Milk samples (n = 186) were collected from supermarkets in four regions (Northeast China and Inner Mongolia, North China, Northwest China, and South China) throughout China in 2021, which included 44 brands and 22 provinces (Supplementary Table S1). The samples were whole and pasteurized bovine milk and were stored at −20°C after collection until analysis. In addition, skim milk powder from local markets was used in the recovery experiments for method validation.

High-performance liquid chromatography-grade n-hexane, methanol, and isopropanol were obtained from Fisher Scientific (Fair Lawn, NJ, USA). NaOH (purity ≥95%), anhydrous sodium sulfate (purity ≥99%), and acetyl chloride (purity >99.5%) were from Macklin (Shanghai, China). Ultrapure water was obtained using a Milli-Q purification system (Millipore, Bedford, MA, USA).

The following standards were used: FA solution GLC 617 (Anpel, Shanghai, China) and FAME solution GLC 674 (Nu-Chek Prep, Inc., Elysian, MN, USA); 6 iso-FAME and 3 anteiso-FAME individual standards were obtained from Larodan (Malmo, Sweden). The C22:3 c13c16c19 FAME individual standard and C17:0 ethyl ester (C17:0 EE) used as internal standards (IS) were obtained from AccuStandard (New Haven, Connecticut, USA). The individual standards of C18:1 c12, C18:4 c6c9c12c15, C19:1 c10, and C20:1 c8 FAMEs, as well as C18:2 c9c11 and C18:2 t9t11 FA individual standards, were obtained from Cayman (Ann Arbor, Michigan, USA). The C5:0, C7:0, C9:0, and C19:0 FAME individual standards were obtained from Dr. Ehrenstorfer (Augsburg, Germany). The C10:1 c3, C10:1 c4, C10:1 c9, C12:1 c5, C12:1 c11, and C16:1 c7 FA individual standards were obtained from Macklin (Shanghai, China). Linoleic acid methyl ester mix Certified Reference Material 47791 and C18:2 c9t11 and C18:2 t10t12 FA individual standards were purchased from Sigma-Aldrich (St. Louis, MO, USA). The triacylglycerol (TAG) individual standard of C11:0 was obtained from Nu-Check Prep (Elysian, MN, USA).

Based on our previous study, a sample pretreatment procedure was developed, including the extraction of lipids from milk and the conversion of FA into FAME (17, 24). Frozen milk samples were first pre-warmed at 40°C in a water bath and then shaken carefully to homogenize them. To extract milk lipids, 2 ml milk samples were mixed with 4 ml of a solution of n-hexane/isopropanol (v/v, 3/2), vortexed, and centrifuged. The upper n-hexane was collected, followed by another extraction with n-hexane, and all of the extracted upper n-hexane were collected and mixed together. Combined n-hexane was mixed with 2 ml solution of methanolic NaOH (2%) with heating at 50°C for 20 min, followed by 2 ml solution of acetyl chlorocarbinol (10%) with heating at 90°C for 150 min for methylation. After cooling to room temperature, 5 ml of ultrapure water was added to the mixture. The upper n-hexane phase was then extracted and diluted. Anhydrous sodium sulfate (0.5 g) was then added, and the mixture was vortexed for 30 s for further dewatering. The supernatant was mixed with an internal standard, diluted with n-hexane, and subsequently analyzed using gas chromatography–mass spectrometry (GC-MS).

Analyses of FAMEs were performed using an Agilent 7890A GC equipped with a 7000 B MS detector system (Agilent Technologies, Santa Clara, California, USA). FAMEs were separated using an Agilent capillary column CP-Sil 88 (100 m × 0.25 mm × 0.20 μm). The proposed method was improved using an applicable GC oven program (Table 1) and by performing in the selected ion monitoring (SIM) mode with 19-time windows (Table 2). Four characteristic ions were selected for the qualification and quantification of each analyte and IS, and the lowest signal-to-noise ratio (S/N) was designated as the quantitative ion. Qualitative analysis of individual FAME in milk samples was performed by comparing their retention times and characteristic ions with those of the corresponding FAME standards. Quantitative analysis of individual FAME was carried out by external standard calculation and IS calibration, that is, using IS calibration, the value of the curve regression equation for each FAME standard was calculated. The FA values were calculated using the stoichiometric factors for the conversion of FAME into FA. The FA composition results are expressed as g/100 g FA.

