The early bactericidal activity of rifabutin measured by sputum viable counts in Hong Kong patients with pulmonary tuberculosis

The early bactericidal activity of rifabutin measured by sputum viable counts in Hong Kong patients with pulmonary tuberculosis S. L. Chan, W. W. Yew, W. K. Ma, D. J. Girling”, Grantham Hospital, Aberdeen, Brompton Hospital, London, and $Department

V. R. Aber*, D. Felmingham+,

Hong Kong, *Medical Research UK, fDepartment

of Bacteriology,

Council Cardiothoracic

of Clinical Microbiology,

Royal Postgraduate

B. W. Allen*, D. A. MitchisonS Epidemiology

Group, Royal

University College Hospital, London, UK

Medical School, London,

UK

S U M M A R Y. Previously untreated patients with smear-positive pulmonary tuberculosis were randomly allocated to treatment with 600,300,150 or 75 mg doses of rifabutin (LM427, ansamycin), 600,300 or 150 mg of rifampicin, 300 mg isoniazid or to no drug daily for 2 days. The fall in viable counts of Mycobacterium tuberculosis in sputum collections during the 2 days, termed the early bactericidal activity (EBA), was estimated from counts of colony-forming units (cfu) on selective 7H-11 agar medium. The EBA for rifabutin ranged from - 0.039 (an increase in counts) to 0.049 log,, cfu/ml/day whereas the EBA increased from 0.071 for 150 mg rifampicin to 0.293 log,, cfu/ml/day for 600 mg rifampicin and was 0.43 log,, cfu/ml/day for 300 mg isoniazid. The difference between the EBAs for rifabutin and rifampicin just attained significance (P = 0.05) suggesting that rifabutin was inactive or less active than rifampicin against the extracellular bacilli in pulmonary cavities. Peak plasma concentrations of rifabutin after the initial doses were found to be proportional to dose size and were approximately 7 times lower than those after the same dose size of rifampicin. The lower EBA of rifabutin as compared to rifampicin is probably due to the low plasma concentrations which are not fully compensated for by slightly greater antituberculosis activity of rifabutin in vitro. R _&S I/Ml?.

Des patients atteints de tuberculose pulmonaire frottis positif non-trait& anterieurement ont CtC repartis au hasard pour un traitement aux doses respectives de 600,300,150 ou 75 mg de rifabutin (LM427, ansamycine), de 600, 300 ou 150 mg de rifampicine, de 300mg d’isoniazide ou a l’absence de traitement quotidien, pendant deux jours. La baisse des numerations viables de Mycobucterium tuberculosis dans les recueils d’echantillons de crachats pendant les 2 jours, nommee activite bactericide precoce (ABP), etait estimee a partir des numerations des unites formant colonies (ufc) sur un milieu en g&lose 7H-11 selectif. L’ABP pour la rifabutin a varie de 4,039 (augmentation des numerations) a 0,049 log,, ufcIml/jour, tandis que I’ABP a augment& de 0,071 pour 150mg de rifampicine a 0,293 log,, ufc/ml/jour pour 600mg de rifampicine, et 0,43 log,, ufcIml/jour pour 300mg d’isoniazide. La difference entre les ABP pour rifabutin et rifampicine a juste atteint sa valeur significative (P = 0,05), ce qui suggere que la rifabutin Ctait inactive ou moins active que la rifampicine cantre les bacilles extracellulaires dans les cavites pulmonaires. Les pits des concentrations plasmatiques de rifabutin aprb les doses initiales Ctaient proportionnels a l’importance des doses et Ctaient approximativement 7 fois plus faibles que ceux obtenues par des doses de rifampicine de m&meimportance. L’ABP plus faible de rifabutin comparee a celle de la rifampicine est probablement due aux concentrations plasmatiques faibles. Celles-ci ne sont pas pleinement compensees par une activite antituberculeuse legrement plus ClevCede la rifabutin in vitro. R ES U M E N.

