Tuberculosis (TB), caused by M. tuberculosis (Mlb) infection, is one of the leading causes of infectious disease
related death globally despite the availability of treatments. Lack of reliable diagnostics for TB contributes to a
significant proportion of the preventable TB deaths as currently only 60% of those who develop TB are
diagnosed. Current diagnostic approaches depend on detection of Mlb in sputum and are either too resourceintensive
to be deployed widely in endemic areas or have low sensitivity, underperforrning especially in those
co-infected with TB and HIV and in children. Use of sputum as a sample itself is a bottleneck as a sizeable
fraction of people with suspected TB cannot produce sputum. Current serological assays, which measure only
the presence or titer of Mlb-specific antibodies, fail to provide the required sensitivity and specificity for TB as
they cannot distinguish active TB disease from latent TB infection (L TBI) which is widespread in endemic regions.
Recent work by the Sarkar lab and others has highlighted the remarkable specific and sensitive ability of
pathogen-specific antibody glycosylation to mark recent and active infections even for endemic pathogens,
including Mlb. Additionally, ii has been shown that integrating this with Fab & Fe region biophysical and functional
properties of antibodies (including antigen-specificity, isotype/sublype, Fe receptor binding) using machinelearning
based analytical approaches defines a multivariate, disease-state specific biomarker. Measuring this for
a large set of Mlb-specific antibodies, however, currently requires intractably large sample volumes, laborious
sample preparation and expensive and complex analytical methods such as mass spectrometry. To resolve this
bottleneck, the Sarkar lab has recently developed sample-sparing, binding-based assays for multiplexed
glycoprofiling of antigen-specific antibodies as well as novel electronic and optical detection microchip formats
for implementing them at the point-of-care (POC). Taken together these findings and technological developments
suggest the exciting prospect of a multiplexed POC 'antibody-omic' profiling chip for TB diagnostics from
a single drop of blood. The goal of this project is to realize this overall vision. First, a microfiuidic multiplexed
antibody profiling chip will be developed to enable coupled measurement of Fab and Fe properties of large
numbers of Mtb-specific antibodies from small sample volumes. Next this will be applied to banked serum from
a clinically well-characterized cohort of HIV- and HIV+ TB patients. Machine-learning based methods will be
applied to resulting high-dimensional data sets to define the antibody Fe signatures that act as a robust
biomarker for TB. Finally, a low-cost, POC detection microchip for TB diagnostics will be developed which
would be applicable in endemic regions including low- and middle-income countries (LMICs). Development of
such a reliable, low-cost diagnostic for TB would have a significant impact on its clinical care and management.
Overall, this project will create broader biomarker discovery and POC diagnostic opportunities beyond TB as
well.
Project Summary/Abstract
