Improving the sensitivity and the cost-effectiveness of a competitive visual lateral flow immunoassay through sequential designs of experiments

• Developing and optimizing competitive visual LFIAs (visual cLFIAs) through the proposed sequential DoE process was convenient. • The Ab:AuNP ratio, amount of the probe, and amount of the competitor were optimized. • Absence/presence of color in presence/absence of cortisol was the parameter to assess sensitivity in visual cLFIAs. • The 4S (start, shift, sharpen, stop) decision process is proposed to interpretate response surfaces of the sequential DoE. • The sensitivity has been 500x improved after 13 experiments for development, and 5000x after 34 experiments. Although visual lateral flow immunoassays (LFIAs) are advantageous analytical tools, they typically suffer from low sensitivity in the competitive format (cLFIA). Optimizing sensitivity in this format is challenging as it depends on the inhibition of a visual signal. To optimize the sensitivity of a cLFIA, an experimental design-based approach is proposed and illustrated for the detection of cortisol. Gold nanoparticles (AuNPs) are used to label monoclonal anti-cortisol antibodies (Ab) to generate the probe. The Ab:AuNP ratio, the amount of probe and competitor on the test line are proposed as relevant parameters for optimization. As discriminatory conditions for sensitivity, the simultaneous disappearance of the color in the presence of a certain level of cortisol and a clearly visible color in its absence were assumed. A decision process called 4S (start, shift, sharpen, stop) is proposed for the interpretation of the response surfaces obtained after multivariate modelling of the data set. Using this approach, sensitivity was improved > 500-fold (visual LOD = 20.5 ± 1.0 ng/mL, LOD = 0.28 ng/mL) after 13 experiments and 10-fold (vLOD = 2.2 ± 0.1 ng/mL, LOD = 0.07 ng/mL) after 21 additional experiments. Reagent consumption was simultaneously reduced 40-fold. The feasibility of cortisol detection in human saliva samples was established with a satisfactory recovery of 70–109 %. The strategy from this work is useful for designing highly sensitive cLFIAs by unravelling the full potential of the immunoreagents and labels employed, with a limited number of experiments and reagent consumption, and without sacrificing the simplicity and affordability of visual, stand-alone LFIAs.