units (cfu) in sputum before the start of treatment, after 2 days and after 5 days of treatment. This design would give information on the bactericidal effect over the first 5 days of administration of each drug. Subsidiary information might be provided by comparing the bactericidal effects during the first 2 days and those during the succeeding 3 days. The study was performed to the same protocol in Durban and in Cape Town. Pharmacokinetic studies, measuring the plasma concentration of the rifamycins and their 25-desacetyl metabolites after a study dose, were done on all of the Cape Town patients. The results of these studies are reported here.
METHODS
Patients
Newly diagnosed, previously untreated patients with strongly smear-positive pulmonary tuberculosis were admitted to the study, organized by the South African Medical Research Council, between October 2000 and September 2002. Of the 123 patients admitted, 63 were in King George V Hospital, Durban, and 60 in Brooklyn Hospital for Chest Diseases, Cape Town. They were aged 18 to 60 years, weighed 40 to 60 kg, and were estimated to produce at least 15 ml of sputum during an overnight collection; female patients were not pregnant. Patients were excluded if they had serious concomitant disease or abnormal renal or hepatic function. Ethical approval for the study was given by the Medical Research Council of South Africa, and by ethics committees of the Universities of Natal and Stellenbosch. Patients gave written, informed consent. Their HIV status was tested by the double ELISA method with pre- and posttest counseling.
Treatment
Treatment was with RPE (Priftin) as 150-mg film-coated tablets or with 150-mg tablets of RMP (Rifadin). The RPE was given as a single dose, at most 15 minutes after taking an "English breakfast" shown to be capable of promoting absorption (13). RMP was given daily, 1 hour before starting the English breakfast. Urine specimens were examined before and during the study for isonicotinic acid (14) and for RMP (15) to detect unprescribed taking of INH or RMP. After 5 days of this treatment, the patients started on standard four-drug combined therapy with INH, pyrazinamide, RMP, and ethambutol.
Sputum Collection and Culture
Sputum was collected for 16 hours, from 4:00 P.M. to 8:00 A.M. the next morning, before the first drug dose (S0 collection), and after 2 (S2 collection) and 5 days (S5 collection). RMP or RPE was given soon after termination of the collection. The sputum was homogenized with a magnetic follower for 30 minutes. To one part (usually 10 ml) of homogenized sputum was added two parts of 1:10 dithiothreitol solution (Sputasol, code SR089A, Oxoid; Basingstoke, UK). After vortex mixing for 20 seconds, it was left at room temperature for 15 to 20 minutes. Serial dilutions were prepared in distilled water and were plated on 7H11 medium made selective by the addition of polymyxin B (200 U/ml), ticarcillin (100 mg/L), trimethoprim (10 mg/L), and amphotericin B (10 mg/L) (16). Plates were incubated with a Mycobacterium phlei plate to provide CO2, and colonies were counted after incubation for 3 weeks.
Pharmokinetic Study
Plasma samples were taken from eight to nine of the Cape Town patients in each treatment group at 0, 1, 2, 4, 6, 8, 12, 24, 48, 72, and 96 hours after the dose of RPE and at 0, 1, 2, 4, 6, 8, 12, and 24 hours after the first and the fifth doses of RMP. They were stored at eC80°C and protected from light, until analyzed for their content of the rifamycin and its 25-desacetyl metabolite (D-RPE and D-RMP) using high-performance liquid chromatography, with ultraviolet detection, measuring down to 0.3 e/ml RMP, 0.2 e/ml D-RMP, and 0.6 e/ml RPE and D-RPE. Reference standards for RPE, D-RPE, and D-RMP were donated by Aventis Pharma (Antony, France). RMP was purchased from Sigma (St. Louis, MO).
Statistical Handling
The Stata package, release 7 (Stata Corp., College Station, TX), was used to obtain summary statistics and for statistical modeling, and the WinNonLin package, version 3.3 (Pharsight Corp., Mountain View, CA), was used to generate pharmacokinetic parameters.
