Mdd Mal Infect. 1996 ; 26 : 904-10 Cellular immune responses in tuberculosis: protection and immunopathology* G.A.W. ROOK *~ and R. HERNANDEZ-PANDO*** SUMMARY Tuberculosis (TB) patients relapse if treatment is not continued for 6 months, because chemotherapy fails to convert the patients' response from the necrotising pattern characteristic of the disease (Koch phenomenon), to the non-necrotising bactericidal function required for immunity. We need to understand the nature of these two immunological states, and we need to learn how to convert one to the other. Recent data indicate that protection requires a Type 1 cell-mediated response. Cytokine TNFc~ is also required, and yet there is evidence to implicate TNFc~ in the pathology of the disease. This paradoxical dual role of TNFc~ may provide the key to understanding TB. TNFc~ causes no tissue damage if injected into an inflammatory site mediated by a "pure" Thl response. In such sites, it acts as an additional macrophage-activating factor. In sharp contrast, TNF(z is toxic if injected into inflammatory sites mediated by mixed Thl+~Th2 (?Th0) T cell responses. DTH response sites in mice with progressive pulmonary TB are also TNFc~-sensitive. Moreover if the immunological state accompanying progressive disease (mixed Thl/Th2 and TNFc~-sensitivity) is deliberately primed in mice before they are infected, they are rendered more susceptible to pulmonary TB than unimmunised control animals. This reminds us of earlier work showing that preimmunised guinea-pigs (so as to have the Koch phenomenon) are more, rather than less, susceptible to deep infection. This state of increased susceptibility to TB can be induced by using a saprophytic environmental mycobacterium. Similarly the same organism can be used to vaccinate (by evoking a "pure" Thl pattern with a dose that is 100-fold lower) or to partially treat established disease in man and mice. Therefore epitopes that are common to M. tuberculosis and to a fast-growing saprophyte (likely to include heat shock proteins) can play key roles in protection and immunopathology. This may explain the effect of contact with environmental species on the variable efficacy of BCG. Thus, it seems that these crucially important common epitopes do not evoke the Koch phenomenon, and in fact skin-test responsiveness to these is lost during progressive TB, while antigen extracts of M. tuberculosis still cause necrosis. This raises questions as to why the common epitopes are handled differently, and an important fact, it means that they are safe to use in human immunotherapy trials. Key-words : Tuberculosis - Immunopathology - Immuno-endocrinology - Cytokines. The key issue for immunologists interested in tuberculosis is the need to understand the difference between the necrotising pattern of response to the antigens of M. tuberculosis found in tuberculosis patients, and bactericidal protective immunity which operates in the majority of people infected with this organism. It was Robert Koch who first observed that guinea- pigs infected 1-2 months previously with virulent tubercle bacilli respond to an injection of further bacilli or bacterial * 45e r6union conjointe organis6e h l'initiative du Comit6 d' Interface entre I'INSERM et la Socidt6 de Pathologie Infectieuse de Langue Frangaise - Paris - 9 novembre 1995. ** Bact6riologie, UCL Medical School, Windeyer Building, 46 Cleveland St. London W1P 6DB - Grande-Bretagne. *** lnstituto Nacional de la Nutricion, Salvador Zubiran, 14000 Mexico DF - Mexique. culture supernatant with a necrotising reaction both at the site of injection, and in the original lesions (1). He at first misunderstood the significance of this finding, because the necrosis resulted in sloughing of skin challenged with live organisms and so superficially resembled immunity. Later authors pointed out that if animals are deliberately pre- immunised so that they express the Koch phenomenon before deep intramuscular challenge, where necrosis cannot result in sloughing of the inoculum, they prove to be more susceptible than unimmunised controls (2). The same point is apparent in the context of human disease. After 2-3 months of standard chemotherapy there are very few live organisms, but if • d . treatment is stopped at this time there is a very high relapse rate (3). Therefore the pattern of response existing in the patient is remarkably inefficient. At best it has some limited ability to wall off the bacilli within a zone of necrosis and 904