Metabolic fingerprinting of saliva samples using nmr

Abstract

The present invention relates to a method for obtaining a 1D 1H-NMR spectrum of metabolites in a bodily fluid sample, the method comprising the step of acquiring 1D 1H-NMR data which involves: a) suppressing signals due to water; and b) suppressing signals due to proteins by using T2 relaxation filtering. The present invention further relates to a method method of determining if a subject is at risk of developing a disease, e.g. cardiovascular disease or diabetes type 2, the method comprising the steps of a. providing concentrations of metabolites determined in the saliva of the subject; and b. determining the risk of developing a cardiovascular disease or diabetes type 2 by using a previously trained mathematical model based on input provided in a.

Description

[0001] Metabolic fingerprinting of saliva samples using NMR

[0002] Field of the invention

[0003] The present invention relates to a method for assessing the activity of a metabolic pathway in a saliva sample of a subject, the method comprising the steps of a. providing concentrations of metabolites in the saliva of the subject by acquiring 1 D1H-NMR data which involves 1) suppressing signals due to water; and 2) suppressing signals due to proteins by using T2 relaxation filtering; b. determining the activity of the metabolic pathway by using a previously trained mathematical model based on input provided in step a, optionally wherein step b is computer implemented.

[0004] Background of the invention

[0005] The global prevalence of metabolic diseases has risen over the past two decades (Chew NWS et al. The global burden of metabolic disease: Data from 2000 to 2019. Cell Metab. 2023 Mar 7;35(3):414— 28). It is estimated that almost one in four adults in Europe has an early form of metabolic diseases, termed metabolic syndrome (Scuteri A, et al. Metabolic syndrome across Europe: Different clusters of risk factors. Eur J Prev Cardiol. 2015 Apr 15;22(4):486— 91 ). In Switzerland, over a third of all deaths are caused by metabolic diseases (Office Federal de la Statistique (OFS). Causes specifiques de deces [Internet]. Statistiques de Sante. 2020 [cited 2022 Sep 5]. Available from: https: / / www.bfs.admin.ch / bfs / en / home / statistics / health / state-health / mortality-causes- death / specific.html). The most effective way of managing metabolic diseases is by improving metabolic health early in the disease progression. It has been found that early lifestyle changes can reduce diabetes risk by 58% (Galaviz KI, et al. Lifestyle and the Prevention of Type 2 Diabetes: A Status Report. Vol. 12, American Journal of Lifestyle Medicine. SAGE Publications Inc.; 2018. p. 4-20). International organizations such as the WHO and UN are emphasizing prevention for reducing the burden of metabolic diseases (WHO. Targets of Sustainable Development Goal 3 [Internet]. Sustainable Development Goals. 2023 [cited 2023 Jun 5]. Available from: https: / / www.who.int / europe / about-us / our-work / sustainable- development-goals / targets-of-sustainable-development-goal-3). In Switzerland, the prevention of non- communicable disease and the promotion of health are key priorities in the 2030 national health strategy (Office Federal de la Sante Publique (OFSP). Strategie nationale Prevention des maladies non transmissibles (MNT) [Internet]. Strategie nationale Prevention des maladies non transmissibles (MNT). 2022 [cited 2022 Sep 5]. Available from: https: / / www.bag.admin.ch / bag / fr / home / strategie-und- politik / nationale-gesundheitsstrategien / strategie-nicht-uebertragbare-krankheiten.html). Unfortunately, the current diagnostic standards for metabolic diseases are to detect the onset of diseases rather than early signs. To effectively manage the metabolic health of their patients and prevent metabolic diseases, healthcare professionals need tools that can detect signs of metabolic dysfunctions that precede the emergence of symptoms. Thus, there is a strong need for innovative tools to track metabolic health.

[0006] A very effective method for assessing metabolic health is metabolic fingerprinting through personalised metabolomics (Buergel T, et al. Metabolomic profiles predict individual multidisease outcomes. Nat Med [Internet]. 2022 Nov 1 ;28(1 1 ):2309— 20. Available from: http: / / www.ncbi.nlm.nih.gov / pubmed / 36138150; Wishart DS. Emerging applications of metabolomics in drug discovery and precision medicine. Nat Rev Drug Discov [Internet]. 2016 Jul 30; 15(7):473— 84. Available from: http: / / www.ncbi.nlm.nih.gov / pubmed / 26965202). This technology relies on the measurement of an array of metabolites using specialised algorithms to correlate the levels of metabolites to the disease condition of an individual. This is a strong improvement over previous approaches, which measure individual biomarkers to signal the presence of the disease, this modern approach looks at the metabolism as a whole and is able to detect deviations in metabolic state, before there is too much cellular damage to be reversed (Newgard CB. Metabolomics and Metabolic Diseases: Where Do We Stand? Cell Metab [Internet]. 2017 Jan 10;25(1):43— 56. Available from: http: / / www.ncbi.nlm.nih.gov / pubmed / 28094011).

[0007] Numerous biofluids have been evaluated for personalized metabolomic measurements aimed at early disease detection. While blood has traditionally served as the gold standard for measuring biomarkers, recent advancements in the analytical sensitivity of instruments and the sophistication of algorithms have opened up new opportunities for frequently overlooked biofluids, such as saliva. Saliva is currently under extensive research and is showing promising results (Hyvarinen E, et al. Salivary metabolomics for diagnosis and monitoring diseases: Challenges and possibilities. Vol. 11 , Metabolites. MDPI; 2021 ; Cuevas-Cordoba B, Santiago-Garcia J. Saliva: A fluid of study for OMICS. Vol. 18, OMICS A Journal of Integrative Biology. 2014. p. 87-97; De Almeida PDV et al. Saliva composition and functions: A comprehensive review. Vol. 9, Journal of Contemporary Dental Practice. 2008; Meleti M et al. Metabolic profiles of whole, parotid and submandibular / sublingual saliva. Metabolites. 2020 Aug 1 ; 10(8): 1 — 1 1 )[10— 13]. The entire metabolome of saliva was measured in 2015, revealing 853 metabolites in saliva (Dame ZT et al. The human saliva metabolome. Metabolomics. 2015 Dec 1 ;11 (6): 1864-83.). Numerous papers have proven the usability of saliva to identify diseases in individuals, such as cardiovascular diseases (Bahbah El, et al. Salivary biomarkers in cardiovascular disease: An insight into the current evidence. Vol. 288, FEBS Journal. John Wiley and Sons Inc; 2021 . p. 6392-405; Gohel V, Jones JA, Wehler CJ. Salivary biomarkers and cardiovascular disease: A systematic review. Clin Chem Lab Med. 2018 Aug 28;56(9): 1432-42), diabetes (Bencharit S et al. Salivary Metabolomics of Well and Poorly Controlled Type 1 and Type 2 Diabetes. Int J Dent. 2022; De Oliveira LRP, et al. Salivary metabolite fingerprint of type 1 diabetes in young children. J Proteome Res. 2016 Aug 5;15(8):2491— 9; Perez-Ros P et al., Changes in salivary amylase and glucose in diabetes: A scoping review. Vol. 11 , Diagnostics. MDPI; 2021), glioblastoma (Garcia-Villaescusa A et al., Using NMR in saliva to identify possible biomarkers of glioblastoma and chronic periodontitis. PLoS One. 2018 Feb 1 ; 13(2)), and sports injuries (Hudson JF et al. “Fuel for the Damage Induced”: Untargeted Metabolomics in Elite Rugby Union Match Play. Metabolites. 2021 ;11 (8)). Salivary is a readily accessible biofluid that can be collected non-invasively and with well-known stability. (Duarte D et al.. Evaluation of saliva stability for nmr metabolomics: Collection and handling protocols. Metabolites. 2020 Dec 1 ;10(12):1— 15) which are mainly the result of bacterial growth in saliva in absence of preservatives (Gardner A, Carpenter G, So PW. Salivary metabolomics: From diagnostic biomarker discovery to investigating biological function. Vol. 10, Metabolites. MDPI AG; 2020).

[0008] Summary of the invention

[0009] Accordingly, the technical problem underlying the present invention may be formulated as the provision of a novel method for assessing the activity of a metabolic pathway in a saliva sample of a subject.

[0010] The problem is solved by the embodiments described herein, and as characterized by the hereto appended claims.

[0011] The present inventors have developed a novel NMR pulse program that is particularly usable in a method for assessing the activity of a metabolic pathway by obtaining a 1 D1H-NMR spectrum of metabolites in a body fluid sample, in particular in saliva. Saliva is an example of body fluid which can be obtained from a subject without the need for any invasive procedures, thus being particularly advantageous for use in any ex vivo diagnostic approaches. Said novel NMR-based method has been demonstrated by the present inventors to be characterized by improved spectral quality, in particular in aliphatic region, thus being particularly well suited for measurements of samples including metabolites. The present inventors have further demonstrated a link between the concentration of certain metabolites in saliva and further physiological parameters of the subject and / or pathological conditions of the subject in the determination if said subject is at risk of developing a disease associated with abnormal activity of a metabolic pathway. It follows, according to the present inventors, that said risk of developing a disease can be determined directly based on the concentration of certain metabolites in saliva. It is noted that the determination developed by the present inventors is further strengthened by innovative NMR-based method of measuring the concentrations of metabolites in saliva.

[0012] The invention will be summarized in the following embodiments.

[0013] In a first embodiment, the present invention relates to a method for assessing the activity of a metabolic pathway in a saliva sample of a subject, the method comprising the steps of a. providing concentrations of metabolites in the saliva of the subject by acquiring 1 D1H-NMR data which involves 1) suppressing signals due to water; and 2) suppressing signals due to proteins by using T2 relaxation filtering; b. determining the activity of the metabolic pathway by using a previously trained mathematical model based on input provided in step a, optionally wherein step b is computer implemented.

[0014] In a second embodiment, the present invention further relates to a method of determining if a subject is at risk of developing a disease associated with abnormal activity of a metabolic pathway, the method comprising the steps of a. assessing the activity of a metabolic pathway in a saliva sample of the subject using the method of the present invention; b. determining the risk of developing a disease associated with abnormal activity of the metabolic pathway by comparing the activity assessed in a with the activity of the metabolic pathway in a saliva sample obtained from a healthy subject.

[0015] Brief description of figures

[0016] The invention will be further illustrated by the following figures. These are however not meant to limit the scope of protection of the present invention (which is determined by the hereto appended claims) in any way. Instead, the presented figures serve merely illustrative purposes.

[0017] Fig. 1 shows schematically the pulse sequence applied in the NMR measurements in the methods of the present invention.

[0018] Fig. 2 shows the pulse sequence of the present invention as employed in Example 2. Fig. 3 shows a comparison of a spectrum collected according to a standard method (dark-gray) and the method of the present invention (light-gray), showing clearly improved quality of the latter spectrum.

[0019] Fig. 4 shows an exemplary saliva collection kit.