RESULTS
Summary Statistics
The patients were allocated at random and equally to the following treatment groups: (1) RPE, 300 mg as a single dose; (2) RPE, 600 mg as a single dose; (3) RPE, 900 mg as a single dose; (4) RPE, 1,200 mg as a single dose; (5) RMP, 150 mg daily; (6) RMP, 300 mg daily; or (7) RMP, 600 mg daily. The EBA was calculated as the fall in log10 cfu/ml of sputum/day, but for simplicity, values are given without units. Three EBAs were calculated as follows: EBA 0eC5 = (log10 S0 cfu/ml sputum eC log10 S5 cfu/ml sputum)/5, EBA 0eC2 = (log10 S0 cfu/ml sputum eC log10 S2 cfu/ml sputum)/2, and EBA 2eC5 = (log10 S2 cfu/ml sputum eC log10 S5 cfu/ml sputum)/3.
Of the 123 patients admitted, 12 were excluded during the study: one patient had negative cultures for M. tuberculosis, five had no sputum collections, two had M. tuberculosis resistant to rifamycins, and four had RMP or INH in their urine. In four patients, individual collections were missing, still allowing calculation of some of the EBAs. There remained 109 EBA 0eC5 values, 109 EBA 0eC2 values, and 107 EBA 2eC5 values for analysis. Summary statistics for each of the three EBAs are set out in Table 1 and Figure 1. With RMP, there is a linear relationship between log-dose size and both the EBA 0eC2 and the EBA 2eC5 (Figure 1). The mean for the EBA 0eC2 (0.121) is slightly higher than the mean for the EBA 2eC5 (0.093), but the difference of 0.028 is not significant (p = 0.25). The results are similar to those obtained in an earlier study (Figure 2) (17). The results with RPE show a smaller response to dose size and a wider interval between the EBA 0eC2 and EBA 2eC5, especially at dose sizes of 600 and 900 mg (Figure 1). The means for EBA 0eC2 and EBA 2eC5 with RPE are 0.257 and 0.194, respectively, a difference of 0.063, which is significant (p = 0.01). The term "therapeutic margin" has been defined as the ratio between the usual therapeutic dose size (600 mg for RMP) and the dose that just fails to yield a measurable EBA (150 mg for RMP). Thus, the therapeutic margin for RMP is 600/150 = 4.0, as found previously (Figure 2) (8, 17). The therapeutic margin for RPE could not be determined because none of the dose sizes was sufficiently low to produce an EBA approaching 0.
Statistical Modeling
The first issues were to establish by analysis of variance whether there were any differences between the two centers (Durban and Cape Town) in either the overall means of their EBA values or in their possible differences in dosage effects and whether the other characteristics of the patients had any similar effect. No such differences in HIV status, age, or body weight were found for each EBA in either the center main effect or the center x dose interaction. However, women had lower values of EBA 0eC5 by 0.074 (p = 0.006), with no evidence that this affected the linear or quadratic treatment components. Thus, the dose response was similar for men and women and the optimal dose of RPE should not depend on sex. The mean of the SDs for the EBA 0eC5 values within each treatment group were similar for Durban (0.140) and for Cape Town (0.146). Each of the mean EBA values was also similar in the 26 patients who were seropositive for HIV infection (mean EBA 0eC5, 0.178) and in the 82 who were seronegative (mean EBA 0eC5, 0.176). The results in the two centers and in HIV status could therefore be combined with confidence. In each analysis of variance, a highly significant treatment effect was found (p < 0.0001 in each; Table 2). The 6 degrees of freedom for treatment (there were seven treatment groups) were then partitioned into (1) the center effect, (2) a log-dose effect, (3) a (log-dose)2 effect to look for curvature of the line relating log-dose size to EBA, (4) a (log-dose)3 effect to look for deviations from the quadratic fit, (5) the difference between RPE and RMP (shown as RMP) and the interactions between RMP and the log-dose effects. Only the log-dose effect was consistently significant with each of the three EBAs (p < 0.0001 in each). The quadratic (log-dose)2 effect was nearly significant (p = 0.052) in the EBA 0eC5, indicating the curvature evident in Figure 1, but it was not significant in the analysis of either of the other two EBAs. Of particular importance, the RMP terms distinguishing the effects of RPE and RMP were clearly nonsignificant (p = 0.46, 0.17, and 0.80 in the analyses of EBA 0eC5, EBA 0eC2, and EBA 2eC5, respectively). The precision of the modeling allowed detection of a difference of 0.1 in the EBAs of the two groups given the same doses (300 and 600 mg) of the two rifamycins. Furthermore, the two interactions between the rifamycin difference and either log dose or (log dose)2 were clearly nonsignificant, so that the response curves for RPE and RMP can be taken as parallel. After removal of the cubic (log-dose)3 term, further regression analyses gave closely similar results.
