WO2024256726A1

WO2024256726A1 - Computer implemented method determining a value of allostatic load of a human being

Info

Publication number: WO2024256726A1
Application number: PCT/EP2024/066814
Authority: WO
Inventors: Sébastien NUSSLÉ; Semira GONSETH NUSSLÉ; Jonviea Danielle CHAMBERLAIN-CETTOU
Original assignee: Genknowme SA
Current assignee: Genknowme SA
Priority date: 2023-06-15
Filing date: 2024-06-17
Publication date: 2024-12-19
Anticipated expiration: 2025-12-15
Also published as: EP4728518A1

Abstract

The method according to the invention determines a value of allostatic load for a human being on the basis of the determination of some CpG levels of a cell sample of this human being with knowledge about CpG levels of a reference population, especially without having discussed with this human subject allostatic load issues or tested him about cardio values or his blood on diabetes, cholesterol issues etc.

Description

TITLE

COMPUTER IMPLEMENTED METHOD DETERMINING A VALUE OF ALLOSTATIC LOAD OF A HUMAN BEING

TECHNICAL FIELD

The present invention relates to a computer implemented method determining a value of allostatic load for a human being, especially a computer-based method.

PRIOR ART

The allostatic load of a person is known to be related to stress and publications as WO 2017/147715 proposes methods for individualized stress management. However, the stress assessment as discloses in said publication, especially in connection with its Fig. 4A, is limited in its functionality.

The article " Sociodemographic, behavioral and genetic determinants of allostatic load in a Swiss population-based study" by Petrovic, D. et al., published in PSYCHONEUROENDOCRINOLOGY, Oxford, GB, vol. 67, 76-85, 13 February 2016, XP029470216, ISSN: 0306-4530, DOI: 10.1016/J.PSYNEUEN.2016.02.003, describes the operationalization of the clinical phenotype of allostatic load which will be mentioned in the present description relating to the use of the so-called SKIPOGH cohort.

Lohman, Trevor, et al. have published the article "Predictors of Biological Age: The implications for Wellness and Aging Research" in GERONTOLOGY AND GERIATRIC MEDICINE, vol. 7, 1 January 2021 , XP055874561 , ISSN: 2333-7214,

DOI : 10.1177/23337214211046419

Chamberlain, Jonviead, et al. have published an article "Blood DNA methylation signatures of lifestyle exposures: tobacco and alcohol consumption" in CLINICAL EPIGENETICS, BIOMED CENTRAL LTD, London, UK, vol. 14, no. 1 , 28 November 2022, pages 1-12, XP021311072, ISSN: 1868-7075, DOI: 10.1186/S13148-022-01376-7 relating to the use of methylation signatures in view of lifestyle related factors. EDES and CREWS have published general comments on "Allostatic load and biological anthropology" in AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY, JOHN WILEY & SONS, INC, US, vol. 162, 20 January 2017, pages 44-70, XP071029781 , ISSN: 0002- 9483, DOI: 10.1002/AJPA.23146.

SUMMARY OF THE INVENTION

Based on this prior art it is an object of the present invention to provide a computer implemented method determining a value of allostatic load of a human being, especially in connection with a specific reference population.

This object is achieved with the features of claim 1 . If a CpG list identified according to claim 1 is available, then the object can be directly achieved with a method having the features of claim 5.

Epigenetic modifications - usually measured via DNA methylation (DNAm) or the addition of a methyl group at cytosine-guanine nucleotides - can be induced in response to external (i.e., environmental exposures) or internal stimuli, resulting in modifications to gene expression. Epigenetic signatures summarize epigenetic modifications (termed CpGs) at different loci and regions of DNA. To this point, CRP measured via DNA methylation has been evidenced to have a superior temporal stability, reflecting rather cumulative exposure in comparison to serum CRP, and was more strongly associated with brain health outcomes. In response to the operationalization issues mentioned above, an epigenetic signature of allostatic load (in short AL) helps to circumvent measurement issues and offer a standardized approach. Therefore, the present invention provides: (1) identification of relevant CpGs associated with a phenotypic measure of allostatic load using epigenomewide association analysis (EWAS) for a reference population; and (2) estimation and validation of an epigenetic signature of allostatic load for a human being.

The human cells used to determine methylation levels for specific methylation sites can be solid tissue, blood, saliva sample that comprises genomic DNA. The determination of methylation levels for specific methylation sites of the reference population can be done on the same or different types of human cells.

