European Respiratory Society

Tuberculosis therapy: past, present and future

M.D. Iseman


The major historical landmarks of tuberculosis (TB) therapy include: the discovery of effective medications (streptomycin and para-aminosalicylic acid) in 1944; the revelation of “triple therapy” (streptomycin, para-aminosalicylic acid and isoniazid) in 1952, which assured cure; recognition in the 1970s that isoniazid and rifampin could reduce the duration of treatment from 18 to 9 months; and the observation in the 1980s that adding pyrazinamide to these drugs allowed cures in only 6 months.

To combat noncompliance, intermittent regimens, twice or thrice weekly, have been proven to cure even far-advanced TB in as few as 62–78 encounters over 26 weeks.

However, these regimens are not sufficiently short or convenient to facilitate effective treatment in resource-poor countries. Therefore, drug-resistant strains have emerged to threaten TB control in various areas of the world, including India, China, Russia and the former Soviet Union. For these reasons, it is vital that new medications are developed to shorten the duration of therapy, increase the dosing interval of intermittent regimens and replace agents lost to resistance. Other special considerations include identifying optimal therapy for persons with acquired immune deficiency syndrome, particularly noting the problems of drug/drug interactions for those receiving antiretroviral treatment.

Finally, the Alchemist's Dream of tuberculosis should be pursued: modulating the immune response to shorten treatment and/or overcome drug resistance.

It is difficult to discuss tuberculosis (TB) therapy, present and future, without reviewing the history of the treatment. Scientific “knowledge” is a continuously evolving process, and recognising how today's paradigms have been arrived at is critical to understanding how advances might be made in the future.


Efforts to treat “phthisis” or “consumption” over the Millennia have been largely tales of tragedy and frustration. A variety of herbal concoctions, dietary interventions and climatic prescriptions were among the more benign remedies offered. By contrast, bleeding and purging probably amplified and accelerated mortality.

At the turn of the 20th century, George Bernard Shaw, via one of the characters in his play “A Doctor's Dilemma”, described the medical treatment of TB in England as “a huge commercial system of quackery and poison”. However, with the discovery of sulfonamides and penicillin in the 1930s, truly effective antimicrobial therapy became a reality. Inspired by observations that soil microbes seemed capable of preventing the growth of other species “in their turf”, Selman Waksman's research in New Jersey lead to the identification of streptomycin (SM) in 1944. In the same year, Jorgen Lehman, working in Sweden, synthesised the para-amino salt of salicylic acid (PAS). Rapidly pressed into use, these two agents had clearly identifiable activity against clinical TB. Serendipitously, due to a shortage SM, the British Medical Research Council (BMRC) performed one of the first randomised clinical trials comparing PAS or SM alone with the combination of both agents 1. The results, which were published in 1950, demonstrated that the combination was more effective at both achieving cures and preventing acquired drug resistance. These insights substantially shaped future treatment trials (table 1).

View this table:
Table 1

Landmarks in tuberculosis (TB) therapy

Gerhard Domagk's research, which led to the discovery of sulfonamides in the 1930s, eventuated in the discovery of the anti‐TB activity isonicotinic acid hydrazide (INH) in 1952. Adding INH to PAS and SM (“triple therapy”) resulted in predictable cures for 90–95% of patients, the Holy Grail. Unfortunately, it required up to 24 months of continuous treatment to achieve this objective 2. This was related to the persistence of viable bacilli in tissues long after sputum cultures had become negative. Perhaps the most fundamentally important principle derived from triple therapy was that such treatment, reliably given, effectively precluded acquired drug resistance.

The replacement of PAS by ethambutol (EMB) in the 1960s had two benefits. EMB was much better tolerated than PAS and it allowed reduction in the duration of treatment to 18 months 3.

The next major advance in therapy was the introduction of rifampicin (RIF). Derived from Streptomyces mediterranei, RIF was studied in early trials by the BMRC in East Africa 4 and Hong Kong 5, demonstrating that practical combinations of INH, SM, EMB and RIF resulted in predictable cures in >95% of cases in just 8–9 months. The particular activity of RIF that facilitated this compression of treatment was its pronounced capacity of rifamycin to kill mycobacteria undergoing sporadic metabolism, the so-called “sterilising effect”.

