Metabolomics in Meat Science from Farm to Fork: A Comprehensive Review of Quality, Safety, and Authenticity Applications

Sample preparation and extraction of metabolites

For sample preparation, quenching is an important step to halt metabolic processes through enzymatic inactivation (Álvarez-Sánchez et al., 2010). In meat studies, this would typically be achieved by snap-freezing in liquid nitrogen. To minimize variability related to water content, samples may also undergo freeze-drying before extraction, particularly when comparing tissues or treatments with differing moisture levels. Following quenching, the next step is the extraction of metabolites (Figure 1) by breaking down complex structures through sonication and extracting metabolites using solvents such as methanol (Windarsih et al., 2022b), chloroform, acetonitrile, or isopropanol, sometimes with additives like formic acid to enhance extraction efficiency (Panseri et al., 2022). Metabolomics studies can differ in multiple attributes based on research objectives and the choice between the untargeted and targeted approaches. Untargeted metabolomics can be applied for metabolic fingerprinting, which compares overall spectral patterns for sample classification, as well as for metabolic profiling, which involves identifying and relatively quantifying specific metabolites or metabolite groups within a sample (Farag et al., 2011). Conversely, targeted metabolomics focuses on specific compounds or classes of compounds that have been predefined as targets in a biological sample (Ibáñez et al., 2013; Maier and Schmitt-Kopplin, 2016). In targeted studies, sample extraction is tailored to selectively isolate a specific metabolite or class of metabolites (e.g., lipid classes, amino acids).

Figure 1.

Schematic overview of the metabolomics workflow for meat samples. The process includes sample preparation, metabolite extraction, chromatographic separation, mass spectrometric detection, and data output in the form of chromatograms and mass spectra.

This figure was created using BioRender.com

Chromatographic separation and detection

Following sample preparation, metabolites from extracts would then be separated primarily by one of 3 methods: gas chromatography (GC), liquid chromatography (LC), or capillary electrophoresis (Cevallos-Cevallos et al., 2009). LC can be further divided into high-performance LC (HPLC) or high-throughput techniques, such as ultra-performance LC (UPLC), capable of detecting more peaks in a short time. Detection of compounds is usually performed using ultraviolet, near infrared spectrometry, nuclear magnetic resonance (NMR), or mass spectrometry (MS), although the latter 2 dominate metabolomics analyses in meat science (Figure 1). NMR does not generally require extensive sample separation and detects a wide range of biological molecules, although the platform lacks the sensitivity of GC-MS and LC-MS (Dunn et al., 2005; Wishart, 2008). More detailed information regarding the instrumentation of metabolomics studies is available in the review by Dunn et al. (2005).

Metabolomics data analysis

Metabolomics data analysis involves several preprocessing steps to ensure a reliable and effective analysis. These steps generally include peak detection (identifies true metabolite signals from raw data), alignment (ensures peaks correspond across samples), and gap filling (fills missing peak values), followed by normalization and scaling to reduce variability (Figure 2). Normalization corrects for sample differences by adjusting metabolite concentrations relative to total ion counts, internal standards, or quality control (QC)-based normalization methods (de Livera et al., 2012). Scaling, on the other hand, ensures that metabolites with vastly different abundance ranges contribute equally to statistical models (van den Berg et al., 2006). Common scaling methods include mean-centering, autoscaling (unit variance scaling), and Pareto scaling, each of which influences the interpretation of metabolomic patterns. Once the data are preprocessed, statistical analyses are applied to extract meaningful biological insights.

Figure 2.

A schematic diagram of metabolomics data analysis.

^aTIC: Total ion current.

^bPQN: Probabilistic quotient normalization.

^cHMDB: Human metabolome database.

^dMANOVA: Multivariate analysis of variance.

^ePCA: Principal component analysis.

^fPLS-DA: Partial least squares discriminant analysis.

^gOPLS-DA: Orthogonal partial least squares discriminant analysis.

^hICA: Independent component analysis.

ⁱHCA: Hierarchical cluster analysis.

^jSVM: Support vector machine.

^kANOVA: Analysis of variance.

Data analysis in metabolomics is broadly categorized into univariate and multivariate approaches (Figure 2). Univariate statistical methods, such as t-tests and analysis of variance, are commonly used to evaluate individual metabolites independently, while fold change is typically applied as a complementary effect-size measure to quantify the magnitude of differences between groups (Saccenti et al., 2014). In contrast, multivariate approaches analyze metabolic profiles holistically, capturing intricate relationships among metabolites. Unsupervised techniques like principal component analysis (PCA) are often used to explore inherent clustering patterns, whereas supervised methods, including partial least-squares discriminant analysis (PLS-DA), orthogonal PLS-DA (OPLS-DA), and random forest analysis, facilitate classification and prediction in relation to a dependent variable. These statistical tools vary in application depending on the objectives of individual studies.

Studies employing metabolomics techniques are often categorized as discriminative, informative, or predictive in nature (Cevallos-Cevallos et al., 2009), although these categories could overlap. Discriminative analyses seek to separate groups of samples based on metabolic fingerprints, e.g., differentiating between forage- versus grain-finished beef (Carrillo et al., 2016; Osorio et al., 2012), and various postmortem aging durations or muscle types (Ma et al., 2017). In such cases, unsupervised clustering, such as PCA, can be used for separation between groups (Brownstein et al., 2003). Informative analyses aim to discover metabolites related to a particular process or attribute, such as tenderness development during postmortem aging (Graham et al., 2010, 2012; Kodani et al., 2017; Muroya et al., 2014), WHC (Bertram et al., 2010; Straadt et al., 2011, 2014; Welzenbach et al., 2016), meat color (Ma et al., 2017; Subbaraj et al., 2016), flavor attributes (Degnes et al., 2017; Sugimoto et al., 2017), and others. Predictive studies utilize statistical models generated from metabolomic data using methods such as PLS-DA, OPLS-DA, random forest analysis, or others to predict an attribute of interest or the classification of new/unknown samples (Carrillo et al., 2016; Osorio et al., 2013). In several applications, these distinctions are not necessarily clear; for example, in a study that employs PCA to find clusters between groups as the main objective, further analysis is often used to identify the metabolites driving the observed clustering. Readers are referred to existing review articles for more detailed discussions of bioinformatics workflows and data analysis strategies for metabolomics (Chen et al., 2022; Ren et al., 2015).

Role of nutrition and feeding strategies on metabolomic profiles of meat

Animal diet affects glycogen reserves, impacting lactic acid production, and influencing ATP depletion, which in turn affects postmortem meat attributes like pH, color, WHC, and tenderness. However, feeding strategies that result in inadequate glycogen reserves at slaughter, such as insufficient dietary energy supply, abrupt feed withdrawal, or poor nutritional balance, coupled with genetic backgrounds and external stressors, can contribute to the development of undesirable meat conditions such as PSE (pale, soft, and exudative) or DFD (dark, firm, and dry) meat. Metabolomics has emerged as a tool to evaluate diet-induced meat quality defects by comparing the metabolomes of normal and impaired meat samples. For instance, Ijaz et al. (2022) used ultra-high-performance liquid chromatography (UHPLC)-based metabolomics to compare the metabolome of normal (pH ≤ 5.70), atypical (5.70 < pH ≤ 6.09), and typical DFD beef (pH > 6.09). This study elucidated the relevance of energy metabolism to meat quality by reporting the lower glycogen and energy metabolites (e.g., glucose-6-phosphate, 3-phosphoglycerate) levels in DFD compared to normal beef. This was further explained by Kiyimba et al. (2024), who compared the metabolome of DFD meat supplemented with glycogen to that of normal meat. They found that upon addition of exogenous glycogen to DFD beef, the metabolome shifted towards the normal beef metabolome with a drop in pH from 6.9 to ∼5.7. Metabolomic profiling through GC-MS further validated this shift through an increase in glycolytic metabolites (glucose, fructose, and their phosphate derivatives, pyruvate, ATP, and lactate) in DFD meat supplemented with glycogen (Kiyimba et al., 2024). Similar outcomes were also reported by Harr et al. (2024), where glycogen content was lower in DFD meat, while alternative fatty acid energy pathways, such as linoleic acid and arachidonic acid metabolism, increased in atypical DFD meat. Such metabolomic evidence confirms that glycogen repletion restores glycolytic rate as well as provides molecular features that distinguish DFD muscle from its normal counterpart, offering a diagnostic framework for diet-related quality defects.

