INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Short-course chemotherapy of tuberculosis has become standardized during the last decade with isoniazid (INH), rifampin (RMP), pyrazinamide, and, usually, ethambutol given in an initial 2-mo intensive phase of therapy often followed by a continuation phase of 4 mo RMP and INH. In patients who do not harbor resistant organisms this regimen will lead to cure without relapse in more than 95% of patients (1). Despite the success of this regimen when it is used under program conditions in developing countries, more than 20% of patients may fail to complete their prescribed 6 mo course of therapy (2). Furthermore, the spread of multi-drug-resistant (MDR) tuberculosis causes concern (3). There is a need for the development and assessment of new antituberculosis agents that might contribute to shortening the length of treatment of patients with initially sensitive organisms and could be effective in those with initially resistant organisms. The fluoroquinolones constitute an important group of new antimycobacterial drugs that are likely to be particularly useful in patients with MDR strains resistant to INH and RMP. Among the fluoroquinolones, ciprofloxacin (CIP) has one of the lowest MICs against Mycobacterial tuberculosis, but it has been little used in treatment (4).

The early bactericidal activity (EBA) of an antituberculosis drug in the lesions of patients with pulmonary tuberculosis may be measured by the daily fall in the number of viable tubercle bacilli in the sputum during the first 2 d of treatment. After the first 2 d, the count of colony-forming units (CFU) falls more slowly as rapidly growing organisms are killed and the variation between the activities of different drugs and different dose sizes of the same drug decreases (5). Furthermore, when a range of dose sizes is tested, one can estimate a drug's therapeutic margin, that is to say, the ratio between the dose size that gives a maximal effect and the size that can be shown just to produce a detectable EBA (6). The therapeutic margin is dependent only on the EBAs found with graded dose sizes. It is an absolute measure of drug activity within tuberculous cavities and thus avoids the guess work inherent in similar estimations of activity made from considerations such as the MIC, the proportion of the drug that is bound to plasma protein, the ability to penetrate into tissues, and known lesional drug concentrations. The therapeutic margin measures the ability of a drug to halt the growth of all bacilli irrespective of the type of lesion and the EBA indicates its ability to kill rapidly multiplying extracellular bacilli found in the walls of cavities during the first 2 d. Because the EBA and the therapeutic margin measure antibacterial activity when the bacterial population is at its greatest, they are also likely to reflect the ability of a drug to prevent the emergence of resistance to a companion drug. In the study reported here the EBA of four dose sizes of CIP was evaluated in patients with smear-positive pulmonary tuberculosis (PTB); these EBAs and the EBA obtained with no drug treatment then allowed calculation of the therapeutic margin. Control groups of patients received 300 mg INH, which has always produced the highest EBA measured.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

The majority of the patients admitted to the study were of mixed race (black with additional Malay, European, and Khoi-San parentage) as previously described (7), and a minority were black. They had newly diagnosed, previously untreated PTB and were producing sputum positive on direct microscopy for acid-fast bacilli (AFB). They were 18 yr of age or older, weighed 40 to 60 kg, and agreed to be admitted to hospital for a period of 3 d. Patients not eligible for study included those with a history of allergy to other carboxyquinolone derivatives or INH, those in poor general condition or having other serious complicating diseases, those pregnant or lactating, and those unable to produce an adequate volume of sputum (5 ml or more). The study was carried out in three phases. In Phase I, 60 patients complying with the inclusion criteria were randomized to receive CIP in daily doses of 250 mg (CIP 250), 500 mg (CIP 500), or 1,000 mg (CIP 1000) or INH 300 mg (INH 300-1) or no drug (Nil). Once safety data for the higher dose size of 1,500 mg CIP were available, a further nine patients were given CIP 1,500 mg (CIP 1500-1) in Phase II. Finally, a further 24 patients were randomly allocated to either CIP 1,500 mg (CIP 1500-2) or to INH 300 mg (INH 300-2) in Phase III.