The next step forward was the recognition that the inclusion of pyrazinamide (PZA) allowed a reduction in the duration required to achieve predictable cures. PZA was found to accelerate the time required to achieve culture negativity and to yield >95% cure rates in 6 months when combined with INH and RIF 57. It has been speculated that the singular role of PZA is activity against tubercle bacilli in the acidic debris in pulmonary cavity walls 8. This is consistent with the observation that PZA exerts all of its beneficial effects in the first 2 months of therapy.


Despite progressive reduction, from 24 to 6 months of the duration of therapy required for cure, noncompliance or abandonment of treatment remain the major impediments to effective therapy. To combat these factors, directly observed therapy (DOT) has been widely endorsed 9. To facilitate such supervision, intermittent (less than daily) regimens have become very important. Multiple studies have shown that 6‐month regimens given thrice weekly throughout 6, 10 or twice weekly following a 2‐week daily induction phase 11 are as efficacious as daily regimens. These regimens involve as few as 62–78 encounters with the patients over 6 months to deliver curative treatment.

Such regimens have allowed the USA to increase the proportion of patients receiving DOT from 4% in 1990 to ∼70% in 2000 12. With the widespread use of DOT, case rates in the USA have steadily declined from 1993 to the present (fig. 1). Indeed, decline of case rates from 1995–2000 averaged 7.8%·yr−1. This was the most rapid rate of reduction over any 5‐yr period since 1953, the beginning of the era of modern chemotherapy.

Fig. 1.—

Reported tuberculosis (TB) cases in the USA between 1980 and 2000. TB cases rose between 1984–1992 due to the effects of human immunodeficiency virus (HIV)‐infection, immigration and deterioration of the public health infrastructure, particularly in urban areas. Throughout the 1990s considerable investment was made into the development of directly-observed therapy short-course (DOTS) programmes. By 2000, >70% of the potentially communicable cases were on DOT. Despite the continued effects of HIV and immigration (which did not abate in the 1990s), case rates have dropped consistently.

Globally, DOT has been endorsed by the World Health Organization (WHO) in a modified model called “DOTS”, directly observed therapy short-course 13. Acknowledging that directly observed treatment for the entire duration may not be feasible, WHO originally stipulated that DOTS should have the following five elements: 1) political commitment to effective treatment; 2) an assured supply of medications; 3) diagnosis by sputum microscopy; 4) analysis of the entire cohort of patients initiated on treatment; and 5) directly-observed treatment, especially during the first 2 months. Most programmes in the developing world do not use intermittent regimens since all doses are not supervised. Rather, they rely on daily self-administered treatment assuming that, by employing various support measures, an adequate number of doses will be taken.

The most controversial element of the DOTS model is reliance upon sputum microscopy, not culture, for diagnosis. The two major drawbacks of microscopy are its insensitivity and inability to identify drug-resistant strains of TB. Microscopy in developing nations is typically performed on unconcentrated sputum using Ziehl-Neelsen staining. Unfortunately, this system only detects patients with very extensive, typically cavitary, lung disease. Thus, while it has been argued that the most advanced cases that are most likely to transmit to others are found, roughly one-half of the patients with active pulmonary TB would not be detected by this approach, and these unrecognised patients would continue to spread TB until death or diagnosis intervene.

Inability to detect drug resistance is the other drawback of microscopy. In the DOTS model, drug resistance is inferred by failure to respond to treatment, typically after 6 months of therapy. Three obvious problems arise from this approach: 1) progressive damage to the lungs, even death, from uncontrolled disease; 2) ongoing transmission of microbes that are extremely difficult to treat; and 3) the possibility of “amplifying” drug resistance, e.g. the patient begins therapy with INH‐ and RIF‐resistant disease and, during treatment, acquires resistance to PZA and/or EMB. The current author believes that it is essential that drug-susceptibility testing be developed for use in less prosperous regions of the world.

The epidemiology of drug resistance has been analysed in two recent international surveys sponsored by the WHO and the International Union Against Tuberculosis and Lung Disease (IUATLD). These surveys documented wide-ranging rates and patterns of drug resistance around the globe 14, 15. Resistance was seen most frequently to INH and SM, the two agents used most extensively over the years. Clinically, the most significant pattern involves resistance to both INH and RIF, so-called “multidrug-resistant TB” (MDR‐TB).