Numerous studies have been conducted to evaluate the effects of feeding conditions on meat quality, including differences in nutrient composition, fat content, and flavor profiles. More recently, metabolomics approaches have provided molecular-level insights into these effects, revealing distinct metabolic signatures and identifying potential quality-related biomarkers that differentiate grass- and grain-finished animals. For instance, a study by Gómez et al. (2022) distinguished beef from feedlot versus pasture-finished cattle. Their findings revealed a higher amount of choline in the meat of feedlot-finished cattle, which is generally associated with higher intramuscular fat and an elevated number of metabolites associated with carbohydrate metabolism, e.g., adenosine monophosphate (AMP), phosphate derivatives of glucose, and succinate, indicating higher glycolytic metabolism in grain-finished beef. Likewise, Spears et al. (2024), through untargeted metabolomics, also validated the clear differences in grass- or grain-fed beef by using LC-MS-based metabolomics. They reported significant differences in the fatty acid composition between grass-fed and conventional grain-fed beef, highlighting the impact of cattle-feeding systems on fat metabolism, along with variation in the amount of branched-chain amino acids. Consistent with these findings, Krusinski et al. (2024) used GC-MS and UHPLC-based metabolomics and reported that 80 fatty acids, along with 90 secondary metabolites such as pyridoxine, alpha-tocopherol, hippuric acid, and gallic acid, can be used to differentiate beef from cattle raised under different feeding systems. They identified higher concentrations of polyunsaturated (C20:3 n-9) and monounsaturated (C16:1 9t) fatty acids in beef from cattle fed pasture/hay than those fed baleage and hulls. Overall, these studies demonstrate that specific metabolites, particularly those related to fatty acid and carbohydrate metabolism, have the potential to differentiate beef produced under grass- and grain-based feeding systems.

Cattle feed is usually supplemented with additives for a variety of reasons, and to better understand how such additives influence meat quality at a molecular level, metabolomics has been increasingly applied to capture diet-induced metabolic changes. Liu et al. (2024) used UHPLC-MS for targeted and untargeted metabolomics to assess the impact of supplementation of rumen-protected sulfur-containing amino acids in the Tibetan sheep diet on meat quality. These additives are used to lower the dietary protein value and support muscle protein synthesis by bypassing the rumen fermentation of amino acids and reducing nitrogen emissions. The data revealed that glutamic acid abundance was positively correlated with decreased lightness (L*), linking amino acid metabolism to darker muscle color, while higher AMP levels were associated with greater shear force, reflecting the role of nucleotide turnover in postmortem energy depletion and rigor development. Similarly, Pedrini et al. (2024) used NMR spectroscopy to investigate the impact of an additive used to reduce methane emission, i.e., 3-nitrooxypropanol (3-NOP), on beef meat quality. While it reduced methane emissions by 38.2%, it minimally impacted carcass characteristics and meat quality (reduced supplemented cattle meat ultimate pH to 5.61 compared to control 5.68). No other detrimental impact on meat quality was reported. Interestingly, this study also revealed that metabolite results showed an enrichment of fatty acids such as myristoleic acid (C14:1), heptadecenoic acid (C17:1), and linolenic acid (C18:3) in supplemented beef, suggesting improved digestive and lipid metabolism that would not have been evident through conventional carcass evaluation. Together, these studies illustrate how metabolomics can mechanistically connect feed additive interventions with postmortem muscle biochemistry, with metabolic responses varying according to the type and purpose of the additive.

Role of genetic makeup on metabolomic profiles of meat

Genetic makeup is one of the key factors to determine meat quality, influencing muscle composition, metabolism, fiber type, collagen content, marbling, muscularity, and carcass traits. Metabolomic profiling enables the linkage between genotype and meat traits through the analysis of metabolites, as Cônsolo et al. (2022) compared the metabolic profiles of beef from Nellore males with either higher (HM) or lower (LM) genetic potential of muscularity by using one-dimensional NMR. The HM group showed higher values of metabolites associated with glycolysis, gluconeogenesis, tricarboxylic acid (TCA) cycle, fatty acid, and amino acid metabolism, i.e., glucose-6-phosphate, inosine, creatinine, lactate, pyruvate, betaine, glycerol, choline, and arginine than the LM group. Higher levels of energy metabolites in HM indicate enhanced substrate utilization, aligning with their phenotypic advantage in muscularity. Compared with HM, LM showed an abundance of inosine monophosphate (IMP), the terminal product of purine nucleotide degradation (Matarneh et al., 2023), suggesting limited nucleotide turnover and reduced energy metabolism, further reinforcing the contrast between HM and LM groups. These differences elucidate the enhanced protein synthesis, muscle growth, and efficient energy metabolism in HM than LM Nellore males. Further, metabolomics has also been used to relate metabolic features to carcass traits. For example, Li et al. (2022), who integrated genomic and metabolomic data in 493 crossbred beef cattle and reported that plasma creatinine, a terminal product of muscle energy metabolism, was positively associated with rib eye area (β = 0.025; 1.87% variance explained), reinforcing its role as a possible biomarker of muscle growth. Similarly, lactic acid was positively related to lean meat yield (β = 2.7 × 10⁻⁴; 2.71%), linking glycolytic metabolism with carcass composition. In contrast, succinic and fumaric acids, key intermediates of the TCA cycle, showed negative associations with carcass marbling score (β = −0.22 and −0.98, respectively), suggesting that enhanced oxidative metabolism may come at the expense of intramuscular fat deposition. Here, the β values denote the regression coefficients that describe the direction and magnitude of these associations, while the proportion of variance explained by individual metabolites was relatively modest (<3%), which is expected for complex carcass traits influenced by multiple genetic and environmental factors.

Breed influences tenderness through differences in muscle fiber characteristics, metabolism, and postmortem proteolysis. Phoemchalard et al. (2022) used NMR spectroscopy for metabolic profiling of the tougher Thai Native breed and the relatively tender Brahman-Thai and Charolais-Brahman crossbreeds (Phoemchalard et al., 2022). This study revealed that the metabolites associated with carbohydrate metabolism pathways, i.e., glycolysis and pyruvate metabolism, were significantly more active in the loin muscle of tender crossbreds compared to tougher Thai Native cattle. This metabolic pattern suggests that greater carbohydrate catabolism supports enhanced postmortem energy turnover, which may contribute to a more optimized pH decline rate and improved tenderness in the crossbred cattle. Additionally, valine, leucine, and isoleucine biosynthesis pathways were more pronounced in Charolais-Brahman beef, indicating a role for branched-chain amino acid metabolism in promoting intramuscular fat deposition and tenderness. Other than muscularity and tenderness, breed variations are also translated into intramuscular fat content, which can be further explored using metabolomics. A study by Muroya et al. (2022) evaluated the metabolic profiling of the loin muscle of Japanese Brown cattle (moderately marbled) and compared it to Wagyu Japanese Black cattle (highly marbled) by applying capillary-electrophoresis time-of-flight mass spectrometry (CE-TOFMS). This study reported higher ultimate pH and lower lactate content in less marbled Brown cattle beef than high-marbled Black Wagyu cattle. Metabolites related to fatty acid metabolism, such as glycine and choline, were more abundant in high-marbled Wagyu Japanese Black cattle loins than in less marbled Japanese Brown cattle loins. In contrast, less marbled Japanese Brown cattle had a higher number of glycolytic-associated metabolites, including guanosine, inosine, and uridine monophosphate. This contrast points to a metabolic shift where high-marbled Wagyu favor lipid-related pathways consistent with marbling, while less marbled Japanese Brown rely more heavily on glycolysis and nucleotide metabolism, aligning with their leaner phenotype and higher meat pH.

