Use of the TB LAM test in people with HIV with CD4 counts below 200 boosted the number of people diagnosed with tuberculosis (TB) by a third, an observational study in Malawi and Mozambique has found. Almost half of TB cases would have been missed if LAM had not been included in the diagnostic algorithm, showing that the test is useful in a far wider population than currently recommended.
The study, carried out by Médecins sans Frontières, is published in the journal PLOS Medicine.
The lateral flow urine lipoaribomannan assay, or LAM, detects a constituent of the cell wall of TB bacteria in urine. The assay can be carried out without electricity at the point of care, takes 25 minutes and costs less than $3.50 a test.
A urine-based assay improves the chances that a sample can be produced for analysis. One study of LAM found that 40% of hospitalised patients were unable to produce a sputum sample but almost all could provide a urine sample. People with advanced HIV disease tend to have more difficulty in producing a sputum sample and smear microscopy using sputum is less likely to produce a positive result in people with advanced HIV disease who have disseminated TB.
Previous studies have shown that the TB LAM assay is better at excluding TB as a cause of illness than it is at correctly identifying people with TB. Nevertheless, use of TB LAM as an additional diagnostic tool reduced the risk of death in hospitalised people with HIV by 17% in a randomised trial. The World Health Organization recommends the inclusion of the TB LAM assay in TB diagnostic algorithms for people who are seriously ill or who have CD4 cell counts below 100 cells/mm3.
Until now, there has been a lack of evidence concerning the value of TB LAM testing in people with CD4 counts below 200 who are not hospitalised. This group of people is at high risk of TB so a point-of-care test that can be used at time of antiretroviral treatment initiation or when a person is identified as having a low CD4 count has the potential to improve TB diagnosis and reduce TB-related mortality.
A prospective observational study was carried out in Malawi and Mozambique. The study recruited 456 consecutive patients with CD4 counts below 200 and at least one TB-related symptom to undergo LAM, microscopy and Xpert MTB/RIF testing (Xpert is a nucleic-acid based platform which can detect mTB DNA).
The median CD4 cell count of participants was 50 cells/mm3, 53% were already on antiretroviral therapy (ART) at the initial study visit and 22% were seriously ill (fever, unable to walk, elevated heart rate or elevated respiratory rate). People in Mozambique were more likely to have a low CD4 cell count or to be seriously ill and less likely to be on ART.
Of the individuals recruited to the study, 205 had laboratory-confirmed TB (defined as a positive result on any of LAM, smear microscopy or Xpert).
Looking at diagnostic yield of the various tests – the percentage of results which enabled clinicians to make a diagnosis – the investigators found that LAM produced a much higher diagnostic yield (82.4%) than sputum microscopy (33.7%) or Xpert (41%). The low yield of sputum microscopy and Xpert in this study reflects the low proportion of people able to produce sputum samples, the poor quality of many samples and difficulties in processing and transporting sputum samples. This finding, in real-world operational conditions, underlines the importance of developing sensitive point-of-care TB tests that do not rely on sputum sampling and transport to laboratories.
Furthermore, the study found:
- 25% of people who were negative on all sputum-based tests were positive by the LAM test
- 39% of people unable to provide a sputum sample were positive by the LAM test
- 34% of people positive by a sputum-based test were negative by the LAM test
- Half of all laboratory-confirmed TB was diagnosed only through the LAM test
- Microscopy and LAM together far outperformed microscopy alone. So did Xpert and LAM together when compared to Xpert alone.
In Malawi, LAM test results were blinded to clinicians for the first nine months at the request of the national TB programme. During this period 19 people were positive for TB by LAM testing but negative by other assays and were not treated. 36.8% died, compared to 12.8% of LAM-positive patients who were treated. Untreated people were two-and-a-half times more likely to die (adjusted risk ratio 2.57, 95% CI 1.29-5.19, p = 0.009).
In this study population, the sensitivity of TB LAM was 65%, in other words, it missed around one-third of TB cases. However, the study authors say that diagnostic yield is a better measure of the practical value of the test given the difficulties with sample collection and processing.
The researchers say that TB LAM use should be expanded to people with CD4 counts in the range 100-199 who have symptoms of TB, pointing out that despite the lower sensitivity of the test, it has the potential to provide useful information when Xpert results are delayed or people cannot produce sputum.