Translate this page into:
How does the Fish Roe Metabolome Compare to that of Sturgeon Caviar using NMR-based Metabolomics? A Case Study from the Mediterranean Sea
* Corresponding author: Dr. Mohamed Farag Department of Pharmacognosy, Cairo Univ, Cairo, Egypt. mfarag73@yahoo.com
-
Received: ,
Accepted: ,
How to cite this article: Baky MH, Wessjohann LA, Mohammad AS, Al-Hammady MA, Abdelrahman AM, Farag MA. How does the Fish Roe Metabolome Compare to that of Sturgeon Caviar using NMR-based Metabolomics? A Case Study from the Mediterranean Sea. Sci Technol Nex 2025;1:58-0. doi: 10.25259/STN_24_2025
Abstract
Objective
Caviar and fish roe are considered valuable marine-derived foods with high nutritional and economic importance.
Material and Methods
1H-NMR spectroscopy coupled with multivariate data analysis was employed to profile the metabolic composition of roe from Sparidae, Sepiidae, Moronidae, and Portunidae in comparison with commercial sturgeon and salmonid caviar.
Results
The analysis revealed that lipids, particularly lysophosphatidylcholines (LPCs), acylglycerides, and fatty acids, were the predominant constituents across all samples, whereas organic acids were detected at significantly lower levels. Distinct differences were observed between commercial caviar and wild roe samples, including the detection of benzoic and citric acids only in commercial products, likely reflecting the addition of preservatives. Multivariate analysis effectively discriminated roe samples by species and, in the case of Sparidae and Sepiidae, by sex.
Conclusion
1H-NMR-based metabolomics provides a powerful approach for evaluating roe authenticity, detecting processing-related additives, and highlighting nutritional diversity among marine species.
Keywords
Caviar
Fatty acids
1H-NMR
Lysophosphatidylcholines
Multivariate data analysis
Sparidae
1. INTRODUCTION
Seafood products, including caviar, have been valued as a delicacy for millennia, with records dating back to ancient Egyptians and Phoenicians who prepared sturgeon roe with salt and vinegar, while the ancient Greeks also prepared caviar as a luxury food.[1] The term “caviar” is derived from the Persian expression Mahi Khaviari, meaning “egg-generating fish”.[2] Caviar is a high-value seafood delicacy made from salt-cured roe (unfertilised eggs) of sturgeon (Acipenseridae), widely consumed worldwide. Caviar is considered one of the most important seafoods with potential commercial and nutritional value owing to its richness in protein, vitamins, and polyunsaturated fatty acids (PUFA).[3]
The family Acipenseridae includes more than 27 species distributed worldwide, with the most valued black caviar traditionally obtained from Huso huso (beluga), Acipenser gueldenstaedtii (ossetra/diamond sturgeon), and Acipenser stellatus (stellate sturgeon), native to the black Sea basin.[4] Low-grade black caviar can be sourced from other Acipenseriformes, such as the American paddlefish (Polyodont spathula). According to the Food and Agriculture Organisation and Codex Alimentarius, true caviar (A label) refers exclusively to roe from Acipenseriformes, while roe from other fish species is classified as caviar substitute (B label).[4,5] Among these, red caviar from salmonids traditionally consumed in Russia and Japan has gained global popularity. Among these, red caviar from salmonids traditionally consumed in Russia and Japan has gained global popularity. Fish roe from the gilthead seabream (Sparus aurata L., family Sparidae) represents a nutrient-dense seafood product, featuring high-quality proteins with notable essential amino acid content, particularly lysine, leucine, and arginine, while also being rich in long-chain omega-3 polyunsaturated fatty acids such as EPA and DHA key components for human nutrition.[6]
Increase in consumer demand coupled with a decline in wild sturgeon populations has driven the search for substitutes from other species, including catfish, herring, mullet, and marine invertebrates such as crustaceans, sea urchins, and sea cucumbers, provided they are correctly labelled.[1] However, the expansion of the caviar substitute market has also increased incidences of food fraud, where lower-grade substitutes are deliberately mislabelled as premium black caviar.[4] Nutritional and biochemical differences between authentic caviar and substitutes, particularly in amino acid and fatty acid composition, offer opportunities for quality control and authentication.[7] Owing to the rapid growth in consumer demand for seafood products, especially caviar, holistic analytical techniques and statistical tools is warranted to unveil their quality characteristics and composition. Recently, metabolomics tools were increasingly employed for profiling and fingerprinting seafood products using both targeted and untargeted approaches to assess their quality and ensure safety.