Most metabolomics studies related to the genetic potential of animals have focused on beef cattle, with relatively limited work in pork and poultry. A study by Liu et al. (2023) used LC-MS/MS-based metabolomics to explore meat quality differences between the Duroc × Landrace × Yorkshire (DLY) and newer disease-resistant Chuanzang Black pig crossbreed. This particular study explained meat quality differences based on metabolite profile and identified breed-specific metabolites, including 6 purine nucleotides, 16 carboxylic acid derivatives, and 17 fatty acyls. Metabolite profile differences between Chuanzang Black and DLY pork include higher levels of central carbon metabolism products (e.g., D-fructose 6-phosphate, inositol) in DLY, which contribute to its lower pH, and elevated IMP in Chuanzang Black, enhancing its umami flavor. Additionally, DLY pork shows increased levels of amino acid and peptide combinations, influencing its unique flavor profile. Similarly, a study by Shi et al. (2022) used untargeted metabolomics through UHPLC to examine meat quality changes in Guangxi Partridge chickens from 2 selective breeding lines, i.e., S-line (selection pressure = faster growth) and D-line (selection pressure = egg production). A total of 151 differential metabolites were identified in the breast and 115 in the thigh muscle, clearly separating the 2 breeding lines in PCA and OPLS-DA models. In the S-line chickens, branched-chain amino acids (valine and leucine) were decreased in breast muscle, reflecting altered valine, leucine, and isoleucine biosynthesis, while pathways related to glycerophospholipid and glutathione metabolism suggested modifications in membrane composition and antioxidant response. In the thigh muscle, higher levels of glucose, galactose, mannose-phosphate derivatives, and 6-phosphogluconic acid indicated activation of amino sugar metabolism and the pentose phosphate pathway, consistent with enhanced glycolysis and oxidative stress in fast-growing birds (Shi et al., 2022). Although this study provides a detailed metabolic map linking selective breeding to muscle quality traits such as pH, drip loss, and tenderness, such applications in poultry remain limited in scale.

Together, these comparative studies illustrate that genetic potential and breed differences in beef are reflected in trait-specific metabolic features. In particular, muscularity can be associated with metabolites such as creatinine and lactate, tenderness with postmortem nucleotide and carbohydrate metabolites including IMP and lactate, and intramuscular fat deposition with lipid-related metabolites such as choline and fatty acids. Although few genotype-focused metabolomics studies exist in pork and poultry, available evidence indicates that selective breeding alters muscle energy and nucleotide metabolism with measurable impacts on meat quality, underscoring the need for further research.

Effect of stress and animal handling on meat metabolome

Assessing animal stress during rearing, transportation, or lairage is challenging, but it can greatly affect meat quality. Metabolomics provides a practical means to detect stress-related metabolic changes, as shown by Tuell et al. (2020) using HPLC-MS metabolomics on broiler M. pectoralis major muscles from broilers reared under long (20 h light) versus short (12 h light) photoperiods. Broilers grow faster under the longer photoperiod treatment, and their muscles exhibited increased hue angle (discoloration) and accumulation of malondialdehyde, a secondary lipid oxidation product. The metabolomics analyses tentatively identified oxidized glutathione, a well-known biomarker of oxidative stress, to be concentrated in longer photoperiod muscles of animals. Additionally, there was a lower concentration of aromatic amino acids and dipeptides (tyrosine, tryptophan, phenylalanine) known to be related to the production of catecholamine hormones and melatonin. While other studies had found increased malondialdehyde concentrations in broiler serum (Guo et al., 2010) and pectoralis muscle with longer photoperiods, the usage of metabolomics provided a deeper understanding of this phenomenon.

Changes in photoperiods can disrupt muscle growth, and when combined with physiological stress, this disruption may contribute to muscle abnormalities such as woody breast (WB) and spaghetti meat (SM). Boerboom et al. (2023) applied LC-MS-based metabolomics to study WB myopathy in broiler chickens, revealing metabolic disruptions impacting meat quality. The study found elevated levels of arginine and taurine, which were relevant to compromised blood circulation, potentially impairing muscle oxygenation and contributing to muscle rigidity in WB. Additionally, reduced levels of anserine and beta-alanine revealed depletion of antioxidant defenses, which could increase oxidative stress, further damage muscle fibers, and impact tenderness. The study also found impaired lipid metabolism, evidenced by lower diet-derived lipids, which may result in reduced intramuscular fat deposition and negatively affect the flavor and juiciness of WB-affected chicken breasts. Similar outcomes were also reported by Liu et al. (2022) as they found ∼40% of lipid species were altered in WB breast, with triacylglycerols elevated and phosphatidylcholine and phosphatidylethanolamine reduced. Monounsaturated fatty acids increased (45.6% versus 38.9% in controls), while polyunsaturated fatty acids and short-chain fatty acids declined, indicating a lipid imbalance consistent with oxidative stress, membrane instability, and meat quality deterioration. In another study, Xing et al. (2020) compared the meat metabolome of WB and normal chicken meat samples by using NMR technology. This study reported that the WB samples had greater drip loss in WB samples (4.0% versus 2.2%; P < 0.05) and elevated oxidative stress markers (malondialdehyde nearly 3-times higher) in concentration. NMR profiling identified increased amino acids (alanine, glutamine, leucine, valine), along with creatine and inosine, while lactate and AMP were reduced. These shifts reflect impaired glycolytic activity and nucleotide turnover, whereas the accumulation of stress-related metabolites such as taurine, glutamine, and alanine indicates activation of antioxidant and cellular stress responses in severe WB. These results were in agreement with Choi et al. (2024), who compared WB and SM meat using UPLC-MS-based metabolomics. Both WB and SM exhibited elevated levels of eicosanoid, such as 15-HETE, lipid mediators derived from arachidonic acid that are closely associated with inflammatory and oxidative stress pathways. At the same time, depletion of critical cofactors such as NADH and folic acid suggested impaired mitochondrial energy metabolism and weakened antioxidant buffering. Together, these alterations indicate that WB and SM share common metabolic signatures of inflammation and disrupted energy/redox balance, offering biological explanations for their poor texture and diminished quality traits. Overall, muscle deformities linked to stress and rapid growth show higher levels of stress metabolites (e.g., taurine and arginine), reduced glycolytic intermediates and AMP/ATP balance, pronounced alterations in lipid classes (increased triacylglycerols, reduced phospholipids), and fatty acid profiles. Taken together, all these metabolic changes, coupled with reduced antioxidant capacity, could contribute to oxidative stress, poor WHC, and inferior texture.

Following rearing, transportation of animals to the slaughterhouse and their rough handling could also trigger the stress response, ultimately affecting the meat quality. In this context, Shaik et al. (2024) used LC-MS/MS-based targeted metabolomics to explore the relationship between epinephrine reactivity after transportation and meat quality in Spanish goats. The goats were categorized into low, medium, and high epinephrine response groups based on plasma epinephrine levels after transportation. Targeted metabolomics (LC-MS/MS) of longissimus muscle at 15 min postmortem revealed higher phosphatidylcholines (C38:0; 40:6) and sphingomyelin (C20:2) in the low epinephrine goats, while dopamine was elevated in the high epinephrine group goats, correlating with epinephrine reactivity. Despite metabolic differences and higher glycogen levels in medium and low epinephrine groups, no significant effects on meat quality were observed, likely due to the short transport time (180 min). Although some studies with transportation duration (>90 min) have reported a negative impact on goat meat quality traits, e.g., color and WHC (Alcalde et al., 2017; Batchu et al., 2021; Nikbin et al., 2016). These discrepancies in transportation impact on meat quality demand more research with recent techniques, such as metabolomics, to further elucidate the regulatory mechanisms.