[0020] Fig. 5 shows overall relationships between parameters (also including concentrations of metabolites, as provided here), models and disease.

[0021] Fig. 6 shows further exemplary sample collection tubes

[0022] Fig. 7 shows further exemplary sample collection tubes

[0023] Detailed description of the invention

[0024] In one embodiment, the present invention relates to a method for assessing the activity of a metabolic pathway in a saliva sample of a subject, the method comprising the steps of a. providing concentrations of metabolites in the saliva of the subject by acquiring 1 D1H-NMR data which involves 1) suppressing signals due to water; and 2) suppressing signals due to proteins by using T2 relaxation filtering; b. determining the activity of the metabolic pathway by using a previously trained mathematical model based on input provided in step a, optionally wherein step b is computer implemented.

[0025] As it is to be understood herein, the metabolites are defined as organic compounds present in said bodily fluid sample. It is to be understood that the metabolites are preferably a composition including several organic compounds. Examples of metabolites are discussed herein.

[0026] The method of the present invention necessarily comprises the step of acquiring 1 D1H-NMR data. Any nuclear magnetic resonance (NMR) spectrometer capable of performing measurements on liquid samples and with a probe suitable for use for proton NMR measurement can be used in the method of the present invention. In principle, the skilled person is capable of performing an 1 D1H-NMR measurement on a sample of bodily fluid.

[0027] The 1 D1H-NMR measurement according to the invention includes two necessary elements in the pulse program: a) suppressing signals due to water; and b) suppressing signals due to proteins by using T2 relaxation filtering.

[0028] The execution of the NMR spectrum acquisition is conducted through pulse programs, which are executed by the NMR spectrometer. As it is apparent to the skilled person, the purpose of a pulse programme is to apply pulses to the sample to excite the molecules in the sample (herein a bodily fluid, for example saliva), generating a radiofrequency decay of various resonances that can be measured and transformed into metabolite concentrations. Accordingly, the present inventors have developed a pulse program including elements a) and b) as outlined above.

[0029] Element a) is required to avoid spectral interference from water present in a sample. The saliva sample typically includes large amounts of water (H2O) which likely interferes with signals of e.g. metabolites in the NMR spectrum. As it is apparent to the skilled person, use of deuterated solvents is one possible solution to avoid issues with interference of the solvent signals. However, in case of metabolite samples from saliva, which are not solid samples made up in an NMR-suitable solvent, but are provided already as water-containing saliva samples, complete exchange of the solvent (i.e., water) to a deuterated solvent would not be practical. Instead, solvent signal, in particular water signal, can be suppressed by specifically designed pulse sequences. Several such sequences are known to the skilled person, for example w5 sequence (also referred to as Watergate sequence) or pulse programs based on presaturation of the solvent signals (e.g. presaturation of the water signal).

[0030] Preferably, suppressing of signals due to water in the NMR measurement of the present invention is performed by presaturating the water signal. This can be done, for example, by using a sequence including a 1 D1H-NOESY-presat pulse sequence. (Mckay, 2011 , How the 1 D-NOESY suppresses solvent signal in metabonomics NMR spectroscopy: An examination of the pulse sequence components and evolution. Concepts in Magnetic Resonance Part A: Bridging Education and Research, 38 A(5), 197-220. https: / / d0i.0rg / l 0.1002 / cmr.a.20223).Thus, as encompassed by the present invention, in the method of the present invention said presaturating of the water signal is preferably done by using 1 D1H-NOESY- presat pulse sequence.

[0031] The skilled person will appreciate that instead of using 1 D1H-NOESY-presat pulse sequence, the presaturating of the water signal can also be done by using 1 D1H-NOESY-presat with gradient pulse sequence. As apparent, a different NMR spectrometer setup may be required for this purpose. The method of the invention further includes element b) suppressing signals due to proteins by using T2 relaxation filtering. The presence of resonances (signals) due to proteins is in principle unavoidable in a saliva sample. These signals may create unwanted background and interfere with the signals due to metabolites. As proteins behave very different in the solution from said metabolites, in particular as they tumble in different way, and accordingly the signals due to proteins have different relaxation properties, it is possible to filter out the signals due to proteins based on their T2 relaxation time.

[0032] Preferably, in the method of the present invention suppressing the signals due to proteins is done by using Carr-Purcell-Meiboom-Gill pulse sequence. Said sequence is also known to the skilled person as CPMG sequence. However, the skilled person is aware of alternative approaches to achieve similar effects.

[0033] Accordingly, the present inventors have reduced their invention to practice and have developed a pulse programme, involving an inventive combination of, for example, a NOESY-presat pulse programme with a CPMG sequence. The present invention is based, at least in part, on the surprising finding that the so arranged pulse program unexpectedly leads to a better resolution of peaks in areas that suffer from broad peaks, making it particularly suitable for studies of metabolities (e.g. metabolic fingerprinting) in natural samples, e.g. in bodily fluids, in particular saliva samples. Figure 1 below shows the schematic representation of this pulse programme. The set-up shown is constructed of a NOESY sequence including a presaturation pulse sequence and a Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence.

[0034] The pulse program as shown in Figure 1 is particularly useful to measure saliva metabolites, according to the present invention.

[0035] Accordingly, in the methods of the present invention, the pulse sequence substantially as shown in the appended Figure 1 may be used. Alternatively, in the methods of the present invention, the pulse sequence substantially as shown in the appended Figure 2 may be used.

[0036] The NOESY experiment leverages a phenomenon called the Nuclear Overhauser Effect, where the magnetization of one nuclear spin (like hydrogen) influences another nearby spin.

[0037] Accordingly, the methods of the invention may, in one embodiment, comprise a combination of a NOESY- presat pulse programme with a CPMG sequence. The resulting sequence may comprise the following elements. Presaturation Pulse: Before the main NOESY sequence begins, a presaturation pulse is applied. This pulse targets and suppresses the signal from a specific set of nuclear spins (often the water signal) that can be very intense and might interfere with the desired signals. By suppressing these unwanted signals, it enhances the clarity and quality of the subsequent NMR data.

[0038] Initial Pulse: The sample, often a metabolite solution, is first subjected to a strong magnetic field. A pulse is applied to excite certain nuclear spins (typically hydrogen nuclei) to a higher energy state.

[0039] Mixing Time: After the initial excitation, a mixing time is made part of the sequence. During this interval, nuclear spins interact with each other, especially those in close proximity due to the Overhauser Effect.

[0040] Detection Pulse: Another pulse is then applied to detect the nuclear spin relaxation. The signals obtained after this pulse provide information about the spatial proximities of the nuclear spins.

[0041] Train of Refocusing Pulses: A series or “train” of refocusing pulses may then be applied. These pulses are specifically timed and designed to correct for the natural dephasing or spreading out of the nuclear spin signals over time due to various interactions in the sample. The series of pulses ensure that the spins remain in phase and don't lose coherence.

[0042] The data obtained according to the method of the present invention are processed using standard approaches known to the skilled person. These involve, at least some of, preferably all of, but are not limited to, Group Delay Correction, apodization, zero filling, Fourier transform, zero order phase correction, baseline correction, negative value zeroing and / or window selection.

[0043] Group delay correction (also referred to as first order phase correction) refers to a Group Delay generated by Bruker's digital filter that is removed during this preprocessing step to enhance data quality. This is also called first order phase correction.

[0044] The Apodization or Weighting Function is applied to the FIDs (free induction decay, which is the format of data typically generated in an NMR experiment) before Fourier Transform to increase the Signal-to-Noise Ratio (SNR). In one preferred example, a factor of 0.3 is employed for exponentiation during data processing.

[0045] Zero filling is performed by adding zeros to the end of the FIDs, enhancing the visual representation of the spectra.

[0046] Fourier Transform is applied to the FIDs, converting the time-domain signals into spectra expressed in the frequency domain. The frequency scale is subsequently converted from hertz into parts per million (ppm). After Fourier Transform, spectra are phased to ensure the real part exhibits pure absorptive mode with positive intensities. This process may also be referred to as zero order phase correction

[0047] Baseline correction is achieved through an Asymmetric Least Squares smoothing algorithm. A smoothing parameter (lambda) and an asymmetry parameter (p) employed. These parameters control the smoothness of the estimated baseline.

[0048] In the process of negative values zeroing, any remaining negative values within the spectra are set to 0 since they lack meaningful interpretation.

[0049] Finally, a spectral window ranging from -0.5 to 9 ppm is selected for further analysis, focusing on the relevant metabolite range.

[0050] In some embodiments of the present invention, the spectra of poor quality, that exhibit a residual water signal are automatically so classified and removed from the analysis. Residual water signal can occur as an artifact in the NMR acquisition due to various factors, such as imperfect water suppression, magnetic field fluctuations, and / or sample impurities.

[0051] The method of the present invention may also include further processing steps that lead to quantifying the metabolites within the sample. As it is apparent to the skilled person, quantification of the metabolites in the spectrum may require interpretation of the spectrum, which predicates on deconvolution of the spectrum. Accordingly and preferably, in the method of the present invention, once the spectra are collected, the signals due to metabolites are deconvoluted. Preferably, upon deconvolution of the signals due to metabolites, said metabolites are quantified.

[0052] The deconvolution of the signals due to metabolites, within the present invention, is based on a library of metabolite spectra. This library has been carefully curated and adapted to the saliva NMR spectrum to get the highest accuracy and precision of the metabolite measurements.

[0053] Sample spectra may be normalized using a peak normalisation algorithm to ensure comparability. This normalises all spectra to the area of the internal standard. Sample spectra are then aligned to account for any variations. Alignment is performed using an alignment algorithm, such as the clustering algorithm developed by Vu et al. (Vu, T.N., Valkenborg, D., Smets, K. et al. An integrated workflow for robust alignment and simplified quantitative analysis of NMR spectrometry data. BMC Bioinformatics 12, 405 (2011)).

[0054] The library spectra may then be fitted to the sample spectra using the ASCIS algorithm developed by Lefort et al., minimising the residuals between the library spectra and the sample spectra, within a defined ppm range. However, alternative algorithms may be used that implement a similar or the same strategy. Metabolite spectra may overlap, producing peaks that ar"e a convolution of multiple metabolites. The spectra are thereby deconvoluted into their individual components before quantification.

[0055] The deconvolution function attempts to optimize the concentration of metabolites in a spectrum by iteratively refining the fit of the included metabolites using a combination of thresholding and constrained linear regression. The optimal fit takes into account all the individual metabolite peaks and fits the best concentration for each metabolite in the mixture. (Lefort G, Liaubet L, Canlet C, Tardivel P, Pere M, Quesnel H, Paris A, lannuccelli N, Vialaneix N, Servien R (2019). ‘ASICS: an R package for a whole analysis workflow of 1 D 1 H NMR spectra .” Bioinformatics, 35(21), 4356-4363.)

[0056] Accordingly, the method of the present invention, preferably further comprises a step of deconvoluting the signals due to metabolites based on a library of metabolite spectra. Said library can be present in any form. Typically said library is presenton a data carrier that is computer-readable, and has been previously compiled.