Analysing the importance of various patterns of drug resistance, Mitchison and Nunn 16 at the BMRC recognised early on the ominous implication of MDR‐TB. Indeed, multiple studies from the BMRC have shown that resistance to INH alone has very little impact on the outcome of therapy 6, 10, 17. Thus, it has become apparent that the rifamycin antibiotics are the keystone of modern, short-course therapy and that future strategies must focus upon conserving susceptibility to those agents.


Looking toward the future of TB treatment, the next section will focus on the issues identified in table 2. These issues will be addressed below.

View this table:
Table 2

Future issues in tuberculosis (TB) therapy

Shorter treatment?

Figure 2 represents a meta-analysis of treatment trials involving regimens of varying duration. The “shoulder” of the sigmoid curve is between 4.5 and 6 months of therapy. Assuming that an appropriate objective for performance of a regimen is 95%+ cure rates, steps that can be taken to “shift the curve to the left” should be considered. Conceptually, this might be accomplished by strengthening either the early bactericidal activity (EBA) or the late sterilising effects (killing bacilli in diminished states of metabolism so that relapses do not occur after treatment is terminated) of the regimen. Early mycobacterial death has been shown to primarily be the product of INH, RIF and PZA 8. Although INH had the most prominent EBA as a single agent in a BMRC study 18, the dramatic killing effect seen in the first week seems not to be vital to curative therapy. Thus, attention shifts to the rifamycins and PZA, and consideration must be given to novel compounds that might enhance either the EBA or sterilising phase.

Fig. 2.—

A meta-analytical representation of cure rates for tuberculosis regimens (reported in trials from around the world) of varying duration and constituents (American College of Chest Physicians, 1995). The 9‐month regimens consisted of isonicotinic acid hydrazide (INH) and rifampicin (RIF), usually with streptomycin (SM) and/or ethambutol (EMB) but not pyrazinamide (PZA). All of the shorter regimens included INH, RIF, PZA, and SM or EMB. The trials were all done under “study conditions” including directly-observed therapy. Thus, they reflect the regimens' capabilities, not the predictably less successful outcomes under “programme conditions”. Data are presented as mean±95% confidence limits.

RIF is the standard rifamycin. It is typically given to adults in daily or intermittent schedules at 600 mg dosage (or 450 mg for those weighing <50 kg). Increasing the dose in daily therapy is not likely to yield improved results: in a USA trial, 750 mg daily did not have a greater effect than 600 mg 19. In intermittent regimens, increasing the dose above 600 mg resulted in increased toxicity 20. Other rifamycins currently in use include rifabutin and rifapentine. Both of these agents have substantially longer half-lives than RIF, but they have significantly different pharmacokinetics and toxicity. Rifabutin's peak serum level following a standard dose of 300 mg is typically in the range of 0.4–0.5 mcg·mL−1 21. In clinical trials, rifabutin has shown no advantage over RIF 22, 23. When the dose of rifabutin is increased to 450 mg, increased toxicity is seen, particularly neutropenia and thrombopenia 24.

The 13‐h half-life of rifapentine is much longer than the half-life for RIF, 2.5–3 h. In addition, maximum serum concentrations for rifapentine are considerably higher than RIF, in the range of 20 mcg·mL−1 versus 8–12 mcg·mL−1 25, and maximum inhibitory concentrations (MICs) for rifapentine are lower than for RIF, 0.015–0.6 versus 0.06–0.25 mcg·mL−1 26. All of these parameters suggest that rifapentine might have therapeutic advantages over RIF. However, rifamycin is only active in the unbound state and rifapentine is 95% protein bound versus 80% for RIF.

Thus far, clinical trials have focussed on rifapentine's long half-life and the potential for once-weekly treatment (see below). However, the current author believes that studies should be performed to determine whether rifapentine in the initial phase might accelerate early killing beyond RIF, thus allowing shorter regimens.

In addition to agents such as the rifamycin which are primarily effective in killing organisms undergoing routine metabolism, attention is being focussed on compounds that are active against tubercle bacilli in a nonmultiplying, semidormant state. Agents that interfere with these “housekeeping” functions are discussed below in the New Drugs? section.

Increased periodicity?