Following transportation, lairage time, or the period animals are held in pens before slaughter to reduce stress levels, can help minimize stress-induced physiological changes (de Oliveira Costa et al., 2019). A study by Lee et al. (2023) used NMR spectroscopy to investigate the metabolomic effects of lairage time on pork loin quality and compared samples from pigs slaughtered with (24 hr) and without lairage time. While no significant differences in physicochemical meat attributes, such as pH and color values, were observed, NMR analysis corroborated these findings by showing no differences in metabolites associated with anaerobic metabolism (e.g., glucose, lactate, acetate), which are key determinants of meat quality. However, the authors suggested that stress from fear or feed deprivation during lairage time could have led to the mixed and inconclusive responses, indicating that further research, metabolomics-based investigations are warranted.

WHC

WHC is a key meat quality attribute that influences sensory characteristics and visual appearance, which in turn impact consumer acceptance and repeat purchase behavior (Van Oeckel et al., 1999; Young et al., 1997). WHC is usually assessed through traditional myo-water loss assays. Renou et al. (1985) were the first to demonstrate that WHC, assessed by imbibition time (the time required for meat juice to be absorbed into pH paper) as an indicator of water retention, correlates with proton NMR relaxation times. Later advances in these analytical techniques enabled the use of metabolomics to map the biochemical changes governing alterations in the WHC of meat. In this context, Zuo et al. (2022) used UHPLC-MS-based metabolomics to study the m. longissimus lumborum of beef during postmortem aging and identified 25 metabolites that were differentially abundant at 0, 0.5, 1, and 2 d postmortem. The study revealed that lactic acid levels plateaued within the first 2 d postmortem, while energy-related metabolites, including ATP, adenosine diphosphate (ADP), and AMP, declined significantly. These changes indicate rapid glycolytic progression and a concomitant decline in WHC during the early postmortem period, followed by partial recovery at later stages. In addition, fructose and mannose metabolism, as well as purine and pentose phosphate pathways, were identified as the most enriched pathways influencing WHC, consistent with findings reported by Muroya et al. (2022). Further, protein degradation can yield peptide fragments with enhanced polarity, thereby improving water binding. For instance, Yu et al. (2024a) investigated the effects of wet-aging on meat quality and exudate metabolites, changes in different beef muscles, and found that prolonged aging time enhanced proteolysis and protein breakdown, as reflected by metabolic pathways related to amino acids and peptide derivatives in the meat exudate. Conversely, in another study, opposite outcomes were reported where pork loins were subjected to multiple freeze-thaw cycles, and exudate was used as an analytical medium for metabolomics (Yu et al., 2024b). Those subjected to 5 freeze-thaw cycles had lower WHC when compared to those that underwent only one freeze-thaw cycle, yet they were overabundant in amino acids and peptides. This study found several amino acid-related pathways were significantly enriched, including arginine-related metabolism, β-alanine turnover, sulfur amino acid metabolism, and aromatic amino acid pathways. Although the authors postulated that this overabundance of peptides and amino acids in exudate could be due to physical disruption of muscle cells rather than enzyme-induced protein degradation, this led to poor WHC. This indicates that metabolomics data should be interpreted cautiously and in consideration of conventional physicochemical and biochemical assays. Moreover, further studies are warranted to evaluate the metabolome shift arising from enzymatic degradation during aging or cryo-induced alterations in relation to WHC.

Color stability

Meat color has been identified as the primary factor influencing consumer purchasing decisions, as consumers utilize this attribute as an indicator of freshness (King et al., 2023). Meat discoloration is primarily the result of oxidation of oxymyoglobin to metmyoglobin (Faustman et al., 2010). This process varies among animals, muscle types, as well as in their ability to regenerate reducing agents, e.g., reduced nicotinamide adenine dinucleotide (NADH) and oxygen availability under different packaging conditions, with metabolomics offering deeper insights into its impact on color stability (Figure 3; Table 2).

Figure 3.

Metabolomics-based overview of key meat quality indicators. The diagram summarizes metabolite markers associated with pre- and postslaughter factors, processed meat taste, and authentication, supporting their roles in evaluating meat quality, safety, and origin.

Muscle fiber type. Muscle fiber type is a critical factor influencing meat quality, especially color stability. The fiber type of a muscle is related to its function and related metabolic needs within the living animal (Faustman and Cassens, 1991; Seyfert et al., 2006). Red muscle fiber types are categorized by higher myoglobin contents and oxidative metabolism, while white muscle fiber types are glycolytic in nature (Choi and Kim, 2009). As such, white muscle fiber types typically exhibit a more rapid pH decline following slaughter due to increased buildup of glycolytic intermediates and end products (Ryu and Kim, 2005). Red muscle fiber types demonstrate a slower postmortem glycolytic metabolism, but more oxidative metabolism, which is known to exhibit a more rapid onset of discoloration. Several studies have found the glycolytic potential of muscles to be closely tied to meat color, as increased accumulation of lactic acid and hydrogen ions can lead to protein denaturation and increased CIE L* values (Hamilton et al., 2003; Moeller et al., 2003).

Ma et al. (2017) utilized HPLC-MS-based metabolomics to categorize metabolite profiles of beef muscles with different fiber types (glycolytic M. longissimus, intermediate M. semimembranosus, and oxidative M. psoas major) during postmortem aging. The more oxidative psoas muscle exhibited a more rapid onset of discoloration, especially with postmortem aging, and was categorized by a higher ultimate pH, myoglobin content, and release of nonheme iron. In terms of metabolite profile, the psoas major muscle was categorized by more abundant carnitine contents, including butyryl-_L-carnitine, pivaloylcarnitine, hexanoylcarnitine, and propionylcarnitine. These intermediates are known to be related to metabolism via the TCA cycle, and energy production in this manner could account for higher oxygen consumption in oxidative muscle types. Conversely, the longissimus muscle had lower nonheme iron, which the authors postulated may be the result of less myoglobin content coupled with metal ion chelation activity of more abundant carnosine and anserine (Brown, 1981; Ma et al., 2017). Similarly, another study (Setyabrata et al., 2023) used UPLC-MS-based metabolomics and reported that the psoas major had higher concentrations of mitochondrial and antioxidant enzymes, i.e., dehydrogenases and superoxide dismutase, than in the longissimus muscle. Whereas the longissimus muscle was overabundant with glycolytic energy metabolism enzymes such as creatine kinase m-type, phosphofructokinase, adenylate kinase isoenzyme 1, and cAMP-dependent protein kinase. Further, this study also reported higher concentrations of lipid-related metabolites, e.g., retinoic acid, stearoylglycerophosphoserine in longissimus muscle. These outcomes were in agreement with the findings of Yu et al. (2024a), who reported longissimus muscle had a higher concentration of short-chain fatty acids than the psoas major. Taken together, these studies demonstrate that different muscle fiber types are metabolically distinct. For example, oxidative muscles such as psoas major have increased mitochondrial activity, carnitine-driven energy metabolism, and greater myoglobin and iron content. In contrast, glycolytic muscles such as longissimus exhibit greater abundances of glycolytic enzymes and are characterized by more metabolites of lipids, short-chain fatty acids, elucidating differences in postmortem quality and color stability.

Availability of reducing agents. Reducing agents such as NADH play a crucial role in meat color stability. They help maintain the balance between metmyoglobin and oxymyoglobin, which directly influences meat color stability (Tuell et al., 2021). NADH, as a cofactor for redox mechanisms, reduces metmyoglobin to deoxymyoglobin through either enzymatic or non-enzymatic systems, delaying discoloration (Bekhit and Faustman, 2005). However, these reducing agents can be influenced by various postmortem factors such as aging or chilling. Subbaraj et al. (2016) discriminated between color-stable and labile ovine longissimus muscles at 1 and 8 weeks of aging using hydrophilic interaction liquid chromatography-mass spectrometry (HILIC-MS) based metabolomics, finding color-stable meat to have higher NADH concentrations even after 8 weeks, indicative of superior metmyoglobin reduction potential. Similarly, Ma et al. (2017) applied HPLC-MS-based metabolomics to determine the effect of aging on meat color, and found NAD⁺, a key metabolite, positively correlated with meat redness (r = 0.672) and negatively correlated with meat discoloration (r = −0.535). This was also reported by Warner et al. (2015) in rapidly chilled lamb carcasses.