[0057] Preferably, in the method of the present invention, said library of metabolite spectra includes the spectra of amino acids, glucose metabolites, short-chain fatty acid, carbohydrates, organic acids, alcohols, amines and / or miscellaneous metabolites.

[0058] As it is to be understood herein, the amino acids are preferably selected from D-glutamic acid, glycine, L- alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L-proline, L-threonine, L-tyrosine, L- valine, taurine, sarcosine, L-glutamine and L-phenylalanine. More preferably, the acids are D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L-proline, L- threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine and L-phenylalanine.

[0059] As it is to be understood herein, the glucose metabolites are preferably selected from lactate and pyruvic acid. More preferably, the glucose metabolites are lactate and pyruvic acid.

[0060] As it is to be understood herein, the short-chain fatty acids are preferably selected from acetate, butyrate, formate and propionate. More preferably, the short-chain fatty acids are acetate, butyrate, formate and propionate. It is to be understood that said short chain fatty acid may be present in their acid form or their salt form, accordingly, the term acetate also refers to acetic acid, the term butyrate also refers to butyric acid, the term formate also refers to formic acid, and the term propionate also refers to propionic acid.

[0061] As it is to be understood herein, the carbohydrates are preferably selected from L-fucose, sucrose, xylose, galactose and glucose. More preferably, the carbohydrates are L-fucose, sucrose, xylose, galactose and glucose.

[0062] As it is to be understood herein, the organic acids are preferably selected from ascorbic acid, succinic acid and 5-aminopentanoic acid. More preferably, the organic acids are ascorbic acid, succinic acid and 5-aminopentanoic acid.

[0063] As it is to be understood herein, the alcohols are preferably selected from ethanol, methanol and propylene glycol. More preferably, the alcohols are ethanol, methanol and propylene glycol.

[0064] As it is to be understood herein, the amines are preferably selected from methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, and choline. More preferably, the amines are methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, and choline.

[0065] As it is to be understood herein, the miscellaneous metabolites are preferably selected from allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine. More preferably, the miscellaneous metabolites are allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine.

[0066] Examples of metabolites include, but are not limited to, 3-Methylhistidine, 4-Hydroxyphenyllactate, 5- Aminopentanoic acid, Acetic acid, Acetoacetic acid, Acetone, Acetylcholine, Acetylglycine, Betaine, Butyric acid, Caffeine, Choline, Citric acid, Creatine, Creatinine, Cresol sulfate, D-Galactose, D-Glucose, Dimethyl sulfone, Dimethylamine, Dimethylglycine, Ethanol, Ethanolamine, Formic acid, Galactitol, Gluconic acid, Glycerol, Glycerophosphocholine, Glycine, Glycolic acid, Hydrocinnamic acid, Hypoxanthine, I ndole-3-acetic acid, Isobutyric acid, Isocaproic acid, Isovaleric acid, L-Alanine, L-Aspartic acid, L-Fucose, L-Glutamic acid, L-Histidine, L-Lactic acid, L-Leucine, L-Lysine, L-Methionine, L- Ornithine, L-Phenylalanine, L-Proline, L-Serine, L-Threonine, L-Tyrosine, L-Valine, Methanol, Methylamine, Methylguanidine, Myo-inositol, Phenylacetic acid, Phenylacetylglycine, Phosphorylcholine, P-Hydroxybenzoic acid, P-Hydroxyphenylacetic acid, Propionic acid, Propylene glycol, Putrescine, Pyroglutamic acid, Pyruvic acid, Sarcosine, Sorbitol, Succinic acid, Taurine, Thymine, Trimethylamine, Uracil, Urea, Valeric acid.

[0067] While the metabolites described hereinabove include the most preferred metabolites used in the method of the present invention, it is to be understood that the methods of the present invention are not meant to be limited to these metabolites, and on practice any metabolites that can be determined in a saliva sample of a subject can be employed in the present invention.

[0068] Thus, the metabolites according to the present invention may be selected from 3-Methylhistidine, 4- Hydroxyphenyllactate, 5-Aminopentanoic acid, Acetic acid, Acetoacetic acid, Acetone, Acetylcholine, Acetylglycine, Betaine, Butyric acid, Caffeine, Choline, Citric acid, Creatine, Creatinine, Cresol sulfate, D- Galactose, D-Glucose, Dimethyl sulfone, Dimethylamine, Dimethylglycine, Ethanol, Ethanolamine, Formic acid, Galactitol, Gluconic acid, Glycerol, Glycerophosphocholine, Glycine, Glycolic acid, Hydrocinnamic acid, Hypoxanthine, I ndole-3-acetic acid, Isobutyric acid, Isocaproic acid, Isovaleric acid, L-Alanine, L-Aspartic acid, L-Fucose, L-Glutamic acid, L-Histidine, L-Lactic acid, L-Leucine, L-Lysine, L- Methionine, L-Ornithine, L-Phenylalanine, L-Proline, L-Serine, L-Threonine, L-Tyrosine, L-Valine, Methanol, Methylamine, Methylguanidine, Myo-inositol, Phenylacetic acid, Phenylacetylglycine, Phosphorylcholine, P-Hydroxybenzoic acid, P-Hydroxyphenylacetic acid, Propionic acid, Propylene glycol, Putrescine, Pyroglutamic acid, Pyruvic acid, Sarcosine, Sorbitol, Succinic acid, Taurine, Trimethylamine, Uracil, Urea, Valeric acid, Thymine, D-glutamic acid, L-isoleucine, L-glutamine, Lactate, Acetate, Butyrate, Formate, Propionate, Sucrose, Xylose, Ascorbic acid, Allantoin, and Acetoin. In one particular embodiment, the metabolites according to the present invention are selected from 3- Methylhistidine, 4-Hydroxyphenyllactate, 5-Aminopentanoic acid, Acetic acid, Acetoacetic acid, Acetone, Acetylcholine, Acetylglycine, Betaine, Butyric acid, Caffeine, Choline, Citric acid, Creatine, Creatinine, Cresol sulfate, D-Galactose, D-Glucose, Dimethyl sulfone, Dimethylamine, Dimethylglycine, Ethanol, Ethanolamine, Formic acid, Galactitol, Gluconic acid, Glycerol, Glycerophosphocholine, Glycine, Glycolic acid, Hydrocinnamic acid, Hypoxanthine, I ndole-3-acetic acid, Isobutyric acid, Isocaproic acid, Isovaleric acid, L-Alanine, L-Aspartic acid, L-Fucose, L-Glutamic acid, L-Histidine, L-Lactic acid, L-Leucine, L- Lysine, L-Methionine, L-Ornithine, L-Phenylalanine, L-Proline, L-Serine, L-Threonine, L-Tyrosine, L- Valine, Methanol, Methylamine, Methylguanidine, Myo-inositol, Phenylacetic acid, Phenylacetylglycine, Phosphorylcholine, P-Hydroxybenzoic acid, P-Hydroxyphenylacetic acid, Propionic acid, Propylene glycol, Putrescine, Pyroglutamic acid, Pyruvic acid, Sarcosine, Sorbitol, Succinic acid, Taurine, Trimethylamine, Uracil, Urea, Valeric acid, Thymine, D-glutamic acid, L-isoleucine, L-glutamine, Lactate, Acetate, Butyrate, Formate, Propionate, Sucrose, Xylose, Ascorbic acid, Allantoin, and Acetoin.

[0069] Accordingly, said library of metabolite spectra may include the spectra of metabolites selected from one or more of 3-Methylhistidine, 4-Hydroxyphenyllactate, 5-Aminopentanoic acid, Acetic acid, Acetoacetic acid, Acetone, Acetylcholine, Acetylglycine, Betaine, Butyric acid, Caffeine, Choline, Citric acid, Creatine, Creatinine, Cresol sulfate, D-Galactose, D-Glucose, Dimethyl sulfone, Dimethylamine, Dimethylglycine, Ethanol, Ethanolamine, Formic acid, Galactitol, Gluconic acid, Glycerol, Glycerophosphocholine, Glycine, Glycolic acid, Hydrocinnamic acid, Hypoxanthine, I ndole-3-acetic acid, Isobutyric acid, Isocaproic acid, Isovaleric acid, L-Alanine, L-Aspartic acid, L-Fucose, L-Glutamic acid, L-Histidine, L-Lactic acid, L- Leucine, L-Lysine, L-Methionine, L-Ornithine, L-Phenylalanine, L-Proline, L-Serine, L-Threonine, L- Tyrosine, L-Valine, Methanol, Methylamine, Methylguanidine, Myo-inositol, Phenylacetic acid, Phenylacetylglycine, Phosphorylcholine, P-Hydroxybenzoic acid, P-Hydroxyphenylacetic acid, Propionic acid, Propylene glycol, Putrescine, Pyroglutamic acid, Pyruvic acid, Sarcosine, Sorbitol, Succinic acid, Taurine, Trimethylamine, Uracil, Urea, Valeric acid, Thymine, D-glutamic acid, L-isoleucine, L-glutamine, Lactate, Acetate, Butyrate, Formate, Propionate, Sucrose, Xylose, Ascorbic acid, Allantoin, and Acetoin. Preferably, said library of metabolite spectra may include one or more of 3-Methylhistidine, 4- Hydroxyphenyllactate, 5-Aminopentanoic acid, Acetic acid, Acetoacetic acid, Acetone, Acetylcholine, Acetylglycine, Betaine, Butyric acid, Caffeine, Choline, Citric acid, Creatine, Creatinine, Cresol sulfate, D- Galactose, D-Glucose, Dimethyl sulfone, Dimethylamine, Dimethylglycine, Ethanol, Ethanolamine, Formic acid, Galactitol, Gluconic acid, Glycerol, Glycerophosphocholine, Glycine, Glycolic acid, Hydrocinnamic acid, Hypoxanthine, I ndole-3-acetic acid, Isobutyric acid, Isocaproic acid, Isovaleric acid, L-Alanine, L-Aspartic acid, L-Fucose, L-Glutamic acid, L-Histidine, L-Lactic acid, L-Leucine, L-Lysine, L- Methionine, L-Ornithine, L-Phenylalanine, L-Proline, L-Serine, L-Threonine, L-Tyrosine, L-Valine, Methanol, Methylamine, Methylguanidine, Myo-inositol, Phenylacetic acid, Phenylacetylglycine, Phosphorylcholine, P-Hydroxybenzoic acid, P-Hydroxyphenylacetic acid, Propionic acid, Propylene glycol, Putrescine, Pyroglutamic acid, Pyruvic acid, Sarcosine, Sorbitol, Succinic acid, Taurine, Trimethylamine, Uracil, Urea, Valeric acid, Thymine, D-glutamic acid, L-isoleucine, L-glutamine, Lactate, Acetate, Butyrate, Formate, Propionate, Sucrose, Xylose, Ascorbic acid, Allantoin, and Acetoin.