The conclusion from these analyses is that there is a strong relationship between log dose and each of the EBAs. There is also evidence of a curved (quadratic) effect with the EBA 0eC5 values (Figure 3). There is no evidence that the responses of RMP and RPE in the model are different; they can therefore be considered as parallel. As the curve for EBA 0eC5 flattens out between the 900- and the 1,200-mg RPE doses, (the curvature term) with the turning point (peak value) at 1,136 mg, there seems to be a maximal bactericidal effect. The equations for the fitted curves are shown in Table 3 together with turning points (peak values) of the curves, which are 981, 1,431, and 1,136 mg for the EBA 0eC2, EBA 2eC5, and EBA 0eC5, respectively. These turning points are all between 900 and 1,200 mg, suggesting that dose sizes of 900 or 1,200 mg have optimal bactericidal effect and that there would not be any advantage in taking the RPE dose above 1,200 mg. However, it should be noted that there is a rather large difference between the mean EBA 0eC2 of 0.364 and the mean EBA 2eC5 of 0.176 at the dose of 900 mg RPE. The difference between these means of 0.188 is 4.5 times the approximate standard error of 0.042 for the difference between any two means (calculated from the residual), so that it should be taken into account in the interpretation of results. At the 1,200-mg dose of RPE, this difference had disappeared so that the EBA 0eC2 and the EBA 2eC5 were similar in size.
Pulses of Rifamycin Activity
The results of the measurements of RMP, D-RMP, RPE, and D-RPE show several expected differences between the two rifamycins and their metabolites (Table 4). Mean time to peak (Tmax) times are longer, whereas mean peak drug concentration (Cmax) and particularly area under the timeeCconcentration curve extrapolated to infinity (AUCI) values are larger for RPE than for RMP. RMP and D-RMP values of AUCI fall between Days 1 and 5, as a result of induction. D-RPE AUCI values are larger relative to those for RPE than are the corresponding values for RMP and D-RMP. The results of the assays of RPE and D-RPE after dose sizes of 600 and 1,200 mg RPE (Figure 4) show typical curves for RPE, with peaks of 12.4 and 22.5 e/ml, respectively, at 8 hours, low concentrations at 72 hours, and just detectable mean concentrations of 0.33 and 0.95 e/ml, respectively, at 96 hours. The curves for D-RPE peaked later at 24 hours and thereafter had concentrations similar to those of RPE. Because D-RPE is less active than RPE against M. tuberculosis, a total active concentration was calculated as the concentration of RPE at each time point plus the concentration of D-RPE divided by 2.972, the geometric mean of the ratio between the MICs of RPE and D-RPE found in 14 pairs of assays by BacTec and 7H11 (Becton-Dickinson, Sparks, MD) plate methods on seven strains of M. tuberculosis (10). These values are shown for dose sizes of 600, 900, and 1,200 mg RPE (Figure 5). With the assumption that 98% of rifapentine is bound to plasma protein (8), leaving only 2% (1/50th) available in its free form to be distributed into the lesions, and that the mean MIC of RPT and D-RPT is 0.12 e/ml (10), a line AB has been drawn on the graph (Figure 5) at a concentration of 6.0 e/ml (0.12 x 50), which represents the lowest concentration of RPE + D-RPE that would have activity in the lesions. It is evident that as the dose of RPE was increased, there was an increase in the duration of the pulse of RPE activity in the lesions from 36.5 to 39 to 64 hours as the dose size was increased from 600 to 900 to 1,200 mg. The size of the pulse also increased with dose size.