The methylation value of a CpG site in a population of human cells can be the average degree of methylation of said CpG methylation site in a population of a sample of cells, usually comprising hundreds up to and over hundreds of thousands of cells.

The method according to the invention allows starting from knowledge about a reference population to determine an allostatic load information of a subject human being only on the basis of the determination of some CpG levels of a cell sample of this human being, especially without having discussed with this human subject allostatic load issues or tested him about cardio values or his blood on diabetes, cholesterol issues etc. .

The invention uses a database of biomarkers. A biomarker or biological marker is a measurable indicator of some biological state or condition. The biomarkers can be arranged in a number of AL concept systems.

The invention now uses one or more AL phenotype variable concepts, which are the options: b1.) wherein a risk threshold value is provided for each biomarker and an allostatic load score for each member is based on the sum of risk values attributed for each biomarker depending on the distance of each biomarker value of said member beyond the corresponding threshold value of said biomarker, b2.) wherein an allostatic load score for each member is based on the sum of risk values attributed for each biomarker depending on the absolute z-score of each biomarker value of said member compared to the corresponding average value of said biomarker for the reference population, b3.) wherein, independently for a predetermined number of AL concept systems, an allostatic load score for each member is based on the sum of latent variable values for such an AL concept system with biomarkers attributed to this AL concept system. b1 .), b2.) and b3.) can be used alone or in combination of these options. It is noted that the number of AL phenotype variable concepts is dependent on the choice of biomarkers within the options or their combination.

Each such AL phenotype variable concept is based on one or more of said different AL concept systems, especially at least three AL concept systems. However, in options b1.) and b2.), the different biomarkers of different AL concept systems are not differentiated, since in b1.) wherein a risk threshold value is provided for each biomarker and an allostatic load score for each member is based on the sum of risk values attributed for each biomarker depending on the distance of each biomarker value of said member beyond the corresponding threshold value of said biomarker, as well as in b2.) wherein an allostatic load score for each member is based on the sum of risk values attributed for each biomarker depending on the absolute z-score of each biomarker value of said member compared to the corresponding average value of said biomarker for the reference population, this calculation does not use the affiliation of a biomarker to such an AL concept system.

However, this affiliation becomes a differentiating feature in b3.) wherein, independently for a predetermined number of AL concept systems, an allostatic load score for each member is based on the sum of latent variable values for such an AL concept system with biomarkers attributed to this concept system. As mentioned above, the biomarkers are arranged in especially three or more AL concept systems. Such AL concept systems can be taken from the group encompassing the cardiovascular system, the neuroendocrine system, the metabolic system and/or the inflammatory system. It is also possible, with other cohorts, to use e.g. telomere length or functional age estimators as such AL concept system, depending which biomarkers are available. Each of these AL concept systems comprise a number of biomarkers, usually more than one.

Then, for each AL concept system, an AL score can be calculated based on the different biomarkers. The score initially depends on the number of biomarkers of such an AL concept system and on the AL phenotype variable concept used. It can be the sum of dichotomized values, or it can be z-score. It can also be a latent variable.

The number of biomarkers available for a specific AL concept system can be very different, from very few as two or three biomarkers in a first AL concept system to e.g. ten or more biomarkers in a second AL concept system. The simple use of addition of dichotomized values (0 or 1) and especially z-scores (from 0 to a value higher than 1 , 2 or 3 depending on the distance of the value to the average) tend to emphasize the importance of one AL concept system having (by chance) a higher number of biomarkers.

The introduction and use of latent variables can limit this influence. A latent variable is defined as an AL score calculated based on the biomarker scores of a specific AL concept system. Therefore, the available biomarkers are defined to be part of a specific AL concept system as e.g. the cardiovascular system or the neuroendocrine system and provide a single score for all biomarkers of said specific system. This enables a more differentiated view on the AL score and prevents an excessive influence from biomarkers from a specific system. In other words, the feature of calculating, independently for a predetermined number of AL concept systems, an allostatic load score for each member based on the sum of latent variable values for such an AL concept system with biomarkers attributed to this concept system, relates to the feature of calculating, for a predetermined number of AL concept systems, especially at least two or three AL concept systems, e.g. the cardiovascular system or the neuroendocrine system, the independently, an allostatic load score for each member based on the sum of latent variable values for such an AL concept system with biomarkers attributed to this concept system. In other words, on one side the calculation of the method uses a database comprising a number of biomarkers and related risk values (dichotomized or as z-score etc.) and on the other side it provides the allostatic load score. According to this embodiment of the invention, latent variables as intermediate allostatic load scores are determined for a number of predetermined AL concept systems as intermediate values which are then combined to the final allostatic load score, as a sum of the different concept values, dichotomized or modified with factors. As an example, instead of using three values for the biomarkers, mean systolic blood pressure, mean diastolic blood pressure, and mean heart rate, one latent variable is calculated for the AL concept system "cardiovascular system".

