Whole Blood Bactericidal Activity during Treatment of Pulmonary Tuberculosis

¹Department of Medicine, University of Medicine and Dentistry of New JerseyNew Jersey Medical School, Newark, ²Department of Medicine, Case Western Reserve University, Cleveland, Ohio, and ³Department of Microbiology, University of Arkansas for Medical Sciences, Little Rock; ⁴Núcleo de Doenças Infecciosas, Universidade Federal do Espíritu Santo, Espíritu Santo, Brazil

Received 5 August 2002; revised 26 September 2002; electronically published 6 January 2003.

    The study protocol was approved by the institutional review boards at the Federal University of Espírito Santo, Case Western Reserve University/University Hospitals of Cleveland, and the University of Medicine and Dentistry of New Jersey. Written informed consent was obtained from subjects. The human experimentation guidelines of the US Department of Health and Human Services and those of the authors' institutions were followed in conducting the clinical research.
     Financial support: National Institutes of Health (grant NO1-AI45244).
     Reprints or correspondence: Dr. Robert S. Wallis, Dept. of Medicine, University of Medicine and Dentistry of New JerseyNew Jersey Medical School, 185 S. Orange Ave., MSB I-503, Newark, NJ 07103 ().

     The worldwide eradication of tuberculosis (TB) has been slowed by the requirement that treatment be continued for many months after symptoms have resolved to prevent relapse. The last major advances in TB therapy occurred >30 years ago, with the introduction of regimens that included rifampin and pyrazinamide [1, 2]. These drugs, which permitted the duration of treatment to be shortened to 6 months from 12 or 18 months, are now characterized as "sterilizing," meaning they hasten sputum culture conversion and decrease the risk of relapse. New, highly sterilizing drugs will be required if the duration of TB treatment is to be shortened further. At present, however, neither the sterilizing activity of new drugs nor the risk of relapse in individual patients can be predicted by simple measures such as MIC measurement or drug-level monitoring [3, 4]. To fill this gap, researchers have sought other predictors of relapse, of which sputum culture status after 8 weeks of treatment has been the most fully evaluated [57]. Other potential early indicators include monitoring sputum concentrations of Mycobacterium tuberculosis antigen 85 and serial determination of sputum colony-forming unit (cfu) counts beyond the first few days of TB treatment [810].

     The requirement that TB treatment be prolonged is thought to reflect the resistance to killing exhibited by a subpopulation of nonreplicating bacilli [11]. Neither the anatomic localization of these persisting organisms nor the biochemical signals responsible for their cessation of replication are known with certainty. Several in vitro models have been proposed to study this phenomenon, but it has not yet been studied in macrophages, an omission due in part to the limited capacity of isolated macrophages to restrict mycobacterial growth. The lack of data in this regard is significant, given the recognized role of the macrophage model in the TB drug discovery process and the potential role of macrophages as a site of persisting nonreplicating infection in vivo.

     We previously have described a whole blood model of intracellular TB infection that may be useful in this respect [12, 13]. Mycobacteria added to whole blood culture rapidly undergo phagocytosis and remain intracellular for the duration of these 72-h cultures. Other components of the immune system interact with infected monocytes in whole blood culture as they do in vivo, conferring additional growth-inhibitory activity, beyond that exerted by isolated macrophages. During TB treatment, drug levels in the whole blood cultures mirror those in the circulation at the time of phlebotomy; bacillary killing in these cultures, therefore, reflects the combined influences of host immunity, strain virulence, and drug effects. We have used the term "whole blood bactericidal activity" (WBA) to describe the activity expressed in these cultures, following the nomenclature of Ison et al. [14]. In most infections, the fractional rate of bacterial killing is dependent on drug concentration. Because drug levels vary throughout the dosing interval, the most accurate assessment of killing is one that measures its total extent throughout the dosing interval. The analyses in this report, therefore, include 2 new measures: total WBA per dose (the area under the killing curve throughout the dosing interval, expressed as log₁₀ cfu-days), and cumulative WBA throughout treatment (the sum of the effects of each individual dose). These parameters were compared with 2 recognized predictors of TB relapse: 8-week sputum culture conversion and serial sputum bacillary counts during the first month of treatment.

METHODS

     Recruitment, initial evaluation, and treatment.     Human immunodeficiency virus type 1 (HIV-1)seronegative patients 1860 years old who had newly diagnosed initial episodes of sputum smearpositive pulmonary TB were recruited from the outpatient TB clinic of the Hospital Universitario Cassiano Antonio de Moraes, a large public teaching hospital in Vitória, Brazil. After giving informed consent for study participation and HIV testing, patients were evaluated by medical history and physical examination, chest x-ray, sputum smear, sputum culture, drug-susceptibility testing, complete blood count, chemistry panel, and stool collection for ova and parasite examination.