Pacientes que presentaban una tuberculosis pulmonar con baciloscopia positiva no tratados previamente fueron asignados en forma randomizada a tratamientos con rifabutina (LM427, ansamicina) a las dosis de 600,300,150 o 75 mg, rifampicina a las dosis de 600,300 o 150 mg, isoniacida a la dosis de 300 mg o ausencia de tratamiento por 2 &as. La disminuci6n de1 recuento de Mycobacterium tuberculosis viables en las muestras de esputo recolectadas durante 10s 2 dias, denominada actividad bactericida precoz (ABP) fue

Correspondence to: Professor D. A. Mitchison, Royal Postgraduate Medical School, Ducane Road, London WI2 ONN, UK. 33

34

Tubercle and Lung Disease

evaluada en base al recuento de unidades formadoras de colonias (cfu) en medio selective de agar 7H-11. La ABP para la rifabutina varid de -0,039 (aumento de 10s recuentos) a 0,049 log,,, cfu/ml/dia, mientras que la ABP aumentd de 0,071 para 150mg de rifampicina a 0,293 log,, cfu/ml/dfa para 600mg de rifampicina ; para 300mg de isoniacida fue de 0,43 log,, cfu/ml/dia. Las diferencias entre las ABPs de rifabutina y rifampicina fueron apenas significativas (P = 0,05) sugiriendo que la rifabutina fue inactiva o menos activa que la rifampicina contra 10s bacilos extracelulares en las cavidades pulmonares. Se observd que las concentraciones plasmziticas mAximas de rifabutina despuks de las dosis iniciales eran proporcionales al tamafio de la dosis y fueron aproximadamente 7 veces m& bajas que aquCllas observadas despuCs de la misma dosis de rifampicina. La ABP m& baja de la rifabutina, comparada con la de la rifampicina, se debe probablemente a las bajas concentraciones plasmaticas las cuales no son completamente compensadas por una actividad antituberculosa in vitro levemente superior de la rifabutina.

INTRODUCTION Rifabutin (LM427, ansamycin) has been used in treatment for AIDS patients with M. avium complex (MAC) infections’-3 and is under study for patients with pulmonary tuberculosis, either when their organisms are rifampicin-resistant’*3,4 or for initial treatment as a possible replacement for rifampicin5 The in vitro MICs of rifabutin against Mycobacterium tuberculosis are reported as being 2-20 times lower than those of rifampicin by conventional titrations in liquid and solid media or radiometrically;6-9 other species of mycobacteria including MAC are also proportionately more sensitive.&* Rifabutin is highly concentrated in tissues but plasma concentrations are lower than those obtained with rifampicin.” It is about 6 times more active than rifampicin in experimental infections of mice with M. tuberculosis6 or M. avium.” Mycobacterial disease in the mouse is essentially an intracellular infection, tissue necrosis only occurring in the late stages of the disease. In contrast, the great majority of the bacilli in untreated human adult pulmonary tuberculosis are extracellular in cavity walls.‘2*‘3 The considerable activity in murine disease may reflect high intracellular concentrations in mouse macrophages while caseating lesions in cavity walls containing numerous acid-fast bacilli, characteristic of human pulmonary tuberculosis, might contain lower concentrations corresponding to those in plasma. Thus it is uncertain whether rifabutin concentrations in these human sites are sufficiently high for effective antimycobacterial activity. A novel way of answering this question was suggested by a previous study in Nairobi on the early bactericidal activity of antituberculosis drugs measured by their effect on the content of viable bacilli in sputum.14 Serial counts of colony forming units (cfu) of M. tuberculosis in the sputum of patients with newly diagnosed, untreated pulmonary tuberculosis showed that the fall in the cfu counts during the first 2 days of treatment varied according to the drug regimen used, whereas in the next 12 days of treatment there was no detectable variation between regimens in the rate of decline. A statistically significant association (P < 0.05) was found between the dose size of rifampicin and the fall in cfu counts during the first 2 days in a total of 19

patients. This finding led to the present study in which a comparison has been made between rifabutin and rifampicin, each drug being given to small groups of patients at several different dose levels. A comparison of the fall in sputum counts could then be used to equate dose sizes of the 2 drugs. In the Nairobi study, isoniazid had the greatest early bactericidal activity of the drugs tested. Control groups of patients were therefore given isoniazid alone or no drug. Since the early bactericidal activity would be estimated during the first 2 days of treatment, only 2 doses of the rifamycins or of isoniazid were necessary. Although the risk of resistance emerging during monotherapy for only 2 days was negligible since it had not occurred during the longer 1Cday period in Nairobi, sensitivity tests were set up on cultures before and after the test. In addition to the sputum counts, plasma concentrations of the rifamycins were measured after the first dose and biochemical tests for drug toxicity were done, as rifabutin had not been widely used in Hong Kong previously.