[0070] Further preferably, the method of the present invention involves a step of quantifying the concentration of selected metabolites based on an internal standard. The term “internal standard” is not particularly limited. Preferably, the internal standard is an alkylsilyl compound, for example tetramethylsilane.

[0071] Library spectra may be fitted to the sample spectra, and quantification may be performed using the ASICS package. The resulting relative concentrations may be normalized to the added TMSP (Tetramethylsilane) concentration, or another internal standard. These concentrations may then be corrected for the peakdependent response factors calculated independently for each metabolite, to calculate the absolute quantification.

[0072] Optionally, when quantifying the signals due to metabolites in the spectra obtained in the methods of the present invention, for each metabolite, a standard addition experiment may be conducted to determine peak-dependent response factors. Pooled saliva samples may be utilized, and the baseline concentration of the metabolite may be assessed. Pure metabolite standards may be spiked into the samples at 1 x, 2x, and 3x the baseline concentration. Quantification may be performed using the established quantification pipeline. Linear regression may be carried out on the resulting data, and the slope of the regression line may be used as the peak-dependent response factor for each metabolite.

[0073] Preferably, in the method of the present invention, the metabolites for which the concentration is quantified are selected from 3-Methylhistidine, 4-Hydroxyphenyllactate, 5-Aminopentanoic acid, Acetic acid, Acetoacetic acid, Acetone, Acetylcholine, Acetyl glycine, Betaine, Butyric acid, Caffeine, Choline, Citric acid, Creatine, Creatinine, Cresol sulfate, D-Galactose, D-Glucose, Dimethyl sulfone, Dimethylamine, Dimethylglycine, Ethanol, Ethanolamine, Formic acid, Galactitol, Gluconic acid, Glycerol, Glycerophosphocholine, Glycine, Glycolic acid, Hydrocinnamic acid, Hypoxanthine, I ndole-3-acetic acid, Isobutyric acid, Isocaproic acid, Isovaleric acid, L-Alanine, L-Aspartic acid, L-Fucose, L-Glutamic acid, L- Histidine, L-Lactic acid, L-Leucine, L-Lysine, L-Methionine, L-Ornithine, L-Phenylalanine, L-Proline, L- Serine, L-Threonine, L-Tyrosine, L-Valine, Methanol, Methylamine, Methylguanidine, Myo-inositol, Phenylacetic acid, Phenylacetylglycine, Phosphorylcholine, P-Hydroxybenzoic acid, P- Hydroxyphenylacetic acid, Propionic acid, Propylene glycol, Putrescine, Pyroglutamic acid, Pyruvic acid, Sarcosine, Sorbitol, Succinic acid, Taurine, Thymine, Trimethylamine, Uracil, Urea, Valeric acid; preferably from D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L- lysine, L-proline, L-threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine L-phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine. More preferably, in the method of the present invention, the metabolites for which the concentration is quantified are selected from D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L-proline, L-threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine L- phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine.

[0074] Alternatively, in the method of the present invention, the metabolites for which the concentration is quantified are selected from 3-Methylhistidine, 4-Hydroxyphenyllactate, 5-Aminopentanoic acid, Acetic acid, Acetoacetic acid, Acetone, Acetylcholine, Acetylglycine, Betaine, Butyric acid, Caffeine, Choline, Citric acid, Creatine, Creatinine, Cresol sulfate, D-Galactose, D-Glucose, Dimethyl sulfone, Dimethylamine, Dimethylglycine, Ethanol, Ethanolamine, Formic acid, Galactitol, Gluconic acid, Glycerol, Glycerophosphocholine, Glycine, Glycolic acid, Hydrocinnamic acid, Hypoxanthine, I ndole-3-acetic acid, Isobutyric acid, Isocaproic acid, Isovaleric acid, L-Alanine, L-Aspartic acid, L-Fucose, L-Glutamic acid, L- Histidine, L-Lactic acid, L-Leucine, L-Lysine, L-Methionine, L-Ornithine, L-Phenylalanine, L-Proline, L- Serine, L-Threonine, L-Tyrosine, L-Valine, Methanol, Methylamine, Methylguanidine, Myo-inositol, Phenylacetic acid, Phenylacetylglycine, Phosphorylcholine, P-Hydroxybenzoic acid, P- Hydroxyphenylacetic acid, Propionic acid, Propylene glycol, Putrescine, Pyroglutamic acid, Pyruvic acid, Sarcosine, Sorbitol, Succinic acid, Taurine, Trimethylamine, Uracil, Urea, Valeric acid, Thymine, D- glutamic acid, L-isoleucine, L-glutamine, Lactate, Acetate, Butyrate, Formate, Propionate, Sucrose, Xylose, Ascorbic acid, Allantoin, and Acetoin.

[0075] Further, in the method of the present invention, the metabolites for which the concentration is quantified are selected from 3-Methylhistidine, 4-Hydroxyphenyllactate, 5-Aminopentanoic acid, Acetic acid, Acetoacetic acid, Acetone, Acetylcholine, Acetyl glycine, Betaine, Butyric acid, Caffeine, Choline, Citric acid, Creatine, Creatinine, Cresol sulfate, D-Galactose, D-Glucose, Dimethyl sulfone, Dimethylamine, Dimethylglycine, Ethanol, Ethanolamine, Formic acid, Galactitol, Gluconic acid, Glycerol, Glycerophosphocholine, Glycine, Glycolic acid, Hydrocinnamic acid, Hypoxanthine, I ndole-3-acetic acid, Isobutyric acid, Isocaproic acid, Isovaleric acid, L-Alanine, L-Aspartic acid, L-Fucose, L-Glutamic acid, L- Histidine, L-Lactic acid, L-Leucine, L-Lysine, L-Methionine, L-Ornithine, L-Phenylalanine, L-Proline, L- Serine, L-Threonine, L-Tyrosine, L-Valine, Methanol, Methylamine, Methylguanidine, Myo-inositol, Phenylacetic acid, Phenylacetylglycine, Phosphorylcholine, P-Hydroxybenzoic acid, P- Hydroxyphenylacetic acid, Propionic acid, Propylene glycol, Putrescine, Pyroglutamic acid, Pyruvic acid, Sarcosine, Sorbitol, Succinic acid, Taurine, Trimethylamine, Uracil, Urea, Valeric acid, Thymine, D- glutamic acid, L-isoleucine, L-glutamine, Lactate, Acetate, Butyrate, Formate, Propionate, Sucrose, Xylose, Ascorbic acid, Allantoin, and Acetoin.

[0076] Saliva samples to be used in the methods of the present invention may be collected using any method known to the skilled person for this purpose. A dedicated saliva collection tube may be used. As specifically exemplified, particularly suitable is Thermo Scientific SpeciMAX™ Saliva Collection Kit. However, any suitable kit may be used, including custom tubes.

[0077] Accordingly, a protocol for optimal sample preparation may comprise the steps of vortexing, centrifugation and addition of a sample buffer. In particular, the step of In view of the foregoing, in one embodiment the step of sample preparation of the present invention comprises the following stepsVortexing of the saliva sample; it is noted that the time of performing said vortexing step is not particularly limited in any way, vortexing the sample can be done for any length of time;

[0078] Centrifugation of the sample, for example at 10’000 x g for 1 hour at 4°C (which is not meant to be limiting in any way); and addition of a sample buffer, the sample buffer comprising a buffer, an NMR standard and an antibacterial agent, e.g. 700 mM potassium phosphate buffer, 50 uM TMSP or other equivalent internal standard and 0.05% (w / v) sodium azide or other equivalent bacterial growth inhibitor.

[0079] In the methods of the present invention, determining the activity of the metabolic pathway by using a previously trained mathematical model based on input provided in step a of the methods of the invention, may be done by using norm-referencing.

[0080] Further, in the methods of the present invention, the mathematical model may be selected from supervised learning methods and unsupervised learning methods.

[0081] In a more particular embodiment, the supervised learning method is selected from linear regression, Random Forest, Support Vector Machines, Neural Networks with limited with single neuron output layer, Gradient Boosting, K-nearest Neighbors.

[0082] In a more particular embodiment, the unsupervised learning method is selected from Principal Component Analysis, Clustering, Independent Component Analysis, autoencoders, self-organizing maps, Association Rule mining), hybrid approaches (semi-supervised learning, transfer learning, ensemble methods.

[0083] The methods of the invention may further comprise the provision of at least one physiological parameter and / or a pathological condition of the subject. The physiological parameter may be any parameter useful to refine the assessment of the activity of a metabolic pathway. Examples within the present invention are height, weight, systolic blood pressure, diastolic blood pressure, heart rate, heart rate variability, activity level, oxygen saturation, body temperature, hydration, stress levels, sleep patterns, circadian rhythm, ECG and saliva speed.

[0084] The pathological condition may be any condition which may be relevant in the assessment of the activity of a metabolic pathway, including, for example, conditions linked to the increased or decreased, abnormal, activity of a metabolic pathway in a subject.

[0085] The present invention is not particularly limited to specific metabolic pathways. Examples of metabolic pathways that may be assessed in the methods of the present invention may be selected from glycolysis, Citric Acid Cycle (Krebs Cycle or TCA Cycle), Oxidative Phosphorylation (Electron Transport Chain), Pentose Phosphate Pathway, Gluconeogenesis, Fatty Acid Beta-Oxidation, Amino Acid Metabolism Pathways (e.g., urea cycle), Lipogenesis (Fatty Acid Synthesis), Cholesterol Biosynthesis Pathway, Purine and Pyrimidine Metabolism (nucleotide synthesis), One-Carbon metabolism.

[0086] In one embodiment, the present invention relates to a method of determining if a subject is at risk of developing a disease.

[0087] In one embodiment, the present invention relates to a method of determining if a subject is at risk of developing a disease associated with abnormal activity of a metabolic pathway, the method comprising the steps of a. assessing the activity of a metabolic pathway in a saliva sample of the subject using the method of the present invention; b. determining the risk of developing a disease associated with abnormal activity of the metabolic pathway by comparing the activity assessed in a with the activity of the metabolic pathway in a saliva sample obtained from a healthy subject.

[0088] Said disease is not particularly limited and according to the present inventors for several different disease it is possible to perform determination if a subject is at risk of developing said disease based on metabolic fingerprinting. As particularly preferred in the present invention, the disease may be a cardiovascular disease or diabetes type 2. However, alternatively, the disease may be liver disease such as NASH or NAFLD or a cognitive disorders such as Alzheimer’s disease or Dementia. The disease may also be selected from cancer, liver diseases, diabetes, cardiovascular disorders, and neurodegenerative diseases.

[0089] Accordingly, in one preferred embodiment, the present invention relates to a method of determining if a subject is at risk of developing a cardiovascular disease or diabetes type 2. The method comprises the steps of a. providing concentrations of metabolites determined in the saliva of the subject; and b. determining the risk of developing a cardiovascular disease or diabetes type 2 based on input provided in a. It is accordingly preferred that said determining based on input provided in a., i.e. based on concentrations of metabolites determined in the saliva of the subject, is done by using a previously trained mathematical model.