     The radiographic extent of disease was determined using published criteria [15]. Treatment consisted of daily doses of isoniazid, rifampin, ethambutol, and pyrazinamide for 60 days, followed by daily doses of isoniazid and rifampin for 120 days. Doses of these 4 drugs were (as specified by the Brazilian TB Control Program, according to body mass), respectively, 300, 300, 600, and 1000 mg for subjects weighing <35 kg; 300, 450, 800, and 1500 mg for subjects weighing 3545 kg; and 400, 600, 1200, and 2000 mg for subjects weighing >45 kg. Treatment was ambulatory and unsupervised, except for those doses for which WBA was determined, which were administered in the hospital after overnight fasting. Doses were administered with a small volume of water. Compliance was monitored by interview and testing of urine samples obtained at regularly scheduled clinic visits for the presence of isonicotinic acid (Mycodyn Uritec; Symcon). Subjects were examined at 2 weeks and 4 weeks after the initiation of anti-TB treatment; monthly, during the remainder of treatment; and at 6 months after completion of therapy.

     Sputum processing and culture.     Sputum was processed as described elsewhere [8], except for the substitution of 0.1% dithiothreitol for N-acetyl cysteine. This modification does not affect sputum microbiology [16] and permitted coenrollment of our subjects in a separate study of sputum induction. Sputum was cultured on supplemented Middlebrook 7H10 agar biplates and Lowenstein-Jensen slants. A BACTEC system (Becton Dickinson) was used for speciation and drug-susceptibility testing [17, 18]. Subjects were to be excluded if resistance to rifampin, ethambutol, or pyrazinamide was found, but no such resistance occurred.

     Determination of sputum bacillary counts.     Quantitative cultures for M. tuberculosis were performed by preparation of serial 10-fold dilutions of the resuspended sediment, using 0.25% Tween 80 in 0.9% saline. Aliquots of 60 L of each dilution were inoculated on Middlebrook 7H10 agar biplates supplemented with oleic acidalbumindextrosecatalase. Plates were sealed, incubated at 37°C in 5%10% CO₂ in air, and examined after 2, 3, 4, and 6 weeks. Colonies were counted on plates with dilutions yielding 1050 visible colonies. The results were expressed as log₁₀ cfu per milliliter of undiluted sputum. Cultures without growth were scored as 0. The mean of the log₁₀ counts of 2 specimens on a single day was taken as the actual value. The rate of decrease in log cfu per milliliter per day was determined by least-squares regression analysis of log cfu values on days 0, 14, and 28. In this analysis, day-28 cultures without growth were censored if day-14 cultures for that subject were also without growth. This approach differs somewhat from that of Brindle et al. [10], in which counts were obtained sequentially on days 2, 7, 14, and 28, and specimens without growth were excluded.

     WBA determination.     WBA was determined as described elsewhere [12, 13]. Readers who wish to review the protocol or to obtain copies of the custom software used to calculate the change in log cfu using BACTEC (R.S.W., University of Medicine and Dentistry of New JerseyNew Jersey Medical School, Newark) are referred to . In brief, cultures consisted of equal volumes of blood and tissue culture medium and were of 72-h duration. The BACTEC 460 TB system was used both to prepare inocula and to determine the extent of killing. For each batch of stock for each isolate, a standard curve was prepared that indicated the relationship between log inoculum volume and days to positivity (DTP). The inoculum volume selected for whole blood culture was that calculated to be positive in 4.5 days, according to its standard curve. This approach results in inocula with 5 × 10⁴ to 10⁵ cfu [13]; the extent of killing is not affected by the extent of variation in inoculum size [12]. At the start of each whole blood culture, a control culture was performed, in which the same volume of mycobacterial stock was placed directly into BACTEC culture. At the conclusion of the whole blood culture, host cells were disrupted by hypotonic lysis. Bacilli were recovered by centrifugation and were inoculated into BACTEC medium. Growth indices of both BACTEC cultures were monitored daily, and the number of DTP was noted. Killing was calculated using the formula  cfu = (final) - (initial), where final and initial represent the volumes of the stock culture indicated by its standard curve to have DTP values equal to those of the completed whole blood culture and its inoculum, respectively. This formula is mathematically equivalent to log₁₀(final/initial), a growth ratio without units. We have used the unit "log₁₀ cfu" for clarity, recognizing that representation without units may also be appropriate.