METHODS Patients The patients in Hong Kong were aged 15 years or more and had newly diagnosed pulmonary tuberculosis with a positive direct smear examination for acid-fast bacilli. They were selected as weighing between 40 and 60kg so as to narrow the range of drug dosages, expressed as mg/kg body weight, resulting from administration of a single size of drug capsule. Particular care was taken to exclude those who might have received even a single dose of previous antituberculosis chemotherapy and also those with serious concomitant disease. They were admitted to the Grantham Hospital, Hong Kong for 4-5 days. The study protocol was approved by the Ethics Committee of the Hong Kong Medical and Health Services. Informed consent to the procedures was obtained from all patients.

Drug regimens On admission to hospital, a urine specimen was tested for acetyl isoniazid15 to exclude concurrent treatment

Early bactericidal

with isoniazid. The patients were then randomly allocated to treatment with one of the following drugs in the dosage specified. Each drug dose was given on 2 successive days, the rifabutin doses after breakfast and the rifampicin doses on an empty stomach. Rifabutin Rifabutin Rifabutin Rifabutin Rifampicin Rifampicin Rifampicin lsoniazid No drug

600 mg 300 mg 150mg 75 mg 600 mg 300 mg 150 mg 300 mg -

(12 mgkg) (6 mgkg) (3 mgkg) ( 1.5 m&g) ( 12 w/kg) (6 mgkg) (3 mgkg) (6 mgkg) _

Intake to the rifabutin 75 mg series was terminated early because of the apparent low early bactericidal activities of rifabutin at all doses.

Sputum and blood samples Sputum collections were made between 16:OO on the day of admission and 08:OO on the next day (Sl collection). The first of the two drug doses was then given. Similarly timed collections were made on the next 2 days (S2 and S3 collections). The sputum was sent to the hospital laboratory where a cfu count was set up within 1 h of completing the collection. A specimen obtained on the day of admission and also on the last day of testing was sent to the Reference Laboratory at the Royal Postgraduate Medical School, London, for sensitivity testing of the culture of M. tuberculosis. Biochemical tests for liver function, renal function and blood counts were done during the study period. Samples of blood were taken immediately before and at 2, 4 and 6 h after the first dose of rifabutin and at 24 h, immediately before the second dose of rifabutin. Plasma from the samples was stored at 20°C and sent by air to University College Hospital, London, where the concentrations of rifabutin or rifampicin, as well as the concentrations of the 25-desacetyl metabolites of these rifamycins were estimated by means of an HPLC method. ”

Cfu counts on sputum In the hospital laboratory, a cfu count was carried out on selective 7H-11 oleic acid agar medium (Difco) containing polymyxin B sulphate (Wellcome) 200 ml, carbenicillin (Beecham) 100 mg/l, trimethoprim (Wellcome, as lactate) 20 mg/I, amphotericin B (Squibb, tissue culture grade) 50 mg/l.” On receipt, the volume of sputum was measured and it was homogenized by shaking with glass beads of about 10 mm diameter. To 2 ml homogenate was added 4 ml 1: 10 dithiothreitol (Sputolysin, Hoechst) and this was vortex mixed for 30 s. After standing at ambient temperature for 15 min,

activity of rifabutin

35

the homogenate was centrifuged (3000 g for 15 min), the supernatant discarded and the deposit resuspended in sterile distilled water to a final volume of 4.8 ml. From this suspension, serial IO-fold dilutions were made and 0.1 ml amounts of the neat suspension and each dilution up to lo-’ were made on one-third sectors of selective 7HlO plates, in duplicate. The plates were sealed in polythene bags and incubated at 37°C for 4 weeks. Early bactericidal activity was calculated as the daily fall in log cfu count, that is (log,, cfulmhday on Sl log,,cfu/ml/day on S3)/2. The technician who did these viable counts received prior training in the method in the London Reference Laboratory. Results were examined by analysis of variance using the SPSS package.

