Diagnostic accuracy of the re-engineered urinary FujiLAM2 assay amongst hospitalized adults with advanced HIV disease.

Background: A nonsputum-based, point-of-care tuberculosis (TB) diagnostic test is a global health priority. The impact of urinary mycobacterial lipoarabinomannan (LAM) testing has been limited by the diagnostic performance of current assays. We assessed the diagnostic accuracy of the re-engineered TB-LAM SILVAMP (FujjLAM2) assay (Fujifilm, Japan) to diagnose TB amongst hospitalized adults living with advanced HIV disease.

Methods: We consecutively enrolled adults presenting with suspected meningitis at two hospitals in Uganda. We implemented a standardized TB diagnostic package: urine Alere TB lipoarabinomannan (TB-LAM), urine Xpert MTB/Rif Ultra, CSF Xpert MTB/Rif Ultra, TB CSF culture, mycobacterial blood culture and chest radiography. We performed FujiLAM2 testing on cryopreserved or fresh urine. We compared diagnostic accuracy against a composite microbiological reference standard of any positive TB test (including Alere-LAM). We assessed 30-day mortality.

Findings: We performed FujiLAM2 testing on urine of 436 hospitalized participants. The median CD4 + cell count was 34 cells/μl [interquartile range (IQR) 11-96]. Using the microbiologic reference standard, FujiLAM2 sensitivity was 34% [95% confidence interval (CI) 25-43%], and specificity was 94% (95% CI 91-96%). When grade-1 Alere TB-LAM positives were excluded, sensitivity was 38% (95% CI 27-50%). Cryopreserved specimens were threefold more frequently positive.

Interpretation: Amongst hospitalized adults with advanced HIV disease, the re-engineered FujiLAM2 urine assay had suboptimal sensitivity but high specificity for diagnosing TB disease. Antigen-antibody/protein complexes may be present accounting for better sensitivity with cryopreserved specimens.