     At weeks 8 and 12, blood specimens for WBA were obtained from each subject immediately before dose administration and 2 and 6 h afterward. Each subject was studied using that individual's isolate and the drug-susceptible M. tuberculosis clinical isolate MP-28 [13]. The total WBA for the dosing interval (in log₁₀ cfu-days) was calculated for each subject as AUC₂₄, using a 5-parameter peak (Weibull) function (SigmaPlot; SPSS) [19]. The function, described by the equation 



(where e is a mathematical constant with the approximate value 2.7183), is particularly appropriate for modeling the pharmacokinetics of oral drugs [2024]. Initial values for a, b, c, x₀, and y₀ (which determine the shape of the curve) were selected on the basis of the range of values in the data set, according to SigmaPlot default parameters. Iterative curve fitting to determine the optimal values of these parameters was performed using the Marquardt-Levenberg algorithm [25]. Steady-state conditions were assumed (i.e., WBA values immediately before successive daily doses were assumed to be equal). Representative curves, using mean 0-, 2-, and 6-h values, are shown in  to illustrate the characteristics of the curve-fitting process. Readers may examine the relationship of the curves to the primary data on-line at . The cumulative WBA of the intensive phase of treatment was calculated as 60 times the total WBA on week 8; that of the continuation phase was calculated as 120 times that obtained on week 12. The cumulative WBA of the entire period of treatment was calculated as the sum of these 2 values.

fig.ommitted

Figure 1.        Population kinetics of whole blood bactericidal activity (WBA) at weeks 8 and 12 of tuberculosis treatment consisting of daily HREZ and HR, respectively. Each subject was tested using that individual's own isolate. Symbols indicate mean values at 0, 2, and 6 h after the dose; values at 24 h were assumed to equal those at 0 h. The cumulative effect of a single dose of treatment was calculated for each subject as the area under the killing curve at 24 h, using a peak (Weibull) function, as illustrated. Cumulative WBA was greater during the intensive phase of treatment than during the continuation phase (-2.32 vs. -1.67 log₁₀ cfu-days, respectively; P < .001, by paired Student's t test). E, ethambutol; H, isoniazid; R, rifampin; Z, pyrazinamide.

     Sterilization occurred in 25% of the whole blood cultures, primarily at 2 h after the dose. These were scored as -6, one-half log greater than the greatest measured effect. The overall conclusions of the study were not affected by the use of other values (from -5 to -8) because of the limited influence of 2-h values on cumulative WBA. The final WBA evaluation occurred at week 28, 2 weeks after treatment ended. Positive values indicate growth; negative values, killing.

     Statistical analysis.     Student's t test or Pearson's product correlation was used, unless otherwise specified (SigmaStat). Two-tailed tests were used throughout.

RESULTS

     Subjects.     From 7 June 2000 to 21 September 2001, 80 individuals were screened for possible participation in this study, of whom 30 were excluded. After enrollment, 14 additional subjects either were removed from the study or had their results censored. The reasons for exclusion or removal are listed in . Twelve subjects were excluded or had treatment discontinued because of an inability to produce sputum or because the results of sputum acid-fast bacilli smears were negative, whereas issues related to whole blood culture prevented the evaluation of only 2 subjects (delayed propagation of the isolate in one case, and failure of growth after thawing on week 12 in a second).

fig.ommitted

Table 1.          Reasons for exclusion or discontinuation of treatment among 80 patients with pulmonary tuberculosis (TB) who were screened for participation in a study of bactericidal activity against Mycobacterium tuberculosis during TB treatment.

     Of the remaining 36 fully evaluable subjects, 8 had 1 sputum culture positive for M. tuberculosis on or after the eighth week of treatment. As shown in , these 8 individuals did not significantly differ from other subjects in any of several baseline characteristics reported to influence the risk of TB relapse [5, 7, 2628], nor did they differ in compliance with treatment (assessed by the proportion of urine specimens that tested positive for isonicotinic acid; 96% of subjects with positive cultures on or after week 8 were compliant, vs. 94% of subjects for whom sputum cultures were negative). All subjects ultimately converted to sputum culture negative during treatment. As of 21 September 2002, all subjects had reached the final clinical evaluation, 6 months after completion of therapy, and no relapses had occurred.

fig.ommitted

Table 2.          Demographic characteristics and laboratory values at baseline for 36 subjects enrolled in a study of bactericidal activity against Mycobacterium tuberculosis in whole blood during tuberculosis treatment.

     Sputum microbiology.     Before treatment, the mean viable sputum bacillary count (±SD) was 5.74 ± 0.7 log₁₀ cfu/mL. After treatment was initiated, the mean bacillary count decreased progressively, to 2.66 ± 1.3 on week 2 and 1.09 ± 1.1 on week 4. The rate of decrease (slope) of log₁₀ bacillary counts during the 4-week interval differed according to 8-week culture status (-0.14 ± 0.04 for subjects with positive cultures vs. -0.19 ± 0.06 log₁₀ cfu/mL/day for subjects with negative cultures; P = .035).