RESULTS Early bactericidal

activity

Of the 138 patients admitted to the study, 16 were excluded because the sputum collections on the Hurst (Sl) or third days (S3), on which the calculation of the early bactericidal activity was based, were negative on culture (8), contaminated (6) or unavailable (2). It proved difficult to find patients producing adequate volumes of sputum. Since the error of the estimation (deviation from linear regression of cfu count during the 3 days) increased with decreasing sputum volume, a further 24 patients were excluded because the sputum volume on either of these days was less than 6 ml. One additional patient was excluded because the cultures for sensitivity testing were contaminated. The analysis was based on the remaining 97 patients with sputum volumes of at least 6 ml. Of these, 4 had cultures resistant to isoniazid but, as none had been allocated to the isoniazid regimen, they were not excluded. No patient had cultures resistant to rifampicin or rifabutin. A comparison of the sensitivity patterns of the cultures obtained before and after the ‘L-day period of monotherapy showed that resistance did not emerge in any patient. Estimates of early bactericidal activity are set out in Table 1. The fall in counts on sputum from the 14 patients in the isoniazid 300mg control group (0.43 log,,/ml/day) was greater than the fall (0.06 log,,/mVday) in the no drug control group (P = 0.03) and was numerically larger than the fall in any other group. In the groups treated with rifampicin, the fall in counts increased with dose size from 0.07 1 log,,,/ml/day (a rise in counts) with the 150mg dose to 0.29 log,,,/ml/day with the 600mg dose, while in those treated with rifabutin all the changes in counts were small and did not increase with dose size. Analysis of variance showed that the difference between the responses for rifampicin and rifabutin (the ‘drug’ main effect) averaged over the 150, 300 and 600 mg close sizes was of borderline significance (P = 0.05), but the dose size main effect and the drug x dose size

36

Tubercle and Lung Disease

Table 1. Early bactericidal activity of rifabutin (RBU), rifampicin (RMP) and isoniazid (INH) Dose size (mg) 75 150 RBU

RMP

EBA* n SEM EBA

-0.0070 5 0.033 _

iEM INH

EBA

No

EBA

drug

n

a.039 12 0.12 0.071 15 0.039

_

iEM -

Table 2. Comparison of early bactericidal activities obtained in previous Nairobi study (N) and in present Hong Kong study (HK) Dose size

Drug 300

600

None

0.041 12 0.11

0.049 10 0.17

-

0.162 10 0.14

0.293 11 0.092

-

0.431 14 0.10

_

_

_

Rifampicin

Isoniazid

Study

No. of patients

Fall in counts (logto cfu/ml/day)

3mgikg

HK

15

Xl.07

5 6 10 12 20

N HK N HK N

3 10 8 11 8

0.06 0.16 0.19 0.29 0.41

N HK

4 14

0.72 0.43

N HK

4 8

-0.02 0.06

!’ ” II *I ‘(

300mg 300mg

No drug

_

0.060 8

SEM

0.11

*EBA = mean early bactericidal activity (loglo cfu/ml/day). n = number of patients. SEM = standard error of mean EBA.

interaction were clearly non-significant (P = 0.5 - 0.6). A separate analysis of the effect of dose size in the 3 rifampicin groups failed to attain significance (P = 0.08). Thus there is a suggestion that rifabutin had appreciably less early bactericidal activity than rifampicin. In planning the study, its size was estimated from the experience gained with the earlier Nairobi study to be adequate to answer the questions posed. In the event, even though the intake was prolonged for several months, the results obtained are inconclusive. The reason for this is that the error term (between patients within groups) against which comparisons between groups are measured was considerably greater than had been anticipated. Thus the standard deviation for error in the Nairobi study was 0.19 (60 d.f.) whereas it was 0.38 (84 d.f.) in the present study. The large error encountered in the present study was not due to high variation in cfu counting in the Grantham Hospital laboratory as

Table 3. Concentrations

of rifabutin, its 25.desacetyl

duplicate counts set up in the middle of the study, without the technician knowing the reason for doing them, showed much smaller variation between the duplicates (mean square = 0.0048, d.f. = 19) than between the sputum specimens (mean square = 5.036, d.f. = 18). A comparison of the estimates of early bactericidal activity obtained in the Nairobi and the present study with different dose sizes of rifampicin, isoniazid 300mg and no drug (Table 2) shows good agreement between corresponding estimates in the two studies. The greatest discrepancy is between the values obtained for isoniazid which were 0.72 log,,/ml/day in the Nairobi study and 0.43 log,,/ml/day in the present study, but the difference is well within the 95% confidence limits for the estimates.

Plasma concentrations Peak plasma concentrations of rifabutin occurred at 4 h and ranged from 0.11 mg/l after the 75 mg dose to 1.14 mg/l after the 600 mg dose (Table 3). The ratio between peak concentration and dose size was

metabolite and rifampicin

in plasma

24

Ratio peak concentration to dose size?