Beside the above mentioned AL concept systems, it is noted that further AL concept systems can be defined and used in connection with the method according to the invention. The list of CpG comprises two further AL concept system definitions: AL-GP and AL-LBC. The method according to the invention was applied for verification on CpGs from two other cohorts. AL-GP are the CpGs that were identified by EWAS to be specific to the allostatic load estimation with the GP method, i.e. the one described in b1). The AL-LBC CpGs are the one estimated with the z-threshold that was initially created using the LBC cohort, i.e. in b2). So methods developed in other cohorts were applied to the reference cohort mentioned in the present application, the SKIPOGH cohort.

An aim of the computer implemented method is to estimate allostatic load based on CpG only. But to build the method, biomarkers in the reference population have to be known, either directly, or to build latent variables. The CpG to be used in the method was taken out of a pool of CpG. For this pool, all CpG at disposal can be used, the one linked to the latent variables, but also the one linked to the other metrics of allostatic load (AL-GP and AI-LBC).

The invention then uses these values in the calculation of feature c.), i.e. running, for at least one of the above chosen and applied AL phenotype variable concepts, an epigenome- wide association study identifying nbf associated CpGs(i) with i = 1 to nbf , wherein all identified CpGs are meeting Bonferonni-level significance independently for each of the above AL phenotype variable. In this context, the feature AL phenotype variable concept is related to the chosen biomarkers, e.g. either the list of biomarkers chosen from the database (i.e. all available or one or more different limited number of biomarkers from the list) or the latent variables, if the list is reduced to the Al concept system(s). Here the method provides the connection between the measured CpGs and the nbf biomarkers, determined to be relevant for the AL score. Then the human user comes into play, or better, the result of the analysis of his probe of human cells, allowing to determine the methylation level of the same nbf CpGs as provided in the database for the reference population. The method caculates the allostatic load for the n_rp members of said reference population using elastic net regression and general population cut-off values of individual biomarkers, wherein the calculated coefficients for each CpG are used to calculate the allostatic load of the human being providing his sample of cells. This provides the connex between the phenotype biomarkers available for the reference population and the CpG values just based on the analysis of some human cells of said human without e.g. asking him a questionnaire relating to the stress level or taking his blood pressure as a value for one part of a concept system or as the determinaton of this cardiovascular system value as latent variable.

The invention also provides a method for validating the phenotype-based method determining a value of allostatic load of a human being based on a sample of cells of this human as mentioned above. In such a method steps a.) to d.) of claim 1 are executed to determine the CpGs followed by: separating the reference population into a first group being a randomised sample of n_ri members of the reference population and a distinct second group of the remaining (n_rp - n_ri) members of the reference population. The allostatic load of a member of the second group is based on his sample of cells through determining the methylation levels of human cells related to the allostatic load for the n_ri members of the randomized sample using elastic net regression and general population cut-off values of individual biomarkers, wherein the calculated coefficients for each CpG are used to calculate the CPG-based allostatic load of the member of the second group providing his sample of cells. This is the same procedure as before. But then, since for the reference population phenotype values are available, the phenotype-based value of allostatic load of this member of the second group is calculated and this phenotype-based value of allostatic load is compared with the CpG-based allostatic load value of this member of the second group to determine the difference between the phenotype-based value and phenotypebased value of allostatic load. Then these steps are repeated for a predetermined number of members of the second group (form one to the maximum number) and the method determines a negative validation result, if the determined differences between the phenotype-based value and phenotype-based value of allostatic load for the different members of the second group exceed a predetermined threshold level for a number of at least one of the predetermined number of members of the second group. This negative result could be given, if only one of the AL phenotype-based values of the second group member differs too much to his CpG determined AL value, showing that the choice of CpGs and/or the choice of AL phenotype variable concepts (especially in number) was not sufficient.