     WBA determination.     WBA was measured during treatment (on weeks 8 and 12) and after completion of treatment (on week 28). Each subject was studied using that individual's own isolate. Because propagation and standardization of these isolates requires 46 weeks, these studies could not be performed before the initiation of treatment. However, several alternative measures were used to estimate the extent of growth of these isolates in the absence of chemotherapy. At week 28 (after completion of treatment), the mean net growth (±SD) in cultures was only 0.12 ± 0.3 log₁₀ cfu, which was not significantly different from zero and was substantially less than the 0.81 log growth expected from virulent isolates during 72-h culture in whole blood or macrophages of healthy individuals [13, 29]. Two additional findings indicate that similar immune antimycobacterial activity exists at earlier time points (before and during treatment). Thirty subjects were studied both before and after completion of treatment, using the attenuated M. tuberculosis strain H₃₇Ra; killing of this isolate did not differ at these 2 time points (-0.63 ± 0.3 vs. -0.52 ± 0.5 log₁₀ cfu, respectively; P = .4, by paired Student's t test). In addition, values obtained before dose administration on week 12 were used to estimate the control of mycobacterial growth in the absence of chemotherapy. This assessment was possible on week 12 because isoniazid and rifampin have little residual intracellular effect 24 h after administration [12]. These values (-0.23 ± 0.5 log₁₀ cfu, using the patient's own isolate) also did not differ significantly from zero or from those measured after completion of therapy. Thus, in the absence of chemotherapy, patient blood cells exerted bacteriostatic but not bactericidal activity. This indicates potential suitability as a model of nonreplicating intracellular infection.

     Bactericidal activity was evident in a time-dependent fashion after each dose of oral chemotherapy. As shown in  and , the maximum effect (-5 to -6 logs) was observed 2 h after the dose, the point at which most TB drugs reach maximum levels in blood. At the 8-week time point, however, 7 of the 36 subjects showed maximum WBA 6 h after the dose, which apparently indicates delayed drug absorption. In contrast, no instances of delayed drug effect occurred at week 12 (P = .02, by ² test). Total WBA per dose at week 8 was strongly correlated with that at week 12 (r = 0.511; P = .002). The mean total WBA (±SD) was greater at week 8, during the intensive phase of treatment, than during the continuation phase (-2.32 ± 0.8 vs. -1.67 ± 0.9, respectively; P < .001, by paired Student's t test). We have reported 0-, 2-, and 6-h WBA values for pyrazinamide of -0.73, -0.68, and -0.65 log₁₀ cfu, respectively, in a healthy volunteer [12]. These values yield a total effect of -0.69 log₁₀ cfu-days/dose, a value very close to the difference between the intensive and continuation phases (-0.65 log₁₀ cfu-days). Because we have demonstrated elsewhere that ethambutol shows only bacteriostatic activity in the model, it is likely that the change in total WBA from week 8 to week 12 represents the effect of pyrazinamide. These findings indicate the internal consistency and reproducibility of the method.

fig.ommitted

Table 3.          Whole blood bactericidal activity in 36 patients receiving treatment for pulmonary tuberculosis, measured by testing each subject with his or her own isolate.

     A full course of treatment in this study consisted of daily doses of isoniazid, rifampin, ethambutol, and pyrazinamide for 60 days, followed by daily doses of isoniazid and rifampin for 120 days. In the present study, the cumulative WBA of a full course of treatment, calculated as the sum of individual doses, was -339 ± 142 log₁₀ cfu-days (mean ± SD). We have reported elsewhere 0-, 2-, and 6-h WBA values for the combination of levofloxacin, pyrazinamide, and ethambutol against multidrug-resistant M. tuberculosis isolates (-0.1, -1.24, and -1.20 log₁₀ cfu, respectively) [12]. The total WBA per dose of this combination of is -0.60 log₁₀ cfu-days, markedly less than standard treatment for drug-susceptible disease. However, its daily administration for 18 months would yield a cumulative effect of -323 log₁₀ cfu-days over the entire course of treatment, a value very close to that of 6 months of standard therapy for drug-susceptible M. tuberculosis.

