A new prototype test for TB lacks sufficient sensitivity as a sole test for TB in a predominately HIV-infected population in Botswana, according to a study published in July 1st edition of Clinical Infectious Diseases. Five other commercially available serodiagnostic TB tests evaluated in this study were also found to be insufficiently sensitive as sole diagnostics.
The study illustrates the challenges facing doctors when attempting to diagnose TB in settings where both HIV and TB are widespread. Results of TB tests may be confused by a high background level of other mycobacterial infections (resulting in false positive results) or extrapulmonary TB infection in the presence of immune deficiency (resulting in false negative results).
The correct diagnosis is important if doctors are to prevent the spread of TB from person to person, particularly where HIV infection is widespread.
Correct diagnosis is also important in patients with HIV beginning antiretroviral therapy. Immune reconstitution illness due to TB may be difficult to manage. It is also possible for TB to develop even after antiretroviral treatment has begun, when the use of antiretroviral drugs may complicate the treatment of TB due to interactions with TB drugs. This is why it is usually recommended that the first two months of TB treatment take place before antiretroviral treatment begins, if at all possible.
A related problem is the need to distinguish active TB from latent TB, since giving the standard treatment for latent TB (isoniazid alone) to someone with active tuberculosis may lead to drug resistance.
Diagnosing TB in HIV-infected populations
HIV infection has contributed to a substantial increase in the incidence of TB. One third of deaths among HIV infected patients are attributed to TB, making it the leading cause of death among people with AIDS. In Africa, where the HIV pandemic is at its worst, TB accounts for over 40% of HIV-attributed deaths, with over 84% of hospitalised patients suspected of having TB in Botswana infected with HIV.
The World Health Organisation promotes the detection of TB by using sputum smear microscopy; however, as a diagnostic tool this has its limitations. Sputum smear microscopy is only 40-60% effective in diagnosing pulmonary TB over three examinations, and this is substantially lower for those with HIV co-infection. This emphasises the need for the development of a far more accurate serological test. This test should not only be capable of discriminating between latent and active TB, but also distinguishing between BCG vaccination and exposure to nontuberculosis mycobacteria. It must be capable of monitoring response to treatment, be simple to use and above all affordable in the context of the developing world.
465 patients with suspected TB (defined as cough lasting more than two weeks) from two hospitals in Botswana were enrolled in this study. Patients were defined as having TB on the basis of any positive smear or culture. A total of 384 patients (83%) were infected with HIV, with 39 (8%) being HIV uninfected and 42 (9%) declining testing or having missing results. Of these patients 175 (38%) were diagnosed with TB by at least positive test result. Among the patients with TB, 72 (41%) had positive results of sputum smear and blood culture, 38 (22%) had positive smear and negative culture results, and 39 (22%) had negative smear and positive culture results. The remainder of the patients with TB (26 (15%)) received diagnosis solely on the basis of positive blood culture results.
This study was designed to prospectively evaluate the use and accuracy of an immunochromatographical strip test (ICS) for diagnosing TB, developed by the Program for Appropriate Technology in Health (PATH; Seattle). The sensitivity and specificity of a further 5 other commercially available products or newly available prototypes for TB serodiagnostic analysis were also evaluated. Among the tests evaluated, the sensitivity was 0%-63%, the specificity was 39%-99%, the positive predictive value was 0-39%, and the negative predictive value was 63%-65%.
Co-infection with HIV reduces the accuracy of TB assays
TB serodiagnostic tests are desperately needed in such settings where patients with suspected TB are hospitalised in areas of high prevalence of HIV infection. Because HIV infection has become such a prominent feature of the TB epidemic in many parts of the world any evaluations of new TB tests should examine the effects of HIV on the assay results. Many assays have proved relatively successful in diagnosing TB in HIV-uninfected subjects, but when used to diagnose TB in HIV-infected subjects they have proved far less effective.
The ICS test was shown to have a good performance rate when analysing banked serum samples in the laboratory-based development phase. Unfortunately, when this was field-tested it was found to lack sufficient sensitivity as the sole test for TB. Both whole blood samples and serum samples were tested. Whole blood samples proved to be far less accurate than using serum samples. The ICA test (in serum) was shown to be far less effective in detecting TB in HIV-infected patients when compared to uninfected patients (26% vs. 40%) and this was further reduced in patients with low CD4 counts. The other assays used techniques developed to detect antibodies to M.tuberculosis rather than the bacteria itself. As there is decreased mycobacterial antibody production in patients co-infected with HIV it is perhaps unsurprising that these assays proved even less reliable than the ICS test. It may be that the tests are not sensitive enough to detect such low antibody production.
The authors conclude that the serodiagnostic tests evaluated in this study are not sufficiently sensitive and specific to routinely diagnose TB in regions where HIV infection is highly prevalent. With the TB epidemic continuing to grow in the face of increased HIV infection rates, a suitable tool for diagnosing TB in HIV-infected patients that can be used effectively in both a developed and developing country context is still desperately needed.
Further information on this website
Treating latent TB - HIV & AIDS Treatment in Practice newsletter September 2003
Two advances in TB diagnostics at South African HIV conference – news story September 2003
New British test may improve TB diagnosis rate – news story April 2003
Need for new TB test highlighted on World TB Day – news story March 2004
References
Talbot EA, Hay Burgess DC, Hone NM et al., Tuberculosis serodiagnosis in a predominantly HIV-infected population of hospitalized patients with cough, Botswana, 2002. CID 39, 2004