Instead of determining the CpGs, it is possible to use the method with a pre-determined list of CpGs.

The method is also capable of taking into account specificities when new cohorts of reference populations become available.

Further embodiments of the invention are laid down in the dependent claims.

BRIEF DESCRIPTION OF THE DRAWINGS

Preferred embodiments of the invention are described in the following with reference to the drawings, which are for the purpose of illustrating the present preferred embodiments of the invention and not for the purpose of limiting the same. In the drawings,

Fig. 1 shows a flow chart of the AL score generation for the reference population;

Fig. 2 shows some entries of a risk threshold table for biomarkers;

Fig. 3 shows some entries of allostatic load scores as risk categorization for the biomarkers shown in Fig. 2;

Fig. 4 shows some entries of allostatic load scores as risk categorization based on the Z-score for the reference population;

Fig. 5 shows a flow chart of the overall principles of the method according to the invention;

Fig. 6 shows a flow chart of the overall principles of a specific embodiment of the method according to the invention; and

Fig. 7 shows a flow chart of a validation method for the method according to an embodiment of the invention. DESCRIPTION OF PREFERRED EMBODIMENTS

Fig. 1 shows a flow chart of the AL score generation 100 for a reference population. The method starts with providing a biomarker database 10 of a reference population comprising biomarker values of a plurality of biomarkers.

It is noted that the methods described herein are computer implemented methods. The databases are stored in memory accessible from a processor in which a software is loaded comprising instructions to execute the method steps. It provides a new approach to determine an allostatic load value of a person based on technical data which was until now not usable to obtain this information.

A “reference population” or “cohort” as used herein refers to sample of a larger population in which participants have been randomly sampled from population registries. It is hypothesized that the reference population is a representative sample of a population and therefore seeks to accurately reflect the characteristics of the larger population. The larger population can be understood as an ethnicity, for example Caucasians, a sub-ethnicity such as Slavic people, a country with several ethnicities, for example Chinese people, a region, or even a continent, the African or South American population for example. Examples of a reference population or cohort comprise the Swiss Kidney Project on Genes in Hypertension study (SKIPOGH).

Such a reference population can be the SKIPOGH group of people. SKIPOGH stands for "Swiss Kidney Project on Genes in Hypertension" and comprises biomarkers for almost 1000 persons. Of course, other cohorts can be used, and it is an aim of the present method to be applicable to any reference population. Such other known databases are Framingham Heart Study, Generation Scotland, Lothian Birth Cohorts. If the person for whom the allostatic load is to be evaluated is from the same or similar reference group, especially similar ethnicity, then specifically adapted results are achieved.

The term "methylation site" as used herein refers to a CpG position that is potentially methylated. Methylation typically occurs in a CpG containing nucleic acid. The CpG containing nucleic acid may be present in, e.g., in a CpG island, a CpG doublet, a promoter, an intron, or an exon of gene. Hyper or hypo-methylation of the methylation sites (e.g., methylation status) can be assessed by detecting methylation status and comparing a value to a relevant reference level. For example, the methylation status of one or more markers can be indicated as a value. The value can be one or more numerical values resulting from the assaying of one or more biological sample(s), and can be derived, e.g., by measuring methylation status of the marker(s) in the sample(s) by an assay, or from a dataset obtained from a provider such as a laboratory, or from a dataset stored on a server. DNA methylation of the methylation markers (or markers close to them) can be measured using various approaches, which range from commercial array platforms (e.g. from Illumina™) to sequencing approaches of individual genes. This includes standard lab techniques or array platforms. A variety of methods for detecting methylation status or patterns have been described in, for example U.S. Pat. Nos. 6,214,556, 5,786,146, 6,017,704, 6,265,171 , 6,200,756, 6,251 ,594, 5,912,147, 6,331 ,393, 6,605,432, and 6,300,071 and US Patent Application publication Nos. 20030148327, 20030148326, 20030143606, 20030082609 and 20050009059, each of which are incorporated herein by reference. Other array-based methods of methylation analysis are disclosed in U.S. patent application 20050196792. For a review of some methylation detection methods, see, Oakeley, E. J., Pharmacology & Therapeutics 84:389- 400 (1999). DNA methylation was determined using the Illumina Infinium MethylationEPIC BeadChip.