     Relationship of WBA to sputum culture results.     The microbiologic response to treatment was assessed in this study as sputum culture status after 8 weeks of treatment and as the rate of decrease (slope) of log₁₀ sputum bacillary counts during the first 4 weeks of treatment. The relationship of cumulative WBA to these 2 parameters is shown in  and . Subjects whose sputum M. tuberculosis cultures converted to negative by the eighth week of treatment showed superior cumulative WBA throughout treatment and during the continuation phase (P = .04 for both periods); trends were apparent in values at other time points. Cumulative WBA throughout treatment also correlated with the rate of decrease of sputum bacillary counts (r = 0.393; P = .018). These findings indicate that subjects whose sputum cultures cleared more rapidly had greater bactericidal activity against isolates in vitro.

fig.ommitted

Figure 2.        Relationship between sputum culture results and cumulative whole blood bactericidal activity (WBA) during tuberculosis treatment. Each subject was studied using that individual's isolate. The Y-axis indicates the rate of change of viable sputum bacillary counts during the first 4 weeks of treatment. Cumulative WBA correlated with the sputum colony-forming units slope (r = 0.39; P = .018) and was greater in subjects whose cultures converted from positive for Mycobacterium tuberculosis to negative (-365 vs. -250 log₁₀ cfu-days; P = .04). Dotted lines indicate 95% confidence interval.

     In contrast, measures of immune control of the patient isolates, either after completion of treatment on week 28 or before administration of the dose on week 12, did not differ according to 8-week sputum culture status (P = .59 and P = .12, respectively), nor did they correlate with sputum cfu slope (P = .96 and P = .31).

     Strain specificity.     Twenty-nine subjects were also studied using the clinical TB isolate MP-28. This isolate differed significantly from the patients' own isolates in that it was more readily controlled by host immune mechanisms (mean WBA [±SD] at week 28, -0.29 ± 0.3 vs. 0.13 ± 0.3, respectively; before administration of the dose on week 12, -0.64 ± 0.4 vs. -0.23 ± 0.5; P < .001 for both, by paired Student's t test). MP-28 also was more readily killed during TB treatment than were other clinical isolates (mean cumulative WBA [±SD] throughout treatment, -404 ± 112 vs. -339 ± 142 log₁₀ cfu-days, respectively; P < .001, by paired Student's t test). Killing of MP-28 was correlated with sputum cfu slope (r = 0.396; P = .034). However, it did not differ according to 8-week sputum culture status, as indicated in . Thus, killing of MP-28 in vitro was associated with only 1 of the 2 sputum markers of relapse risk.

fig.ommitted

Table 4.          Whole blood bactericidal activity in 29 patients receiving treatment for pulmonary tuberculosis, measured by testing with the clinical isolate MP-28.

DISCUSSION

     The main finding of this study is that of inferior bactericidal activity in whole blood cultures of sputum from patients with TB who had delayed sputum sterilization. Before this report, all treatment-related predictors of TB relapse involved sputum. An early report by Aber and Nunn [5] established a relationship between sputum culture status at 3 months after initiation of treatment and relapse risk. An extensive review by Mitchison in 1993 [6] established that regimens with higher relapse rates were less likely to sterilize sputum by the eighth week of treatment. Recent data from Study 22 of the Centers for Disease Control and Prevention Tuberculosis Trials Consortium confirm and extend this observation [7]. In that report, one-half of all relapses occurred in individuals who had positive sputum cultures at 8 weeks; the risk of relapse in that subset was 6 times that among culture-negative subjects. The present finding, therefore, indicates that the whole blood model may have a role as the first in vitro assessment of sterilization in TB. This is supported by the additional correlation of WBA with sputum cfu slope during the first 4 weeks of treatment; such evaluations appear to provide an assessment of sterilizing activity superior to that of early bactericidal activity by emphasizing effects on bacilli that persist later during treatment [10]. All 3 markers may be valuable as indicators of the potential sterilizing activity of new drugs; however, when used singly, none is likely to have predictive power sufficient to form the basis for clinical decisions. For example, the positive predictive value of 8-week culture positivity in Study 22 was only 18% [7]. Further studies are therefore warranted to determine whether the combined use of blood and sputum markers in tandem might increase the accuracy of predictions of TB relapse and improve the reliability of estimates of the sterilizing activity of new TB drugs or regimens.

     Some caution must be exercised when comparing 2 surrogate markers in lieu of clinical end points, because the potential exists for both surrogates to be influenced by factors unrelated to outcome. For example, 2 tests performed on sputum may both be affected by sputum viscosity and cough strength, which have no bearing on risk of relapse. It is significant that the whole blood model shares no recognized confounding factors with the sputum microbiologic markers with which it was compared. Indeed, its inability to reflect anatomic factors known to affect relapse risk (such as the presence of cavitary disease) increases the biologic importance of the correlation we observed.

