In silico optimization of a quartz crystal microbalance immunosensor for CYFRA 21–1 detection: a comparative analysis of Langmuir and first-order binding kinetics for early non-small cell lung-cancer screening

Abstract

Non-small cell lung cancer (NSCLC) makes up the bulk of lung malignancies and is most often identified only after the disease has advanced, when curative options are scarce and prognosis is poor. CYFRA 21–1, a soluble fragment of cytokeratin 19 shed into the bloodstream during epithelial tumor-cell turnover, is a well-established serum indicator of NSCLC and an attractive target for minimally invasive screening. In this work, we construct an in silico model of a quartz crystal microbalance (QCM) immunosensor for CYFRA 21–1 and use it to probe how the choice of binding kinetics and the density of immobilized antibody together shape the sensor response. Capture of the antigen was simulated at the following analyte concentrations (1, 5, 10, 25, 50, 75 and 100 ng/mL) and three fractions of the theoretical maximum antibody density (10%, 50% and 100%) under two kinetic descriptions, the Langmuir adsorption model and a first-order binding model, producing 84 runs in total in Python 3.9. The Langmuir formulation, which imposes a finite pool of binding sites and reversible association, reproduced surface saturation and returned frequency-shift predictions in line with the physical behavior of a QCM, whereas the first-order model, having no saturation constraint, consistently overstated the response. At 10% antibody coverage, the modeled device reached a limit of detection of 0.014 ng/mL over a linear window of 1–100 ng/mL (R² > 0.9999). Fractional site occupancy stayed under 2% throughout, keeping the sensor within the Henry region of the isotherm and guaranteeing a proportional, predictable signal even at trace analyte levels. Collectively, these findings supply a quantitative design basis for inexpensive, reusable QCM platforms intended for early NSCLC detection.