Sensitivity, linearity, accuracy, and precision were involved in validating the method (25). Sensitivity was calculated from the concentration with an S/N ratio of 10 and expressed as the limit of quantitation (LOQ) (15). Linearity was assessed using the coefficient of determination (R²) of the standard curves of each FAME at five different concentration levels (concentration ratio between FAME and IS vs. the peak area ratio). Recovery experiments were spiked with three different concentrations of FA solution (GLC 617) and triacylglycerol (C11:0) in a solution of skim milk powder, and the results were used to demonstrate the accuracy. The intra- and inter-day precisions were determined by testing six parallel samples three times within a day and once a day for three consecutive days, respectively.

The peak areas were obtained using the Agilent MassHunter Workstation (Agilent Technologies, Santa Clara, California, USA). The results were preliminarily sorted using Microsoft Excel 2019. Before significance analysis, the Kolmogorov–Smirnov test was applied to determine whether the results coincided with the normal distribution obtained using the univariate procedure in SAS 9.4. The Kruskal–Wallis test was performed on the FA data that did not comply with the normal distribution using the NPAR1WAY procedure in SAS 9.4. Multiple comparisons were performed on the transformed RANK of the original data with Duncan's method using the generalized linear model (GLM) procedure in SAS 9.4. A P-value of < 0.05 was considered statistically significant.

Some FA indices, such as n-6/n-3 polyunsaturated fatty acid (PUFA), PUFA/SFA ratio (P/S), AI, TI, health-promoting index (HPI), and hypo/hypercholesterolemic ratio (h/H), were calculated as described previously (20), as follows:

To calculate the FA intake from retail milk, we assumed that the FA composition of all dairy products produced in China was consistent with that of the milk samples analyzed in this study. The FA intake via retail milk consumption was estimated (26, 27) as follows:

FA intake (mg/d) = 300 (milk intake, g/d) × 1, 000 × 3.8 (contribution of fat from whole milk, %) ÷100 × 93.3 (the proportion of FA in milk fat, %) ÷100 × milk FA concentration (the proportion of FA in total FA, %) ÷100.

Using the proposed method, 82 FAs were detected simultaneously by chromatographic and spectrographic optimization, including 11 ECSFAs, 10 OCSFAs, nine BCSFAs, 30 MUFAs, and 22 PUFAs (Figure 1).

Regarding chromatographic separation, the capillary column and oven temperature program were optimized by virtue of the differences in boiling point and polarity among the abovementioned FAs. Several capillary columns, including DB-WAX and CP-Sil 88, were attempted, and most of the FAME peaks presented less co-elution, excellent peak symmetry, and rational retention times on the CP-Sil 88 column; therefore, we selected it for further study. It was noticed that an oven temperature of ~180°C gave better resolution among FAMEs with <C18, and an oven temperature of ~210°C gave better resolution among FAMEs with >C18. Separation behaviors of cis- and trans-isomers of C18:1 are greatly affected by column and oven temperatures (28, 29). Thus, the column temperature was divided into three temperature change stages by setting an appropriate rate of increase in the oven temperature to achieve baseline separation of most FAMEs in the total ion chromatogram (TIC; Figure 1). Three types of co-elution problems were observed in the TIC which made the quantification of co-eluted FA almost impossible without the SIM mode. (i) the co-elution of methyl BCFA and MUFA, as shown in Figures 2A, B; (ii) the co-elution of methyl MUFA and PUFA, as shown in Figures 2C–F; (iii) the co-elution of methyl PUFA and SFA, as shown in Figure 2E. In previous reports, the problem of co-elution is common in the analysis of fatty acids by gas chromatography, including the interference of methyl BCFA and MUFA and the co-elutions of cis- and trans-isomers of both C16:1 and C18:1 (12, 30). To eliminate interference between the peaks of partial and complete co-elution, we analyzed the characteristics of the unique fragment ions of these co-elutions using SIM (14). Identification and quantification were based on the SIM mode shown in Table 2. For example, C15:0 iso methyl and C14:1 t9 methyl co-eluted in the TIC and were distinguished by specific fragment ions at 256 and 208 m/z, respectively, in the extracted ion chromatogram, which provided good resolution (Figure 2A). Similarly, the SIM parameters for the analysis of other co-eluted FAMEs in the TIC were also optimized (Figures 2A–F).

