Whole Blood-Based Single-Tier Serology Assay for Lyme Disease Detection at the Point-of-Care

Introduction: : Lyme disease (LD), caused by Borrelia burgdorferi, is the most prevalent vector-borne illness in the United States. Yet, early and accurate diagnosis remains a significant clinical challenge. Current diagnostic protocols rely on multi-step, laboratory-based serologic testing that is time-consuming, complex, and often inaccessible in point-of-care (POC) settings. These limitations can lead to delayed or missed diagnoses, increasing the risk of serious neurological, cardiac, or rheumatologic complications. Further, a key limitation of existing POC platforms is the requirement for plasma or serum separation, which typically necessitates a centralized lab environment.
To address this unmet need, we developed a single-tier, paper-based vertical flow assay (VFA) capable of detecting LD directly from unprocessed whole blood. Building on our previous serum-based VFA, we reengineered a redesigned assay top case with an expanded sample inlet and an integrated, multi-layered membrane stack for efficient plasma filtration. To mitigate matrix variability from hematocrit and hemolysis, we optimized the wash buffer by incorporating a surfactant, improving flow distribution, reducing nonspecific binding, and enhancing signal uniformity. The assay features a multiplexed panel of nine synthetic peptides—six from our prior study and three newly designed di-peptides—selected for epitope specificity and diagnostic performance. This low-cost, single-measurement, rapid diagnostic platform supports decentralized testing by enabling accurate, rapid testing directly from whole blood, offering a promising alternative to current LD diagnostics.

Materials and

Methods: : To support whole-blood compatibility, we reengineered the VFA top case with a widened inlet and tested fluidic performance for uniform flow and minimal membrane clogging. We evaluated multiple separation membranes, including asymmetric plasma separation membranes (GX and GR) and glass fiber membranes (GF/C, Fusion 5, GF/DVA). Based on blood volume capacity, flow uniformity, and minimal clogging, a stacked configuration combining GR and GX membranes was selected. To minimize matrix-related signal variability, we formulated an optimized wash buffer containing surfactants to reduce hemolysis-induced background and nonspecific binding, while improving signal consistency across hematocrit levels. The multiplexed nitrocellulose membrane incorporated nine synthetic peptides derived from conserved, immunodominant B. burgdorferi antigens, six from our previous serum-based assay, and three new di-peptides designed to enhance classification accuracy. Analytical optimization was performed using whole blood samples spiked with Lyme antibodies. For clinical validation, the assay was tested using 60 clinically characterized whole blood samples (30 LD-positive, 30 LD-negative) from the Lyme Disease Biobank and 35 healthy donor samples from the UCLA system. All samples were run in duplicate, and signal intensities were quantified via a smartphone-based optical reader.

Results, Conclusions, and Discussions:: The optimized VFA demonstrated 100% concordance with the serum-based format in spiked whole blood samples. The redesigned top case, combined with the stacked membrane configuration, enabled efficient whole blood processing in under 20 minutes, comparable to our traditional serum-based assay. The assay maintained high analytical reproducibility, with an inter-spot coefficient of variation (CV) below 10%. For binary classification of LD-positive or LD-negative status based on peptide response, two diagnostic algorithms were evaluated: (1) a threshold-based scoring system using t–test–selected peptides, and (2) a neural network classifier trained with feature-selected peptide intensities. Notably, both approaches achieved >98% sensitivity, exceeding current testing standards. The threshold-based method yielded 100% sensitivity and 81% specificity in the training dataset. In comparison, the neural network–based classifier, using sequential forward feature selection, further improved overall diagnostic performance, achieving 98% sensitivity and 97% specificity. This improvement is likely due to the model’s ability to capture complex, non-linear patterns in the repertoire of epitope-specific antibody responses, enabling more accurate classification and reducing misclassification compared to the threshold-based method.
These results demonstrate that this peptide-based VFA enables rapid and accurate Lyme disease testing directly from whole blood compatible with point-of-care operation. By eliminating the need for lab-based blood separation and simplifying assay operations, the platform reduces diagnostic complexity. With performance comparable to serum-based assays, this low-cost, paper-based single-tier device supports decentralized testing with minimal infrastructure. Beyond Lyme disease, this approach has the potential to advance point-of-care diagnostics more broadly by enabling whole-blood-based testing for other conditions, ultimately improving accessibility, simplicity, and timely disease management.

Acknowledgements and/or References (Optional)::