[8] NMR-based metabolomics provides a rapid, robust, and non-destructive approach for the identification and quantification of food constituents.[9] This technique encompasses a variety of measurements, including one-dimensional (1HNMR) and two-dimensional experiments such as HSQC, TOCSY, and HMBC, which facilitate the resolution of overlapping peaks observed in 1HNMR spectra and enable more confident metabolite identification (Mahrous and Farag, 2015). Considering the complex data sets generated from metabolomics analysis, multivariate data analyses including unsupervised principal component analysis (PCA) and supervised orthogonal partial least squares discriminant analysis (OPLS-DA), can ease sample classification and identify markers.[10] In the present study, a novel integrative NMR-based metabolomic approach that combines 1H-NMR fingerprinting with quantitative profiling and multivariate data analysis was applied to profile roe from Mediterranean species belonging to Sparidae, Moronidae, Sepiidae, and Portunidae. Furthermore, principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA) were applied to classify samples and identify discriminant metabolites as potential chemical markers for authenticating caviar substitutes. This strategy allows for comprehensive characterisation and discrimination of fish roe and caviar samples, demonstrating the modern potential of NMR in food metabolomics beyond its conventional applications. Moreover, the current work provides a scientific basis for authenticity verification, quality control, and understanding the metabolic variations among different roe and caviar types.
2. MATERIAL AND METHODS
Fish roe was obtained from Gilthead seabream (Sparus aurata L. Sparidae, 4 samples), European seabass (Dicentrarchus labrax L. Moronidae, one sample), and common cuttlefish (Sepia officinalis L., Sepiidae, two samples). Roe was also obtained from crustaceans, namely blue swimmer crab (Portunus pelagicus, Portunidae, 2 samples). A summary of sturgeon caviar and different roes-producing fish species under investigation is provided in Table 1.
| Code | Scientific name | English name | Family | Origin | Caviar/roe photos |
|---|---|---|---|---|---|
|
D1 (male) 1 DS (female) 1 DS2 (male) 2 D2 (female) 2 |
Sparus aurata (Linnaeus, 1758) | Gilthead seabream | Sparidae (Porgies) | Alexandria, Mediterranean Sea |
|
| OS1 (female) | Dicentrarchus labrax (Linnaeus, 1758) | European Seabass | Moronidae (Temperate basses) | Alexandria, Mediterranean Sea |
|
|
S2 (male) S1 (female) |
Sepia officinalis (Linnaeus, 1758) | Common cuttlefish | Sepiidae | Alexandria, Mediterranean Sea |
|
|
K1 K2 |
Portunus pelagicus (Linnaeus, 1758) | Blue swimmer crab | Portunidae | Alexandria, Mediterranean Sea |
|
| KFR | Caviar Red | Local market |
|
||
| KB | Caviar Black | Local market |
|
2.1. Extraction for NMR analysis
Each dried and deep-frozen roe sample was extracted with 5 ml 100% MeOH using a Turrax mixer (11000 RPM) for five 20 s periods. To prevent heating, a period of 1 min separated each mixing period. Extracts were then vortexed vigorously and centrifuged at 3000×g for 30 min to remove any debris. For NMR analysis, 4 ml were aliquoted using a syringe and the solvent was evaporated under a stream of nitrogen till dryness. Dried extracts were resuspended with 800 µL 100% methanol-d4 containing HMDS. After centrifugation (13,000×g for 1 min), the supernatant was transferred to a 5 mm NMR tube. All 1H NMR spectra for multivariate data analysis were acquired consecutively within a 48h time-interval with samples prepared immediately before data acquisition. Repeated control experiments after 48h showed no additional variation.
2.2. NMR fingerprinting and quantification
Agilent VNMRS 600 NMR spectrometer was used to record all NMR spectra operating at a 1H NMR frequency of 599.83 MHz and combined with implemented Varian VNMRJ 2.2C spectrometer software.[9] For 1H NMR experiments, the instrument’s parameters were adjusted to a relaxation delay = 17.95s; pulse angle = 90°, pulse width (PW) = 6.25μs and number of transients = 120. For 2D-NMR experiments, 45° pulse width was used within the standard CHEMPACK 4.1 pulse sequences (COSY, TOCSY, HSQC, HMBC). The heteronuclear single quantum coherence spectroscopy (HSQC) experiment was adjusted to 1JCH = 146 Hz with DEPT-like editing and 13C-decoupling during acquisition time. In addition, the heteronuclear multiple bond correlation (HMBC) experiment was optimised for a long-range coupling of 8 Hz; a two-step 1JCH filter was utilised (130–165 Hz).