The results of the method validation are shown in Supplementary Tables S2, S3. The limit of quantitation (LOQ) of FAMEs ranged from 1.9 to 990.1 μg/L corresponding to 0.0439 μg/ml to 23.6832 μg/ml of FAs from milk samples. The regression equations of all the FAME standard curves covered a sufficiently wide linear quantitative range, with determination coefficients >0.9991. Recoveries of FAs ranged from 81.4 to 108.3%, while those of TAGs were 101.9 to 106.3%. Intra-day and inter-day relative standard deviations (RSDs) were <5.0% based on the determined FAs in the milk samples.

In general, method optimization involves the choice of polar column, column temperature program, and characteristic ions that influence separation efficiency along with the accurate determination of all FAs from bovine milk samples, particularly UFAs that have received little or no attention in the FA analyses, including n-1, n-3, n-5, n-6, n-7, n-9, and n-12 UFAs. Method validation showed that the proposed method presented competitive sensitivity, suitable linearity, acceptable accuracy, and precision for the determination of milk FAs. Thus, a high-throughput GC-MS method was applied to determine milk FAs, with good separation in the chromatogram and good resolution in the SIM mode, which is the basis of a comprehensive understanding of milk FA composition.

The FA profiles of bovine retail milk from China are shown in Figure 3. The ratio of SFA to UFA in retail milk in China is ~7:3 (Figure 3A). Previous studies have reported that the ratio of SFA and UFA in bovine retail milk in the United Kingdom ranged from 73:27 to 68:32 (22, 31), while the ratio in the United States ranged from 70:30 to 68:32 (18). This is inconsistent with our findings because FA composition is affected by many factors, including diet and season (32, 33). In addition, some studies did not consider as many FAs as possible and only included the high content ones in the analysis; therefore, the results of these studies were biased compared to our results. In addition to the major ECSFAs, OCSFA, and BCSFA are important components of milk SFAs. C14:0, C16:0, and C18:0 were the most abundant ECSFAs, whereas C15 and C17 FAs with straight and branched chains were the most abundant OCSFAs (Figure 3B). This observation is consistent with that of other studies and is mainly determined by the synthesis of milk FAs in dairy cows, primarily from the de novo synthesis of FAs with carbon chain length <C16 and the intake of FAs with carbon chain length >C16 from the blood in the mammary glands (34, 35). n-9, n-6, and n-7 UFAs were the main components of milk UFAs, whereas other UFAs were relatively minor, accounting for <1% of the total FAs (Figure 3A). In our study, oleic acid (OA, C18:1 c9) was the most dominant n-9 UFA in milk, accounting for 96% of the total n-9 UFAs. LA and C18:2 c12 were the main n-6 UFAs in milk, accounting for ~60 and 13% of the total n-6 UFAs, respectively. Alpha-linolenic acid (ALA, C18:3 c9c12c15), docosapentaenoic acid n-3 (DPA n-3, C22:5 c7c10c13c16c19), eicosapentaenoic acid (EPA, C20:5 c5c8c11c14c17), and eicosatrienoic acid (ETA, C20:3 c11c14c17) were the main n-3 UFAs in milk, accounting for 49%, 18%, 16%, and 10% of the total n-3 UFA, respectively. This is because LA (C18:2 c9c12) and ALA are the precursors of n-6 UFA and n-3 UFA synthesis, respectively, in dairy cows, and these precursors synthesize the corresponding UFAs through carbon chain elongation and desaturation (36). In addition, we found that C10:1 c9 was the only n-1 UFA in milk, and C14:1 c9, C16:1 c9, and C18:1 c6 were the main n-5, n-7, and n-12 UFAs in milk, respectively. Previous studies have reported the functions of n-9, n-6, and n-3 UFAs; therefore, the content of these UFAs in milk has been widely studied. The results of our study provide information on the content of other minor UFAs in milk, which could provide reference data for subsequent research on the function of milk FAs.