In certain aspects of the invention measuring methylation status comprises, performing methylation specific PCR (MSP), real-time methylation specific PCR, methylation-sensitive single-strand conformation analysis (MS-SSCA), quantitative methylation specific PCR (QMSP), PCR using a methylated DNA-specific binding protein, high resolution melting analysis (HRM), methylation-sensitive single-nucleotide primer extension (MS-SnuPE), base-specific cleavage/MALDI-TOF, PCR, real-time PCR, Combined Bisulfite Restriction Analysis (COBRA), methylated DNA, immunoprecipitation (MeDIP), a microarray-based method, pyrosequencing, or bisulfite sequencing.

Usually, the methylation status will be expressed as a beta-value, i.e., the percentage of methylated DNA string at a given location. This is the percentage of methylation in a tissue. Each cell has only two DNA strings at one location, so the methylation is either 0%, 50% or 100%. A tissue cell can be differentially methylated (typically in blood).

The Swiss Kidney Project on Genes in Hypertension (SKIPOGH) is a Swiss family-based population cohort with two waves of cross-sectional data collection, for which participants were recruited between December 2009 and April 2013. Individuals were eligible for inclusion if they were (a) 18 years or older, (b) of Caucasian descent, (c) had at least one first-degree family member willing to participate, and (d) provided written consent for study inclusion. Clinical biomarkers were collected at one of three study centers, i.e. Lausanne, Bern or Geneva, while standardized questionnaires were filled-out by the participant at home. Whole blood samples were collected during the second follow-up survey (October 2012-December 2016).

In the SKIPOGH cohort, whole blood samples were collected and stored at -80°C. DNA methylation was measured by first denaturing DNA (~1.2ug) with sodium bisulfate, and then amplifying the DNA by PCR. For 250 SKIPOGH participants, DNA methylation was assayed using the Illumina Infinium HumanMethylation450 BeadChip, while 721 participants had DNA methylation assayed using the Illumina Infinium MethylationEPIC BeadChip. Blood cell counts were estimated using the Housemann method.

The operationalization of the clinical phenotype of allostatic load for the SKIPOGH cohort has been described in Petrovic, D. et al. Sociodemographic, behavioral and genetic determinants of allostatic load in a Swiss population-based study. Psychoneuroendocrinology 67, 76-85 (2016). Briefly, included individuals were determined to be “at risk” according to cut-off values identified in extant literature. A sum-score was estimated as the sum of the dichotomized biomarkers; thus, creating an AL score ranging from 0-14 (AL-GP). In a sensitivity analysis, an alternative operationalization of AL was used; a weighted sum-score (including additional biomarkers) based on study populationspecific risk (ALSA). Risk for the weighted sum-score was determined based on the distribution of each individual biomarker within the SKIPOGH population; individuals were classified as “at-risk” if they fell within the lower or upper quartile of the distribution (biomarker-dependent). Biomarkers were then added together within each system, and then divided by the total number of system-specific biomarkers. The system-specific scores were then summed together to create the final AL score. The ALSA score ranged from 0-4.

CpGs were selected for validation based on a Bonferroni-adjusted threshold (a = 10-7). Then, in a second step, the CpGs identified in the first EWAS were considered validated if they reached a significance level of p<0.05.

Relating to the epigenetic signature, like the methodology described above for the discovery and validation EWAS, SKIPOGH data were divided into a training and validation set, also called being a first and a second group of the reference population in the database. The training set included all individuals with DNA methylation data obtained using the EPIC platform (N=705), while the validation group included individuals with DNA methylation obtained using the HM450K platform (N=249). To estimate the epigenetic signature of allostatic load (MethAL), CpGs confirmed in the validation EWAS were included in a stepwise regression model using the MASS package in RStudio. Relevant CpGs were selected based on the Akaike’s Information Criterion (AIC).

In a discovery EWAS, CpGs were associated with the clinical phenotype of AL at Bonferroni- level significance (p<0.1^A-8) and 32 at BH-level significance in Manhattan and Volcano plots. In comparison, only one CpG was identified to be associated with the ALSA score. Of the CpGs identified in the discovery EWAS, 10 were negatively correlated with ALcp. The top CpG site was cg06690548, located on the SLC7A11 gene (effect size= -0.18, p- value=8.4^A-13). Of the original 13 CpGs, 9 were validated in the validation set.

Fig. 3 shows a plot of the correlation between phenotypic AL and the epigenetic signature. Fig. 4 shows a plot of odds ratios and 95% confidence intervals for probability of history of CVD corresponding to different measures of AL.