The relationship between the pulse sizes and the EBA 0eC5 in those given the same sized doses of RMP and RPE are explored, with and without the effects of plasma binding, in Table 5. The best estimate of the antibacterial effect of a pulse was taken as the AUCI/MIC from experience gained with mouse pharmacokinetics of RMP (18). Values of the single-dose AUCI are taken from Table 4, with the means of the values for RMP at Days 1 and 5. Total AUCI was calculated for RPE by summing the AUCIs for RPE and D-RPE, each divided by its MIC. For RMP, a corresponding procedure was performed on the average AUCIs at Days 1 and 5. When no further correction was made for plasma binding, the total AUCI/MIC for RPE is four to seven times higher than for RMP. However, when it was assumed that plasma binding leaves only free drug available for antibacterial activity, estimated as 14% for RMP (9) and 2% for RPE (8), there is now good agreement between the AUCI/MICs of RMP and RPE. Furthermore, these values associate well with the EBA 0eC5 in the four dosage groups.
DISCUSSION
What does a comparison of the EBAs on RMP and RPE measure The EBA 0eC5 measures killing effects on a mixture of rapidly growing and persisting bacilli (19), and because it does not measure a pure effect on persisters, it is not a good indicator of the long-term sterilizing activity of a drug. However, RMP and RPE differ in their binding to plasma protein, their pharmacokinetics, and their penetration into macrophages. EBAs measure the effect these factors have on bacterial killing within cavitary lesions. They are characteristics that remain more or less constant throughout therapy during the change from activity to persistence in the bacterial population, so that the ratios between EBAs for RMP and RPE during the first week should remain throughout, regardless of the rate of growth and therefore the persistence of the bacilli.
The statistical modeling of the EBA results suggests that two smooth curves (Figure 3) can represent the doseeCresponse curve of the EBA 0eC5 for RPE and RMP. These curves are parallel, although the curve for RPE lies a little above the curve for RMP, possibly indicating slightly greater potency. The curve for RPE reaches a maximum at about a dose size of 900 to 1,200 mg RPE, suggesting that these are optimal dose sizes, and that no further increase would be beneficial. A similar flattening out of the EBAeCdose size curve has been found with INH (20). The analysis also implies that results with a single weekly dose of 600 mg RPE should have been at least as good as those with daily 600 mg RMP. This conclusion does not, however, agree with the findings of the two large-scale clinical trials, in which relapse rates were higher with 600 mg RPE than with 600 mg RMP, implying that the bactericidal activity of RPE was less good during the continuation phase (5, 6). How can we account for this disparity
One reason is that the experimental period of the EBA study was 5 days, whereas there were 7 days between the doses of RPE in the clinical trials. With RMP given daily, the EBA 2eC5 was only slightly lower than the EBA 0eC2, a nonsignificant difference of 0.028, whereas the difference of 0.063 between the corresponding means for RPE was larger and significant. Because plasma concentrations after the single dose of RPE peak during the first day after ingestion and then decrease during the next 3 days (Figure 4), the falling concentrations in plasma, and presumably also in lesions, might well account for lower EBAs toward the end of the 5-day period.