     For other bacterial infections, such as bronchitis and pneumonia, bacterial killing and clinical outcome can be predicted by pharmacokinetic parameters such as the ratio of the maximum concentration of drug (C_max) to MIC, time above MIC, and the ratio of AUC₂₄ to MIC [3034]. No such predictive models exist for TB, despite the frequent occurrence of low blood levels of key drugs, particularly rifampin [35, 36]. Two large studies illustrate this paradox. Indonesian researchers recently reported a study in which only 2 of 62 patients with TB had peak rifampin blood levels in the recommended range of 824 g/mL [4]. More than one-half had peak levels <4 g/mL and ratios of C_max to MIC <10, which are thought to be subtherapeutic [37]. However, they found that these parameters had no influence on 2-month sputum culture conversion or cure. The second study, in which 25 relapses occurred in a cohort of 188 Florida patients with TB, found no association of relapse with lower-than-expected levels of isoniazid or rifampin or with any other clinical, microbiologic, or epidemiologic parameters [3]. The authors concluded that the factors that determine the risk of relapse remained to be defined.

     The reasons for the failure of these studies are not clear. One possible explanation lies in the ability of M. tuberculosis to adapt to hostile environmental conditions by entering a nonreplicating state. Researchers have used several in vitro models of nonreplicating persistence to examine its biologic basis, including hypoxia, nutrient deprivation, and prolonged drug exposure [3842]. Some of these models appear to show clinical relevance. We found, for example, that phenotypic tolerance in a model of prolonged drug exposure was associated with increased risk of relapse [42]. Tolerance was not drug specific and was unrelated to drug concentration and MIC. These characteristics may be anticipated in nonreplicating bacilli and may explain the failure of pharmacokinetic models based on MIC to predict relapse. However, some of the findings in these models appear to be model specific and not readily generalized. Metronidazole, for example, shows activity under hypoxic conditions but not after nutrient deprivation or in animal models [38,4346]. The activity of rifampin in these models has been reported with similar variability [38, 47]. These differences indicate that it may be difficult to develop simple in vitro models that accurately reflect the complex environment in vivo during TB treatment.

     The ability of whole blood culture to serve as a "physiologic" model of nonreplicating growth may have contributed to its accuracy as a surrogate marker for relapse. The relative contributions of various immune mechanisms to this control of growth are not yet known. Similar immune activation of blood cells in vivo may account for the relative rarity of mycobacteremia and disseminated disease in adult TB; these mechanisms have been reported to be impaired in whole blood cultures of children and HIV-infected patients with TB [48], among whom dissemination is more common. It is likely that the restriction of mycobacterial replication by host immune responses in TB affects profiles of mycobacterial gene expression and metabolic activity, enhancing the activity of some drugs, and diminishing that of others. Further studies will be required to test this hypothesis.

     The finding that WBA and sputum bacillary clearance were most closely related when patients were studied using their own isolates indicates that unique strain characteristics have a potentially important role in the model for phenotypic drug tolerance. Further studies also are warranted to elucidate the physiology of tolerance and the nonreplicating state more generally, so that these may be specifically targeted by new drugs and therapeutic regimens.

     TB clinicians have long held that a "weak" regimen may be strengthened by compensatory prolongation. This view is reflected in current recommendations for prolonged treatment of multidrug-resistant TB and in recent proposals for prolongation of treatment for patients with drug-susceptible disease and delayed sputum culture conversion. The present observation that standard regimens and those aimed at multidrug-resistant organisms are equal when cumulative WBA throughout treatment is assessed, even though the effects per dose differ substantially, supports this concept. Further studies are warranted to determine whether cumulative WBA can be used to determine the optimal duration of TB chemotherapy, as well as optimal composition. Studies also are needed to determine the accuracy of the Weibull equation in modeling cumulative effect [22, 24, 49].

     In summary, bactericidal activity in whole blood culture was found to correlate with clearance of viable bacilli from sputum during treatment of pulmonary TB. This indicates that this marker may have a role in the evaluation of new drugs for TB and even in the management of TB in selected patients. Further studies to compare the accuracy of these markers will require clinical trials with sufficient power to study relapse.

References

1. 

Fox W, Mitchison DA. Short-course chemotherapy for pulmonary tuberculosis. Am Rev Respir Dis 1975; 111:32553.

2. 

Fox W, Mitchison DA. Short-course chemotherapy for pulmonary tuberculosis . Am Rev Respir Dis 1975; 111:8458.

3. 

Narita M, Hisada M, Thimmappa B, et al. Tuberculosis recurrence: multivariate analysis of serum levels of tuberculosis drugs, human immunodeficiency virus status, and other risk factors. Clin Infect Dis 2001; 32:5157.

4. 

van Crevel R, Alisjahbana B, de Lange WC, et al. Low plasma concentrations of rifampicin in tuberculosis patients in Indonesia. Int J Tuberc Lung Dis 2002; 6:497502.

5. 

Aber VR, Nunn AJ. Short term chemotherapy of tuberculosis: factors affecting relapse following short term chemotherapy. Bull Int Union Tuberc 1978; 53:27680.

