Tuberculosis (TB) remains a global health challenge, requiring rapid and accessible diagnostics to guide preventive therapy. C-X-C motif chemokine ligand 10 (CXCL10) mRNA is a promising biomarker of TB infection, yet current assays rely on long stimulation times and complex instrumentation. Here, we present a hydrogel-based reverse transcription loop-mediated isothermal amplification (HRT-LAMP) assay that partitions amplification reactions within a porous hydrogel matrix, enabling digital-like quantification of CXCL10 mRNA using only a heater and smartphone imaging. A hand-held 3D-printed imaging device was developed to capture fluorescence signals, which were subsequently quantified by a convolutional neural network (CNN). The assay demonstrated a detection limit of 10 copies per reaction with excellent linearity (101-105 copies/test, R2 = 0.998), supporting a dynamic range up to 105 copies/test via high-resolution microscopy and 10 to 104 copies/test using the portable smartphone platform. In 79 samples used to establish diagnostic thresholds, HRT-LAMP reached 93.6% sensitivity and 100% specificity. In an independent validation cohort of 91 samples, it achieved 91.2% agreement with clinical diagnosis with sensitivity 92.3%, specificity 89.7%, κ = 0.82, and area under the curve (AUC) = 0.918, outperforming interferon-γ release assays (IGRA) used in clinics. Moreover, accurate results were maintained with as little as 50 μL of blood (95.7% agreement). HRT-LAMP thus enables rapid (CXCL10 mRNA responses, offering a practical platform for TB diagnosis in resource-limited and pediatric settings.
Wáng et al. (Fri,) studied this question.