[0090] The method of the present invention includes a step of a. providing concentrations of metabolites determined in the saliva of the subject. It is highly preferred that the method of the present invention is an ex vivo method, and accordingly the method does not include any invasive steps of acquiring the samples from the human body. Preferably, step a is performed by the NMR-based measurements of saliva. It is particularly preferred that the measurements are performed according to the present invention and particular according to the method for obtaining a 1 D1H-NMR spectrum of metabolites in a bodily fluid sample as described hereinabove. Thereby, NMR spectra of sufficient quality are obtained that are suitable for determination of the risk of developing a disease according to the method of the present invention.

[0091] Preferably, in the method of the present invention the metabolites determined in saliva of the subject are selected from D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L- lysine, L-proline, L-threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine L-phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine.

[0092] More preferably, in the method of the present invention the metabolites determined in saliva of the subject are D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L- proline, L-threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine L-phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine.

[0093] Alternatively, in the method of the present invention, the metabolites determined in saliva of the subject are selected from 3-Methylhistidine, 4-Hydroxyphenyllactate, 5-Aminopentanoic acid, Acetic acid, Acetoacetic acid, Acetone, Acetylcholine, Acetyl glycine, Betaine, Butyric acid, Caffeine, Choline, Citric acid, Creatine, Creatinine, Cresol sulfate, D-Galactose, D-Glucose, Dimethyl sulfone, Dimethylamine, Dimethylglycine, Ethanol, Ethanolamine, Formic acid, Galactitol, Gluconic acid, Glycerol, Glycerophosphocholine, Glycine, Glycolic acid, Hydrocinnamic acid, Hypoxanthine, I ndole-3-acetic acid, Isobutyric acid, Isocaproic acid, Isovaleric acid, L-Alanine, L-Aspartic acid, L-Fucose, L-Glutamic acid, L- Histidine, L-Lactic acid, L-Leucine, L-Lysine, L-Methionine, L-Ornithine, L-Phenylalanine, L-Proline, L- Serine, L-Threonine, L-Tyrosine, L-Valine, Methanol, Methylamine, Methylguanidine, Myo-inositol, Phenylacetic acid, Phenylacetylglycine, Phosphorylcholine, P-Hydroxybenzoic acid, P- Hydroxyphenylacetic acid, Propionic acid, Propylene glycol, Putrescine, Pyroglutamic acid, Pyruvic acid, Sarcosine, Sorbitol, Succinic acid, Taurine, Trimethylamine, Uracil, Urea, Valeric acid, Thymine, D- glutamic acid, L-isoleucine, L-glutamine, Lactate, Acetate, Butyrate, Formate, Propionate, Sucrose, Xylose, Ascorbic acid, Allantoin, and Acetoin.

[0094] Preferably, in this alternative embodiment of the method of the present invention, the metabolites determined in saliva of the subject are 3-Methylhistidine, 4-Hydroxyphenyllactate, 5-Aminopentanoic acid, Acetic acid, Acetoacetic acid, Acetone, Acetylcholine, Acetylglycine, Betaine, Butyric acid, Caffeine, Choline, Citric acid, Creatine, Creatinine, Cresol sulfate, D-Galactose, D-Glucose, Dimethyl sulfone, Dimethylamine, Dimethylglycine, Ethanol, Ethanolamine, Formic acid, Galactitol, Gluconic acid, Glycerol, Glycerophosphocholine, Glycine, Glycolic acid, Hydrocinnamic acid, Hypoxanthine, I ndole-3-acetic acid, Isobutyric acid, Isocaproic acid, Isovaleric acid, L-Alanine, L-Aspartic acid, L-Fucose, L-Glutamic acid, L- Histidine, L-Lactic acid, L-Leucine, L-Lysine, L-Methionine, L-Ornithine, L-Phenylalanine, L-Proline, L- Serine, L-Threonine, L-Tyrosine, L-Valine, Methanol, Methylamine, Methylguanidine, Myo-inositol, Phenylacetic acid, Phenylacetylglycine, Phosphorylcholine, P-Hydroxybenzoic acid, P- Hydroxyphenylacetic acid, Propionic acid, Propylene glycol, Putrescine, Pyroglutamic acid, Pyruvic acid, Sarcosine, Sorbitol, Succinic acid, Taurine, Trimethylamine, Uracil, Urea, Valeric acid, Thymine, D- glutamic acid, L-isoleucine, L-glutamine, Lactate, Acetate, Butyrate, Formate, Propionate, Sucrose, Xylose, Ascorbic acid, Allantoin, and Acetoin.

[0095] The method of the present invention further comprises step b. determining the risk of developing a cardiovascular disease or diabetes type 2 based on input provided in a. It is accordingly preferred that said determining based on input provided in a., i.e. based on concentrations of metabolites determined in the saliva of the subject, is done by using a previously trained mathematical model. However, the skilled person will recognize that other approaches are also possible, in particular based on statistical correlations.

[0096] Said mathematical model is defined preferably as a model capable of predicting input based on the output, which has been previously trained, for example by using known correlations between the input and output variables. Exemplary mathematical models and their training are discussed herein.

[0097] The present inventors have postulated and demonstrated the correlation between the metabolite concentrations in the saliva and the risk of certain diseases, e.g. cardiovascular diseases and / or diabetes type 2. Accordingly, the present invention is based, at least in part, on a surprising discovery of the present inventors that the concentrations of metabolites in saliva on a subject are useful in predicting the risk of developing a disease (in particular a cardiovascular disease or diabetes type 2) by said subject.

[0098] According to the present invention, the determination is to be performed based on input in a. However, the present invention further encompasses embodiments wherein further input may be used to supplement the input in a.

[0099] Accordingly, the present invention further relates to an embodiment of the method of the present invention, further comprising step a1 of providing blood parameters of the subject, wherein the determination in step b is further based on input provided in a1. Preferably, the blood parameters of the subject are selected from glycosylated hemoglobin, fasting blood glucose, cholesterol, high-density lipoprotein, low-density lipoprotein and triglycerides. More preferably, the blood parameters of the subject are glycosylated hemoglobin, fasting blood glucose, cholesterol, high-density lipoprotein, low-density lipoprotein and triglycerides. These parameters are known to the skilled person and can be determined without undue burden.

[0100] The present invention further relates to an embodiment of the method of the present invention, further comprising step a2 of providing physical parameters of the subject, wherein the determination in step b is further based on input provided in a2. Preferably, the physical parameters of the subject are selected from height, weight, systolic blood pressure, diastolic blood pressure, heart rate and saliva speed. More preferably, the physical parameters of the subject are height, weight, systolic blood pressure, diastolic blood pressure, heart rate and saliva speed. These parameters are known to the skilled person and can be determined without undue burden.

[0101] Accordingly, the determination in step b) may be performed based on input from step a) alone, based on combined input of step a) and a1), based on combined input of step a) and a2), or based on combined input of step a), a1) and a2).

[0102] The present invention further relates to an embodiment, wherein the determination step, i.e. the step b) is computer-implemented. The skilled person nevertheless appreciates that the technical character of the method of the present invention goes beyond the computer implementation of said step b), and also refers to the use of technical means for obtaining concentrations of metabolites, in particular NMR measurements for obtaining the same.

[0103] Determination in b) using the previously trained mathematical model can be done e.g. per analogy to the methods described in the Examples section.

[0104] As provided in the present invention, the previously trained mathematical model preferably has been trained based on a training dataset including for each subject a correlation between the metabolites determined in saliva of the subject, blood parameters of the subject, physical parameters of the subject and a risk of developing a disease by said subject, in particular cardiovascular disease or diabetes type 2.

[0105] Preferably, as referred to herein, the metabolites determined in saliva of the subject in said training dataset are selected from D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L-proline, L-threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine L-phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine. More preferably, the metabolites determined in saliva of the subject in said training dataset are D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L-proline, L- threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine L-phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine.

[0106] Alternatively, the metabolites determined in saliva of the subject in said training dataset are selected from 3-Methylhistidine, 4-Hydroxyphenyllactate, 5-Aminopentanoic acid, Acetic acid, Acetoacetic acid, Acetone, Acetylcholine, Acetyl glycine, Betaine, Butyric acid, Caffeine, Choline, Citric acid, Creatine, Creatinine, Cresol sulfate, D-Galactose, D-Glucose, Dimethyl sulfone, Dimethylamine, Dimethylglycine, Ethanol, Ethanolamine, Formic acid, Galactitol, Gluconic acid, Glycerol, Glycerophosphocholine, Glycine, Glycolic acid, Hydrocinnamic acid, Hypoxanthine, I ndole-3-acetic acid, Isobutyric acid, Isocaproic acid, Isovaleric acid, L-Alanine, L-Aspartic acid, L-Fucose, L-Glutamic acid, L-Histidine, L-Lactic acid, L- Leucine, L-Lysine, L-Methionine, L-Ornithine, L-Phenylalanine, L-Proline, L-Serine, L-Threonine, L- Tyrosine, L-Valine, Methanol, Methylamine, Methylguanidine, Myo-inositol, Phenylacetic acid, Phenylacetylglycine, Phosphorylcholine, P-Hydroxybenzoic acid, P-Hydroxyphenylacetic acid, Propionic acid, Propylene glycol, Putrescine, Pyroglutamic acid, Pyruvic acid, Sarcosine, Sorbitol, Succinic acid, Taurine, Trimethylamine, Uracil, Urea, Valeric acid, Thymine, D-glutamic acid, L-isoleucine, L-glutamine, Lactate, Acetate, Butyrate, Formate, Propionate, Sucrose, Xylose, Ascorbic acid, Allantoin, and Acetoin.

[0107] Preferably, in this alternative embodiment, the metabolites determined in saliva of the subject in said training dataset are 3-Methylhistidine, 4-Hydroxyphenyllactate, 5-Aminopentanoic acid, Acetic acid, Acetoacetic acid, Acetone, Acetylcholine, Acetyl glycine, Betaine, Butyric acid, Caffeine, Choline, Citric acid, Creatine, Creatinine, Cresol sulfate, D-Galactose, D-Glucose, Dimethyl sulfone, Dimethylamine, Dimethylglycine, Ethanol, Ethanolamine, Formic acid, Galactitol, Gluconic acid, Glycerol, Glycerophosphocholine, Glycine, Glycolic acid, Hydrocinnamic acid, Hypoxanthine, I ndole-3-acetic acid, Isobutyric acid, Isocaproic acid, Isovaleric acid, L-Alanine, L-Aspartic acid, L-Fucose, L-Glutamic acid, L- Histidine, L-Lactic acid, L-Leucine, L-Lysine, L-Methionine, L-Ornithine, L-Phenylalanine, L-Proline, L- Serine, L-Threonine, L-Tyrosine, L-Valine, Methanol, Methylamine, Methylguanidine, Myo-inositol, Phenylacetic acid, Phenylacetylglycine, Phosphorylcholine, P-Hydroxybenzoic acid, P- Hydroxyphenylacetic acid, Propionic acid, Propylene glycol, Putrescine, Pyroglutamic acid, Pyruvic acid, Sarcosine, Sorbitol, Succinic acid, Taurine, Trimethylamine, Uracil, Urea, Valeric acid, Thymine, D- glutamic acid, L-isoleucine, L-glutamine, Lactate, Acetate, Butyrate, Formate, Propionate, Sucrose, Xylose, Ascorbic acid, Allantoin, and Acetoin.