2.3. NMR analysis
The 1H NMR spectra were automatically Fourier transformed, phase and baseline corrected with MestReNova software (9.0.1, Mestrelab Research, Santiago, Spain). The spectra were referenced to internal HMDS at 0.062 ppm for 1H NMR and to internal CD3OD signals at 49 ppm for 13C-NMR, respectively. Spectral intensities were then reduced to integrated sections of identical width (0.04 ppm), referred to as buckets, within the region of 10.94 to 0.0 ppm. The regions belonging to residual water between 5.0 and 4.7 ppm and methanol signals 3.26–3.20 ppm were removed prior, and the data were subjected to multivariate analyses. For metabolites quantification using 1H NMR spectroscopy (qNMR), the equation previously described by19 was followed. Manual integration of the peak areas of selected proton signals assigned to compounds and internal standard HMDS was performed.
2.4. Multivariate data analysis of NMR datasets
For NMR dataset, spectral intensities were reduced to integrated regions, referred to as buckets, of equal width (0.04 ppm) for all spectral (δ 0.4-11.0 ppm) and aromatic (δ 5.5-11.0 ppm) regions. The spectral regions corresponding to the residual solvent signals; δ 4.90–4.80 (water) and δ 3.33–3.28 ppm (methanol), were removed before multivariate analyses. This binning method allowed us to evaluate the absolute quantification of the identified metabolites. The table of bins was imported into SIMCA-P version 13.0 (Umetrics, Umea, Sweden) and hierarchical cluster analysis (HCA), principal component analysis (PCA), and orthogonal projection to latent structures-discriminant analysis (OPLS-DA) were performed with all variables mean centred and scaled to Pareto variance. Validation of the developed OPLS-DA models were verified using permutation tests and CV-ANOVA (ANOVA of cross-validated residuals). Metabolite markers were then recognised by analysing the S-plot, which was declared with covariance (p) and correlation (pcor) in addition to the variable influence in the projection (VIP).
3. RESULTS AND DISCUSSION
1HNMR spectroscopy was employed to generate comprehensive metabolic fingerprints of 9 roe samples derived from Mediterranean species, together with two reference commercial products black sturgeon caviar and red salmonid caviar which are considered premium benchmarks in the global market. The inclusion of both wild-caught Mediterranean roe and established commercial caviar allowed direct comparison between local alternatives and internationally recognised standards. This approach not only facilitated the identification of common metabolites but also highlighted unique features that may serve as chemical markers of authenticity, origin, and processing.
3.1. NMR metabolites fingerprinting of caviar extracts
The 1H-NMR spectrum of all samples was characterised by high-intensity peaks of fatty acids, cholesterol, and phospholipids, as shown in Figure 1, Supplementary Table 1, and suggesting that roes are richer in lipids. Amino acid signals were also detected but at much lower levels, being mostly overlapped by higher intensity peaks of lipids, the major metabolites in all analysed NMR spectra. Signals derived from preservatives such as benzoic and citric acids were detected only in commercial caviar samples. All signals were assigned based on examination of 2D spectra (1H-1H COSY, 1H-1H TOCSY, 1H-13C HSQC, and 1H-13C HMBC), considering less complexity in NMR signal overlap. Owing to the similarity in structures, fatty acids of different chain lengths (C18 vs. C20) could not be differentiated from the 1H-NMR spectrum, albeit a clear distinction was observed between n-6 and n-3 fatty acids due to the slight downfield shift of the terminal methyl group in the latter δH=0.96 ppm. Similarly, linolenic acid (omega-3 fatty acid) was differentiated from linoleic acid by the chemical shift of its allylic methylene atδH= 2.86 ppm, while that of linoleic acid was detected at δH= 2.81. The assignment of specific NMR signals in each metabolite is discussed below in more details mostly aided by 2D-NMR spectra.
