Determination and Risk Assessment of Flavor Components in Flavored Milk

This study aimed to determine chemical composition and assess exposure in flavored milk among Chinese residents, based on risk assessment methodologies of acceptable daily intake (ADI) and toxicological concern threshold (TTC). Esters (32.17%), alcohols (11.19%), olefins (9.09%), aldehydes (8.39%), and ketones (7.34%) comprised the majority of the flavoring samples. Methyl palmitate (90.91%), ethyl butyrate (81.82%), and dipentene (81.82%) had the highest detection rates in flavor samples. This study screened fifteen flavor components of concern and discovered that 2,3,5-trimethylpyrazine, furfural, benzaldehyde, and benzenemethanol were detected in 100% of flavored milk samples. Benzenemethanol was found in the highest concentration (14,995.44 μg kg−1). The risk assessment results revealed that there was no risk for Chinese residents in consuming flavored milk, and the maximum per capita daily consumption of 2,3,5-trimethylpyrazine, furfural, and benzenemethanol were 226.208 g, 140.610 g, and 120.036 g, respectively. This study could provide guidelines for amounts of flavor additive ingredients in milk.


Introduction
According to the Chinese Dairy Industry Quality Report (2019), ultra-high temperature sterilized milk, flavored milk, fermented milk, and pasteurized milk accounted for 40.6%, 28.1%, 21.3%, and 10% of total liquid milk consumption, respectively. Flavored milk is a type of sterilized liquid milk that comprises at least 80% raw bovine (caprine) milk or reconstituted milk (GB 25191-2010), and serves as a nutritional alternative to plain milk [1]. Distribution of volatile compounds is directly related to food flavor [2]. Flavor can be added in appropriate proportions and unlimited amounts according to GB 2760-2014 "National Standard for Food Safety Food Additive Use". However, maltol directly stimulates Cyp1a1 gene expression [3]. Maltol has harmful effects on the skin, eyes, and respiratory system [4]. Mutagenesis and genotoxicity are linked to furfural compounds [5]. Moreover, the safety of flavor components is frequently neglected, with the characteristic of self-limiting and low dosage.
The Flavor and Extract Manufacturers Association (FEMA) has clearly stated the safe amount of flavor additives in soft drinks, candies, baked goods, puddings, and meat, as shown in Table 1. Unfortunately, the amount added to dairy products was rarely mentioned. Flavored dairy products, as the most popular dairy products among children, should be given greater attention. Acceptable daily intake (ADI) was initiated by the Joint FAO/WHO Expert Committee on Food Additives (JECFA) in 1961 and is accessible for toxicological evaluation [6]. When estimated daily intake (EDI) is smaller than ADI, it does not cause  [7]. Toxicological concern threshold (TTC) is also a useful screening and prioritizing measure for assessing food safety [8]. Each substance is examined and categorized based on chemical structure, and divided into three human exposure thresholds (1800, 540 and 90 µg p −1 d −1 ). When a substance's human exposure is lower than the threshold value, the potential safety risk is negligible. Risk exposure assessment needs to combine the concentration of chemicals and the amount of food consumed [9]. A common means of determining this is to interview and record by questionnaire, such as investigating and analyzing the correlation between dairy product consumption and cardiovascular diseases [10], serum vitamin D deficiency [11], ACEN [12], cultural factors and purchasing behavior [13]. Hence, it is an excellent choice that this study adopts a questionnaire method to investigate the consumption of flavored milk in different age groups.
Currently, flavor extraction and determination procedures in dairy products include solid phase micro-extraction (SPME) [14,15], supercritical CO 2 fluid extraction (SFE), and dynamic headspace (DHS) [16]. The flavor compounds of infant milk powder [17], reducedfat dairy products [18], raw goat cheese [19], pea protein beverages [20], camel milk [21], cattle and sheep milk [22], mascarpone cheese [23], and sweet condensed milk [24] were mainly determined by gas chromatography-mass spectrometry (GC/MS). We investigated and measured the consumption of flavored milk in various age groups and established a risk assessment model for flavored milk. This risk assessment has been carried out by taking the mean value as the food consumption data and the maximum detected value as the substance concentration data. This study was necessary and important to provide additive standards for flavor ingredient amounts in milk.

