In this prospective, clinic-based study, the Determine™ TB LAM urine test had low sensitivity, or a high false negative rate, as a screening test for tuberculosis among newly-diagnosed HIV-infected adults in a tuberculosis-endemic region. However, the urine LAM test had comparable test sensitivity to sputum AFB, and, when combined with sputum AFB testing, added 20% to the true positive rate, though at the expense of identifying an additional 9.6% as false positives. Urine LAM testing was easily performed by trained nurses in an outpatient clinic, returned results within an hour of sample collection, and did not require electricity, reagents, specimen transport, or a highly-trained laboratory microscopist. Our data support a recent suggestion that the urine LAM test be used in conjunction with, and not in place of, current diagnostic tests for active tuberculosis .
Diagnostic accuracy of the Determine™ TB LAM test has been reported in laboratory- and hospital-based studies. In a retrospective, laboratory-based study using frozen urine samples, Lawn et al. found nearly identical test sensitivity (28.2%) for detecting culture-positive pulmonary tuberculosis, but a higher test specificity (98.6%), when using urine LAM as a screening test among HIV-infected adults in Cape Town . Their higher test sensitivity may be a result of performing two sputum cultures for most (94.6%) participants, as well as performing urine LAM testing on frozen samples in a controlled laboratory-environment. In a study of HIV-infected TB suspects, 34% of whom were receiving antiretroviral therapy, a urine LAM testing had a sensitivity of 44.8% and specificity of 90.1% . While the test specificity was similar to our findings, the higher test sensitivity was likely due to inclusion of patients with higher bacillary loads, as 34% of their participants had mycobacterium detected in blood cultures, as well as sicker hospitalized patients. We performed 2 urine LAM tests per sample, which improved sensitivity, but reduced overall test specificity. A second study of hospitalized HIV-infected patients reported test specificity of 90% when using the manufacturer’s threshold of positive versus negative, which was similar to our findings . In contrast to these study, our results represent diagnostic accuracy when trained nurses use Determine™ TB LAM testing as a screening test in newly-diagnosed HIV-infected adults in a real-world outpatient clinical setting, where patients generally have less advanced disease compared to hospitalized patients.
Although the rapid urine LAM test was comparable to sputum AFB for detecting active tuberculosis, when combined the two tests improved detection of tuberculosis. In a retrospective study by Lawn et al., adding urine LAM testing to sputum AFB testing increased overall sensitivity from 28.2% to 43.5%, while having minimal impact on test specificity . While the change in sensitivity was similar to our study results, we found a large decrease in test specificity, or a higher false negative rate. Other studies have found the laboratory-based ELISA urinary LAM test capable of detecting extra-pulmonary tuberculosis . The addition of urine LAM to existing screening strategies would be relatively inexpensive, and the rapid urine LAM test has been reported to be cost-effective when used to diagnose HIV-infected adults with CD4 < 100/mm3 and symptoms of tuberculosis . In our cohort, urine LAM detected one-third of participants with extrapulmonary TB. The price for one LAM test, including a disposable pipette, was US $3.05, or $6.10 for our two-test screening strategy.
Utilizing rapid urine testing for active tuberculosis is appealing for several reasons. First, as we demonstrated in this study, the Determine™ TB LAM urine test is a true clinic-based, point-of-care test that can be used by trained nurses in peripheral clinics or remote settings without electricity or reliance on laboratory infrastructure. Second, participants were 3 times more likely to produce a urine sample than a sputum sample, which has been similarly reported in another study . This difference could have additional clinical benefit for urine LAM testing, but would need to be evaluated in an operational study . Third, obtaining urine samples from tuberculosis-infected patients carries a lower risk of transmission to health care workers than sputum specimens containing live, active bacilli. Finally, urine LAM has the potential for detection of extrapulmonary tuberculosis and may be a valuable biomarker of tuberculosis resolution during anti-tubercular therapy.
Our study had several limitations and strengths. Trained nurses performed specimen collection and urine LAM testing in an outpatient clinic to assess diagnostic accuracy when used at the clinical, not hospital, point-of-care in a real-world setting. While trained nurses may not interpret test results as accurately as a certified laboratory technicians, these results are more consistent with the intended use of the rapid, point-of-care test. If nurses had interpreted the appearance of a weak, faint test line as positive, this could have resulted in reduced test specificity. We obtained one sputum sample for the gold standard test of mycobacterial culture, while some studies of diagnostic accuracy included two or three specimens [23, 24]. Sputum culture is an imperfect gold standard test and non-differential misclassification could lead to reduced diagnostic accuracy. We did not perform Xpert MTB/RIF testing, obtain mycobacterial blood cultures, perform testing on those unwilling to share their HIV status, score the positive urine LAM tests, obtain user performance evaluation data, obtain data on clinical reasons for initiating TB treatment, evaluate causes of extrapulmonary tuberculosis, measure inter-observer variability, or measure the bacillary burden of tuberculosis. Finally, we evaluated urine LAM as an outpatient screening test among HIV-infected adults with and without tuberculosis symptoms in a tuberculosis-endemic region, and these results may not be generalizable to other populations or areas with low tuberculosis rates.