Optical Spectral Fingerprinting Enables Sensitive Detection of Anthracycline Chemotherapeutics in Synthetic Clinical Biofluids.

Anthracycline chemotherapeutics are common chemotherapeutics that have substantial toxicities. There is substantial interpatient pharmacokinetic variability, though there is no method to quantify organ or tumor exposure. Here, we exposed an optical nanosensor array to detect each of four anthracyclines. We screened 12 ssDNA sequences paired with seven single-walled carbon nanotube (n,m) species against several concentrations of doxorubicin, daunorubicin, idarubicin, and epirubicin. Complex spectral responses were used to develop machine-learning-based classification models to quantify each anthracycline. The optimized extreme gradient boosting model classified high levels of each anthracycline with 100% accuracy. Principal component analysis distinguished low (≤5 μM) and high concentrations of each anthracycline. Finally, we validated selected ssDNA-(n,m) pair performance in synthetic urine and sweat. Our findings deliver a generalizable optical spectral fingerprinting methodology for hard-to-detect analytes. Their use in clinical biofluids portends the preclinical and potentially clinical pharmacokinetic measurement of anthracyclines to improve efficacy and reduce toxicities.