Clinical utility of a commercial LAM-ELISA assay for TB diagnosis in HIV-infected patients using urine and sputum samples

Abstract

Background: The accurate diagnosis of TB in HIV-infected patients, particularly with advanced immunosuppression, is difficult. Recent studies indicate that a lipoarabinomannan (LAM) assay (Clearview-TB(R)-ELISA) may have some utility for the diagnosis of TB in HIV-infected patients; however, the precise subgroup that may benefit from this technology requires clarification. The utility of LAM in sputum samples has, hitherto, not been evaluated.

Methods: LAM was measured in sputum and urine samples obtained from 500 consecutively recruited ambulant patients, with suspected TB, from 2 primary care clinics in South Africa. Culture positivity for M. tuberculosis was used as the reference standard for TB diagnosis.

Results: Of 440 evaluable patients 120/387 (31%) were HIV-infected. Urine-LAM positivity was associated with HIV positivity (p = 0.007) and test sensitivity, although low, was significantly higher in HIV-infected compared to uninfected patients (21% versus 6%; p<0.001), and also in HIV-infected participants with a CD4 <200 versus >200 cells/mm(3) (37% versus 0%; p = 0.003). Urine-LAM remained highly specific in all 3 subgroups (95%-100%). 25% of smear-negative but culture-positive HIV-infected patients with a CD4 <200 cells/mm(3) were positive for urine-LAM. Sputum-LAM had good sensitivity (86%) but poor specificity (15%) likely due to test cross-reactivity with several mouth-residing organisms including actinomycetes and nocardia species.

Conclusions: These preliminary data indicate that in a high burden primary care setting the diagnostic usefulness of urine-LAM is limited, as a rule-in test, to a specific patient subgroup i.e. smear-negative HIV-infected TB patients with a CD4 count <200 cells/mm(3), who would otherwise have required further investigation. However, even in this group sensitivity was modest. Future and adequately powered studies in a primary care setting should now specifically target patients with suspected TB who have advanced HIV infection.

Figure 1. Flow chart of patients recruited to the study stratified by patient subgroup, smear microscopy, HIV status, and CD4 T cell count.

Figure 2. The sensitivity and specificity (and 95% confidence intervals (%); top panel of table) and positive and negative predictive value (middle panel of table) of smear-microscopy alone, urine LAM alone, and a combination of urine LAM and smear-microscopy.

The bottom panel of the table shows the test sensitivity in smear negative culture positive TB patients. For sensitivity calculations the definite TB group (n = 141) was used whilst specificity calculations were performed using the non-TB group (n = 172) as a reference.

Figure 3. LAM positivity (> zero OD units =  positive for LAM after subtraction of the negative control i.e. cutpoint is zero) in cultures of oral mouth flora (oral flora) and in organism-specific cultures (various Actinobacteria, including different strains of Nocardia and Streptomyces, and C. albicans, T. paurometabolum, and C. neoformans inoculated into normal broth culture [containing yeast extract] and Todd-Hewitt culture media [without yeast extract].

Normal oral flora from six different healthy control subjects was also cultured).