Sample Pretreatment
The method of measuring flavor samples was that used by Li Ning et al. [21]. Fifteen flavor components of concern were configured into mixed standard solutions and stored in a 4 • C refrigerator. An external standard approach was used for quantitative analysis. Flavored milk samples were put in solid phase extraction bottles for investigation. The 50/30 µm DVB/CAR/PDMS fiber head was aged at 250 • C until the baseline was stable [25]. The extraction temperature of the solid phase micro extraction platform was set to 60 • C and rotation speed to 800 rpm, then balanced for 10 min after inserting the aged fiber extraction head. A distance of 1.5 cm was kept between the fiber head and the liquid surface. The extraction temperature was kept at 55 • C for 50 min, and the extraction fiber head was put into the GC-MS injection port at 250 • C for 5 min of analysis.

GC/MS Conditions
A DB-WAX column was adopted for GC separation. The heating procedure was as follows: the initial temperature was 40 • C for 5 min, and the temperature rose to 150 • C at 3 • C/min, then to 230 • C at 6 • C/min and held for 5 min. The inlet temperature was set to 250 • C and flow rate set to at 1.0 mL/min. No shunt injections and solvent was delayed for 3 min. MS settings were as follows: electron ion source (EI), no solvent delay, SCAN mode, mass scanning range m/z 35~450 u. NIST, Wiley 9, and other libraries were searched for flavor compounds. Those compounds with an SI above 750 and Total Score above 90 were taken as preliminary screening results. Retention Index (RI), as calculated by the instruments, with RI obtained from the retrieval database, was used to further determine flavor compounds.

Exposure Risk Assessment Methods in Flavored Milk
The estimated daily intake (EDI) and per capita daily intake (PCI) of flavor components of concern in flavored milk were calculated according to Equations (1) and (2).
PCI (µg/person/day) = F × C. in which F is daily intake of dairy products per capita in kg/person/day; C is maximum flavor additive content in µg kg −1 ; and W is average weight in kg.

Data Analysis
The data was shown by mean ± SD. Non-parametric tests were used to analyze significant differences by IBM SPSS Statistics 26. The bar plot diagram, heat map, and principal component analysis were drawn in Tutools (https://www.cloudtutu.com/ accessed on 9 March 2023).

Distribution of Sample Characteristics
There were 2108 valid questionnaires obtained, with a questionnaire efficiency of 90.36%. The sociological characteristics of participants are shown in Table S2, including genders, ages, occupations, region of residence, and flavor preferences. In this study, the distribution of gender was lopsided, with females (60.2%) more represented than males (39.8%). This gender ratio is consistent with the study investigating the impact of society and lifestyle factors on dairy consumption [26]. The average weight (kg) and daily consumption per capita (g) from different ages and regions were shown in Table 2. There was a significant difference in the daily per capita intake of flavored milk between males and females (p < 0.01). Males consumed more than females among all age groups, but the results were not consistent with Marek Kardas et al. [27]. Teenagers (<18 years old) had the highest dairy consumption (males 77.57 ± 89.34 g, females 55.14 ± 52.44 g), followed by persons aged 18~24 years (males 57.66 ± 89.59 g, females 50.57 ± 56.66 g). Previous studies indicated teenagers consumed more flavored milk than plain milk in order to enhance their consumption of sugar and fat [28]. Chinese teenagers (66 g) consumed more flavored milk per day than adolescents (50 g) in the United States [1].
This study also analyzed flavored milk consumption in seven regions, namely eastern, southern, central, northern, northwestern, southwestern, and northeastern China. Males from northern China (55.13 ± 133.73 g) and northeastern China (64.33 ± 87.32 g) had higher flavored milk consumption than other regions. In the northeast, the consumption of males (64.33 ± 87.32 g) was much higher than that of females (36.12 ± 48.83 g). Gender, grade, and region all had an impact on the intake of milk. Some results found that children in northern schools were more likely to consume milk than children in southern schools in the United States [29].
In addition, we investigated flavor preferences for flavored milk in Figure 1. Results indicated that people preferred strawberry flavor, milk flavor, chocolate flavor, wheat flavor, red date flavor, mango flavor, and yellow peach flavor, while raspberry flavor, red bean flavor, passion fruit flavor, and pineapple flavor were liked by fewer people. Taste had a significant impact on children's flavored milk consumption, and a high association with brands and emotions was shown. Some studies discovered that 50% of Belgian children (8~13 years old) preferred chocolate flavor first and fruit flavors second [30] and removing the option of chocolate flavored milk significantly reduced intake of milk [31]. red bean flavor, passion fruit flavor, and pineapple flavor were liked by fewer people. Taste had a significant impact on children s flavored milk consumption, and a high association with brands and emotions was shown. Some studies discovered that 50% of Belgian children (8~13 years old) preferred chocolate flavor first and fruit flavors second [30] and removing the option of chocolate flavored milk significantly reduced intake of milk [31].