-

1.5

Interval after dose (h) Drug Rifabutin

Dose (mg)

2

75

-

150 300 600

4

6

0.21* (0.114.38) 0.56 (0.3 l-l .OO) 0.73 (0.262.06)

0.11 (0.06-0.23) 0.27 (0.22-0.33) 0.69 (0.57-0.83) 1.14 (0.90-1.43)

0.10 (0.060.18) 0.20 (0.17-0.24) 0.46 (0.424.52) 0.96 (0.82-1.13)

0.13 (0.06-0.26)

0.20 (0.160.26)

0.18 (0.13-0.26)

1.96 (0.79-4.87) 3.21 (1.44-9.03) 9.25 (7.66-11.18)

1.80 (0.78-4.16) 3.27 (1.34-8.01) 9.53 (7.81-11.63)

1.38 (0.63-3.04) 3.03 (2.014.55) 6.93 (5.5Ck8.73)

1.8 0.18 (0.14-0.23)

2.3 1.9

25-desacetyl rifabutin Rifampicin

600 150 300 600

*Geometric mean (mg/l) 95% confidence limits in parentheses. i-Highest concentration (mg/l)/dose size (8).

-

1.3 1.1

-

1.6

Early bactericidal

approximately constant over the range of dose sizes indicating that there was no apparent saturation of absorption, metabolism or excretion with the higher doses. The half-life calculated from the concentrations at 6 and 24 h after the 600 mg dose was 6.9 h, a value likely to be an overestimate as absorption was probably not complete at 6 h. It should be noted that estimates of plasma concentration by a sensitive HPLC method have shown a phase of rapid excretion during the first 12 h changing to a much slower phase from 24 onwards, having a terminal half-life of 45 h but with concentrations so low (below 100 ng/ml) as unlikely to be of clinical significance.‘* The 25-desacetyl metabolite of rifabutin, which retains microbiologic activity, was estimated to be present at 18% of the concentration of rifabutin during the period of 2-6 h after a 600 mg dose. The peak concentrations of rifampicin were about 7 times higher than those following an equal sized dose of rifabutin.

Adverse effects Of the total 138 patients admitted to the study, 57 received rifabutin in dosages of 600mg (14 patients), 300 mg (17 patients), 150 mg (18 patients) and 75 mg (8 patients); 49 received rifampicin in dosages of 600 mg (17 patients), 300 mg (16 patients) and 150 mg (16 patients); 16 received 300 mg of isoniazid and 16 received no drug. Only 3 complained of adverse effects. The first, who received 150 mg doses of rifabutin, had a mild itchy maculopapular rash which subsided spontaneously; the second, who received 75 mg doses of rifabutin, became febrile after each of the two doses, and the third, who received 150 mg doses of rifampicin, experienced mild pruritus which subsided spontaneously without a rash developing.

DISCUSSION The results of the sputum viable counts tempted us to think that rifabutin, even at a high dose of 600 mg (12 mg/kg) has littie or no early bactericidal activity at the start of treatment. However, only the comparison between rifabutin and rifampicin over the dose sizes of 150, 300 and 600 mg attained statistical significance and even then the value of P was only 0.05. We can conclude that rifabutin very probably is less active than rifampicin at the same dose size, but we cannot say, except in the the vaguest terms, by how much nor can we say that rifabutin is inactive. An interpretation of this finding depends on consideration of the site of the tubercle bacilli, the minimal inhibitory concentrations (MICs) of rifabutin and rifampicin and the plasma levels of the two rifamycins. Virtually all tubercle bacilli in the sputum at the start of treatment must have come from cavity contents, with the bacilli lying extracellularly in