The FA composition of retail milk across four regions of China is presented in Table 3. Statistical differences were observed in the concentrations of some individual FAs for different geographic regions, including C4:0, C15:0, C15:0 anteiso, VA, LA, conjugated linoleic acid n-7 (CLA n-7, C18:2 c9t11), ALA, arachidonic acid (ARA, C20:4 c5c8c11c14), and FAs (including OCSFA, BCSFA, n-6 PUFA, n-1 UFA, n-5 UFA, n-6 UFA, n-7 UFA, and n-12 UFA). There were no significant differences in the concentrations of individual FAs, including C16:0, C18:1 c9, n-3 UFA, EPA, DPA, DPA n-6, and FA, including SFA, MUFA, and PUFA. Differences in rumen fermentation due to the diet and management of dairy cows may partly explain the inconsistent content of the abovementioned OBCFA and UFA among the regions (37, 38). In this study, where milk was collected at the retail level, it was not possible to collect detailed information at the farm level, and, therefore, we cannot provide a more detailed explanation of the subtle variations in some FAs. In general, although the minor FAs showed numerically smaller significant differences among the regions, the major FAs mostly showed no significant differences. Thus, the overall composition of milk FAs among different regions was numerically similar, and the minor FAs showed little difference.

Some FA indexes have been used to evaluate the nutritional quality of milk and are listed in Table 4.

n-6 and n-3 PUFAs induce proinflammatory and anti-inflammatory responses, respectively, and their ratios are related to a balanced diet (39–41). A balanced n-6/n-3 PUFA ratio allows for progressive brain, eyes, and heart development while reducing the risk of coronary heart disease and neurodegenerative disorders (42, 43). In the study, the n-6/n-3 PUFA content of milk in Northwest and South China was lower than that of milk from other regions. The decreased ratio indicated a desirable alteration between regional milk and benefited adults. To enable proper neuronal development and prevent most chronic disorders, a 1 or 2 n-6/n-3 PUFA ratio should be maintained (44, 45). In this study, the n-6/n-3 PUFA ratio of milk in different regions is 5.51–6.63, which is still a certain distance from the recommended balanced n-6/n-3 PUFA ratio, thus, further regulation of this composition is needed.

The P/S ratio is commonly used to assess the effect of dietary FA on CVH. The index is based on the research that dietary PUFAs are associated with decreased serum cholesterol and low-density lipoprotein content, while dietary SFAs are associated with increased serum cholesterol content (46). Therefore, the higher the P/S ratio, the more significant the inhibitory effect on the increase of blood cholesterol, and the more beneficial to cardiovascular health. In this study, there was no significant difference in the P/S ratio of milk from different regions, indicating that milk intake from different regions had no significant effect on CVH. The P/S ratio of milk ranged from 0.02 to 0.07 previously reported in the literature (22, 47). Our results were within the expected range of P/S index values, ranging from 0.06 to 0.07.

AI index characterized the atherogenic potential of dietary FA. The index shows the correlation between the pro-atherogenic FA (C12:0, C14:0, and C16:0) and the anti-atherogenic FA (UFA) (48). Thus, the lower the values of AI, the greater the proportion of anti-atherogenic fatty acids present in milk, and the more conducive to the maintenance of CVH. In this study, there was no significant difference in the values of AI of milk from different regions, indicating that intake of milk from different regions had no significant effect on the anti-atherosclerosis effect. The values of the AI index ranged from 1.37 to 5.13 reported in the previous literature (47, 49). Our results were within the expected range of AI index values, ranging from 2.66 to 2.84.