[0108] Preferably, as understood herein, the blood parameters of the subject in said training dataset are selected from glycosylated hemoglobin, fasting blood glucose, cholesterol, high-density lipoprotein, low-density lipoprotein and triglycerides.

[0109] More preferably, as understood herein, the blood parameters of the subject in said training dataset are glycosylated hemoglobin, fasting blood glucose, cholesterol, high-density lipoprotein, low-density lipoprotein and triglycerides.

[0110] Preferably, as understood herein, the physical parameters of the subject in said training dataset are selected from height, weight, systolic blood pressure, diastolic blood pressure, heart rate and saliva speed.

[0111] More preferably, as understood herein, the physical parameters of the subject in said training dataset are height, weight, systolic blood pressure, diastolic blood pressure, heart rate and saliva speed.

[0112] Preferably, for the purposes of model training, the risk of developing a cardiovascular disease or diabetes type 2 by said subject is determined by using Diabetes 8-year risk probability score and / or Framingham's 10-year cardiovascular risk score.

[0113] Framingham’s 10 year Cardiovascular Risk score is based on Age, Cholesterol, HDL, systolic blood pressure, smoking, and if a patient is classified as diabetic or not.

[0114] Diabetes 8 year risk probability score is based on BMI, blood pressure, HDL, Triglycerides, fasting blood glucose, and age.

[0115] Further according to the present invention, the previously trained mathematical model may be a machine learning model, for example selected from Linear regression, Random forest, Support vector machines, Neural networks, Gradient boosting (i.e. XGBoost or LightGBM) and K-nearest neighbors.

[0116] As previously mentioned, it is preferred that in the method of the present invention the risk of cardiovascular disease or diabetes type 2 is determined. Accordingly, in one embodiment of the method of the present invention the risk of cardiovascular disease is determined. In a further embodiment of the method of the present invention the risk of diabetes type 2 is determined.

[0117] Said determining the risk may be expressed in different ways. It may for example relate to a chance of getting a disease within a particular time span. However, it may also refer to determining if said subject is suffering from a disease. For example, in one embodiment, the subject at risk of developing a cardiovascular disease or diabetes type 2 is a subject suffering from cardiovascular disease or diabetes type 2, respectively.

[0118] As previously mentioned, the present invention is not meant to be limited to cardiovascular diseases or diabetes type 2. In one embodiment, the disease is a liver disease, for example NASH or NAFLD. In such an embodiment, the NAFLD fibrosis score is used instead of Framingham’s cardiovascular risk score or diabeters 8 year risk score. Alternatively, in one embodiment, the disease is a cognitive disorder such as Alzheimer’s disease or Dementia. In this case, clinical dementia rating scale (CDR), known to the skilled person, is used for the model training purposes.

[0119] The present invention also provides methods for monitoring a disease in a subject, the method comprising the steps of a. assessing the activity of a metabolic pathway in a saliva sample of the subject using the method of the present invention; b. monitoring a disease associated with abnormal activity of the metabolic pathway by comparing the activity assessed in a with the activity of the metabolic pathway in a saliva sample obtained at a different time point.

[0120] The present invention also provides methods for assessing treatment success of a disease in a subject, the method comprising the steps of a. assessing the activity of a metabolic pathway in a saliva sample of the subject using the method of the present invention; b. determining the treatment success of a disease associated with abnormal activity of the metabolic pathway after administering an agent to treat the disease by comparing the activity assessed after administration of the agent with the activity of the metabolic pathway in a saliva sample obtained from the subject without administering the agent. Further particularly preferred embodiments of the present invention are provided in the following numbered items.

[0121] 1 . A method for obtaining a 1 D1H-NMR spectrum of metabolites in a bodily fluid sample, the method comprising the step of acquiring 1 D1H-NMR data which involves: a) suppressing signals due to water; and b) suppressing signals due to proteins by using T2 relaxation filtering.

[0122] 2. The method of item 1 , wherein suppressing of signals due to water is performed by presaturating the water signal.

[0123] 3. The method of item 2 wherein said presaturating of the water signal is done by using 1 D1H- NOESY-presat pulse sequence.

[0124] 4. The method of any one of items 1 to 3, wherein suppressing the signals due to proteins is done by using Carr-Purcell-Meiboom-Gill pulse sequence.

[0125] 5. The method of any one of items 1 to 4, wherein the pulse sequence substantially as shown in the appended Figure is used.

[0126] 6. The method of any one of items 1 to 5, further comprising the step of deconvoluting the signals due to metabolites based on a library of metabolite spectra.

[0127] 7. The method of items 6, wherein said library of metabolite spectra includes the spectra of amino acids, glucose metabolites, short-chain fatty acid, carbohydrates, organic acids, alcohols, amines and miscellaneous metabolites.

[0128] 8. The method of item 7, wherein the amino acids are D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L-proline, L-threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine and L-phenylalanine; glucose metabolites are lactate, and pyruvic acid; short-chain fatty acids are acetate, butyrate, formate and propionate; carbohydrates are L-fucose, sucrose, xylose, galactose and glucose; organic acids are ascorbic acid, succinic acid and 5-aminopentanoic acid; alcohols are ethanol, methanol and propylene glycol; amines are methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, and choline; and miscellaneous metabolites are allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine. The method of any one of items 1 to 8, further involving the step of quantifying the concentration of selected metabolites based on an internal standard, preferably wherein the internal standard is an alkylsilyl compound. The method of any one of items 1 to 9, wherein the metabolites for which the concentration is quantified are selected from D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L- isoleucine, L-leucine, L-lysine, L-proline, L-threonine, L-tyrosine, L-valine, taurine, sarcosine, L- glutamine L-phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine; preferably wherein the metabolites for which the concentration is quantified are D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L-proline, L- threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine L-phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine. The method of any one of items 1 to 10, wherein the bodily fluid sample is a saliva sample. A method of determining if a subject is at risk of developing a cardiovascular disease or diabetes type 2, the method comprising the steps of: a. providing concentrations of metabolites determined in the saliva of the subject; b. determining the risk of developing a cardiovascular disease or diabetes type 2 by using a previously trained mathematical model based on input provided in a. 13. The method of item 12, wherein the metabolites determined in saliva of the subject are selected from D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L- lysine, L-proline, L-threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine L-phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine, preferably wherein the metabolites determined in saliva of the subject are D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L-proline, L-threonine, L- tyrosine, L-valine, taurine, sarcosine, L-glutamine L-phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine.

[0129] 14. The method of item 12 or 13, wherein step a is performed by the NMR-based measurements of saliva.

[0130] 15. The method of any one of items 12 to 14, wherein step a is performed according to any one of claims 1 to 11 .

[0131] 16. The method of any one of items 12 to 15, further comprising step a1 of providing blood parameters of the subject, wherein the determination in step b is further based on input provided in a1 .

[0132] 17. The method of item 16, wherein the blood parameters of the subject are selected from glycosylated hemoglobin, fasting blood glucose, cholesterol, high-density lipoprotein, low-density lipoprotein and triglycerides, preferably wherein the blood parameters of the subject are glycosylated hemoglobin, fasting blood glucose, cholesterol, high-density lipoprotein, low-density lipoprotein and triglycerides.

[0133] 18. The method of any one of items 12 to 17, further comprising step a2 of providing physical parameters of the subject, wherein the determination in step b is further based on input provided in a2. 19. The method of item 18, wherein the physical parameters of the subject are selected from height, weight, systolic blood pressure, diastolic blood pressure, heart rate and saliva speed, preferably wherein the physical parameters of the subject are height, weight, systolic blood pressure, diastolic blood pressure, heart rate and saliva speed.

[0134] 20. The method of any one of items 12 to 19, wherein step b is computer-implemented.

[0135] 21 . The method of any one of items 12 to 20, wherein the previously trained mathematical model has been trained based on a training dataset including for each subject a correlation between the metabolites determined in saliva of the subject, blood parameters of the subject, physical parameters of the subject and a risk of developing a cardiovascular disease or diabetes type 2 by said subject.

[0136] 22. The method of item 21 , wherein the metabolites determined in saliva of the subject in said training dataset are selected from D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L- isoleucine, L-leucine, L-lysine, L-proline, L-threonine, L-tyrosine, L-valine, taurine, sarcosine, L- glutamine L-phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine, preferably wherein the metabolites determined in saliva of the subject in said training dataset are D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L- proline, L-threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine L-phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine.

[0137] 23. The method of item 21 or 22, wherein the blood parameters of the subject in said training dataset are selected from glycosylated hemoglobin, fasting blood glucose, cholesterol, high-density lipoprotein, low-density lipoprotein and triglycerides, preferably wherein the blood parameters of the subject in said training dataset are glycosylated hemoglobin, fasting blood glucose, cholesterol, high-density lipoprotein, low-density lipoprotein and triglycerides.

[0138] 24. The method of any one of items 21 to 23, wherein the physical parameters of the subject in said training dataset are selected from height, weight, systolic blood pressure, diastolic blood pressure, heart rate and saliva speed, preferably wherein the physical parameters of the subject in said training dataset are height, weight, systolic blood pressure, diastolic blood pressure, heart rate and saliva speed.

[0139] 25. The method of any one of item s21 to 24, wherein the risk of developing a cardiovascular disease or diabetes type 2 by said subject is determined by using Diabetes 8-year risk probability score and / or Framingham's 10-year cardiovascular risk score.

[0140] 26. The method of any one of items 12 to 25, wherein the risk of cardiovascular disease is determined.

[0141] 27. The method of any one of items 12 to 26, wherein the risk of diabetes type 2 is determined.

[0142] 28. The method of any one of items 12 to 27, wherein the subject at risk of developing a cardiovascular disease or diabetes type 2 is a subject suffering from cardiovascular disease or diabetes type 2, respectively.

[0143] The invention will be further illustrated in the following examples. These examples are not to be construed as limiting in any way to the scope of protection, which is defined by the hereto appended claims. Accordingly, the examples hereinbelow serves solely illustrative purposes.

[0144] Examples

[0145] Example 1 - Collection of saliva samples

[0146] Samples for this metabolomics analysis were collected using a dedicated saliva collection tube, for example the Thermo Scientific SpeciMAX™ Saliva Collection Kit. The kit is shown in Figure 4. The tubes shown in Figures 6 / 7 can also be used.