A final selection of seven CpG sites were included in the epigenetic signature for AL (MethAL). Among SKIPOGH participants, the MethAL was positively associated with the clinical AL phenotype (R²=0.5, p<0.001).

Although both the clinical phenotype of AL and the MethAL signature were associated with a higher probability of reported CVD history (adjusting for age and sex), a stronger association was observed for the MethAL signature. For example, a one-unit increase in ALCP was associated with a 30% increased probability of CVD history (OR=1.31 , 95% Cl=1 .17-1.47), while a one-unit increase in Meth-AL was associated with a nearly 90% increased probability of CVD history (OR=1.88, 95% CI=1.50-2.36). A similar association was observed for diabetes, with a stronger association observed for the Meth-AL signature compared to the ALCP (OR=2.19, 95% CI=1.63-2.94; OR=1.64, 95% Cl=1.39-1.93, respectively).

Starting from this biomarker database 10 an AL score is generated for each member of the reference population. Here, three different ways to calculate such a score are shown, allowing to calculate six different AL scores, since the AL score generation 130 based on latent variables can be calculated for four different concept systems as explained below.

According to a first concept, the biomarker database 10 can also comprise a risk threshold value table 111 or cut-off value table for each biomarker. Fig. 2 shows some entries for such a risk threshold table 111 for biomarkers. Systolic blood pressure, Insulin and C-reactive protein are shown as three examples. A plurality of other biomarkers can be used. Here the risk threshold values are usually cut-off values for the general population but can also be chosen specifically for a cohort.

Fig. 3 shows some entries of allostatic load scores as risk categorization for the biomarkers shown in Fig. 2. When the biomarker of a person of the reference population is higher ('>') than this risk threshold value, the allostatic load is determined as 1 or another unity value and the allostatic load related to this biomarker is determined as 0. For some biomarkers, the risk threshold is the lowest acceptable value of the biomarker and then the allostatic load is determined as 1 if the biomarker is lower ('<') than this threshold. Here the AL score with five biomarker values of which three are exemplified is AL = 2.

In a generalized embodiment for this step, the allostatic load score is calculated as a distance value and can thus have other values than natural numbers. Usually, this distance value is calculated based on thresholds for higher risk values and can start from 0 or 1 if the threshold value is reached.

This first concept approach provides an allostatic load score for each member being based on the sum of risk values attributed for each biomarker depending on the distance of each biomarker value of said member beyond the corresponding threshold value of said biomarker.

According to a second concept of calculation of AL score 120, the biomarker database 10 is used as a closed system and the biomarker values of the reference population are organized as Z-scores 121 for each different biomarker. In other words, the biomarker database values of the database 10 are arranged as Z-scores, i.e. the average value for any biomarker receives the value 0 and value of plus or minus one standard deviation receives the value +1 and the value of plus or minus two standard deviations is +2 etc. . Fig. 4 shows some entries of allostatic load scores as risk categorization based on the Z- score for the reference population and the allostatic load score is AL = 4. The first concept uses scientifically approved values of the cut-off or threshold values, whereas the allostatic load values following the second concept are determined as deviation from the average value of the reference population.

Four further concepts are shown with the entry 130, where specific biomarkers are chosen from here four different AL concept systems 131. The four different AL concept systems 131 are Inflammation, Neuroendocrine, Metabolic and Cardiovascular.

The biomarkers for the inflammation concept are: IL-6, CRP, IFN-y, TNF-a and IL-10, for the neuroendocrine concept are: Dehydroepiandtrosterone (24hr), Cortisol (24hr), Androsterone (24hr) and an assigned stress level from 1 to 10; for the Metabolic concept are: Ins2s, ALAT, GGT, Glucose, TG and Ura; and finally for the cardiovascular concept are: Diastolic BP, Systolic BP and Heart Rate.

Independently for each of the AL concept systems, an allostatic load score for each member is based on the sum of latent variable values for the correspondingly chosen AL concept system with biomarkers attributed to this concept system, so for four AL concept systems, four additional ALs are generated. In total it is possible to generate up to six AL scores for each member of the reference population.

Fig. 5 shows a flow chart of the overall principles of the method according to the invention.