A 16-h collection of sputum was made between 4:00 P.M. on the day of admission to the study and 8:00 A.M. the next day (S1). After the first collection was complete, the first drug dose was administered at least 60 min before breakfast, and this procedure was repeated twice more on subsequent days to give S2 and S3 sputum collections. The second of the daily drug doses was given after the S2 collection. As soon as the study protocol had been completed with the S3 collection, the patients were started on treatment with a combined preparation of INH, RMP, and pyrazinamide, as was then standard practice in South Africa. The study protocol was approved by the Ethical Committee of the Medical Faculty of the University of Stellenbosch.

Microbiologic Methods

Sputum in the S1, S2, and S3 collections was examined conventionally by direct smear, culture, and sensitivity testing. CFU counts on the sputum collections were carried out as described previously (7), except that after mixing a sample of 2 ml of homogenized sputum from each collection with 3 ml 1:10 dithiothreitol (Sputolysin; Hoechst, Cape Town, South Africa), 20 µl of the dilutions were set up without preliminary centrifugation on thirds of triplicate plates of selective 7H10 medium. Drug resistance did not develop during the 3 d of the study.

Statistical Methods

The EBA was calculated from the mean daily fall in CFU/ml sputum during the 2-d period of drug administration (S3 log₁₀ CFU/ml  S1 log₁₀ CFU/ml sputum) (5, 7). The means, their 95% confidence limits, and the standard deviations (SD) for each treatment group were calculated using EXCEL worksheets (Microsoft Corp.). Differences between the EBAs of treatment groups and multiple regression analysis was done using EPIINFO 6 (8). The dosage effect of CIP in Phase I was determined with a nonparametric trend test across ordered groups (9) in the STATA package (release 4, Stata Statistical Software, TX). In addition to calculation of the EBA, standardized CFU counts, without logarithmic transformation, were calculated as the S3/ S1 ratio and were reported as percentages. For comparison of the various treatment regimens with the control (Nil) regimen, point estimates of the ratio of their standardized counts together with nonparametric 95% confidence limits were calculated using a SAG macro in the SAS statistical package (1995).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A total of 93 patients were admitted to the study. Of these, 13 were excluded from the analyses for the following reasons: two patients produced less than 5 ml of sputum on one or more occasions; in four patients, one or more of the sputum cultures were contaminated and in four further patients negative cultures were obtained; three patients harbored tubercle bacilli resistant to INH, and one patient had received concomitant medication with amoxicillin/clavulanic acid, a drug with possible antimycobacterial action. Characteristics of the 80 patients whose data were finally analyzed are summarized in Table 1. The distributions of age, weight, sex, and extent of disease were similar in the three phases of the study, as they were in the randomly allocated treatment groups within each phase (data not shown). All had multicavitary disease, with the diameter of the largest cavity < 2 cm in nine patients (11%), between 2 and 4 cm in 27 patients (34%), and > 4 cm in 44 patients (55%). The mean CFU counts in the S1, S2, and S3 sputum collections from the eight treatment groups are shown in Table 2. The mean EBA (fall in log₁₀ CFU/ml sputum/d) in each treatment group with the standard deviation and 95% confidence limits is set out in Table 3. In Phase I, as was found in earlier studies, the SD increased in step with the EBA, from 0.05 for the Nil group to 0.21 for the INH 300-1 group (Bartlett's ² = 18.4, p = 0.001, indicating heterogeneous variance). The mean EBA increased from 0.011, indicating a slight increase in CFU counts, in the Nil group to 0.046, 0.092, and 0.121 in the CIP 250, 500, and 1,000 groups. In view of the heterogeneity of variance in the different treatment groups, a nonparametric test for the trend to increasing EBAs (8) was applied to the four groups and was highly significant (z = 3.31, p < 0.001). The mean EBA of 0.554 in the INH 300-1 group was much higher than the EBA of 0.121 in the CIP 1000 group (p = 0.0001). The weak association between the EBA and the S1 CFU count found previously (7) was not evident in the Phase 1 CIP groups.