[0147] Individuals participating in the study were instructed to sit quietly during saliva collection. The collection tube was placed in the front of the mouth, positioned in front of the teeth. To ensure an even distribution of saliva from the three major saliva glands, participants were asked to lean forward slightly. At rest, the parotid gland contributes approximately 20% of total saliva production, the submandibular gland contributes approximately 65%, and the sublingual and minor glands contribute approximately 15%.

[0148] The stability of saliva samples is a critical consideration. Samples can be stored at room temperature for a certain period, but they are extremely stable at -80°C.

[0149] To prepare the saliva samples for metabolomics analysis, the following steps were carried out:

[0150] Vortexinq: Each saliva sample is vortexed for 1 minute at maximum power. This step is essential for breaking down glycoproteins, and homogenises viscosity between samples.

[0151] Centrifugation: After vortexing, the samples are centrifuged at 10,000 x g, at 4°C for 1 hour. This centrifugation step is crucial for separating particulate matter and large proteins from the sample, ensuring a cleaner and more homogeneous solution for analysis.

[0152] A buffer is added to each prepared sample. This buffer contains specific compounds in pure deuterium (D2O), which is then diluted 10 times in saliva to achieve a 10% D2O solution. The composition of the buffer includes:

[0153] Potassium Phosphate Buffer: This buffer is added to a final concentration of 70 mM, maintaining a pH of 7.4. This adjustment is essential to prevent large chemical shift differences between samples, ensuring consistency in the metabolomics analysis.

[0154] Sodium azide: Sodium azide is added to a final concentration of 0.05%. This addition is crucial for preventing bacterial growth, which could potentially influence metabolite levels. It is particularly important to mitigate the impact on metabolites like acetate and trimethylamine.

[0155] Internal Standard: An internal standard sodium 3-(Trimethylsilyl)propionic-2,2,3,3-d4 acid (TMSP) is included at a final concentration of 50 pM. The internal standard serves as a reference and is used for quantification and normalization during the metabolomics analysis. Example 2 - the pulse program

[0156] The pulse program used in the present invention is shown in Figure 2, and is also presented as coded for Bruker NMR spectrometers in the following.

[0157] ;Maven Health Pulse Programme

[0158] ;avance-verslon (12 / 01 / 11)

[0159] ;1D version of noesyphpr

[0160] ;with presaturation during relaxation delay and mixing time

[0161] ;replace pl by p31 and pl9 by pl39 to avoid getprosol prevents override of Increased PLW9

[0162] ;$CLASS=HighRes

[0163] ;$DIM=1D

[0164] ;$TYPE=

[0165] ;$SUBTYPE=

[0166] ;$COMMENT=

[0167] Sinclude <Avance.lncl>

[0168] Sinclude <Grad.incl>

[0169] Sinclude <Delay.incl>

[0170] "dll=30m"

[0171] "dl2=20u"

[0172] "acqt0=-p31*2 / 3.1416"

[0173] "p3=p31"

[0174] "p2=2*p31"

[0175] 1 ze

[0176] 2 30m dl2 pl39:fl dlO dl cw:fl ph29

[0177] 4u do:fl dl2 pll:fl p31 phi

[0178] 4u p31 ph2 dl2 pl39:fl d8 cw:fl

[0179] 4u do:fl dl2 pll:fl p31 ph3 go=2 ph31

[0180] 30m me #0 to 2 FO(zd) exit phl=02 ph2=00000000 2 2 2 2 2 2 2 2 ph3=002 2 1 1 3 3 ph4=l 1 1 1 2 2 2 2 ph5=l 1 3 3002 2 ph29=0 ph31=0 2 2 0 1 3 3 1 2002 3 1 1 3

[0181] ;$ld: noesyprld,v 1.12 2012 / 01 / 31 17:49:28 ber Exp $

[0182] Example 3 - Acquisition of NMR spectra of metabolites

[0183] In the following, the typical methods for collecting NMR spectra of metabolites according to the method of the present invention are included.

[0184] Once per week, an offset calibration is performed. This calibration involves measuring a routine sample at different offsets ranging from to ensure optimal performance. Specifically, the 01 p (carrier frequency) is set to achieve the optimal water suppression signal.

[0185] A total of 600 pL of the final sample solution was carefully pipetted into 5 mm NMR tubes. These sample tubes were then placed in a Bruker SampleXpress autosampler for automated acquisition. The NMR magnet was locked onto the deuterium signal, and probes are auto-tuned and matched. Shimming is performed automatically to optimize the magnetic field homogeneity.

[0186] Spectra were acquired using a custom pulse program, as shown in Example 2, as depicted in Figures 1 and 2, and as described herein. The set-up shown was constructed of a NOESY sequence with a presaturation pulse sequence and a Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence.

[0187] Each acquisition consists of 16 dummy scans followed by 256 scans, with a total of 16 loops of the T2 filter. This acquisition process takes approximately 80 minutes and 23 seconds. Additional acquisition parameters include:

[0188] Total Data Points (TD): 32,768

[0189] Spectral Width (SW) in ppm: 15,6218 Relaxation Delay (D1) in seconds: 15

[0190] The pulse sequence disclosed herein combines the advantages of a standard pulse sequence with a CPMG pulse sequence, offering a novel and highly effective approach for NMR metabolomics of saliva metabolites. The specific benefits of this innovative pulse program are as follows. The subsequent application of the CPMG sequence after the NOESY-presat sequence reduces the interference from protein signals, allowing for the detection and quantification of low-concentration metabolites that might otherwise be obscured by the dominating protein signals. This is important for gaining insights into less abundant metabolites that could play significant roles in biological processes. Furthermore, the combination effectively eliminates broad signals from proteins, enhancing baseline quality. This feature is particularly advantageous in complex biological samples, where protein interference often poses a challenge in metabolite identification. Figure 3 shows a comparison between the standard noesyprld sequence and our proprietary pulse sequence. The regions highlighted in yellow are the regions with the most significant improvement of signal quality. In figure 4 we show these regions in detail with specific peak clusters that show the highest signal quality improvement. The effect of the CPMG pulses is adaptable and can be finetuned specifically for the sample analysed and specific NMR instrumentation used.

[0191] Example 4 - Comparison of NMR spectra of metabolites acquired using the method of the invention and standard approaches

[0192] Figure 3 shows a comparison of a standard spectrum and a spectrum acquired according to the present invention.

[0193] Example 5 - Proof of concept through a human research project.

[0194] Over the period of June to August 2023 inventors performed a research project, which involved collecting parameters from healthy participants. This project had as purpose to benchmark the saliva metabolite profiles versus base parameters like blood pressure, heartrate etc. and blood markers. During a 45-minute visit, participants undertook all tests and where in principle healthy (defined by not being actively medicated for any diabetes or cardiovascular health related medical issues).

[0195] The following parameters were collected:

[0196] * D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L- proline, L-threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine L-phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine. The biological interpretation in formulated in two main parts: 1) creation of the individual interpretation of saliva metabolite values and 2) the direct relation and correlation of multiple saliva metabolites with individual or multiple blood and meta parameters.

[0197] The individual metabolite values determined according to the method of the present invention, or provided according to the method of the present invention, are used for creation of reference values.

[0198] The approach is to build reference ranges using various techniques, namely:

[0199] - Statistical descriptors (normal distribution, mean, median, standard deviation, etc.)

[0200] - Percentile descriptors (e.g. 5th, 95thpercentile)

[0201] - Box-and-whiskers (boxplot); Box (Interquartile Range), Whiskers (1 .5x the interquartile range), outliers

[0202] - Machine Learning approaches (Logistic regression, decision trees, random forests, gradient boosting with shallow trees, support vector machines, k-nearest neighbors, interpretable neural networks; with or without regularization techniques)

[0203] These approaches can be aided and supplemented by additional pre-processing steps dependent on the type of distribution a single or group of metabolites approaches, including, but not limited to:

[0204] - Shapiro-Wilk test

[0205] - Anderson-Darling test

[0206] - Kolmogorov-Smirnov (With additions like Li lliefors test for smaller sample sizes)

[0207] - Chi-Square test

[0208] - Additional parametric and non-parametric tests

[0209] Potential existing distributions of data can be found in the following table: Depending on the assessed state of the distribution preprocessing steps can be taken to create a normal distribution, apply one of the abovementioned reference range techniques and then convert back to the original distribution.

[0210] The result is a reference range with classifications that can be represented as such:

[0211] - Out-of-range (low-end; left hand of the distribution)

[0212] - Sub-optimal (low-end; left hand of the distribution)

[0213] - Optimal

[0214] - Sub-optimal (high-end; right hand of the distribution)

[0215] - Out-of-range (high-end; right hand of the distribution)

[0216] Correlations between saliva metabolites, blood parameters, base and meta parameters can be assessed and modelled in two main approaches, 1) statistical correlations or 2) (Machine learning) modelling approaching in various formats. The features are the 47 saliva metabolites and optionally a set of base and / or meta parameters. The Target are the blood parameters and optionally a set of base and / or meta parameters.

[0217] Derivatives of the blood, base and meta parameters are used for creation of proxy Target variables. The following is a non-exhaustive list of proxy target variables:

[0218] - Framingham’s 10 year Cardiovascular Risk score o Components: Age, Cholesterol, HDL, systolic blood pressure, smoking, diabetic patient

[0219] - Diabetes 8 year risk probability score o Components: BMI, blood pressure, HDL, Triglycerides, fasting blood glucose, age These proxy targets can be individually used as target variables in the modelling approach or combined in a linear additive or non-linear additive score to train and predict the features on.

[0220] Correlations and relationships can be described in various modes:

[0221] - One-to-One

[0222] - One-to-Many

[0223] - Many-to-One

[0224] - Many-to-Many

[0225] Various statistical tests can be applied to assess the correlations.

[0226] 1) Statistical correlations (One-to-One)

[0227] Pearson’s correlation coefficient - Spearman’s rank correlation coefficient

[0228] - Kendall’s Tau

[0229] 2) Modelling approaches (One-to-Many, Many-to-One, Many-to-Many) a. Supervised Learning methods i. Linear regression ii. Random forest ill. Support vector machines iv. Neural networks (limited with single neuron output layer; therefore explainable and part of supervised) v. Gradient boosting (i.e. XGBoost or LightGBM) vi. K-nearest neighbors b. Unsupervised Learning methods i. Principal component analysis ii. Clustering (e.g. K-means, Hierarchical clustering etc.) ill. Independent component analysis iv. Autoencoders v. Self-organizing maps vi. Association rule mining c. Hybrid approaches i. Semi-supervised learning ii. Transfer learning ill. Ensemble methods

[0230] Model performance assessment has general methods and some model specific approaches. General approaches rely on:

[0231] - Cross-validation

[0232] - Confusion matrix

[0233] - Learning curves

[0234] - Grid search and hyperparameter tuning

[0235] - Receiver Operating Characteristic (ROC) curves

[0236] - Precision-Recall curves

[0237] The specific method approaches rely on: a. Supervised learning methods a. Classification metrics: i. Accuracy ii. Precision iii. Recall (Sensitivity or True Positive Rate) iv. F1 Score v. Area Under the Receiver Operating Characteristic (ROC-AUC) b. Regression metrics: i. Mean Absolute Error (MAE) ii. Mean Squared Error (MSE) iii. Root Mean Squared Error (RMSE) iv. R-squared (Coefficient of Determination) b. Unsupervised learning (clustering) i. Silhouette score ii. Calinski-Harabasz index iii. Homogeneity, Completeness and V-Measure iv. Density-Based Spatial Clustering of Applications with Noise (DBSCAN)

[0238] Example 6 - Correlation between saliva parameters and other relevant parameters.