6. 

Mitchison DA. Assessment of new sterilizing drugs for treating pulmonary tuberculosis by culture at 2 months . Am Rev Respir Dis 1993; 147:10623.

7. 

The Tuberculosis Trials Consortium. Rifapentine and isoniazid once a week versus rifampin and isoniazid twice a week for treatment of drug-susceptible pulmonary tuberculosis: a randomized clinical trial. Lancet 2002; 360:52834.

8. 

Wallis RS, Perkins M, Phillips M, et al. Induction of the antigen 85 complex of Mycobacterium tuberculosis in sputum: a determinant of outcome in pulmonary tuberculosis treatment. J Infect Dis 1998; 178:111521.

9. 

Wallis RS, Perkins M, Phillips M, et al. Predicting the outcome of therapy for pulmonary tuberculosis. Am J Respir Crit Care Med 2000; 161:107680.

10. 

Brindle R, Odhiambo J, Mitchison DA. Serial counts of Mycobacterium tuberculosis in sputum as surrogate markers of the sterilising activity of rifampicin and pyrazinamide in treating pulmonary tuberculosis. BMC Pulm Med 2001; 1:2.

11. 

Mitchison DA. The action of antituberculosis drugs in short-course chemotherapy. Tubercle 1985; 66:21925.

12. 

Wallis RS, Palaci M, Vinhas S, et al. A whole blood bactericidal assay for tuberculosis. J Infect Dis 2001; 183:13003.

13. 

Cheon SH, Kampmann B, Hise AG, et al. Bactericidal activity in whole blood as a potential surrogate marker of immunity after vaccination against tuberculosis. Clin Diagn Lab Immunol 2002; 9:9017.

14. 

Ison CA, Anwar N, Cole MJ, et al. Assessment of immune response to meningococcal disease: comparison of a whole-blood assay and the serum bactericidal assay. Microb Pathog 1999; 27:20714.

15. 

Falk A, O'Connor JB, Pratt PC, et al. Classification of pulmonary tuberculosis. In: Diagnostic standards and classification of tuberculosis. 12th ed. New York: National Tuberculosis and Respiratory Disease Association, 1969:6876.

16. 

Bhowmik A, Seemungal TA, Sapsford RJ, et al. Comparison of spontaneous and induced sputum for investigation of airway inflammation in chronic obstructive pulmonary disease. Thorax 1998; 53:9536. 

17. 

Siddiqi SH, Hwangbo CC, Silcox V, et al. Rapid radiometric methods to detect and differentiate Mycobacterium tuberculosis/M. bovis from other mycobacterial species. Am Rev Respir Dis 1984; 130:63440.

18. 

Siddiqi S. Radiometric (BACTEC) tests for slowly growing mycobacteria. In: Isenberg HD, ed. Clinical microbiology procedures handbook. Washington, DC: American Society for Microbiology, 1992:5.14.

19. 

Press WH, Flannery BP, Teukolsky SA, et al. Numerical recipes: the art of scientific computing. Cambridge, UK: Cambridge University Press, 1986.

20. 

Ahlstrom H, Alvero J, Alvero R, et al. Pharmacokinetics of bambuterol during oral administration to asthmatic children. Br J Clin Pharmacol 1999; 48:299308.

21. 

Allen GD, Coates PE, Davies BE. On the absorption of clavulanic acid. Biopharm Drug Dispos 1988; 9:12736.

22. 

Heikkila HJ. New models for pharmacokinetic data based on a generalized Weibull distribution. J Biopharm Stat 1999; 9:89107.

23. 

Rietbrock S, Merz PG, Fuhr U, et al. Absorption behavior of sulpiride described using Weibull functions. Int J Clin Pharmacol Ther 1995; 33:299303.

24. 

Piotrovskii VK. The use of Weibull distribution to describe the in vivo absorption kinetics. J Pharmacokinet Biopharm 1987; 15:6816.

25. 

Marquardt DW. An algorithm for least squares estimation of parameters. J Soc Indust Appl Math 1963; 11:43141.

26. 

Mitchison DA, Nunn AJ. Influence of initial drug resistance on the response to short-course chemotherapy of pulmonary tuberculosis. Am Rev Respir Dis 1986; 133:42330.

27. 

Tripathy SP. Relapse in tuberculosis. Ind J Tuberc 1981; 28:4557.

28. 

Tam CM, Chan SL, Lam CW, et al. Rifapentine and isoniazid in the continuation phase of treating pulmonary tuberculosis: initial report. Am J Respir Crit Care Med 1998; 157:172633.

29. 