A second and more important reason lies in the possible regrowth of persisting bacilli during the last few days of the week, between RPE doses. Such regrowth would not be detected in EBA measurements. Regrowth of sensitive persisters would account for the higher relapse rates in the RPE arms of the two clinical trials. It could also account for the occurrence of rifamycin monoresistance in the relapse cultures of patients with seropositive HIV (7). Strains with rifamycin monoresistance must arise from rifamycin-resistant mutants. However, it is highly improbable that there would be any rifamycin-resistant mutants in the entire bacillary population at the end of the 8-week initial phase of treatment in the trials because of the very low frequency of mutations to resistance. The highest proportion of rifamycin mutants expected at the start of treatment has been calculated, using Luria and Delbruck fluctuation tests, as 1 in 3 x 10eC8 (21). Even if one assumed a very high estimate of 1010 viable bacilli in the entire initial bacterial population, the size of the population would be reduced to below 109 by the initial bactericidal action of INH and would soon be reduced substantially by the sterilizing action of pyrazinamide and INH (19) to bring it well below the number required for a mutation to be present. Further evidence that sufficient mutants would not be available without regrowth is shown by the rarity of the emergence of resistance to INH when continuation phase treatment, preceded by a 2-month, four-drug initial phase, was given with INH alone in a Tanzanian study (22), despite the higher proportion of mutant bacilli resistant to INH pretreatment (3 x 10eC6) than to RMP in a population of sensitive organisms. Without regrowth, no mutants would be available so that a companion drug would not be necessary at all to prevent RMR, except to guard against the occurrence of gross irregularity in drug taking. However, with regrowth, mutations would occur and mutants would multiply, particularly in immunocompromised patients; they might then develop rifamycin monoresistance (7). Failure of the companion drug INH could also lead to relapse, as occurred in a recent clinical trial (23).
However, regrowth would be unlikely to occur if the RPE concentrations estimated in the lesions were increased. The postantibiotic lag has been found to be 5 days after a pulse of 1 e/ml RMP or RPE lasting 96 hours and approximately 1 day after a pulse lasting 24 hours (12). Although a 6-hour exposure to 1 e/ml RPE was not followed by any lag (12), a recent study (24) indicates that lags increase with the higher RPE concentrations found when dose size was increased (Figure 4). The total length of the period of inhibition after a dose of RPE would then be the duration of the pulse for 3 to 4 days (Figure 4), with a peak concentration well above 1 e/ml, and a postantibiotic lag of at least 5 days (Table 6). Thus, bacterial multiplication would be inhibited throughout the week between doses.
We propose that the pulse of RPE must be much smaller than would be estimated from the measured plasma concentrations, presumably because of 98% plasma binding. This has been shown by the close correlation between AUCI/MIC and EBA 0eC5 for both RMP and RPE (Table 5). That only the 2% free portion of RPE is available in lesions has also been indicated by experience with RMP (8). The finding in the present and in a previous study (17) that the therapeutic margin for RMP is 4, whereas it is approximately 20 for INH (see Figure 2) (20), with similar ratios between the peak plasma concentrations and the MICs of both drugs, has suggested that only 20% (4/20) of the plasma concentration is available in lesions. This estimate is in reasonable agreement with the proposition that the proportion of RMP that is active in the lesions is only the unbound 14% of the dose (8). If only the 2% of free RPE is available in tuberculous cavities, the size and duration of the RPE pulse reaching the bacilli would be smaller than what appear from plasma estimates, and the lag after the pulse would also be shorter (Tables 5 and 6, with duration of pulse from Figure 4). The lag after a pulse is difficult to estimate from the data on only three pulse sizes (12), even when supplemented by more recent data on much higher RPE concentrations (24). The estimates suggest that the pulse after a 600-mg dose and the lag after it would last 3 days so that there would be 4 days available for regrowth to occur. Using our pharmacokinetic and postantibiotic lag data, an increase in the dose of RPE to 1,200 mg would increase the duration of the pulse to 2.5 days and also the lag period after it, leaving only 1.5 days for regrowth. However, the size and height of the pulse are substantially increased as the dose size is raised from 600 to 1,200 mg with a probable increase in the lag period, which is then estimated to be at least 3.5 days (Table 6). There would then be a maximum of only 1 day, allowing only one bacterial division each week, insufficient to create rifamycin monoresistance. Furthermore, the EBA 2eC5 increased substantially when the RPE dose was increased from 900 to 1,200 mg, suggesting better bactericidal activity toward the end of the week. We suggest that the balance between the bactericidal effect of a pulse from a 600 mg dose and subsequent regrowth at the end of the treatment cycle may be insufficient to create an overall bactericidal effect in all patients, but the balance might move to a greater overall effect, without appreciable regrowth, when a larger dose is given. Further estimates of the effect of pulse size and height on postantibiotic lag would be helpful.
