|Home | About | Journals | Submit | Contact Us | Français|
Drugs that increase the risk of diabetes include corticosteroids, thiazides and beta blockers.1 In addition certain antipsychotic2 and antiretroviral drugs3 have been associated with hyperglycaemia and the metabolic syndrome. Such effects have not been reported with antituberculosis chemotherapy.
A Somalian woman of 38 was referred to the chest clinic in August 2001 because of cough, night sweats, haemoptysis and bilateral cervical lymphadenopathy. She had been living in the UK for 4 years. Her only medication was cyclical oestrogen for irregular periods. 2 years previously her general practitioner had noted a slightly raised random blood glucose (10.1 mmol/L) but an oral glucose tolerance test had been normal (fasting glucose 4.6 mmol/L; 2 hour, 6.7 mmol/L). A chest radiograph showed mild elevation of the left hilum suggestive of previous tuberculosis but no signs of active disease. The erythrocyte sedimentation rate was 14 mm/h, C-reactive protein slightly raised at 6 mg/L. All other blood results, including liver function tests, calcium and full blood count were within normal limits (fasting glucose 4.5 mmol/L). Sputum cultures were negative for Mycobacterium tuberculosis but excision biopsy of a cervical lymph node showed caseating granulomata highly suggestive of active tuberculosis.
In December 2001 she was started on Rifinah-300 (two tablets, each containing rifampicin 300 mg, isoniazid 150 mg), ethambutol 1200 mg, pyrazinamide 2 g and pyridoxine 10 mg, all daily. Pyrazinamide was stopped after two months and the other treatments were continued for a further four months. She improved symptomatically and her cervical lymphadenopathy lessened. The tuberculosis nurses thought she had adhered well to the treatment, and chemotherapy was stopped in June 2002. A year later, she returned to the chest clinic because of increasing cervical lymphadenopathy, fever and sweats. The chest X-ray showed increased shadowing in the left upper zone, and sputum microscopy was positive for acid-fast bacilli. Treatment was resumed with Rifinah 300 and ethambutol in the previous dosage. Six months into treatment, in January 2004, she reported polyuria, polydipsia and weight loss. The antituberculosis treatment was her only medication. Random plasma glucose was 17.9 mmol/L and she had heavy ketonuria. On referral to the diabetes clinic she was started on subcutaneous mixed insulin twice daily; a dietitian found her diet to be good, with little excess refined carbohydrate or fat. On 30 units of mixed insulin daily good glycaemic control was achieved. Tests for glutamic acid decarboxylase and islet-cell antibodies were negative. Her tuberculosis chemotherapy was stopped three months later when she developed signs of ethambutol ocular toxicity, and soon afterwards she was troubled by frequent unpredictable hypoglycaemic symptoms. Under the guidance of the diabetes nurses, her insulin doses were reduced weekly. Within three weeks she was able to stop her insulin, and home glucose monitoring then showed preprandial values in the range 4-5 mmol/L, postprandial 6-7 mmol/L. In an oral glucose tolerance test, fasting glucose was 5.3 mmol/L, 2 hour glucose 4.9 mmol/L.
Abnormalities of glucose tolerance are found in around 20% of patients with active tuberculosis,4,5 and indeed patients with diabetes are thought to have a 6-7-fold excess risk of developing tuberculosis.6 Little has been published on hyperglycaemia due to antituberculosis drugs, though the British National Formulary does list this as a potential adverse effect of isoniazid. We have found no previous report of insulin-requiring diabetes developing during tuberculosis chemotherapy and resolving on cessation of treatment. In the present case, cause and effect seem likely, though other factors cannot be excluded. So far as we could ascertain, the patient's dietary habits did not change during or after treatment, and she was not taking any other therapy that might have influenced glucose metabolism (including 'alternative' therapies). There are case reports of rifampicin being associated with abnormalities of glucose tolerance. An increased requirement for oral hypoglycaemics7,8 has been reported, probably due to the effect of enzyme induction, which speeds the metabolism of these drugs. In one noteworthy case, a patient with type 1 diabetes required more insulin when treated with rifampicin and returned to previous dosage when rifampicin was stopped.9
How might tuberculosis chemotherapy induce hyperglycaemia? In patients taking rifampicin for tuberculosis Japanese investigators obtained normal results for intravenous glucose tolerance but hyperglycaemia after oral glucose.10 No such effect was seen with other agents including isoniazid, ethambutol and streptomycin. These workers suggested that rifampicin either augments intestinal absorption of glucose or reduces insulin sensitivity.