In contrast, an NMR-based metabolomics study of muscle tissues from 3 species (beef, lamb, and venison) did not report NADH levels (Kanokruangrong et al., 2025), but it did observe succinate depletion correlating with reduced color values during display (Figure 3). This suggests that other substrates that are related to NADH supplementation, such as succinate, may also play a role in color stability. Setyabrata et al. (2023) utilized proteomics and metabolomics analyses to investigate biochemical changes in wet-aged beef tenderloin (psoas major) exudate, reporting increased levels of malate dehydrogenase and lactate dehydrogenase enzymes. Both enzymes catalyze reversible reactions involving malate and lactate, which are key intermediates in NADH regeneration. The increased abundance of these enzymes suggests higher redox activity involving malate and lactate, which may alter the balance of reducing equivalents and limit the availability of NADH needed to maintain myoglobin in its reduced form, thereby affecting color stability. These findings underscore the utility of metabolomics in providing a comprehensive understanding of the myoglobin redox system influencing meat color stability.

Packaging conditions and oxygen availability. Given the well-documented impacts of oxygen availability under different packaging conditions on meat quality attributes (Kim et al., 2012; Mancini and Hunt, 2005; Suman et al., 2014), it is reasonable to postulate that packaging may influence postmortem metabolic processes. This hypothesis was examined by Subbaraj et al. (2016), who applied HILIC-MS-based metabolomics to ovine longissimus muscles subjected to different aging periods (1 or 8 wk) and packaging conditions (vacuum or high-oxygen modified atmosphere packaging (MAP)). The authors found that succinic acid, a TCA cycle intermediate, was related to the observed differences in color stability, with higher concentrations in MAP samples. However, glutathione, a well-known antioxidant, was found to be more abundant in vacuum-packaged samples than in MAP-packaged samples, possibly due to lower oxidation stress under vacuum conditions. Likewise, Yi et al. (2015) found a benefit of adding TCA cycle intermediates for stabilizing ground meat color, although their eventual metabolism promoted lipid and myoglobin oxidation. This phenomenon mirrors what is observed between muscle types (i.e., oxidative muscle types typically have higher myoglobin content/redness but poorer color stability) and high-oxygen MAP packaging (Church and Parsons, 1995; Kim et al., 2012; McMillin, 2008). However, Seyfert et al. (2007) found that high-oxygen MAP packaging improves color stability up to 7 d of display regardless of glycolytic or oxidative muscle type. This may be a result of differences in oxygen-related oxidative processes between fresh and ground products, where intact muscle cuts would be less susceptible to discoloration. Ercolini et al. (2011) reported differences in beef metabolite concentrations by packaging type through ¹H NMR analysis. They found metabolites involved in cellular respiration, such as ADP, IMP, and other metabolites, degraded faster in modified atmosphere (60% O₂, 40% CO₂) and air packaging than in vacuum packaging. Taken together, these studies show that packaging conditions affect postmortem metabolism and lead to differences in energy-related metabolites, antioxidants, and nucleotides. The use of metabolomics gives a general understanding of these differences and why packaged meats exhibit varied color stability.

Tenderness

Metabolomics provides a broader analytical framework to lay out biochemical changes that determine meat tenderness (Figure 3). King et al. (2019) applied untargeted metabolomics by using UPLC-MS and GC-MS to profile metabolites of the tough and tender (slice shear force greater or lower than 19 kg, respectively) loin steaks. The study found that 102 metabolites out of 2,562 were correlated (r > 0.5 or < −0.5) to either slice shear force or desmin degradation, of which 28 were annotated. Amino acids such as phenylalanine, isoleucine, methionine, valine, tyrosine, and threonine were negatively correlated with slice shear force (r ∼ −0.5 to −0.62). These amino acids were also positively correlated with desmin degradation (r ∼ 0.72 to 0.82), reflecting proteolysis of myofibrillar muscle proteins, which was also translated into instrumental tenderness of steaks. In contrast, malic acid, glycerol-3-phosphate, and 3-phosphoglyceric acid were more abundant in tough steaks and positively correlated (r > 0.5) with slice shear force, indicating persistence of glycolytic and TCA intermediates that constrained tenderization. The glycolysis and TCA cycles play a critical role in muscle pH changes, which eventually affect the proteolysis and muscle tenderness.

Similarly, in another study by using ¹H NMR spectroscopy, Antonelo et al. (2020) compared the metabolomic profiles of tough and tender beef (average WBSF was 54.7 and 28.7 N, respectively). This study also showed that some amino acids (such as β-alanine, glutamine, and valine) and energy metabolism intermediates (such as acetyl-carnitine, adenine, and fumarate) were higher in tender beef (P < 0.05) and negatively correlated with WBSF (r ∼ −0.3 to −0.5). Similar outcomes were also reported by Kodani et al. (2017), where tender beef after aging had high concentrations of glutamine and alanine amino acids. Likewise, evidence from Johnson et al. (2024) in pork loin chops supports these findings, as GC-MS-based metabolites profiling revealed that tender chops (star probe ∼ 4.2 kg) had 22 differentially concentrated metabolites. Of these, glycolytic intermediates such as glucose-6-phosphate, mannose-6-phosphate, and fructose were less abundant in tender chops (fold change difference ∼ −1.64 to −0.44). Like previous studies (Antonelo et al., 2020; King et al., 2019), these pork tender chops also had greater amino acid release, such as serine, leucine, valine, tyrosine, and methionine (fold change difference ∼ 0.31 to 0.59) compared to tough chops (star probe ∼ 6.2 kg). Together, these results establish amino acid accumulation and reduced glycolytic metabolite persistence as consistent biochemical features of tender meat.

Aging is a well-known method to improve meat tenderness. Using multiple metabolomics platforms, including ¹H NMR, GC-MS, and HPLC, studies have also demonstrated that concentrations of specific amino acids increase with extended aging time, consistent with tenderness trendlines and reflective of ongoing proteolytic activity. Liu et al. (2025) applied untargeted LC-MS/MS-based metabolomics and reported that all aging methods (wet-, dry-, and bag-aging) lowered shear force with aging, but wet-aged beef consistently showed higher MFI values, indicating more extensive proteolysis. This was also reflected metabolically by elevated concentrations of free amino acids and small peptides in wet-aged samples, aligning with the established role of proteolysis in weakening myofibrillar structure. In contrast, dry- and bag-aged beef showed greater accumulation of fatty acids and hydroxy acids, indicative of lipid oxidation associated with surface dehydration. Despite similar tenderness outcomes, metabolomics revealed distinct biochemical routes, suggesting that amino acid enrichment predominates in wet-aging while lipid oxidation characterizes dry-aging. These outcomes are consistent with Yu et al. (2024a), who reported enriched pathways related to amino acid metabolism, peptide derivatives, and lipid oxidation, reflecting ongoing protein degradation, proteolysis, and oxidative processes during wet-aging, as evidenced by the presence of amino acids, peptides, oxidized fatty acids, and hydroxy fatty acids in long-term aged meat exudates.

Along with aging methods, aging duration and temperature are additional factors that alter proteolytic activity and thereby influence tenderness. In a study by Hernandez et al. (2025a), longissimus steaks were wet-aged at −2, 0, and 4°C for 14 to 56 d, and metabolite profiles were obtained by applying untargeted GC-MS-based metabolomics. The most tender phenotype exhibited the lowest shear force, greater proteolysis of desmin and troponin-T, and the highest free amino acid content (35.20 versus 11.51 nmol/kg), particularly branched-chain and aromatic residues (isoleucine, leucine, valine, phenylalanine, and methionine). Importantly, the abundance of amino acid enrichment in tender phenotypes is consistent with earlier metabolomics observations (Antonelo et al., 2020; King et al., 2019), reinforcing amino acid metabolism as a central biochemical feature of beef tenderization across diverse postmortem conditions.