In conclusion, it seems reasonable to assume that the most effective dose size of RPE would be 1,200 mg and that this dose size is likely to inhibit growth in lesions for approximately 6 of the 7 days between doses. The increase in dose size should be sufficient to prevent the development of rifamycin monoresistance, whether or not an effective companion drug is given. Thus, there seems no necessity to find a better companion drug than INH if the dose size is increased. It does not seem from this study that any benefit in terms of preventing relapse would be likely to accrue from raising the dose size above 1,200 mg, because our study indicates that there would be no further increase in bactericidal action.
Rifapentine EBA Collaborative Study Group:
D. P. Parkin, A. Venter, M. Bester, Stellenbosch University; J. Allen, T. C. P. Mthiyane, Unit for Clinical and Biomedical Research, Medical Research Council, Durban; A. S. Pala, A. Ramjee, I. H. Master, King George V TB Hospital, Durban; J. L. S. Pillay, Umlazi Chest Clinic, Prince Mshiyeni Hospital, Durban; A. Olowolagba, T. Chinappa, Ethekweni Municipality, Durban: S. Siwendu, Brooklyn Hospital for Chest Diseases, Cape Town; P. Smith, University of Cape Town, Cape Town, South Africa.
Acknowledgments
The authors acknowledge the excellent assistance of the staff of the King George V Hospital, Durban, and the staff of Brooklyn Hospital for Chest Diseases, Cape Town.
Members of the Rifapentine EBA Collaborative Study Group are listed at the end of the article.
REFERENCES
Arioli V, Berti M, Carniti G, Randisi E, Rossi E, Scotti R. Antibacterial activity of DL473, a new semisynthetic rifampicin derivative. J Antibiot 1981;34:1026eC1032.
Keung A, Eller MG, McKenzie KA, Weir SJ. Single and multiple dose pharmacokinetics of rifapentine in man: part II. Int J Tuberc Lung Dis 1999;3:437eC444.
Truffot-Pernot C, Grosset J, Bismuth R, Lecoeur H. Activitee de la rifampicine adminutesistreee de manieere intermittente et de la cyclopentyl rifamycine (our DL473) sur la tuberculeuse experimental de la souris. Rev Fr Maladies Respir 1983;11:875eC882.
Dhillon J, Dickinson JM, Guy JA, Ng TK, Mitchison DA. Activity of two long-acting rifamycins, rifapentine and FCE 22807, in experimental murine tuberculosis. Tuber Lung Dis 1992;73:116eC123.
Tam CM, Chan SL, Kam KM, Sim E, Staples D, Sole KM, Al-Ghusein H, Mitchison DA. Rifapentine and INH in the continuation phase of a 6-month regimen. Interim report: no activity of INH in the continuation phase. Int J Tuberc Lung Dis 2000;4:262eC267.
Tuberculosis Trials Consortium. Once-weekly rifapentine and INH versus twice-weekly rifampin and INH in the continuation phase of therapy for drug-susceptible pulmonary tuberculosis: a prospective, randomised clinical trial among HIV-negative persons. Lancet 2002;360:528eC534.
Vernon A, Burman W, Benator D, Khan A, Bozeman L, for the Tuberculosis Trials Consortium. Acquired rifamycin mono-resistance among patients with HIV-related tuberculosis treated with supervised once weekly rifapentine and INH. Lancet 1999;353:1843eC1847.
Mitchison DA. Development of rifapentine: the way ahead. Int J Tuberc Lung Dis 1998;2:612eC615.
Boman G, Ringberger V-A. Binding of rifampicin by human plasma proteins. Eur J Clin Pharmacol 1974;7:369eC373.