The results in Phase II were surprising. The mean EBA in the CIP 1500-1 group of 0.033 was less than the EBA in the previous CIP 250 group (0.121) despite the greater dose size. Furthermore, the SD was higher than in any other CIP group, with the lowest individual EBA (0.35) much less than the lowest in the Nil group (0.064). Because of the unreliability of these results, Phase III was carried out in a further group treated at random with either 1,500 mg CIP daily (CIP 1500-2) or with 300 mg INH (INH 300-2). The results in the CIP 1500-2 group seem more consistent with Phase I results than those in the CIP 1500-1 group, as the mean EBA continued the trend with dosage seen in Phase I, the SD was smaller and the control INH 300-2 group had a mean EBA similar to that found in Phase I in the INH 300-1 group (Table 3). Furthermore, the means of the CIP 1500-1 and the CIP 1500-2 differed significantly (Kruskal-Wallace nonparametric test: p = 0.04). The mean EBA values in the CIP groups of the study have been plotted with the CIP 250, 500, and 1,000 results joined by a solid line and a dotted line (signifying doubt) joining the means of the CIP 1000 and the CIP 1500-2 groups (Figure 1); the EBA of the Nil group is indicated by a horizontal line. On this graph are also plotted the EBAs obtained with graded doses of RMP and INH in previous studies (7, 10). It is evident that CIP, although clearly active, is less bactericidal than RMP even in the high dose of 1,000 mg daily, and it is much less bactericidal than INH. The therapeutic margin for CIP in the high dose size of 1,000 mg was estimated to be about four-fold (assuming that a dose lower than 250 mg would have an EBA close to the Nil value). This is much the same as the margin for RMP but much less than the margin of greater than 16-fold for INH (the ratio of dose sizes comparing the usual therapeutic dose of 300 mg to the 18-mg dose, which just has an EBA greater than the Nil value). Also set out in Table 3 are comparisons of the treatment groups calculated without logarithmic transformation as standardized counts, which give the S3 CFU counts as percentages of the S1 count. Treatment comparisons then express, as a rate, the standardized count for groups treated with CIP or INH as a percentage of the standardized count for the Nil group. It is evident that the two approaches yield similar results. The comparisons of the CIP 250 and the CIP 1500-1 groups with the Nil group are not statistically significant, whereas all other groups differ significantly, as is evident from inspection of the confidence limits.

View larger version (12K): [in this window] [in a new window]   Figure 1.   The early bactericidal activities (EBA) of graded doses of ciprofloxacin (CIP), rifampin (RMP), and isoniazid (INH).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The highly significant (p < 0.001) rise in values of the EBAs of CIP as the dose size was increased from Nil to 250 mg and then to 1,000 mg daily, together with the probable further increase in EBA on giving 1,500 mg CIP, clearly indicates that CIP has activity against M. tuberculosis in pulmonary cavities. Furthermore, it seems that CIP should be given in a high dosage of at least 1,000 mg daily so that a small therapeutic margin of about four-fold would be available to allow for variation in drug concentrations in different lesions and different patients.

