Determining the performance and accuracy of SILVAMP TB LAM and cost-effectiveness of diagnostic testing for TB meningitis

Project Summary
Tuberculous meningitis (TBM) is the second most common cause of meningitis in Sub-Saharan Africa.
Neurologic disability and mortality are common, mortality is at least 50% in people with HIV. TBM diagnosis
remains difficult and diagnostic delay/missed diagnosis are major contributors to poor outcomes. Acid fast
bacilli smear of cerebrospinal fluid (CSF) is cheap and fast but with sensitivity of only ~10% in most settings.
CSF culture has improved sensitivity (~50-60%) but is slow, up to six weeks. Our studies on GeneXpert
MTB/Rif and the re-engineered GeneXpert MTB/Rif Ultra showed improved sensitivity (50-80%) with these
rapid (~2hrs) tests. Yet, these tests have inadequate negative predictive value to rule-out TBM, require
expensive instruments and cartridges, and their availability is inconsistent across the areas with the highest TB
incidence. Thus, alternative or additional tests for TBM remain crucial needs to improve outcomes.
A previous lipoarabinomannan (LAM) antigen test (Alere) had only ~20% sensitivity in CSF. Our study of
the SILVAMP TB LAM (FujiLAM) assay in CSF found 52% sensitivity in definite or probable TBM compared to
55% for Xpert Ultra yet this study was small and requires confirmation. Of the 58 cases of definite or probable
TBM, six were positive by FujiLAM but not Xpert Ultra. Eight were positive by Xpert Ultra but not FujiLAM. This
study was unable to systematically and thoroughly address cases that were possibly false by FujiLAM. Further,
formal cost-benefit analysis for this test, and other important tests for TBM has not been done. Given that cost
remains a major limitation in accessing TB diagnostic tests, the lack of research in this area is problematic.
Our overall objective is to reduce mortality and morbidity due to TBM by improving diagnostic accuracy,
rapidity, and cost-effectiveness. To accomplish these objectives, the aims of this proposal are to: 1) Determine
the accuracy of SILVAMP TB LAM (FujiLAM) in CSF to diagnose TBM in comparison to uniform TBM case
definitions; 2) Determine whether positive SILVAMP TB LAM (FujiLAM) tests without corroboration by other
TBM tests are false or true positives, using autopsy and metagenomics next generation sequencing; and 3)
Determine the cost and cost-effectiveness of TBM diagnostic testing strategies, including FujiLAM and
GeneXpert MTB/Rif Ultra.
The first two aims focus on better defining the diagnostic accuracy of FujiLAM, an easy to use, rapid test,
that requires limited technological infrastructure or expertise. The third aim focuses on cost-effectiveness of
this test and other commonly used tests. These studies will impact clinical practice by better informing our
understanding of the diagnostic tools for TBM. This proposal has the potential to shift the paradigm of TBM
diagnosis to two rapid tests, FujiLAM and Xpert Ultra, influencing international WHO guidelines while providing
valuable costing data for stake holders and ministries of health to consider investment and implementation.

Grant Number: 3R01AI170158-05S1
NIH Institute/Center: NIH
Principal Investigator: Nathan Bahr