Postmortem chilling regimes substantially alter the metabolomic landscape linked to beef tenderness. Using untargeted LC-MS/MS-based metabolomics, Chen et al. (2024) compared superchilled (−3°C within 5 h) and conventionally chilled (∼4°C within 24 h) beef longissimus. Superchilling lowered shear force values and accelerated actomyosin dissociation, changes that coincided with distinct metabolite shifts. 4-hydroxyproline, a collagen-specific amino acid, was consistently higher in superchilled samples and negatively correlated with initial shear force (r = −0.86; after 10 h postmortem), suggesting enhanced solubilization of connective tissue, favoring improved tenderness. Purine metabolites, including cytidine-5-monophosphate and adenylosuccinic acid, were significantly more abundant (after 5 h postmortem) in superchilled beef and inversely associated with shear force (r = −0.94 and −0.84, respectively). These findings indicate that superchilling accelerates both collagen solubility and nucleotide catabolism, biochemical routes that reduce structural rigidity and promote early tenderization, thereby distinguishing it from conventional chilling.

In addition to postmortem handling factors, intrinsic muscle characteristics such as marbling have also been linked to tenderness, although the relationship remains inconclusive. Jeong et al. (2020) applied NMR-based metabolomics to compare high and low-marbled beef, reporting higher sensory tenderness scores (4.58 versus 3.71) and lower shear force values (33.7 versus 49.0 N) in high-marbled samples. The metabolite profile distinguishing marbling groups included creatine, carnosine, lactate, glycine, methionine, and tyrosine. While these compounds were primarily interpreted in relation to flavor and juiciness, their overlap with amino acid and energy metabolism pathways that consistently emerge in tenderness-focused studies is noteworthy. Nevertheless, direct associations between marbling-specific metabolites and tenderization mechanisms remain unresolved, highlighting the need for further metabolomics studies to determine whether marbling contributes to tenderness biochemically or primarily through its effect on juiciness and overall eating quality.

Flavor and aroma

Meat flavor is a subjective measurement of organoleptic properties that impacts consumer purchasing and satisfaction, influenced by various factors including taste, odor, mouthfeel, and others (Dwivedi and Brockmann, 1975). The flavor of meat is influenced by the concentration of precursor molecules such as sugars, amino acids/dipeptides, nucleotide degradation products, and others (Macleod, 1994). During the heating process, these flavor precursors could participate in Maillard and lipid oxidation reactions to generate volatile flavor compounds (Khan et al., 2015; Macleod, 1994). The composition of flavor compounds can vary depending on aging duration, aging methods, muscle types, marbling, and animal species. Recent advancements in metabolomics have provided a deeper understanding of these processes, enabling more precise analysis of how each factor contributes to the development of flavor (Figure 3). The following sections examine these key factors individually to highlight their specific roles in shaping the final flavor profile of meat.

Impact of postmortem aging on flavor. As previously discussed, metabolomic analyses have shown that concentrations of free amino acids, dipeptides, and other metabolites increase with aging duration (Graham et al., 2010; Kodani et al., 2017; Muroya et al., 2022). Kodani et al. (2017) reported that metabolomics analyses showed that as the wet-aging period increases, the abundance of flavor-related precursor compounds, such as acetic acid, alanine, glutamic acid, and other amino acids, significantly increases, which could potentially impact the final meat flavor quality.

Along with the aging period, the type of aging process applied (i.e., dry versus wet-aging) has been shown to alter the volatile composition related to meat flavor. The application of the metabolomics approach has been shown to provide further information related to the alteration of the flavor precursor profile and to provide insights into the mechanisms potentially responsible for releasing those flavor precursors. The study by Kim et al. (2016) employed ¹H NMR to discriminate between dry- and wet-aged beef samples for 21 d. Those authors found several free amino acids, including glutamate, to be higher in dry-aged samples compared to wet-aged samples. Glutamate is a well-known flavor compound related to the perception of umami, and the increased abundance of this compound in the dry-aged samples was suggested by the authors to explain the higher flavor and overall liking score of the consumer panel. Notably, wet-aged samples exhibited higher IMP levels. Since IMP is typically degraded to ribose, inosine, and hypoxanthine, limited degradation in wet-aged meat may account for inferior flavor development by reducing the availability of Maillard-reactive sugars. Similar results were observed by other authors, revealing greater accumulation of amino acids/protein-derived metabolites in dry-aged products regardless of species and breed (Bischof et al., 2021; Setyabrata et al., 2021, 2022; Zhang et al., 2021a), providing further evidence on the importance of protein-derived metabolite compounds in the formation of dry-aged flavor.

Furthermore, the metabolite profiling also exposed the presence of other metabolites, such as vitamin degradation products (pyridoxine phosphate and ascorbyl stearate) and porphyrin rings that could contribute to the meat flavor development process (Setyabrata et al., 2021). Water-soluble vitamins could be degraded further into thiazole products that could react with amino acids during cooking to generate a meat-like aroma and flavor (Khan et al., 2015; Yu and Zhang, 2010). Those authors also reported that pathway analyses based on metabolomics profiling showed an alteration in terpenoid and polyketides metabolism, vitamin B6 metabolism, glutathione metabolism, and alanine, aspartate, and glutamate metabolism. Similarly, Setyabrata et al. (2021) identified the presence of shikimic acid in dry-aged pork through metabolomics analysis. The metabolite is an intermediate in the biosynthesis of aromatic amino acids and is only observed in microbial metabolism pathways, exhibiting the potential role of microorganism involvement in the liberation of dry-aging flavor precursors. This shows that metabolomics analysis can elucidate the underlying mechanisms responsible for meat flavor development. This is particularly important as flavor discrepancies between eating quality outcomes of the postmortem aging, especially the dry-aging process is common (Ha et al., 2019; Li et al., 2021b; Sitz et al., 2006) and the use of metabolomics could expose underlying mechanisms influenced by different aging factors such as duration, temperature, humidity, airflow, and microbial growth/composition to further understand and optimize the flavor generation process.

Intramuscular fat and cooked flavor development. Along with variations in Maillard reaction precursors, such as peptides, amino acids, and pentoses, intramuscular fat is a critical factor influencing flavor perception through its impact on lipid-derived volatiles and shaping muscle-, animal, and species-specific flavor (Calkins and Hodgen, 2007). Marbling can also contribute majorly to flavor and aroma (Table 2). A study used NMR technology to show differences between high- and low-marbled loin muscle (Jeong et al., 2020). High-marbled meat had higher levels of taste compounds like betaine, glycine, tryptophan, carnosine, creatine, glucose, and lactate compared to low-marbled meat. These metabolites were found to be related to flavor-affecting pathways such as protein or carbohydrate metabolism. Given that lipid content and profile vary among species, a study by Wang et al. (2022) used non-targeted metabolomics to compare muscle tissue from chicken, duck, pork, and beef to identify their contribution to meat flavor. This study confirmed that lipid metabolism plays a crucial role in generating species-specific meat flavor, with higher fat content in beef and pork enhancing lipid-derived volatiles, while the leaner composition of chicken emphasizes umami-related compounds.

Similarly, a study by Hicks et al. (2023) on ground beef patties with different lean sources and fat contents showed that lean sources did not significantly impact flavor metabolites in both raw and cooked beef patties. However, the fat content (10% versus 20%) of beef patties affected the fatty acid content, volatile aroma compounds, and metabolite distribution, influencing the final meat flavor. Different from Hicks et al. (2023), Setyabrata et al. (2024) reported that while the inclusion of dry-aged lean and fat trim in ground beef patties impacted the flavor precursors of the raw products, dry-aged lean trim significantly altered the metabolite profile following cooking. The authors demonstrated clear clustering of the samples based on the lean sources used, revealing a distinct separation between samples that included dry-aged lean trim, while the control samples and those with only dry-aged fat trim grouped together into the same cluster.