To facilitate DOT, once-weekly treatment schedules have been studied employing rifapentine. Three major studies have been performed in Hong Kong 27, South Africa/USA 28 and USA/Canada 29, employing slightly different protocols. All trials concluded with 4 months of once-weekly rifapentine and INH, compared to twice- or thrice-weekly RIF and INH. In all three trials, the RIF arms modestly outperformed the rifapentine arms. However, for patients in the Centres for Disease Control and Prevention (CDC) whose sputum cultures became negative after 2 months of treatment, outcomes were comparable 30. The current author's interpretation of these trials suggests that the once-weekly rifapentine regimens could be strengthened by increasing the dose of rifapentine, adding additional agents, such as a potent fluoroquinolone (see below), or improving INH delivery.

An alternative approach would be to employ rifapentine throughout therapy, including the twice-weekly continuation phase treatment. The current author believes that the sustained high-level exposure to a rifamycin, which would be realised with twice-weekly rifapentine might reduce the duration required to assure 95%+ cure rates to the range of 4–4.5 months.

Rifalazil (KRM‐1648) is a novel rifamycin that has an extremely long half-life, possibly related to its propensity for concentration within macrophages 31. Its activity has been demonstrated in the mouse model 3236 and may have potential in the treatment of active disease or of latent infection. However, it is far away from human trials.

New drugs?

The following questions should be asked about new medications. Can they replace medications lost to resistance? Can they improve the performance of conventional regimens versus drug-susceptible disease?

The fluoroquinolones (FQNs) are by far the most promising agents in both of these dimensions. Wild strains of Mycobacterium tuberculosis are predictably susceptible in vitro to FQN 37, 38, and various FQNs have been demonstrated to be active in murine models 39, 40. The most potent of the currently available drugs in descending order of in vitro activity are moxifloxacin, gatifloxacin, levofloxacin, ofloxacin and ciprofloxacin 4044. Considerable experience in human disease documents the utility of ciprofloxacin, ofloxacin and levofloxacin against TB including MDR strains 4549. Indeed, data from Hong Kong 49 and Turkey 50 indicated that the outcome of treatment of MDR‐TB was substantially better with in vitro susceptibility to the FQNs.

While recent studies suggest that moxifloxacin is particularly active, the long-term tolerability and safety of third-generation FQNs like moxifloxacin or gatifloxacin have not been established as they have for ofloxacin and levofloxacin 51, 52, and given the unanticipated toxicity of such agents as temafloxacin, trovafloxacin, sparfloxacin, and grepafloxacin, this should not be regarded lightly. Nonetheless, these drugs merit careful study both as agents to augment conventional therapy and as major drugs for MDR‐TB.

The oxazolidinones are a novel group of antimicrobials 53. One member of the family, linezolid, has been introduced into clinical medicine primarily for the treatment of drug-resistant Gram-positive coccal infections. Based on in vitro and murine activity versus TB 54, the current author and others have used it in the management of extensive drug-resistant TB (INH, RIF, and multiple other first-line and retreatment agents); it has shown some apparent efficacy (unreported data). However, its utility is compromised by toxicity (haematological disturbances and painful peripheral neuritis) and extreme expense. Another member of the family, designated as PNU 10048, apparently has more activity than linezolid versus M. tuberculosis with MICs ranging from 0.03–0.5 mg·mL−1 55. However, plans for its clinical development are not apparent.

Compounds that have potential activity versus nonmultiplying bacilli include congeners of metronidazole and isocitrate lyase inhibitors. Nitroimidazopyrans have been shown to be active compared to static M. tuberculosis populations, and replicate bacilli 56. Metronidazole itself has limited activity against nonreplicating bacilli, but analogues may be more active 57. Recent studies of tubercle bacilli living in a semidormant state in lung tissue have identified the importance of the glyoxylate shunt, particularly involving the enzymatic pathway isocitrate lyase 58. Research is underway to find agents capable of disruption of this system under the premise that this might dramatically improve “sterilisation”.

Other novel drugs with potential utility against TB were recently reviewed by Agrawal et al. 59. Promising agents included thiolactomycins that have activities somewhat analogous, albeit less potent, to INH 60. Additionally, analogues to pyrazinamide or morphazinamide that may be both more active than these agents and not cross-resistant with PZA have been described by a research consortium including Cynamon et al. 6163 and L. Heifets (National Jewish Medical and Research Centre, Denver, CO, USA, personal communication).

Optimal therapy for tuberculosis in persons with acquired immune deficiency syndrome

The human immunodeficiency virus (HIV) has greatly augmented the TB epidemics in sub-Saharan Africa as well as focal populations of South-East Asia, Latin America, the Indian subcontinent, Russia, and urban, industrialised communities.