We can compare the direct estimate of the therapeutic margin of CIP with indirect estimates obtained from measurements of the MIC and serum concentrations. The in vitro MIC of CIP for M. tuberculosis lies between 0.25 and 2.0 µg/ml (4). Mean peak serum concentrations of CIP for doses of 250, 500, and 1,000 mg have been reported to be 1.7 µg/ml (± 0.64 µg/ml), 2.29 µg/ml (± 0.92 µg/ml), and 5.92 µg/ml (± 1.64 µg/ml), respectively (11). No data are available for serum concentrations after a dose of 1,500 mg, although several publications have confirmed the safety of dosages of as much as 1,500 or 2,000 mg daily (12, 13). Thus, a dose of 1,000 mg CIP yields peak concentrations well above the MIC, but experience with other drugs suggests that the overall antituberculosis activity, defined as the ability to prevent the emergence of resistance to a companion drug by the complete inhibition of bacterial growth, is more likely to be related to a lower concentration, closer to lesional concentrations such as the 3-h concentration than to the peak concentration. Such relationships would be influenced by the lag in regrowth of M. tuberculosis after a drug pulse of CIP (bacteriopausal effect), but we know little about these effects for CIP or indeed other quinolones. As far as it is possible to estimate the therapeutic margin from the MIC and serum concentrations, the range of its values therefore agree reasonably well with the direct estimate from the EBA study.

Only two reports, one being a previous EBA study, provide limited evidence of the efficacy of CIP in pulmonary tuberculosis (14, 15). Two further reports are of lesser value because of other drugs given with CIP (16) or because of the small number of patients studied (17). Clinical evidence of efficacy is available for ofloxacin, which has similar pharmacokinetics and antituberculosis activity in vitro (18, 19). Unfortunately, no comparison was included with ofloxacin, for commercial reasons, in the EBA study design. Although firm evidence of the clinical efficacy of CIP is lacking, the finding that the EBA of even 1,000 mg CIP daily was lower than the EBA obtained with the usual dose of 600 mg RMP and much lower than the 300 mg INH does not imply that ciprofloxacin is an inferior drug since useful drugs may have low EBAs. The size of the EBA depends on the mechanisms by which bacilli are killed, which may not reflect the therapeutic value of the drug. What matters is the therapeutic margin, which is far more important in treatment than the absolute value of the EBA at the usual dose size. The situation with CIP can be compared with corresponding findings with RMP, which also has a small therapeutic margin of about four-fold, estimated from its EBA/log concentration curve (Figure 1). RMP has been shown to be less active early in treatment when the dose was reduced from 600 to 450 mg daily (20), so that dosage is also critical, as would be expected from the therapeutic margin of only four. The superiority of rifampin in treatment over the quinolones is due to the sterilizing activity of rifampin, which is postulated to kill semidormant organisms in the lesions more effectively than other drugs (21). Sterilizing activity is not measured by the therapeutic margin or by the EBA. Thus, pyrazinamide has a very low EBA (5, 22), but it is still a potent sterilizing drug, and RMP has a much lower EBA than does INH (5, 7), but it is again much better at sterilizing lesions. It is among CIP's assets that it is concentrated in macrophages and that the levels achieved in sputum in patients with cystic fibrosis are equal to or greater than the maximum serum concentrations in the same patients (23). Although this property is unlikely to influence the EBA because the bacilli being sampled in the sputum are derived from extracellular sites in cavity walls, improved ability to kill bacilli inside macrophages might result in increased sterilizing activity in later stages of treatment when bacilli may be more often intracellular (24). However, CIP has been shown to be less effective than pyrazinamide, and, therefore, probably rifampin as well, as a sterilizing drug, at least in HIV-positive patients (16). It should also be noted that sparfloxacin, a quinolone with considerable bactericidal activity in vitro, failed to increase sterilizing activity when administered with INH plus RMP or RMP and pyrazinamide in experimental murine tuberculosis (25).

The early bactericidal activity of CIP in pulmonary tuberculosis has been estimated previously and found equal to that of INH (14). Unfortunately, the sputum in the study was decontaminated before culture with NaOH rather than plated directly onto selective culture medium as we have done in all our work. NaOH selectively kills precisely those actively multiplying bacilli that are killed at different rates by different drugs and different dose sizes (26). As a result, the estimates were characteristic of the bactericidal action on less actively metabolizing bacilli that occurs after the first 2 d when all drugs kill at about the same rate (5), and therefore failed to detect the very large difference between the EBAs of CIP and INH. The study was nevertheless of value in indicating the activity of CIP.