Rastogi N, Goh KS, Berchell M, Bryskier A. Activity of rifapentine and its metabolite 25-O-desacetylrifapentine compared with rifampicin and rifabutin against Mycobacterium tuberculosis, Mycobacterium africanum, Mycobacterium bovis and M. bovis BCG. J Antimicrob Chemother 2000;46:565eC570.
Dhillon J, Mitchison DA. Activity in vitro of rifabutin, FCE22807, rifapentine and rifampin against Mycobacterium microti and M. tuberculosis and their penetration into mouse peritoneal macrophages. Am Rev Respir Dis 1992;145:212eC214.
Dickinson JM, Mitchison DA. In vitro properties of rifapentine (MDL473) relevant to its use in intermittent chemotherapy of tuberculosis. Tubercle 1987;68:113eC118.
Chan SL, Yew WW, Porter JHD, McAdam KPWJ, Allen BW, Dickinson JM, Ellard GA, Mitchison DA. Comparison of Chinese and Western rifapentines and improvement of bioavailability by prior taking of various meals. Int J Antimicrob Agents 1994;3:267eC274.
Ellard GA, Greenfield C. A sensitive urine-test method for monitoring the ingestion of isoniazid. J Clin Pathol 1977;30:84eC87.
Mitchison DA, Allen BW, Miller AB. Detection of rifampicin in urine by a simple microbiological assay. Tubercle 1970;51:300eC304.
Mitchison DA, Allen BW, Carrol L, Dickinson JM, Aber BR. A selective oleic acid albuminutes agar medium for tubercle bacilli. J Med Microbiol 1972;5:165eC175.
Sirgel FA, Botha FJH, Parkin DP, Van de Wal BW, Donald PR, Clark PK, Mitchison DA. The early bactericidal activity of rifabutin in patients with pulmonary tuberculosis measured by sputum viable counts: a new method of drug assessment. J Antimicrob Chemother 1993;32:867eC875.
Jayaram R, Gaonkar S, Kaur P, Suresh BL, Mahesh BN, Jayashree R, Nandi V, Bharat S, Shandil RK, Kantharaj E, et al. Pharmacokinetics-pharmacodynamics of rifampin in an aerosol infection model of tuberculosis. Antimicrob Agents Chemother 2003;47:2118eC2124.
Jindani A, Doree CJ, Mitchison DA. The bactericidal and sterilising activities of antituberculosis drugs during the first 14 days. Am J Respir Crit Care Med 2003;167:1348eC1354.
Donald PR, Sirgel FA, Botha FJ, Seifart HI, Parkin DP, Vandenplas ML, Van de Wal BW, Maritz JS, Mitchison DA. The early bactericidal activity of isoniazid related to its dose size in pulmonary tuberculosis. Am J Respir Crit Care Med 1997;156:895eC900.
David HL. Probability distribution of drug-resistant mutants in unselected populations of Mycobacterium tuberculosis. Appl Microbiol 1970;20:810eC814.
Tanzania/British Medical Research Council. A controlled trial of a 4-weekly supplement of rifampicin, pyrazinamide and streptomycin in the continuation phase of a 7-month daily chemotherapy regimen for pulmonary tuberculosis. S Afr Med J 1996;86:960eC965.
Weiner M, Burman W, Vernon A, Benator D, Peloquin CA, Khan A, Weis S, King B, Shah N, Hodge T, and the Tuberculosis Trials Consortium. Low isoniazid concentrations and outcome of tuberculosis treatment with once-weekly isoniazid and rifapentine. Am J Respir Crit Care Med 2003;167:1341eC1347.
Chan C-Y, Au-Yeang C, Yew W-W, Leung C-C, Cheng AFB. In vitro postantibiotic effects of rifapentine, isoniazid, and moxifloxacin against Mycobacterium tuberculosis. Antimicrob Agents Chemother 2004;48:340eC343.(Frik A. Sirgel, P. Bernar)