Going deeper in lipolysis: Prediction of individual free fatty acid contents in milk by mid-infrared spectroscopy

Christophe O., Reding R., Leblois J., Grelet C., Pittois D., Soyeurt H., Guignard C., Dehareng F.

Journal of Dairy Science, vol. 108, n° 12, pp. 13085-13098, 2025

Abstract

The dairy sector deals with a recurring issue: a taste alteration due to degradation of fat, commonly called lipolysis. Lipolysis occurs after milking, through the physical shocks induced by freezing, pumping, transfer, and storage of the milk. Physical breaking of fat globules makes triglycerides accessible to enzymes and degrades into free fatty acids (FFA). Among them, the volatile short-chain FFA lead to organoleptic issues through undesired tastes. Currently, methods to measure FFA are complex and time-consuming. Easier methods provide incomplete information on the total amount of FFA, or a combination of these. To bring a new way of preventive and corrective action for dairies and farmers, this study aims to develop predictive models based on milk Fourier transform mid-infrared spectroscopy (FT-MIR) to quantify individual FFA, with a special focus on short-chain FFA. A total of 792 milk samples were collected from individual cows and in bulk tanks in 4 different countries (Luxembourg, Belgium, France, and Germany) and analyzed by FT-MIR spectroscopy as well as GC coupled with tandem mass spectrometry. When performing a random sampling of individual cows, and therefore relying only on spontaneous lipolysis, most of the samples showed that short-chain FFA under the limit of quantification and derived FT-MIR models had poor performance. In a second step, lipolysis was induced by 4 different mechanical processes of milk. Storage after milk homogenization demonstrated a clear increase of short-chain FFA values leading to improved predictive performance of models. The predictions of short-chain FFA could be used for rough screening on the dairy population, especially C4 and total short-chain FFA as their models had a validation R<sup>2</sup> above 0.7. Unfortunately, the mid- and long-chain FFA models showed poor predictive performance. However, those results can still be considered positive because they may enable a better monitoring of taste defect mainly occurring from short-chain FFA.

People

PITTOIS Denis

Elemental and Molecular Analysis

Send an email

GUIGNARD Cédric

Elemental and Molecular Analysis

Send an email

How can we help you?

By content type (optional)