Composition Analysis of Seven Flavor Samples
Strawberry flavor, milk flavor, chocolate flavor, wheat flavor, red date flavor, mango flavor, and yellow peach flavor were analyzed, as shown in Figure 2b-d. A total of 168 compounds were identified, esters being the main components among the seven flavor samples. The strawberry flavor had the highest proportion of esters, as the most important category [34], including propyl decalactones, ethyl 2-methylbutyrates, ethyl butyrate, methyl cinnamate, phyllyl acetate, and ethyl caproate. These compounds were also detected in previous studies [35]. The PCA (Figure 2c) showed that mango and strawberry had the most similar composition. Mango had a higher percentage of ethers, anhydrides, alcohols, and olefins [36,37].
The chocolate flavor had a complex composition of aldehydes, pyrazines, alcohols, esters, ketons, furans, acids, and phenols [38]. According to this study, chocolate had the most aldehydes, while other studies discovered pyrazines were the major volatile and key odor compounds in chocolate flavor. It is possible that pyrazines produced by the Maillard reaction were the most important compounds that contributed to the final chocolate flavor [39]. The PCA diagram revealed that chocolate and wheat flavors had a similar composition. Wheat flavor had the largest amount of pyrazines and alkanes. According to the heat map (Figure 2d), wheat flavor mostly consisted of pyrazines, thia-

Composition Analysis of Seven Flavor Samples
Strawberry flavor, milk flavor, chocolate flavor, wheat flavor, red date flavor, mango flavor, and yellow peach flavor were analyzed, as shown in Figure 2b-d. A total of 168 compounds were identified, esters being the main components among the seven flavor samples. The strawberry flavor had the highest proportion of esters, as the most important category [34], including propyl decalactones, ethyl 2-methylbutyrates, ethyl butyrate, methyl cinnamate, phyllyl acetate, and ethyl caproate. These compounds were also detected in previous studies [35]. The PCA (Figure 2c) showed that mango and strawberry had the most similar composition. Mango had a higher percentage of ethers, anhydrides, alcohols, and olefins [36,37].
The chocolate flavor had a complex composition of aldehydes, pyrazines, alcohols, esters, ketons, furans, acids, and phenols [38]. According to this study, chocolate had the most aldehydes, while other studies discovered pyrazines were the major volatile and key odor compounds in chocolate flavor. It is possible that pyrazines produced by the Maillard reaction were the most important compounds that contributed to the final chocolate flavor [39]. The PCA diagram revealed that chocolate and wheat flavors had a similar composition. Wheat flavor had the largest amount of pyrazines and alkanes. According to the heat map (Figure 2d), wheat flavor mostly consisted of pyrazines, thiazoles, and furan compounds. Milk flavor had the highest proportion of alcohols relative to other flavor samples and was mainly composed of alkanes, phenols, and acids. Yellow peach flavor was mainly made up of esters, alcohols, aldehydes, and ketones. Additionally, yellow peach flavor had a higher content of peach aldehyde (26.26%), benzyl acetate (17.71%), and hexyl acetate (16.64%) which was considered key odorants influencing the flavor quality of peach fruit [40].
In the PCA analysis, the red date flavor had a unique flavor composition. The largest class of aroma-impact compounds was esters, including ethyl laurate, ethyl palmitate, methyl hydroxyacetate, and isopentyl acetate. Another important class of odor-active chemicals was aldehydes, and three aroma-impact aldehyde compounds were found in the samples. 5-methylfuranal, furfural, and benzaldehyde were among them [41]. Flavor is a complex mixture of volatile compounds, and the composition was specific to the species and variety of fruits [42].