Silver RF, Li Q, Ellner JJ. Expression of virulence of Mycobacterium tuberculosis within human monocytes: virulence correlates with intracellular growth and induction of tumor necrosis factor alpha but not with evasion of lymphocyte-dependent monocyte effector functions. Infect Immun 1998; 66:11909.

30. 

Meinl B, Hyatt JM, Forrest A, et al. Pharmacokinetic/pharmacodynamic predictors of time to clinical resolution in patients with acute bacterial exacerbations of chronic bronchitis treated with a fluoroquinolone. Int J Antimicrob Agents 2000; 16:27380.

31. 

Ambrose PG, Grasela DM, Grasela TH, et al. Pharmacodynamics of fluoroquinolones against Streptococcus pneumoniae in patients with community-acquired respiratory tract infections. Antimicrob Agents Chemother 2001; 45:27937.

32. 

Moise PA, Forrest A, Bhavnani SM, et al. Area under the inhibitory curve and a pneumonia scoring system for predicting outcomes of vancomycin therapy for respiratory infections by Staphylococcus aureus. Am J Health Syst Pharm 2000; 57(Suppl 2):S49.

33. 

Highet VS, Forrest A, Ballow CH, et al. Antibiotic dosing issues in lower respiratory tract infection: population-derived area under inhibitory curve is predictive of efficacy. J Antimicrob Chemother 1999; 43(Suppl A):5563.

34. 

Goss TF, Forrest A, Nix DE, et al. Mathematical examination of dual individualization principles. II. The rate of bacterial eradication at the same area under the inhibitory curve is more rapid for ciprofloxacin than for cefmenoxime. Ann Pharmacother 1994; 28:8638.

35. 

Pillai G, Fourie PB, Padayatchi N, et al. Recent bioequivalence studies on fixed-dose combination anti-tuberculosis drug formulations available on the global market. Int J Tuberc Lung Dis 1999; 3(Suppl 3):S30916.

36. 

Taylor B, Smith PJ. Does AIDS impair the absorption of antituberculosis agents? Int J Tuberc Lung Dis 1998; 2:6705.

37. 

Douglas JG, McLeod MJ. Pharmacokinetic factors in the modern drug treatment of tuberculosis. Clin Pharmacokinet 1999; 37:12746.

38. 

Betts JC, Lukey PT, Robb LC, et al. Evaluation of a nutrient starvation model of Mycobacterium tuberculosis persistence by gene and protein expression profiling. Mol Microbiol 2002; 43:71731.

39. 

Dick T. Dormant tubercle bacilli: the key to more effective TB chemotherapy? J Antimicrob Chemother 2001; 47:1178.

40. 

Dick T, Lee BH, Murugasu-Oei B. Oxygen depletion induced dormancy in Mycobacterium smegmatis. FEMS Microbiol Lett 1998; 163:15964.

41. 

Michele TM, Ko C, Bishai WR. Exposure to antibiotics induces expression of the Mycobacterium tuberculosis sigF gene: implications for chemotherapy against mycobacterial persistors. Antimicrob Agents Chemother 1999; 43:21825.

42. 

Wallis RS, Patil S, Cheon SH, et al. Drug tolerance in Mycobacterium tuberculosis. Antimicrob Agents Chemother 1999; 43:26006.

43. 

Brooks JV, Furney SK, Orme IM. Metronidazole therapy in mice infected with tuberculosis. Antimicrob Agents Chemother 1999; 43:12858.

44. 

Dhillon J, Allen BW, Hu YM, et al. Metronidazole has no antibacterial effect in Cornell model murine tuberculosis. Int J Tuberc Lung Dis 1998; 2:73642.

45. 

Lim A, Eleuterio M, Hutter B, et al. Oxygen depletioninduced dormancy in Mycobacterium bovis BCG. J Bacteriol 1999; 181:22526. 

46. 

Wayne LG, Sramek HA. Metronidazole is bactericidal to dormant cells of Mycobacterium tuberculosis. Antimicrob Agents Chemother 1994; 38:20548.

47. 

Herbert D, Paramasivan CN, Venkatesan P, et al. Bactericidal action of ofloxacin, sulbactam-ampicillin, rifampin, and isoniazid on logarithmic- and stationary-phase cultures of Mycobacterium tuberculosis. Antimicrob Agents Chemother 1996; 40:22969.

48. 

Tena NG, Kampmann B, Nicol M, et al. Growth of BCG in whole blood of children is dependent on immune competence [abstract P123]. In: Abstracts of the Fifth International Conference on the Pathogenesis of Mycobacterial Infections (Stockholm). Stockholm: Smittskyddsinstitutet, 2002:168.

49. 

Polli JE. Dependence of in vitroin vivo correlation analysis acceptability on model selections. Pharm Dev Technol 1999; 4:8996.