-
1H-NMR spectrum of commercial black caviar sample in methanol-d4 at 600 MHz. Characteristic signals are assigned by analysing corresponding 2D-spectra and are labelled as a: cholesterol, b: Fatty acids, c: n-3 fatty acids, d: Unsaturated fatty acids, e: LPC, f: PLE, g: TAG, h: Benzoic acid, i: Formic acid, NMR: Nuclear magnetic resonance, LPC: Lysophosphatodylcholines, PLE: Phosphatodylcholines, TAG: Triacylglycerides
3.2. Identification of cholesterol
Cholesterol was the only major sterol identified in 1H-NMR of fish roe samples, indicating its presence at a relatively high level. Other cholesterol derivatives may be present but are below the detection limits of NMR, considering its low sensitivity. The most characteristic signal of cholesterol was that of C-18 methyl at δH/c 0.716/12.17, representing the most upfield signal in the 1H-NMR spectrum. Its assignment to cholesterol was supported by its key HMBC correlations to C-12, C-14, and C-17 at δc of 40.8, 58.2, and 56.5, respectively. Other discriminating signals for cholesterol were those of C-3 at δH/c 3.38/72.5 where the 1H-NMR chemical shift at 3.38 ppm confirmed the presence of free cholesterol and not cholesterol esters. The chemical shift of C-6 at δH/c 5.23/122.5 confirmed the presence of one double bond in the cholesterol thus excluding the presence of a 7-dehydrocholesterol. 1H-NMR quantification of cholesterol in caviar samples revealed that cholesterol level range 1.3 -11.9 µg/g, with the highest level was detected in sample DS2 at 11.95 µg/g and the lowest level was detected in DS at 1.34 µg/g. These levels are considerably lower than those reported in the literature which indicates that roe typically contains cholesterol within the range of 2–15%, though levels may decrease during processing. Certain fish species, such as salmonids, channel catfish, smelt, and kahawai, are particularly rich in cholesterol, ranging from 550 to 640 mg/100 g.[1] Several factors could affect the level of cholesterol in fish roe including differences in species, processing conditions, and the analytical methodology employed.[11] Notably, NMR possesses a higher detection threshold compared to LC-MS or GC-MS, potentially underestimating trace cholesterol levels.[9] Moreover, processing methods such as salting, heating, or fermentation can significantly alter cholesterol levels in roe.[12] From a nutritional perspective, cholesterol remains an essential structural lipid and precursor of steroid hormones, vitamin D, and bile acids. However, excessive dietary intake is associated with cardiovascular risk, making accurate profiling of cholesterol in high-value foods such as caviar and fish roe particularly relevant for both food safety and consumer awareness. The relatively modest cholesterol levels detected in this study suggest that Mediterranean roe could provide a rich source of beneficial lipids while contributing less cholesterol than traditionally expected for commercial caviar products.
3.3. Identification of phosphatodylcholines
Phosphatidylcholines (PCs) constituted one of the most abundant lipid classes detected across all fish roe samples.[13] Signals of phosphatidylcholine amounted for some of the abundant classes detected in all investigated fish roe samples, especially those of the three identical methyl groups of the choline residue at δH/c 3.22/56.1, which showed HMBC correlations to two methylene groups at δH/c 3.64/68.7 and 4.27/61.6, consistent with the choline moiety. The downfield shift of the hydroxymethylene at δH= 4.27 was revealed for esterification with the phosphate group as in phosphatidylcholines. The signals of the glycerol moiety attached to the phosphate ester appeared slightly downfield from other acyl glycerol, with the hydroxymethylene being directly attached to the phosphate group appearing at δc = 67.2. The chain length of the esterifying fatty acids could not be unambiguously determined using 1H-NMR, as all fatty acids share the same overlapping signals of their aliphatic chains at δH =1.25-1.30. Phosphatidylcholine is broadly recognised as the predominant phospholipid in fish tissues, including roe. In various marine species such as salmon, tuna, and rainbow trout PC constitutes approximately 38–55% of total phospholipids.[14] Their abundance is functionally significant, as PCs are critical components of biological membranes, influencing fluidity, signalling pathways, and lipoprotein assembly. Moreover, phosphatidylcholines serve as reservoirs for polyunsaturated fatty acids (PUFAs), particularly eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), which are recognised for their cardiovascular and neuroprotective benefits. The presence of PCs as a dominant lipid class also provides an important discriminating feature for metabolomic profiling and can contribute to species-specific lipid signatures relevant to authentication of caviar substitutes.