activity of rifabutin

37

caseous material.‘* Since intracellular concentration is not involved, we would expect that the ratio between levels in the plasma and in cavity walls would be similar for rifabutin and rifampicin particularly as there is close correspondence between assays of plasma and cavitary the other rifampicin” and concentrations of antituberculosis drugs streptomycin2’ and isoniazid.*’ Using conventional assay methods, the MIC of rifabutin against M. tuberculosis has been found to range from 2 to 20 times lower than that for rifampicin.6-9 A more realistic method in which the MIC is determined after only a 3-day period of contact with the rifamycin suggests that the rifabutin MIC is only 2.5 times lower,** an estimate in fair agreement with the ratio of 4 obtained by the radiometric method.’ If we assume that the geometric mean of these two values (a ratio of 3.2) is correct, we can deduce from our finding of peak plasma levels of rifabutin about 7 times lower than those produced by an equal dose of rifampicin, that the antibacterial activity of the 600mg dose of rifabutin should have about the same early bactericidal activity as 300 mg of rifampicin, a dose size that is probably inadequate.23 The more usually prescribed 300 mg dose of rifabutin would be equivalent to 150 mg of rifampicin, an even less adequate dose size. The study was planned to obtain a more accurate estimate of the early bactericidal activity than in fact proved possible because of unexpectedly large variations between patients in their sputum counts. Improving accuracy would have meant prolonging the study for a considerable period beyond the 15 months that it took to complete. The finding of a much larger variation between patients in Hong Kong than in Nairobi, despite a reproducible counting technique in the Hong Kong laboratory, strongly suggests that the variability among the Hong Kong patients was due to characteristics of their pulmonary lesions. It seems possible that the African patients in the Nairobi study had acute lesions with a likelihood that separate cavities would be connected and that the sputum samples would therefore be representative of the whole. The more chronic lesions in the Hong Kong patients, from whom large volumes of sputum could not easily be obtained, might be better separated and so lead to more variable sampling of their contents. If this is the reason, work on early bactericidal activity would be best carried out in Africa or in other areas where the acute type of disease characteristic of African patients is encountered. An estimate of early bactericidal activity could be of great value in bridging the gap between the laboratory development of a new drug and a full scale clinical trial in man, since the technique we have described offers the possibility of comparing the effect of different dose sizes in relatively small groups of patients. While an estimate of early bactericidal activity may help to decide how active a drug is against extracellular tubercle bacilli, we do not know whether its action against organisms in these sites adequately assesses its value in chemotherapy. Rifampicin is of value as a

38

Tubercle and Lung Disease

sterilizing drug probably because it kills a special population of semi-dormant persisting bacilli as soon as they begin to metabolise.24 Conditions within a macrophage are ideal for the creation of such a population. Antituberculosis immunity appears to create macrophages which do not kill tubercle bacilli within them but inhibit their growth” to an extent that may well protect them against the slow onset, bactericidal activity of isoniazid, as has been demonstrated in the essentially intracellular infection of experimental murine tuberculosis26. If the special population is indeed mainly intracellular in man, the greater concentration of rifabutin than of rifampicin inside cells would be an advantage. It would seem to be logical to treat tuberculosis with a mixture of rifampicin, aimed at the predominantly extracellular bacilli present at the start of treatment and at any part of the special population of persisters that is extracellular, and rifabutin, aimed at persisters lying inside macrophages. The rifamycins could be given on alternate days or sequential periods with rifampicin first and rifabutin second. Thus, the role of rifabutin in the treatment of pulmonary tuberculosis is not yet clear. This study cannot be used as evidence that it is ineffective as a sterilizing drug in short-course chemotherapy but only that it has advantages and disadvantages when compared to rifampicin.

Acknowledgements This work was supported by research grants from the World Health Organization and Messrs Farmitalia Carlo Erba Ltd. who also provided the rifabutin.

References 1. Centers for Disease Control, US Department of Health and Human Services. Diagnosis and management of mycobacterial infection and disease in persons with human immunodeficiency virus infection. Ann Intern Med 1987; 106: 254-256. 2. Agins B D, Berman D S, Spicehandler D, El-Sadr W, Simbcrkoff M S, Rahal J J. Effect of combined therapy with ansamycin, clofazimine, ethambutol, and isoniazid for Mycobacterium avium infection in patients with AIDS. J Infect Dis 1989; 159: 784-787. 3. O’Brien R J, Lyle M A, Snider Jr D E. Rifabutin (Ansamycin LM 427): A new rifamycin-S derivative for the treatment of mycobacterial disease. Rev Infect Dis 1987; 9: 519-530. 4. Hong Kong Chest Service/British Medical Research Council. A controlled study of rifabutin and an uncontrolled study of ofloxacin in the retreatment of patients with pulmonary tuberculosis resistant to isoniazid, streptomycin and rifampicin. Tubercle and Lung Disease 1992; 73: 6&68. 5. Felten H K. Preliminary experience with rifabutin (ansamycin LM427) in patients with newly diagnosed pulmonary tuberculosis (PTB) and chronic disease. Am Rev Respir Dis 1987: 135: A136.