TI index was used to evaluate the tendency of dietary FA to thrombosis in blood vessels. This index is defined as the relationship between the pro-thrombotic FA (C12:0, C14:0, and C16:0) and the anti-thrombotic FA (MUFA, n-3 PUFA, and n-6 PUFA) (48). Specifically, the lower the values of the TI index, the greater the anti-thrombotic effect, and the better for CVH. In this study, there was no significant difference in the TI of milk from different regions, indicating that intake of milk from different regions had no significant effect on the anti-thrombosis effect. The values of the TI index ranged from 2.23 to 4.03 reported in the previous literature (50). Our results were within the expected range of AI index values, ranging from 3.30 to 3.52.

h/H ratio was commonly used to assess the cholesterolemic effect of dietary FA. The index is based on the research that dietary C18:1 and PUFA are associated with decreased serum cholesterol content, while dietary C12:0, C14:0, and C16:0 are associated with increased serum cholesterol content (51). Therefore, nutritionally speaking, the higher the h/H ratio, the more obvious the effect of inhibiting the increase of serum cholesterol content, and the more beneficial to CVH. In this study, there was no significant difference in the h/H ratio among the different regional milk samples, indicating that milk intake from different regions had no significant effect on human cholesterol metabolism. The h/H ratio range of milk reported in the literature is 0.32–0.74 (50). Our milk results were within the expected range of h/H ratio, ranging from 0.49 to 0.52.

HPI is the reciprocal of AI, which is mainly used to evaluate the impact of the fatty acid composition of dairy products on cardiovascular health (52). Therefore, the higher the HPI index, the more beneficial to CVH. In this study, there was no significant difference in the HPI among the different regional milk samples, indicating that the nutritional implications for CVH were similar among retail milk from different regions. Previous literature reported that the values of the HPI index ranged from 0.16 to 0.68 (20, 52). Our results were within the expected range of HPI index values, ranging from 0.36 to 0.38.

To determine whether the differences in milk FA composition from different geographic regions caused nutritionally meaningful differences, FA intake from retail milk was assessed in different regions (Table 5). Overall, FA intake was within similar ranges regardless of the region. The World Health Organization (WHO) has recommended SFA intake of <10% of energy intake, and the Chinese Nutrition Society has set an acceptable macronutrient distribution range (AMDR) for SFA intake as <8% of energy intake in children 4–17 years and <10% in adults over 18 years of age (21, 53). SFA from retail milk would provide either 33% of total SFA intake (based on recommended values by the WHO) or 33%−41% of total SFA intake (based on recommended values by the Chinese Nutrition Society). Stergiadis (22) reported that dairy fat contributed to approximately one-third of the maximum recommended intake of SFA in adult consumer diets across the UK dairy production systems, which is consistent with our findings. Although the Chinese Nutrition Society has not set a maximum intake of TFA, the WHO has recommended a TFA intake of <1% of energy intake (21, 53). Estimated TFA intake from retail milk throughout China is 6%−9% of the maximum recommended TFA daily intake, which would be well below the maximum values recommended by the WHO.

Based on recommended UFA intake by the WHO, UFAs from retail milk in China would provide ~7% of the total UFA intake for consumers (21, 53). Furthermore, the Chinese Nutrition Society has suggested an AMDR of n-6 and n-3 PUFAs, expressed as % of energy intake, from 2.5 to 9% and from 0.5 to 2%, respectively. According to the results of the present study and current milk intake in China, retail milk throughout China would provide an intake of n-6 PUFA and n-3 PUFA from 2 to 7% and 2 to 6 % of AMDR, respectively. Additionally, the Chinese Nutrition Society has advised that the adequate intake (AI) of LA and ALA is 4 and 0.6% of energy intake, respectively. The consumption of retail milk would contribute ~3% of AI. The Chinese Nutrition Society has set the AMDR for EPA + docosahexaenoic acid (DHA) between 250 mg/d and 2,000 mg/d. Our study showed that the consumption of retail milk would provide an intake of EPA + DHA from 0.6 to 4.9% of AMDR. Due to the low content of EPA and DHA in retail milk and the low conversion efficiency of dietary ALA in humans (54), the supply of EPA and DHA from milk is extremely low; therefore, an additional supply of EPA and DHA from other foods is essential. Nutritional recommendations refer to the total diet rather than individual foods (21, 26). Although the current study estimated potential changes in FA intake from dairy products, any potential effects on human health are influenced by FA intake from other foods. Future research should investigate the nutritional roles of the individual components of milk.