3.4. Acylglycerides
Acyl glycerides were identified by the presence of characteristic signals of the glycerol moiety, especially of C-2 detected at 5.07 and for diacylglycerides (DAG) and at 5.25 for triacylglycerides (TAG). Meanwhile, C-1 and C-3 for glycerol appeared at δH= 4.3-4.1 upon esterification with fatty acids versus unesterified appeared at δH= 3.5-3.7. The esterification of glycerol was further proven from HMBC key correlations between glycerol protons and carbonyl carbon at δc = 173-174. The esterifying fatty acids were characterised by the α-methylene signal next to the carbonyl carbon which shows long-range HMBC correlations to C-2 of glycerol at δc = 72-73 and C1 or C3 δc = 63-64. These assignments are in line with previous NMR studies on glycerolipids in natural oils and fish lipids.[15] Detailed assignments are summarised in Supplementary Table 1. From a nutritional perspective, di- and triacylglycerides represent the major storage lipids in fish roe, serving as a rich source of energy and essential fatty acids, including long-chain polyunsaturated fatty acids such as DHA and EPA, which are crucial for human health and contribute significantly to the food value of caviar and fish roe.[16] The relative distribution of DAGs versus TAGs may vary among fish species and could serve as an additional discriminating parameter in roe profiling. For instance, salmonid roe is particularly rich in TAGs, whereas some marine invertebrates show higher proportions of structural lipids. This variation has potential implications for authentication, since characteristic acylglyceride patterns, combined with other lipid signatures, may help distinguish premium caviar from regional substitutes. Moreover, processing conditions such as salting, storage, and maturation may alter the TAG/DAG ratio, suggesting that glycerolipid profiling could also serve as an indicator of processing history and freshness.
3.5. Fatty acids
Fatty acids were present in the fish roe samples either in free form or esterified as acylglycerides and phosphatidylcholines.[17] 1H-NMR could not differentiate between free and esterified forms due to signal overlapping of the free acid α methylene at 2.24 ppm and that of the fatty acid esters at 2.30 ppm. The distinction could be made in the HMBC spectra where the free carboxylic carbons appeared at δc = 179-180 versus esterified form at δc = 173-174. Based on the signal intensity of these cross peaks between the carboxylic carbon to both α and β methylenes, it could be concluded that majority of fatty acids in caviar are in the form of esters.[15] Although most fatty acids share overlapping aliphatic resonances, certain structural features could be assigned from their well-resolved signals. For example, terminal methyl groups of n-3 fatty acids were observed at δH 0.96, while allylic protons of polyunsaturated fatty acids (PUFAs) resonated at δH 2.82–2.87. Additional characteristic resonances of unsaturated fatty acids included δH 2.02 and δH 5.32, corresponding to bis-allylic and olefinic protons, respectively. The high intensity of the allylic proton signals at δH 2.82–2.87 indicated a notable enrichment of roe samples with PUFAs, particularly long-chain omega-3 fatty acids such as EPA and DHA. 1H-NMR quantification of fatty acids in caviar samples revealed that fatty acids level range 2.01 -14.2 µg/g, with the highest level was detected in sample DS2 at 14.2 µg/g and the lowest level was detected in DS at 2.01 µg/g. The enrichment of roe samples with PUFA was evident from the high signal intensity of the allylic protons observed at δH= 2.82-2.87 suggest for their potential as a healthier lipid source, of potential for use in cardiovascular diseases.[1] The abundance of omega-3 PUFAs in fish roe highlights its value as a functional food. These fatty acids are well known for their cardioprotective, anti-inflammatory, and neuroprotective properties.[16] Consequently, the enrichment of roe with PUFAs underscores its potential as a healthier lipid source and a valuable dietary component for the prevention and management of cardiovascular disease. Beyond health benefits, the fatty acid composition also provides important markers for species differentiation and product authentication, as specific PUFA profiles are often characteristic of marine taxa.
