Intended for healthcare professionals


Diagnosing multidrug resistant tuberculosis in Britain

BMJ 1998; 317 doi: (Published 07 November 1998) Cite this as: BMJ 1998;317:1263

Clinical suspicion should drive rapid diagnosis

  1. F A Drobniewski, Director
  1. Public Health Laboratory Service Mycobacterium Reference Unit, Dulwich
  2. Public Health Laboratory, King's College Hospital, London SE22 8QF

    Each year an estimated 8 million new cases of tuberculosis occur, leading to 3 million deaths; and almost a third of the world's population is infected by the causative bacterium, Mycobacterium tuberculosis. 1 2 The incidence of tuberculosis is rising in Britain too, and London has seen two recent outbreaks of multiple drug resistant tuberculosis—that is, resistance to at least isoniazid and rifampicin—in patients coinfected with HIV.3 This is one of several dramatic outbreaks of multidrug resistant tuberculosis described recently in Europe and the United States. Although overall the problem of resistance remains small in the UK, lessons from recent cases can help clinicians suspect multidrug resistant tuberculosis and enable their laboratory colleagues to identify it quickly. Updated guidelines on the management of cases of tuberculosis have recently been published.4

    The UK drug resistance surveillance system MYCOBNET, coordinated by the Public Health Laboratory Service, shows that levels of multidrug resistant tuberculosis in the UK are relatively low but the incidence is higher in London than elsewhere. In 1993-6 rates of multidrug resistant tuberculosis in initial isolates rose from 0.6% to 1.7% overall; in 1993-6 1.9% of isolates from greater London were multidrug resistant compared with 0.9% outside the capital.5

    As shown in the recent London outbreaks, patients with HIV are particularly at risk of developing multidrug resistant tuberculosis. In immunocompromised patients such as those with HIV tuberculosis often presents atypically, and almost any chest x ray abnormality is compatible with tuberculosis. Acid fast bacilli are less likely to be seen on a sputum smear, but if they are present a presumptive diagnosis of tuberculosis should always be made, with treatment targeted against M tuberculosis (and Mycobacterium avium-intracellulare if the patient is HIV positive). Early diagnosis of multidrug resistant tuberculosis followed by treatment with three drugs to which the organism is susceptible reduces mortality significantly in both HIV negative and HIV positive patients. 6 7 So what is the best way of getting a rapid definitive diagnosis?

    Molecular techniques can speed diagnosis and detect resistance within days but, as well as being expensive to use, they are neither necessary nor appropriate in all cases. Molecular amplification techniques have sensitivities and specificities of 90-95% for smear positive sputa, but, although specificity remains high in smear negative samples, sensitivity is low (40-77%)8 and positive and negative predictive values—the probability that a patient does or does not have tuberculosis with a positive or negative result respectively—will depend on the degree of clinical suspicion that a patient has tuberculosis. Amplification tests, if done correctly, are expensive owing to the number of concurrent controls that must be run and should be used only when the result is likely to change management. When done badly high rates of false positivity have been recorded in coded samples.9

    In the UK 90-95% of rifampicin resistant isolates are resistant to isoniazid, so rifampicin resistance is predictive of multidrug resistant tuberculosis. The PHLS mycobacterium reference unit can identify M tuberculosis in smear positive patients and predict rifampicin resistance in over 90% of cases within three to four days (and offers this service nationally), but it would be inappropriate to perform this analysis on every specimen, as over 90% of new cases are drug sensitive and will be successfully treated with empirical therapy. Good communication between clinicians and laboratories is thus essential to identify those patients with risk factors for resistance, for whom such rapid testing is appropriate.

    The mechanism of action of drugs other than rifampicin has not been completely defined at the genetic level, so if the best available techniques were used to predict isoniazid, aminoglycoside, or fluoroquinolone resistance, roughly 10%, 40%, and 25% of resistant isolates, respectively, would be missed. Culture based systems are likely to be required for some time, and non-radiometric rapid culture systems offer a compromise between slower solid culture systems and faster molecular amplification methods until further research reveals new mechanisms of drug resistance and methods of diagnosis.10

    In London and comparable cities, which have the greatest problems with multidrug resistant tuberculosis, the Public Health Laboratory Service mycobacterium reference unit has proposed a streamlined model system which integrates clinical information with laboratory systems to use the most appropriate means of testing sputum samples from smear positive patients.

    Hospital laboratories currently perform microscopy and usually culture specimens on to solid media; over 95% of hospitals refer these cultures to the mycobacterium reference unit for identification and drug susceptibility testing. If reference units use faster methods to identify and determine the drug resistance of these cultures this will have only a modest impact on public health as the sending laboratories using solid media cultures are likely to take weeks to isolate mycobacteria and to fail to culture some significant isolates. The new plan is that hospitals would continue to examine direct sputum smears but should send the first positive sputum sample from a patient for rapid culture at the reference laboratory.

    In the vast majority of patients this sample will be cultured in a rapid culture system. Cultures positive for acid fast bacilli detected in an average of 14 days would then be analysed using DNA hybridisation probes for the presence of M tuberculosis or M avium-intracellulare. M tuberculosis positive specimens would be reported and inoculated into a rapid culture system for isoniazid and rifampicin detection and on to solid media for detecting resistance to other first line drugs. Results for isoniazid and rifampicin susceptibility would be reported in a further seven days (see figure). Thus in most cases a diagnosis of tuberculosis and sensitivities could be provided simultaneously to microbiologists, physicians, and consultants in communicable disease control within 30 days, the standard proposed by the Centers for Disease Control in the United States


    An integrated model for reducing transmission of tuberculosis in London

    For patients with a high clinical risk of resistant tuberculosis molecular amplification techniques would be used to identify tuberculosis and predict rifampicin resistance (see figure), and these results would be available in three to four days. This model should improve diagnosis, communication, and collaboration between key parties involved, surveillance at local level, and preserve local laboratory skills.


    View Abstract