Extra-virgin olive oil authentication: triacylglycerol profiling and machine learning using the Shimadzu MALDI-8020/MALDI-8030 and eMSTAT Solution

Importance of the Topic

Extra-virgin olive oil (EVOO) is prized for its unique sensory and health properties, commanding a premium price. Its high value makes it a frequent target for adulteration with lower-cost vegetable oils, compromising quality, safety, and consumer trust. Reliable, rapid authentication methods are essential to protect industry integrity and public health.

Objectives and Study Overview

This work develops a streamlined analytical workflow for EVOO authentication using matrix-free MALDI-TOF mass spectrometry and machine learning. Sunflower oil was selected as a model adulterant, spiked into EVOO at 5, 10, and 20% (v/v). The aims were to demonstrate minimal sample preparation, fast automated analysis on a benchtop Shimadzu MALDI-8020/8030 instrument, and accurate classification with the eMSTAT Solution software.

Methodology and Instrumentation

Sample preparation involved:

• Dissolving EVOO, sunflower oil, and their mixtures in chloroform with tricaprin as internal standard for mass alignment
• Pre-coating the MALDI target with sodium trifluoroacetate (NaTFA) to promote formation of sodiated triacylglycerol species ([M+Na]+)
• Spotting 0.5 µL of each solution onto the target without additional matrix

Analysis was performed on a Shimadzu MALDI-8020 (positive mode) and MALDI-8030 (dual polarity) linear TOF mass spectrometer. Data processing and statistical treatment used Shimadzu’s eMSTAT Solution software, employing principal component analysis (PCA) and RandomForest machine learning with K-fold cross-validation.

Key Results and Discussion

MALDI spectra revealed distinct triacylglycerol (TAG) profiles: EVOO samples were dominated by palmitic and oleic acid combinations (e.g., OPO, POO, OOO), whereas sunflower oil exhibited high linoleic TAGs (e.g., LLL, PLL). Adulterated mixtures displayed progressive shifts in TAG ratios; notably, the LLL peak at m/z 901 was detectable even at 5% sunflower oil.
PCA score plots achieved clear separation of pure EVOO, adulterated EVOO, and sunflower oil. PCA loading identified m/z 901 and 907 as the most discriminant markers, confirmed by ROC analysis with sensitivity and specificity values ≥97% and AUC values approaching 1.0.
RandomForest classification of a blinded validation set (26 EVOO, 52 adulterated EVOO) yielded 100% accuracy, correctly identifying all low-level (5%) adulterations.

Benefits and Practical Applications

• Minimal solvent use and no matrix preparation reduce labor and environmental impact
• Rapid, automated benchtop MALDI-TOF analysis enables high throughput
• User-friendly eMSTAT Solution software facilitates intuitive PCA and machine learning workflows
• Highly sensitive detection of low-level adulteration enhances quality control in production and regulatory laboratories

Future Trends and Potential Applications

• Extending the approach to detect other vegetable oil adulterants such as palm, canola, or hazelnut oils
• Incorporation of dual polarity analysis for broader metabolite profiling
• Integration into online quality-control systems for continuous monitoring
• Development of expanded machine learning libraries to classify multiple adulteration types and grades

Conclusion

The described matrix-free MALDI-TOF method combined with machine learning provides a rapid, robust, and highly accurate tool for EVOO authentication. With minimal sample preparation, low solvent consumption, and full automation, it is ideally suited for routine quality assurance and anti-fraud applications.

References

• Salivo S. Extra-virgin olive oil authentication: triacylglycerol profiling and machine learning using the Shimadzu MALDI-8020/MALDI-8030 and eMSTAT Solution. Shimadzu Application News. 2022.
• European Commission Regulation (EEC) No 2568/91 on the characteristics of olive oils and olive-pomace oils and their analytical methods.