3.6. Organic acids
Only a few signals belonging to organic acids were observed in the NMR spectra of caviar samples, but at much lower signal intensity than those of lipid molecules. One amino acid observed at appreciable level was glutamine assigned based on its α CH at δH/c 3.82/54.8 and the HMBC correlation between the β-methylene proton at 2.45 to two carbonyl groups at 174 and 177 ppm. A distinct signal at δH/c 8.46/170.2 indicated the presence of formic acid, while three aromatic signals at δH= 7.97, 7.51, 7.4 were assigned to benzoic acid found only in commercial caviar samples. Additionally, aliphatic signals at 2.66 and 2.75 were assigned to citric acid based on their HMBC correlation to two carboxylic groups at δC= 176.6 and 180.7 and were likewise only observed in the commercial samples of probably as added preservatives. The detection of organic acids, though minor relative to lipids, is of nutritional and technological relevance since they may influence flavour, stability, and shelf life of caviar products.[18] The detection of benzoic and citric acids provides important insights into product processing and authenticity. Their absence in wild Mediterranean roe, contrasted with their presence in commercial caviar, highlights the potential of 1H-NMR metabolomics for verifying the addition of preservatives and distinguishing natural roe from industrially processed products.
3.7. Multivariate data analysis of caviar samples 1H-NMR spectra
Fish roe from Sparidae, Sepiidae, Moronidae, and Portunidae were compared to commercial sturgeon and salmonoid caviar where samples, were mainly clustered based on their content of lysophosphatodylcholines (LPC) and saturated fatty acids as revealed from loading plot. PCA score [Figure 2a] revealed that black and red caviar samples were the most separated from Moronidae and female Sparidae roe, with the latter being more enriched in long-chain fatty acids and LPC, leading to positive PC1 values [Figure 2b]. Moderate segregation was also observed with Sepiidae samples along the PC2 axis, which was mainly influenced by signals derived from LPC δH= 3.22, 3.64, and 4.27 ppm and acylglycerides δH= 5.07, 5.25, and 4.1-4.3 ppm. Meanwhile, other samples did not show enough segregation. Similar results were observed in a supervised OPLS model [Figures 2c and d]. Due to Sparidae samples being the most segregated, we developed an OPLS model of Sparidae modelled against all other samples, where the abundance of 1H-NMR signals at δH at 2.02, 2.25 and 2.81 indicated enrichment of Sparridae roe with mono and polyunsaturated fatty acids. Modelling of Sparridae roe samples separately using PCA model clearly separated male from female fish roe along PC1 which accounted for 72.6% of observed variability. The 1H-NMR signals contributing the most belonged to fatty acids including aliphatic signals indicating chain length and methyl signals indicating the fatty acids δH= 2.24, and 2.30 ppm which appeared more abundant in female Sparidae roe, resulting in positive PC1 values. Similarly, the OPLS-DA model of female versus male Sparidae roes (R2 progression value of 0.89, p< 0.005) identified signals of saturated fatty acids at 0.85 ppm (terminal methyl) and 1.25-1.29 (repeated methylenes) as the most discriminatory signals in male Gilthead seabream [Figure 3]. Another PCA model (P < 0.005) [Figure 3c] was developed to compare male and female Sepiidae roe, which were clearly segregated along PC1 axis influenced mainly by 1H-NMR signals of δH= 3.22, 3.64, and 4.27 ppm and acylglycerides δH= 5.07, 5.25, and 4.1-4.3 ppm. These modelling results clearly indicated that variability among roe samples was attributed to lipids as the most abundant metabolites.[1] However, owing to the similarity of structural features of lipid molecules, NMR failed in determining the exact series homologs associated with this variability in terms of the exact chain length, number, alongside the position of double bonds. However, due to structural similarities among lipid species, 1H-NMR alone could not fully resolve homologous lipid series in terms of chain length, degree of unsaturation, or double-bond positions, warranting complementary LC-MS/MS approaches for precise annotation.[19]
- Multivariate analysis of all investigated fish roe samples based on their 1H-NMR spectra (0.0-9.5ppm). (a) PCA score plot. (b) loading plot of 11 samples as indicated in the figure legend. (c) OPLS-D score plot and (d) loading plot of the same data, respectively. PCA: Principle component analysis, OPLS-DA: Orthogonal projection to least square -Discriminant analysis
- Multivariate data analysis of Sparidae samples based on their 1H-NMR (0.0-9.5ppm). (a) OPLS-DA score plot. (b) loading plot of Saparidae samples modelled against other fish roes. (c) OPLS-DA score plot and (d) loading plot of all female Saparidae modelled against male Saparidae. OPLS-DA: Orthogonal projection to least square discriminant analysis
3.8. Technology development perspective
