Determination of Patulin in Apple Juice and Apple-Derived Products Using a Robotic Sample Preparation System and LC-APCI-MS/MS

Patulin, a toxic mycotoxin, can contaminate apple-derived products. The FDA has established an action level of 50 ppb (ng/g) for patulin in apple juice and apple juice products. To effectively monitor this mycotoxin, there is a need for adequate analytical methods that can reliably and efficiently determine patulin levels. In this work, we developed an automated sample preparation workflow followed by liquid chromatography–atmospheric pressure chemical ionization tandem mass spectrometry (LC-APCI-MS/MS) detection to identify and quantify patulin in a single method, further expanding testing capabilities for monitoring patulin in foods compared to traditional optical methods. Using a robotic sample preparation system, apple juice, apple cider, apple puree, apple-based baby food, applesauce, fruit rolls, and fruit jam were fortified with 13C-patulin and extracted using dichloromethane (DCM) without human intervention, followed by an LC-APCI-MS/MS analysis in negative ionization mode. The method achieved a limit of quantification of 4.0 ng/g and linearity ranging from 2 to 1000 ng/mL (r2 > 0.99). Quantitation was performed with isotope dilution using 13C-patulin as an internal standard and solvent calibration standards. Average recoveries (relative standard deviations, RSD%) in seven spike matrices were 95% (9%) at 10 ng/g, 110% (5%) at 50 ng/g, 101% (7%) at 200 ng/g, and 104% (4%) at 1000 ng/g (n = 28). The ranges of within-matrix and between-matrix variability (RSD) were 3–8% and 4–9%, respectively. In incurred samples, the identity of patulin was further confirmed with a comparison of the information-dependent acquisition-enhanced product ion (IDA-EPI) MS/MS spectra to a reference standard. The metrological traceability of the patulin measurements in an incurred apple cider (21.1 ± 8.0 µg/g) and apple juice concentrate (56.6 ± 15.6 µg/g) was established using a certified reference material and calibration data to demonstrate data confidence intervals (k = 2, 95% confidence interval).