Lab tests that detect antibodies to TB in blood serum do not perform well in Botswana according to a study presented at the 1st National HIV and AIDS (NHASORC) last month in Gaborone. The study looked at several different types of antibody tests and found each lacking.
One of the greatest hurdles to more effective management of TB is the lack of an inexpensive and easy to use technique to diagnose active infection in patients. The gold standard for TB diagnosis is to culture the organism in a laboratory, either from sputum or blood. Culturing is both highly specific and highly sensitive but requires dedicated equipment and takes up to 8 weeks to yield results.
Thus, the World Health Organization recommends that sputum smear microscopy be used for diagnosis, since it is much faster. However, while the technique is very specific for the condition (a positive result is correct over >99% of the time) it has only limited sensitivity (a range of only 40-60%) which means that about half of the time infection may be present even if the laboratory technician sees nothing. Sensitivity is even lower in children and people with HIV. A chest radiograph is commonly recommended but it is not specific for TB (other conditions can look similar to TB on X-ray) and is not always available in resource-limited settings.
An inexpensive, highly sensitive and specific diagnostic test for TB is greatly needed. One alternative could be serodiagnostic tests, rapid tests that are used to test blood and serum for TB antibodies. These tests are rapid and inexpensive and do not necessarily need a typical laboratory for processing. According to Dr. Letsatsi "an ideal serodiagnostic test should be rapid and with results available in 25 minutes or less; capable of testing few or many patients; simple and safe to use by lab techs without extensive training and affordable at (<$1) per test."
She presented the results of a study examining four types of serodiagnostic tests. She spoke mostly about one type, called the immunochromatographic strip, or ICS. This is a strip test for TB using blood or serum to determine antibodies unique to TB and the results are available within an hour. The sensitivity of the ICS in previous studies (from 284 HIV-negative adult patients in Brazil, India, and four African countries) showed a sensitivity of 72% on average, ranging from 62 to 100%.
"The ICS is simple to use outside the laboratory, and is more suitable to field conditions than Elisa-based tests, which require a specific laboratory set-up," she said. It was hoped that the integration of the ICS test in to existing TB diagnostic investigations (smear and culture) would improve case detection.
However, her team first needed to know how the test would perform in Botswana, particularly because it has such a high prevalence of HIV. They also evaluated several other serodiagnostic tests including those from Mycodot, American Bionostica, Osborne Scientific, and Omega Diagnostic. Another goal of the study was to investigate the usefulness of mycobacterial blood cultures for TB diagnosis, which are not routinely performed in Botswana.
TB suspects (persons reporting cough for 2 weeks or more) were identified at two sites. Persons who had recently received a blood transfusion or TB treatment were excluded from the study. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were calculated for the study tests of interest.
A total of 465 patients were enrolled in the study. The median age was 36 years (range 21-89), 53% were male and 83% were HIV infected. 175 patients (38%) were diagnosed with confirmed tuberculosis. Of these, 93 (53%) had TB diagnosed from sputum; 57 (33%) were diagnosed from sputum and blood; and finally, 26 (15 %) were diagnosed from blood culture only. "In this last group," Dr. Letsatsi noted, "mycobacterial blood culture was the sole source for TB diagnosis. Mycobacterial blood cultures should therefore be considered a viable adjunct to diagnosing TB in settings with a high volume of HIV and TB."
Demographic and clinical characteristics among TB cases and non-cases were compared. Characteristics such as weight loss, diarrhoea, fever, chest pains and cough were evaluated but were not significantly different between those with TB and without TB. Abnormal chest x-rays were common to both groups. Those who had TB tended to be younger patients. Fewer complained of headaches but were more likely to suffer from night sweats. Patients with TB were more likely to be infected with HIV and also to have a CD4 count of less than 200 cells. However, HIV infection was common among both groups. Patients with TB were also more likely to be anaemic and had a lower platelet count.
Unfortunately, the ICS test did not perform up to expectations.
ICS to serum showed a sensitivity of 27% and specificity of 75% while ICS to whole blood performed even worse with a sensitivity of 11%. Results were similar when stratified by HIV status, CD4>200, sputum status and other factors.
Unfortunately, "all the serodiagnostic tests performed poorly in this population," said Dr. Letsatsi. The relative sensitivities, specificities, positive predictive and negative predictive values for the tests are as follows. For the ICS: 27%, 75%, 39%, and 63%; Osborne Scientific 37%, 63.%, 36%, and 36%; MycoDot 3%, 99%, 55%, 55%; and American Bionostica 0, 99%, 0, and 0.
“The currently evaluated serodiagnostic tests were not sufficiently sensitive and specific for routine application in the diagnosis of TB in Botswana or similar settings with high prevalence of HIV," concluded Dr. Letsatsi. She noted that part of the tests' failure in this setting could have been due to the fact that specimens were transferred to the national referral lab for testing, and that this delay prior to testing may have led to poor test performance. However, a member in the audience pointed out that just as patients with HIV and latent TB are often anergic (lacking an antibody response to TB), so too the antibody response in some patients with active TB may also be too weak for such tests to detect.
References
Letsatsi P et al. Tuberculosis serodiagnosis in HIV infected persons, Botswana, 2002. First National HIV/AIDS/STI/Other Related Infectious Diseases Research Conference, Gaborone, Botswana, abstract WBT53-9.