Setyabrata et al. (2024) also found that cooking significantly reduced the variation of flavor metabolites regardless of the formulation of beef patties. The authors reported that more umami-related metabolite compounds (such as 5'-S-methylthioadenosine, AMP, and glutathione) were found in greater abundance in cooked samples compared to raw counterparts, potentially contributing to meat flavor by directly impacting the taste perception. This observation showed that metabolomics profiling could potentially be utilized to uncover flavor precursor fate during the cooking process and identify their contribution to meat flavor generation.

Metabolomic profiling of processed meat for quality assessment

Processed meat usually undergoes curing, fermentation, smoking, or the use of additives to improve shelf life, sensory traits, or safety. These interventions could alter the underlying biochemical reactions, difficult to fully capture by traditional assays. Because metabolomics can map these changes at the molecular level, it provides critical insights into how processing conditions and ingredients shape product quality (Table 3). For instance, Yoo et al. (2016) utilized GC-MS-based metabolic profiling to evaluate fermented sausages supplemented with pineapple (a natural sugar source) to expedite fermentation. This study revealed clear metabolic differences, as pineapple-supplemented sausage had increased amino acid levels compared to controls (no pineapple addition), indicating active fermentation by Lactobacillus spp. This was further supported by the elevated level of lactic acid in supplemented sausage, depicting enhanced fermentation in pineapple-added sausages compared to controls, favoring the use of natural sugar sources in sausage fermentation. This study also reported a greater reduction in monosaccharides, i.e., glucose and fructose, compared with sucrose, a disaccharide, indicating the bacteria prefer monosaccharides for fermentation. However, the fermentation rate and sensory quality of processed meat may not only depend on the substrate sugars, but starter cultures of bacteria could also play a role. As demonstrated by Rocchetti et al. (2023), in Italian salami, the starter cultures (Latilactobacillus sakei, Staphylococcus xylosus) and glucose supplementation significantly influenced metabolomic fingerprints, with glutamyl peptides enhancing kokumi taste and lipid oxidation products differentiating sensory outcomes.

Like fermentation, the sensory quality traits of processed meat could vary depending on factors such as ripening time, salinity, and smoking practices. Traditionally, sensory traits are generally evaluated through consumer sensory evaluation, which is inherently subjective and cannot provide biochemical insight, highlighting the necessity of more objective approaches to evaluate product quality. To address these gaps, Sugimoto et al. (2017) used CE-TOFMS-based metabolomics to analyze dry-cured hams produced with different recipes, i.e., smoked ham, unsmoked ham, and processed ham. This study identified 203 charged metabolites and demonstrated significant correlations between metabolite profiles and sensory properties, including redness and fat whiteness. The smoked ham had a higher number of amino acids, including glutamic acid, than unsmoked and processed ham. A similar pattern was also observed in peptides; however, no differences were found in organic acids or nucleotides, explaining that the sensory development in hams was driven by proteolysis rather than energy metabolism.

Likewise, Huang et al. (2020) used NMR-based metabolomics to profile no-added-nitrite Chinese bacon during processing, identifying 21 metabolites across various stages, including marinating, air-drying, fermentation, and baking. A continuous increase in branched-chain (isoleucine, leucine, and valine) amino acids and organic acids throughout processing stages was observed. Baking, in particular, promotes taste formation, resulting in higher levels of metabolites (i.e., glutamate, AMP, and lactate) associated with umami, sweetness, and sourness, respectively. Similarly, Zhou et al. (2021) used ¹H NMR-based metabolomics to compare the metabolome of traditional versus modern style Jinhua ham. This study revealed that modern processing methods elevated taste-active compounds such as glutamic acid, lactate, and anserine, which correlated with higher sensory scores compared with traditional methods. In addition to processing methods, the length of ripening plays a decisive role in shaping sensory outcomes in dry-cured hams. Sugimoto et al. (2020) profiled dry-cured ham over 680 d of ripening and found that the highest umami aftertaste score occurred on day 540, coinciding with peak concentrations of aspartic and glutamic acids relative to total amino acids. Taken together, these studies illustrate that the metabolomics approach can discriminate the methods adopted for processing, and the generation or reduction of characteristic metabolites, e.g., sugar, lactic acid, and taste-developing amino acids and AMP during the process, and relate them to processed meat quality.

Characterizing microbe-driven metabolites in processed meat

Along with formulation and recipes, microbial communities play a critical role in determining the quality of processed meats. Recent studies have used metabolite profiling, including metabolomics-based approaches, to investigate microbial-derived compounds involved in fermentation, spoilage, and flavor development in processed meats. For instance, Han et al. (2021) investigated low-temperature sausages cooked at 70 to 80°C and stored at 20°C, revealing that Bacillus velezensis (spoilage bacteria) were the dominant bacterial species after 6 d of storage. They found that lipids, organoheterocyclic compounds, and organic acids were the primary non-volatile metabolites, with accumulation of acylcarnitines, a product of β-lipid oxidation. Similarly, Mu et al. (2020) used the GC-MS-based metabolomics to determine the metabolically active bacteria during the different stages of Panxian ham fermentation. Among the total of 31 significant metabolites, the majority belonged to amino acids (16) and fatty acids (6). Differentially abundant metabolites were involved in 30 metabolic pathways, including 6 of the 9 essential amino acid pathways. Their microbial analysis revealed Psychrobacter as the dominant bacterial genus (16.28%), and it further increased to 31.57% after salting. In Fungi, Debaryomyces and Aspergillus were the most dominant, linked to the production of amino acids, fatty acids, and organic acids during the fermentation process. In another study, Qin et al. (2024) determined the microbial diversity and metabolome of 4 different ham varieties from western Yunnan to compare the underlying differences. Microbiome data showed these 4 ham types had 97% similarity, and the remaining 3% was responsible for distinct microbe-driven differences in their metabolite profiles. Like Mu et al. (2020), this study also reported that among the identified 422 different metabolites, primarily amino acids (71) and fatty acids (27) were involved in various metabolic pathways. Moreover, the correlation analysis revealed Lentibacillus, Serratia, and Izhakiella had positive correlations with 3,4-dimethoxy-benzaldehyde (r ≥ 0.4) and gibberellin A36 (r ≥ 0.4). Acinetobacter and Marinococcus were negatively correlated with 2,4-dihydroxybenzophenone (r ≤ −0.6). Given the preliminary nature of these findings, the implications of these microbial and metabolic changes on processed meat quality are largely unknown, highlighting the need for further research.

Metabolomic profiling of artificial or plant-based meat

Artificial or plant-based meat is an emerging meat alternative; however, consumer skepticism regarding its nutritional value is a concern (Kombolo Ngah et al., 2023). van Vliet et al. (2021) used GC-MS-based untargeted metabolomics to compare the metabolome of grass-fed beef with plant-based meat products. Although the nutritional facts (label on products) were comparable, they found 171 differed significantly (false discovery rate < 0.05). Notably, 22 metabolites were distinctly present in beef, and 31 were unique to plant-based meat products. Beef samples predominantly had antioxidant or inflammatory metabolites (e.g., anserine, cysteamine), whereas plant-based meat was characterized by a higher abundance of phenolic metabolites. These findings suggest that while plant-based and beef meat can serve complementary roles in the diet, they are not nutritionally interchangeable. Likewise, Hernandez et al. (2025b) reported the metabolic differences in ground beef and plant-based meat, validating the findings of van Vliet et al. (2021). Ground beef had higher concentrations of amino acids, dipeptides, and their derivatives, while plant-based meat was rich in plant-derived compounds such as isoflavones (e.g., daidzein, genistein) and phenolic acids. These findings were also corroborated by Kaczmarska et al. (2021), who used GC-MS and LC-MS-based metabolite profiling to compare the volatile and non-volatile metabolites in traditional meat versus meat substitutes, e.g., tempeh, natto, and tofu. This study also revealed that traditional meat had elevated levels of protein precursors such as dipeptides, carnitine derivatives, and amino acids, as previously reported (van Vliet et al., 2021; Hernandez et al., 2025b). However, fermented plant-based products were enriched with free amino acids (e.g., tyrosine, methionine), and unique plant-related flavor-active volatiles like pyrazines, including bioactive peptides (e.g., glutathione, γ-glutamyl peptides). In parallel, cultured meat is also gaining popularity, and Park et al. (2025) compared the metabolite profiles of lab-grown (cultured) chicken muscle with conventional chicken meat through LC-MS/MS-based untargeted metabolomics. Multivariate analysis (PCA and PLS-DA) revealed distinct metabolic compositions, where conventional chicken meat exhibited relatively higher abundance of essential amino acids (e.g., leucine, isoleucine, valine), nucleotide-related compounds (e.g., inosine, AMP), and energy metabolism intermediates. Conversely, the cultured meat had a higher number of metabolites involved in cellular signaling or neurotransmission, e.g., diethanolamine and acetylcholine. These findings indicate that although cultured meat uses animal stem cells, it is metabolically distinct from conventional meat. Taken together, these studies show that plant-based and cultured meats exhibit distinct metabolic signatures compared with conventional meat, characterized by differences in amino acid, peptide, nucleotide, lipid-related, and plant-derived metabolites, indicating that while these products may complement traditional meat in the diet, they are not nutritionally or metabolically interchangeable.