The application of 1H-NMR-based metabolomics to fish roe and caviar authentication presents significant technological implications for the seafood industry, particularly in quality control, food safety, and fraud prevention. The demonstrated ability of NMR, combined with chemometric analyses, to distinguish between authentic caviar and substitutes, detect preservatives, and identify sex- or species-specific metabolic signatures paves the way for the development of standardised authentication platforms. Such approaches could be translated into rapid screening technologies for regulatory agencies, aquaculture operations, and commercial markets, thereby safeguarding consumer trust and enhancing traceability systems. Nonetheless, current technological gaps persist, most notably the limited resolution of 1H-NMR in distinguishing structurally similar lipid homologues, such as fatty acids differing in chain length or double-bond positions. This limitation underscores the need for integrating complementary high-resolution platforms, such as LC-MS/MS-based lipidomics or advanced NMR modalities, to achieve deeper molecular annotation. Looking forward, future technological development could focus on miniaturised and automated NMR devices coupled with machine learning–driven metabolomic databases, allowing on-site authentication and nutritional profiling. Such progress would not only accelerate caviar quality assessment but also enable broader applications in marine biotechnology, functional food development, and nutraceutical innovation. From an interdisciplinary perspective, the proposed outlook bridges food science, analytical chemistry, marine biology, and aquaculture, with potential extensions into pharmacology, where bioactive lipids identified in roe could inspire therapeutic or dietary interventions. Furthermore, integration with digital technologies such as blockchain for supply-chain authentication could enhance transparency and consumer confidence. Thus, this study not only advances analytical methodologies for seafood authentication but also sets the stage for multi-sector technological innovations with societal, nutritional, and economic impact.
4. CONCLUSION
The present work highlights 1H-NMR spectroscopy as a robust tool for comparative metabolomic profiling of fish roe and commercial caviar. Lipids were confirmed as the primary contributors to metabolic variation, consistent with their role as the dominant nutritional class in fish eggs. Multivariate data analysis revealed clear differences according to species and sex, particularly in Sparidae and Sepiidae roe, which were mainly attributed to fatty acid chain composition and LPC content. The exclusive detection of benzoic and citric acids in commercial caviar underscores the method’s potential in detecting preservatives and ensuring product authenticity. However, the limited resolution of 1H-NMR in distinguishing lipid homologues suggests that complementary LC-MS/MS-based lipidomics is necessary for deeper structural annotation. Overall, this integrated approach provides novel insights into the biochemical diversity of marine roe and supports its application in food authentication, quality control, and nutritional evaluation.
Ethical approval
Experiments were conducted with approval from the Ethics Committee for the use of animal subjects, Faculty of Pharmacy, Ain Shams University, Cairo, Egypt (approval number REC 413).
Declaration of patient consent
Patient’s consent not required as patients identity is not disclosed or compromised.
Financial support and sponsorship
Nil
Conflicts of interest
Dr. Mohamed A. Farag and Dr. Mostafa H. Baky are on the Editorial Board of the Journal.
Use of artificial intelligence (AI)-assisted technology for manuscript preparation
The authors confirm that there was no use of artificial intelligence (AI)-assisted technology for assisting in the writing or editing of the manuscript and no images were manipulated using AI.
References
- Trends in Food Science Technology. 2021;110:405-417.
- Journal of Applied Ichthyology. 2011;27:169-175.
- PloS one. 2013;8:e71552.
- [CrossRef] [PubMed] [PubMed Central]
- Reviews in Fisheries Science Aquaculture. 2021;29:753-768.
- Journal of Applied Ichthyology. 2019;35:257-266.
- Mar Drugs. 2020;18
- International Journal of Food Properties. 2008;11:669-677.
- Applied Sciences. 2024;14:10755.
- RSC Advances. 2024;14:10799-10813.
- [CrossRef] [PubMed] [PubMed Central]
- Food Chemistry. 2022;383:132374.
- [CrossRef] [PubMed]
- Food Chemistry. 2018;264:73-80.
- [CrossRef] [PubMed]
- Foods. 2020;9:540.
- [CrossRef] [PubMed] [PubMed Central]
- Journal of Food Composition Analysis. 2024;136:106782.
- Journal of Agricultural Food Chemistry. 2020;68:14286-14296.
- Trends in Food Science Technology. 2001;12:328-338.
- Annual Review of Food Science Technology. 2018;9:345-381.
- [CrossRef] [PubMed]
- Journal of Oleo Science. 2010;59:515-519.
- [CrossRef] [PubMed]
- Food Analytical Methods. 2016;9:3428-3438.
- Aquatic and Seafood Products. 2024;14:10755.