6. Della Bruna C, Schioppacassi Cl, Ungheri D, Jabes D, Morvillo E, Sanfilippo A. LM 427, a new spiropiperidylrifamycin: in vitro and in vivo studies. J Antibiot 1983; 36: 1502-1506. 7. Dickinson J M, Mitchison D A, In vitro activity of new rifamycins against rifampicin-resistant M. tuberculosis and MAIS-complex mycobacteria. Tubercle 1987; 68: 177-182. 8. Truffot-Pemot C, Giroir A M, Grosset J. A study of the minimal inhibitory concentrations of rifabutin (ansamycin LM427) for M. tuberculosis, M. avium-intracellulare and M. nenopi. Rev Ma1 Resp 1988; 5: 401- 406. 9. Heifets L B, Lindholm-Levy P J, Iseman M D. Rifabutin : minimal inhibitory and bactericidal concentrations for Mycobacterium tuberculosis. Am Reg Respir Dis 1988; 137: 719-721. 10. Fanfani A, Riva F, Sanfillipo A, Sardi A. Rifabutin: LM427 ansamycin. Milan: Farmitalia Carlo Erba. December 1985. 11. Saito H, Sato K, Tomioka H. Comparative in vitro and in vivo actively for rifabutin and rifampicin against Mycobacterium avium complex. Tubercle 1988; 69: 187-192. 12. Canetti G, The tubcrcle bacillus in the pulmonary lesions of man. New York: Springer, 1955: pp 62-68. 13. Bates J H, Transmission, pathogenesis, pathology and clinical manisfestations of tuberculosis. In Kubika G P, Wayne L G, eds. The mycobacterial. A sourcebook. Part B. New York: Marcel Dekker, 1984: p 997. 14. Jindani A, Aber V R, Edwards E A, Mitchison D A. The early bactericidal activity of drugs in patients with pulmonary tuberculosis. Am Rev Respir Dis 1980, 121: 939-949. 15. Eidus L, Hamilton E J. A new method for the determination of Nacetyl isoniazid in urine of ambulatory patients. Am Rev Respir Dis 1964; 89: 587-558. 16. Gillespie S H, Baskerville A J, Davidson R N, Felmingham D, Bryceson A D M. The serum rifabutin concentrations in a patient successfully treated for multi-resistant Mycobucterium tuberculosis infection. J Antimicrob Chemother 1990; 25: 490-491. 17. Mitchison D A, Allen B W, Carrol L, Dickinson J M, Aber V R. A selective oleic acid albumin agar medium for tubercle bacilli. J Med Microbial 1972; 5: 165-175. 18. Strolin Benedetti M, Efthymiopoulos C, Sassella D, Moro E, Repetto M. Autoinduction of rifabutin metabolism in man. Xenobiotica 1990; 20: 1113-I 119. 19. Canetti G, Parrot R, Porven G, Le Lirzin M. Les taux de rifampicine dans le poumon et dans les lesions tuberculeuses de l’homme. Acta tuberc pneumol Belg 1969; 60: 315-322. 20. Canetti G, Grumbach F. Diffusion de la streptomycine dans les lesions caseeuses des tuberculeux pulmonaires. Ann Inst Pasteur 1953; 85: 38@386. bestimmte Isoniazid21. Bartmann K, Freise G. Mikrobiologisch Konzentrationen im Gewebe von Tier und Mensch. Beitr Klin Tuberk 1963; 127: 546-560. 22. Dhillon I, Mitchison D A. Activity in vitro of rifabutin, FCE 22807, rifapentine and rifampicin and their penetration into mouse peritoneal macrophages. Am Rev Respir Dis 1992; 145: 212-214. 23. Long M W, Snider D E, Farer L S. US Public Health Service cooperative trial of three rifampin-isoniazid regimens in treatment of pulmonary tuberculosis, Am Rev Respir Dis 1979; 119: 879-894. 24. Dickinson J M, Mitchison D A. Experimental models to explain the high sterilizing activity of rifampicin in the chemotherapy of tuberculosis. Am Rev Respir Dis 1981; 123: 367-371. 25. Lowrie D B. Is macrophage death on the field of battle essential to victory, or a tactical weakness in immunity against tuberculosis? Clin Exp Immunol 1990; 80: 301-303. 26. Dhillon J, Mitchison D A. Influence of BCG-induced immunity on the bactericidal activity of isoniazid and rifampicin in experimental tuberculosis of the mouse and guinea-pig. Br J Exp Path01 1989; 70: 103-l 10.