[0239] The overall relationships between parameters models and disease are shown in Figure 5.

[0240] The two models highlighted Framingham and diabetes probability models are known strong proxy’s for CVD and diabetes type 2 risk estimation.

[0241] - Framingham’s 10 year Cardiovascular Risk score o Components: Age, Cholesterol, HDL, systolic blood pressure, smoking, diabetic patient

[0242] - Diabetes 8 year risk probability score o Components: BMI, blood pressure, HDL, Triglycerides, fasting blood glucose, age o

[0243] The models were built up from base parameters and blood parameters and the inventors could show that there are relationships between saliva parameters and base / blood parameters. This indicates that Saliva parameters can be used for the assessment of CVD risk scores and diabetes type 2 risk scores. A subset of the 121 cohort was selected for a simplified analysis (dropped all NA’s; i.e. LDL in blood at times can be below detection limit, thus ‘NA’). The present inventors consider the resulting 103 participants for the following correlations. Currently various statistically significant observations between saliva parameters and blood parameters are observed, namely:

[0244] **Blood parameters are in mg / dL and Saliva parameters in mmol / L With this data it has been demonstrated that using the NMR method described herein one can effectively measure saliva metabolites and that these saliva metabolite measurements are statistically relevant to important blood markers in use for disease risk models thus showcasing the indirect link between saliva metabolites and Cardiovascular disease and Diabetes type 2, as examples of diseases associated with activity of metabolic pathways. The inventors have thus demonstrated that the herein described methods can be used to assess activity of metabolic pathways and to use this activity for assessing the risk of developing such a disease.

Claims

CLAIMS1. A method for assessing the activity of a metabolic pathway in a saliva sample of a subject, the method comprising the steps of a. providing concentrations of metabolites in the saliva of the subject by acquiring 1 D1H-NMR data which involves1 ) suppressing signals due to water; and2) suppressing signals due to proteins by using T2 relaxation filtering; b. determining the activity of the metabolic pathway by using a previously trained mathematical model based on input provided in step a, optionally wherein step b is computer implemented.

2. The method of claim 1 , wherein suppressing of signals due to water is performed by presaturating the water signal, by using 1 D1H-NOESY-presat pulse sequence.

3. The method of claim 1 or 2, wherein suppressing the signals due to proteins is done by using Carr- Purcell-Meiboom-Gill pulse sequence.

4. The method of any one of claims 1 to 3, wherein the pulse sequence substantially as shown in the appended Figure 2 is used.

5. The method of any one of claims 1 to 4, further comprising a step of deconvoluting the signals due to metabolites based on a library of metabolite spectra.

6. The method of claim 5, wherein said library of metabolite spectra includes the spectra of metabolites selected from 3-Methylhistidine, 4-Hydroxyphenyllactate, 5-Aminopentanoic acid, Acetic acid, Acetoacetic acid, Acetone, Acetylcholine, Acetylglycine, Betaine, Butyric acid, Caffeine, Choline, Citric acid, Creatine, Creatinine, Cresol sulfate, D-Galactose, D-Glucose, Dimethyl sulfone, Dimethylamine, Dimethylglycine, Ethanol, Ethanolamine, Formic acid, Galactitol, Gluconic acid, Glycerol, Glycerophosphocholine, Glycine, Glycolic acid, Hydrocinnamic acid, Hypoxanthine, lndole-3-acetic acid, Isobutyric acid, Isocaproic acid, Isovaleric acid, L-Alanine, L- Aspartic acid, L-Fucose, L-Glutamic acid, L-Histidine, L-Lactic acid, L-Leucine, L-Lysine, L- Methionine, L-Ornithine, L-Phenylalanine, L-Proline, L-Serine, L-Threonine, L-Tyrosine, L-Valine,Methanol, Methylamine, Methylguanidine, Myo-inositol, Phenylacetic acid, Phenylacetylglycine, Phosphorylcholine, P-Hydroxybenzoic acid, P-Hydroxyphenylacetic acid, Propionic acid, Propylene glycol, Putrescine, Pyroglutamic acid, Pyruvic acid, Sarcosine, Sorbitol, Succinic acid, Taurine, Thymine, Trimethylamine, Uracil, Urea, Valeric acid; preferably D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L-proline, L-threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine L- phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine; preferably wherein said library of metabolite spectra includes the spectra of D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L-proline, L- threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine L-phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine.

7. The method of any one of claims 1 to 6, further involving a step of quantifying the concentration of selected metabolites based on an internal standard, preferably wherein the internal standard is an alkylsilyl compound, wherein the metabolites for which the concentration is quantified are selected from 3- Methylhistidine, 4-Hydroxyphenyllactate, 5-Aminopentanoic acid, Acetic acid, Acetoacetic acid, Acetone, Acetylcholine, Acetyl glycine, Betaine, Butyric acid, Caffeine, Choline, Citric acid, Creatine, Creatinine, Cresol sulfate, D-Galactose, D-Glucose, Dimethyl sulfone, Dimethylamine, Dimethylglycine, Ethanol, Ethanolamine, Formic acid, Galactitol, Gluconic acid, Glycerol, Glycerophosphocholine, Glycine, Glycolic acid, Hydrocinnamic acid, Hypoxanthine, I ndole-3-acetic acid, Isobutyric acid, Isocaproic acid, Isovaleric acid, L-Alanine, L-Aspartic acid, L-Fucose, L- Glutamic acid, L-Histidine, L-Lactic acid, L-Leucine, L-Lysine, L-Methionine, L-Ornithine, L- Phenylalanine, L-Proline, L-Serine, L-Threonine, L-Tyrosine, L-Valine, Methanol, Methylamine, Methylguanidine, Myo-inositol, Phenylacetic acid, Phenylacetylglycine, Phosphorylcholine, P- Hydroxybenzoic acid, P-Hydroxyphenylacetic acid, Propionic acid, Propylene glycol, Putrescine, Pyroglutamic acid, Pyruvic acid, Sarcosine, Sorbitol, Succinic acid, Taurine, Thymine, Trimethylamine, Uracil, Urea, Valeric acid;preferably D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L-proline, L-threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine L- phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine; preferably wherein the metabolites for which the concentration is quantified are D-glutamic acid, glycine, L-alanine, L-aspartic acid, L-histidine, L-isoleucine, L-leucine, L-lysine, L-proline, L- threonine, L-tyrosine, L-valine, taurine, sarcosine, L-glutamine L-phenylalanine, lactate, pyruvic acid, acetate, butyrate, formate, propionate, L-fucose, sucrose, xylose, galactose, glucose, ascorbic acid, succinic acid, 5-aminopentanoic acid, ethanol, methanol, propylene glycol, methylamine, putrescine, hypoxanthine, dimethylamine, methylguanidine, trimethylamine, choline, allantoin, acetoin, dimethyl sulfone, uracil, urea and caffeine.

8. The method of any one of claims 1 to 7, wherein in step b biostatistical norm-referencing is used.

9. The method of any one of claims 1 to 8, wherein at least one physiological parameter and / or a pathological condition of the subject is provided.

10. The method of claim 9, wherein the physiological parameter of the subject is selected from height, weight, systolic blood pressure, diastolic blood pressure, heart rate, heart rate variability, activity level, oxygen saturation, body temperature, hydration, stress levels, sleep patterns, circadian rhythm, ECG and saliva speed.11 . The method of any one of claims 1 to 10, wherein the metabolic pathway is selected from glycolysis, Citric Acid Cycle (Krebs Cycle or TCA Cycle), Oxidative Phosphorylation (Electron Transport Chain), Pentose Phosphate Pathway, Gluconeogenesis, Fatty Acid Beta-Oxidation, Amino Acid Metabolism Pathways (e.g., urea cycle), Lipogenesis (Fatty Acid Synthesis), Cholesterol Biosynthesis Pathway, Purine and Pyrimidine Metabolism (nucleotide synthesis), One-Carbon metabolism.

12. A method of determining if a subject is at risk of developing a disease associated with abnormal activity of a metabolic pathway, the method comprising the steps of: a. assessing the activity of a metabolic pathway in a saliva sample of the subject using themethod of any one of claims 1 to 11 ; b. determining the risk of developing a disease associated with abnormal activity of the metabolic pathway by comparing the activity assessed in a with the activity of the metabolic pathway in a saliva sample obtained from a healthy subject.

13. The method of claim 12, wherein in b a previously trained mathematical model is used.

14. The method of claim 13, wherein the mathematical model is selected from supervised learning methods and unsupervised learning methods.

15. The method of claim 14, wherein the supervised learning method is selected from linear regression, Random Forest, Support Vector Machines, Neural Networks with limited with single neuron output layer, Gradient Boosting, K-nearest Neighbors.

16. The method of claim 14, wherein the unsupervised learning method is selected from Principal Component Analysis, Clustering, Independent Component Analysis, autoencoders, self-organizing maps, Association Rule mining), hybrid approaches (semi-supervised learning, transfer learning, ensemble methods.

17. The method of any one of claims 12 to 16, further comprising step a1 of providing blood parameters of the subject, wherein the determination in step b is further based on input provided in a1 , preferably wherein the blood parameters of the subject are selected from glycosylated hemoglobin, fasting blood glucose, cholesterol, high-density lipoprotein, low-density lipoprotein and triglycerides, more preferably wherein the blood parameters of the subject are glycosylated hemoglobin, fasting blood glucose, cholesterol, high-density lipoprotein, low-density lipoprotein and triglycerides; and / or further comprising step a2 of providing physical parameters of the subject, wherein the determination in step b is further based on input provided in a2, preferably wherein the physical parameters of the subject are selected from height, weight, systolic blood pressure, diastolic blood pressure, heart rate and saliva speed, more preferably wherein the physical parameters of the subject are height, weight, systolic blood pressure, diastolic blood pressure, heart rate and saliva speed.

8. The method of any one of claims 12 to 17, wherein the disease associated with abnormal activity of the metabolic pathway is selected from cancer, liver diseases, diabetes, cardiovascular disorders, and neurodegenerative diseases.

Images