Most authorities suggest that regimens that are effective among non‐HIV‐infected TB patients are comparably efficacious in those with acquired immune deficiency syndrome (AIDS) 64. However, there are problems in this arena. Several reports suggest that relapse rates are higher in the presence of AIDS, albeit modestly so 6567, and as persons with AIDS live longer due to antiretroviral therapy and opportunistic infection prophylaxis, it is plausible that they will survive long enough to allow more relapses. Furthermore, RIF, the mainstay of modern therapy, has such profound pharmacological effects (via induction of the hepatic Cytochrome P‐450 pathways) that it is not compatible with contemporary antiretroviral regimens 68, 69. Rifabutin, which does not have analogous effects on the cytochrome system, has been used as an alternative 69. However, it is not free from complicating drug/drug interactions 69. Rifapentine is intermediate between RIF and rifabutin in terms of its effect on the cytochrome system 70. However, its use in persons with AIDS has been discouraged due to the evolution of rifamycin monoresistance in a small number of patients in a USA Centres for Disease Control trial 71.

Other problems that loom in this field include possible malabsorption of TB medications 68 and dealing with “the immune reconstitution system” or paradoxical late-worsening of TB when antiretroviral therapy restores immunologically mediated inflammation 7276.


In Shaw's “Doctor's Dilemma”, the same character who lambasted drug therapy advised that people must “stimulate the phagocytes. Drugs are a delusion”. Presumably, inspired by his friend, the Nobel-prize winning immunologist Sir Thomas Almroth-Wright, Shaw anticipated the 20th Century preoccupation with a vaccine to prevent or cure TB.

The bacillus of Calmette and Guérin (BCG), a living vaccine derived from an attenuated strain of M. bovis (a very close relative of M. tuberculosis), has been given to nearly 5 billion persons since its widespread use in the 1930s. Although it does appear to lessen disease, particularly extrapulmonary TB, in young people following primary infection, it has had very meager effects upon adult pulmonary disease 7780. Thus, transmission and deaths rage on largely uncontrolled despite immense programmes of BCG vaccination in developing nations.

BCG presumably exerts its benefits by lessening dissemination through accelerated immune defenses following primary pulmonary infection. Considerable work is being directed at finding a vaccine that is more efficient or durable at this task than BCG 81, 82.

There is also consistent interest in developing an immunomodulating agent, “vaccine” or otherwise, that might help control active disease. This concept is based broadly on the notion that part of the “pathogenic strategy” of TB is the corruption of the human immune response. As tuberculous disease is initiated and advances in the body, there appears to be a shift from a T‐helper cell type 1 (Th1) protective response to a T‐helper cell type 2 (Th2) pathway which is less effective and more injurious to tissue. Vaccines derived from M. vaccae, a saprophytic rapid-growing mycobacterium, have been studied in this regard with highly variable results 83, 84.

Other agents that have been studied or are under consideration include interferon‐γ 85, 86, interferon‐α 87, imiquimod 88, interleukin‐12 89, granulocyte-macrophage colony-stimulating factor 9092, levamisole 93, 94 or transfer factor 9599.


Although the usual case of drug-susceptible TB can be predictably cured in 6 months with a reasonably nontoxic, economical regimen involving as few as 62–78 encounters, novel methodologies must be established if TB is going to be controlled in the decades ahead.

Given the drugs/drug-groups available now and in the near future, it is highly unlikely that the duration of therapy will be able to be reduced to <4 months. To shorten therapy below this level will take either novel agents that are active versus the semidormant, sporadically multiplying microbes left behind after the initial dramatic killing effects, or an immunological modulator that substantially enhances the host's cellular immunity.

Unfortunately, despite the massive burden that tuberculosis constitutes, there are not good economic incentives for the pharmaceutical industry to invest in these endeavors. In response, the “Working Alliance for Tuberculosis Drug Development” was formally launched in Cape Town, South Africa in February 2000 100. It pledged to accelerate the development of new tuberculosis drugs by a system of partnerships, formulation of plans, reassessment of pharmaco-economics, and advocacy. Sponsored by an array of nongovernmental organisations, private philanthropy and governmental agencies, this “Global Alliance” has embarked upon a vital but formidable task.

  • Received January 25, 2002.
  • Accepted March 13, 2002.


View Abstract