Quantitative Analysis of Flavor Concerned Components in Flavored Milk
Fifteen components of concern were quantitatively analyzed in flavored milk samples and shown in Table 4. Benzenemethanol, 2,3,5-trimethylpyrazine, furfural, and benzaldehyde were all detected in 100% of flavored milk samples. Benzenemethanol is a colorless liquid with a mild pleasant aromatic odor naturally produced by fruits and teas. Pyrazines are nitrogen-containing heterocyclic compounds that contribute significantly to the flavor of various grilled, roasted, and similarly cooked foods, including baked potatoes, nuts, and meats [46]. Differences in furfural and benzaldehyde levels in flavored milk were generated by the Maillard reaction and the protein denaturation reaction. Benzaldehyde is an aromatic aldehyde bearing a single formyl group and an almond odor, and can be extracted naturally and is widely utilized in the production of aniline dyes, perfumes, flavorings, and medicines. In addition, the detection rates of 5-methylfurfural (96.4%), maltol (96.4%) and 1-hexanol (92.9%) were higher than 90%. Maltol is one of the byproducts of sugar degradation. There were differences in the content of flavor components of concern in various flavored milk brands, which were influenced by additive amount and manufacturing techniques. Maltol concentrations ranged from 0.83 µg kg −1 to 1682.11 µg kg −1 . The maximum content of benzenemethanol (14,995.44 µg kg −1 ) was determined, which was significantly higher than 2,3,5-trimethylpyrazine (2387.18 µg kg −1 ), furfural (3840.42 µg kg −1 ), and linalool (4958.30 µg kg −1 ).

Risk Exposure Assessment of Flavored Milk
The maximum EDI of different age groups was found to be considerably smaller than the ADI in the risk evaluation. The findings suggested that the flavor components had nothing exposure risks or health threats to the Chinese people. The detailed data is showed in Table 5. However, EDI values were different among age groups. Among people < 18 years old, EDI (0.023~20.27 µg kg −1 , bw d −1 ) was much higher than other age groups, which was closely related to the high flavored milk consumption by teenagers. The EDIs of benzenemethanol (5.231~20.27 µg kg −1 , bw d −1 ), furfural (1.34~5.19 µg kg −1 , bw d −1 ), and linalool (1.73~6.70 µg kg −1 , bw d −1 ) were much higher than other components.
The PCI of different age groups was less than or significantly less than TTC. The results of risk assessment are shown in Table 6 and indicated that there was no exposure risk to human health. The results showed that PCI of 2,3,5-trimethylpyrazine, furfural, linalool, and benzenemethanol were the highest across different groups, particularly among people under 18 years old. The PCI of furfural (297.92 g p −1 d −1 ) and benzenemethanol (1163.26 g p −1 d −1 ) in males (>18 years old) was closed due to their toxicological concern thresholds of 540 and 1800, respectively. Therefore, children should pay closer attention to the consumption of flavored milk. The maximum daily consumption was estimated using the TTC of the concerned components. The daily maximum intake of three flavor components (2,3,5-trimethylpyrazine, furfural, benzenemethanol) was less than one box of milk (250 g). The highest linalool and maltol consumption was less than two boxes of milk (250 g). This might serve as a starting point for additional research into maximal exposure and utilization in flavored milk.

Conclusions
Through the investigation of the intake of flavored milk in different regions, ages, and demographics, it was found that males (<18 years old) in southwestern regions had the highest intake of flavored milk. At the same time, 285 components of different flavor components were determined, and 15 flavor components of concern were screened for risk assessment. Two risk assessments confirmed that Chinese residents' intake of flavors in flavored milk was safe for their bodies. In addition, the maximum intake of 2,3,5-trimethylpyrazine (226.21 g), furfural (140.61 g), and benzenemethanol (120.04 g) was less than 250 g, which can provide a reference value for flavor additive amounts in milk. The result of per capita dairy product consumption in the questionnaire survey was a rough estimate and all the estimates were based on one-week records of flavored milk consumption. Although this may not represent the usual intake, it was enough for estimating people's average intake.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/foods12112151/s1. Table S1: Residents flavored milk consumption questionnaire; Table S2: Sample characteristics of flavored milk consumption questionnaire; Table S3: The basic information of 285 flavor compounds; Table S4: Linear equation, R2 and detection limit of flavor concerned components; Table S5: Recovery and precision of 15 concerned components.