The T-cell response of human donors to secreted antigen fractions of Mycobacterium tuberculosis was investigated. The donors were divided into five groups: active pulmonary tuberculosis (TB) patients with minimal and with advanced disease, Mycobacterium bovis BCG-vaccinated donors with and without contact with TB patients, and nonvaccinated individuals. We found that patients with active minimal TB responded powerfully to secreted antigens contained in a short-term culture filtrate. The response to secreted antigens was mediated by CD4 ؉ Th-1-like lymphocytes, and the gamma interferon release by these cells was markedly higher in patients with active minimal TB than in healthy BCG-vaccinated donors. Patients with active advanced disease exhibited depressed responses to all preparations tested. The specificity of the response to secreted antigens was investigated by stimulating lymphocytes with narrow-molecular-mass fractions of short-term culture filtrate obtained by the multielution technique. Considerable heterogeneity was found within the donor groups. Patients with active minimal TB recognized multiple secreted targets, but interestingly, six of eight patients demonstrated a predominant recognition of a low-mass (<10-kDa) protein fraction which induced high levels of gamma interferon release in vitro. Only a few of 12 previously characterized secreted antigens were recognized by T cells isolated from TB patients, suggesting the existence of a number of as yet undefined antigenic targets among secreted antigens.

Most antituberculous drugs, with the notable exception of streptomycin, are prone to cause liver injury. The hepatotoxic potential of isoniazid given alone is well established,' while data on the hepatotoxicity of rifampicin, pyrazinamide, and ethambutol are difficult to interpret since these drugs are almost always used in different combinations.2 The evidence supporting possible hepatotoxic interaction between rifampicin and isoniazid is circumstantial.3The requirement of a hepatotoxic drug interaction is the absence of hepatotoxicity when the drugs are administered separately and the presence ofhepatotoxicity when given together. We report the first case where a hepatotoxic drug interaction between rifampicin and isoniazid is firmly documented. This interpretation proved to be essential for optimal treatment of our patient.Case report A 35 year old black woman from Somalia was admitted to our department with a seven month history of dry cough, night sweats, and weight loss of 15 kg. A chest radiograph revealed diffuse fine nodular shadowing, and a diagnosis of miliary tuberculosis was confirmed by the presence of acid fast bacilli in the bronchoalveolar lavage fluid.Treatment was initiated with the Danish four drug routine regimen of isoniazid 300 mg, rifampicin 450 mg, ethambutol 1 g, and pyrazinamide 1 5 g (body weight 46 kg) daily. Five days before the start of treatment the serum level of aspartate aminotransferase (ASAT) was slightly raised at 59 units/l (upper limit of normal (ULN) 35 units/l), whereas the bilirubin level was normal at 11 jtmol/l (ULN 17 1imol/l) (fig 1). After four days of antituberculous treatment the patient developed abdominal pain, nausea, and malaise and ASAT levels started to increase (fig 1). Bilirubin rose to 26,mol/I and alkaline phosphatase to 1-5 times ULN. The patient was not on any other potentially hepatotoxic drug and did not consume alcohol. IgM anti-HAV, HBsAg, and IgM anti-CMV were negative. An ultrasound scan of the liver was normal.Drug treatment was discontinued and bilirubin and ASAT levels fell (fig 1). Treatment was reinstituted seven days later with a daily dose of 100 mg isoniazid, increased to 300 mg after three days. The liver profile remained stable and a dose of 150 mg rifampicin was added. After four days (the day after the dose of rifampicin was doubled to 300 mg) a steep increase in ASAT levels was observed (fig 1). Bilirubin levels reached 31 itmol/l and alkaline phosphatase rose to almost twice the ULN. The patient developed abdominal pain, nausea, and vomiting. Rifampicin was withdrawn and isoniazid treatment was supplemented by a daily dose of streptomycin 1 g, followed one week later by ethambutol. ASAT, bilirubin, and alkaline phosphatase levels returned to normal (fig 1).The general condition of the patient did, however, gradually worsen, and 26 days after rifampicin withdrawal (58 days after starting antituberculous therapy) the isolated Mycobacterium tuberculosis proved to be resistant to isoniazid. Accordingly, isoniazid was withdrawn and rifa...

Recent information from several laboratories points to proteins secreted from live Mycobacterium tuberculosis as being involved in protective immunity. We have studied protein release from M. tuberculosis during growth and have defined a short-term culture filtrate enriched with secreted antigens and with a minimal content of autolytic products. Several components in this filtrate have been characterized and are presently being studied in animal models of tuberculosis. In the mouse model, protection against tuberculosis is mediated by a long-lived population of memory T cells directed against secreted molecules, and the antigens ESAT-6 and the 85 complex are important key targets for this response. The specificity of the protective immune response in humans is less well defined, but our studies indicate that antigen-specific CD4+ cytotoxic T cells and memory T cells secreting interferon-γ (IFN-γ) are involved. The IFN-γ response is directed against a range of the culture filtrate proteins with a preferential recognition of low molecular mass antigens.