Fluorescent aptamer sensing of clenbuterol and salbutamol in artificial urine and construction of a proof-of-concept screening model.

A proof-of-concept differential dual-modal aptamer array was developed for clenbuterol (CLB) and salbutamol (SAL) screening in artificial urine. A CLB-selective aptamer and AP-Ago, a broad β2-agonist aptamer, were coupled to graphene-oxide fluorescence resonance energy transfer and to methylene-blue/ferrocene-labelled gold electrodes. Their four responses formed an analyte fingerprint by combining CLB-selective recognition with the broader response of AP-Ago. Seventy-two blind-coded samples represented six matrices, four screening classes and three independent preparation batches. Batches 1-2 (n = 48) supported model development, whereas batch 3 (n = 24; six samples per class) was locked for external assessment. Weighted calibration gave provisional procedural limits of detection of 0.083 ng mL-1 for CLB and 0.164 ng mL-1 for SAL. Mean external prediction-to-reference ratios were approximately 96% for CLB and 81% for SAL, distinct from controlled spike recoveries; intraday precision was 3.7% and interbatch precision was 5.8%. Against isotope-dilution LC-MS/MS, concordance correlation coefficients were 0.939 for CLB and 0.793 for SAL. A regularized multinomial logistic model achieved external balanced accuracy of 0.833, macro-F1 of 0.836 and Matthews correlation coefficient of 0.791. The reported bootstrap 95% confidence interval for balanced accuracy was 0.656-0.969, reflecting the small external set. The results support a candidate front-end triage strategy within the tested artificial matrices: differential signals identify presumptive CLB, SAL-associated high-response and mixed patterns, while any presumptive positive is referred for confirmatory LC-MS/MS rather than used for an anti-doping adjudication.