Characterization of spoilage and freshness in meat by metabolomics

Spoilage of meat or contamination by pathogenic bacteria poses significant challenges to the meat industry, compromising both food safety and economic value. Metabolomics can potentially provide a rapid, relatively broad, and sensitive approach to detect meat spoilage, addressing the limitations of traditional methods (Table 4). Studies across various meats, such as beef (Argyri et al., 2015; Ercolini et al., 2011), pork (Zhao et al., 2022), chicken (Zhang et al., 2020), and sheep meat (You et al., 2018), have demonstrated the application of metabolomics in identifying potential spoilage indicators to enable the quality control and meat safety to be more efficient.

Ercolini et al. (2011) used headspace solid phase microextraction GC-MS, and ¹H NMR to analyze volatile organic compounds (VOCs) and microbial metabolites during beef spoilage over 0–45 d under various packaging systems. VOCs like acetoin and butanoic acid were linked to spoilage microbial growth, while ¹H NMR identified amino acids such as histidine, proline, and leucine as prospective beef spoilage biomarkers. Similarly, another study used these techniques to assess minced beef spoilage under varying environmental conditions (Argyri et al., 2015). Moreover, a study by Zhao et al. (2022) used exudate as an analytical medium for metabolomics and proteomics to study pork spoilage at −2, 4, 10, and 25°C. Spoiled pork was characterized by the presence of juniper acid and isopropylmalic acid, whereas fresh pork contained higher levels of ascorbic acid, tiglylcarnitine, carnitine, hypoxanthine, and IMP. Zhang et al. (2020) employed UHPLC-MS/MS-based non-targeted metabolomics to identify biomarkers reflecting chilled chicken freshness. Amino acid derivatives, including serylphenylalanine, phenylmercapturic acid, and indole-3-carboxaldehyde, along with gluconic acid, tyramine, and uridine monophosphate, were identified as potential indicators of meat freshness and spoilage. Similarly, You et al. (2018) applied GC time-of-flight MS-based metabolomics to determine sheep meat freshness and identified 27 statistically significant metabolites, including amino acids (e.g., phenylalanine and methionine), sugars (e.g., glucose-1-phosphate and isomaltose), sugar alcohols (e.g., ribitol), and organic acids (e.g., D-glyceric acid and lysine). Moreover, metabolites such as phosphorylated compounds (e.g., D-glycerol phosphate) showed strong negative correlations with established freshness quality indicators, including total viable counts (r = −0.76), total volatile basic nitrogen (r = −0.79), and pH levels (r = −0.76). While amide derivatives such as asparagine were positively correlated with total viable counts (r = 0.90), total volatile basic nitrogen (r = 0.83), and pH levels (r = 0.91), indicating the utility of metabolomics to find out metabolites, which could be related to meat freshness and quality.

Pathogenic bacteria, unlike spoilage bacteria, present a major challenge to the meat industry due to their low infectious dose. Chen et al. (2023) employed targeted and non-targeted metabolomics to identify biomarkers for Salmonella enteritidis, including acetylcholine, proline, and valine, with the highest accuracy (area under the curve = 0.95). Similarly, Xu et al. (2010) demonstrated that pork samples could be discriminated by natural spoilage or Salmonella typhimurium contamination using GC-MS. Several metabolites (e.g., valine, creatinine, tetradecanoic acid, hexadecenoic acid, and octadecanoic acid) were identified as potential biomarkers, with differences detectable within 24 h of contamination by Salmonella or spoilage microorganisms.

Overall, current applications of metabolomics in meat spoilage and pathogen detection remain preliminary, with most studies being untargeted in approach and limited in scale. While promising potential biomarkers of freshness and contamination have been reported across species, further large-scale, standardized investigations are needed to validate these markers (targeted metabolomics) and narrow down robust panels. Such validated signatures could then be integrated into rapid detection platforms or predictive models, offering practical tools for quality assurance and food safety monitoring in the meat industry in the future.

Authentication, adulteration, and fraud detection in meat using metabolomics

Studies have reported the usage of metabolomics to identify adulteration or authenticity of meat products (Table 4), such as the inclusion of pork in other meat products (Trivedi et al., 2016; Windarsih et al., 2022b), dog muscles in sausages (Windarsih et al., 2024a), mislabeling of organic meat (Robson et al., 2022), and the authenticity of Halal or Kosher meats (Suratno et al., 2023; Windarsih et al., 2024b; Windarsih et al., 2022a).

Robson et al. (2022) used rapid evaporative ionization MS (REIMS) combined with high-resolution time-of-flight mass spectrometry to analyze lipid profiles from different cuts of meat (neck, rump, and shin) taken from organic and conventionally raised cattle. The mass spectra (m/z 600–1000) were processed, and PCA and linear discriminant analysis models were used to distinguish between production systems and meat cuts. The model correctly identified the meat cuts (98% accuracy) and differentiated organic from conventional meat (84% accuracy). Similarly, Zanardi et al. (2015) employed ¹H NMR-based metabolomics to differentiate irradiated from non-irradiated beef, identifying glycerol, lactic acid esters, and tyramine as key discriminating metabolites. This capability is important given ongoing consumer concerns regarding irradiated meat, which, despite being tasteless and odorless, requires mandatory labeling.

Along with mislabeling, meat adulteration is another concern for the meat industry that can be detected through metabolomics (Figure 3). A study done by Trivedi et al. (2016) used untargeted metabolomics and lipidomics through UHPLC-MS and GC-MS in an effort to detect the adulteration of different grades of minced beef with pork. PCA data showed clearly distinguished profiles with different percentages of pork addition to beef, with some metabolites, e.g., 3-Oxohexadecanoic acid, decanoylcholine, heptadecane, and malic acid significantly increased with the change in percentage of pork adulteration to beef. Among the identified metabolic pathways, glutathione and inositol metabolism differed considerably between beef and pork. Similarly, studies have used metabolomics to validate or discriminate meat/muscles from different species (Akhtar et al., 2021; Zhang et al., 2023) and to detect adulteration in high-value meat-like Pangasius hypopthalmus and beef sausages with pork meat (Windarsih et al., 2022b) or dog muscles (Windarsih et al., 2024a), respectively. Since muscles from certain species are strictly prohibited in some religious meats, such as Halal or Kosher, studies have used metabolomics to determine their authenticity (Suratno et al., 2023; Windarsih et al., 2024b; Windarsih et al., 2022a). These studies applied LC-MS or GC-MS-based untargeted profiling to distinguish pork or dog components from allowed meat types by identifying species-specific metabolite markers, as mentioned above, such as dacysteine, coumarone, and pyrophosphoric acid.