These 1 to 6 values of AL are taken as the basis to run, for each of the above applied AL phenotype variable concepts, an epigenome-wide association study identifying associated (number equal to nbf) CpGs(i) with i = 1 to nbf all meeting Bonferonni-level significance independently for each of the above AL phenotype variable concepts which is shown as box 200. Therefore, this box 200 is executed for every AL score determined as mentioned in box 100 and therefore a loop is formed with a decision 205 if further phenotype concepts are to be used. Of course, some of the identified CpGs from different phenotype concepts can be identical to other identified CpGs. The CpGs from all EWAS are compiled into one single CpG list as shown in box 210.

Now the measured CpG values of cells provided by the human being as subject come into play to calculate 210 the allostatic load of the above mentioned human being based on his sample of cells through determining the methylation levels (CpG(i, 1 ), ... .,CpG(i,n_rp) for i = 1 to ribf Of the predetermined methylation sites (CpG(1)... .CpG(nbf) ) of human cells related to the allostatic load for the n_rp members of said reference population using elastic net regression and general population threshold or cut-off values of individual biomarkers, wherein the calculated coefficients for each CpG are used to calculate the allostatic load of the human being providing his sample of cells.

The following table 1 shows possible biomarkers to be used. The biomarkers used depend on the level of data, the database 10 of the reference population contains.

The entry "Stress" is defined as "On a scale from 1-10, what is your level of daily stress?"

Table 2 provides a list of identified CpGs with the SKIPOGH cohort. Table 2

Fig. 6 then shows a flow chart of the overall principles of a specific embodiment of the method according to the invention when such a list of identified CpGs is available as box 300.

The method then starts also with the AL score generation for a number of AL phenotypes variable concepts and then the AL score of the human being is calculated as above with the CpG taken as a sub-group from the above Table 2.

The thresholds for the general population (clinical cut-offs) are used to estimate the phenotype-based measure of allostatic load. These thresholds are used to define whether someone is “at risk” for a given biomarker. The epigenetic signature for AL is derived by identifying the CpGs corresponding to the phenotype-based AL score, using elastic net regression, that result in the best fit. And therefore, the individual AL scores are the best ones given the reference population and the CpGs identified. It is a “local best fit”. The method is built on the insight that knowledge of methylation levels of identified CpGs is sufficient to determine a physical value relation to the corporal and mental situation of a human being normally related to specific phenotype-based measurement values. However, the result depends on the choice and availability of the one or more AL phenotype variable concepts. Therefore, the invention is accompanied by a validation method validating the approach within predetermined boundaries. Fig. 7 shows a flow chart of a validation method for the method according to an embodiment of the invention.

The reference population is separated into a first group being a randomized sample of n_ri members of the reference population and a distinct second group of the remaining (n_rp - n_ri) members of the reference population as reflected in box 300. The second validation group can be chosen smaller than making up the total number of the cohort.

The method for validating the phenotype-based method determining a value of allostatic load of a human being based on a sample of cells of this human being is starting with the steps 100 for each member of the first group, the steps 200 for every phenotype concept as checked in box 205 followed by compiling the identified CpGs into a list as in box 210. These steps can also be executed based on the entire reference population, but it is preferred to only use the first group. Then, this specific choice depending on available measurements and defining choice of applied concepts as well as thresholds is validated with and for the second group of the reference population.

The allostatic load of a member of the second group is based on his sample of cells through determining the methylation levels (CpG(i, 1), ... .,CpG(i,n_ri) for i = 1 to nbf of the predetermined methylation sites (CpG(1)... .CpG(nbf) ) of human cells related to the allostatic load for the n_ri first group members of the randomized sample using elastic net regression and general population cut-off values of individual biomarkers, wherein the calculated coefficients for each CpG are used to calculate the CPG-based allostatic load of the said member of the second group providing his sample of cells as shown in box 310.

The phenotype-based value of allostatic load of the member of the second group is calculated as shown in box 320 and compared this phenotype-based value of allostatic load with the CPG-based allostatic load of the member of the second group to determine the difference between the phenotype-based value and phenotype-based value of allostatic load. These steps 310 and 320 are repeated 325 for a predetermined number relating to the number of or a portion of the members of the second group and determine in Box 330 a negative validation result, if the determined differences between the phenotype-based value and phenotype-based value of allostatic load for the different members of the second group exceed a predetermined threshold level for a number of at least one of the predetermined number of members of the second group.

If there is no excess as questioned in box 335 than the validation is successful, and the method ends in box 340. Otherwise, if one or more of the phenotype-based values, which are available for the reference population, calculated for a member of the second group exceeds the value of the deducted and derived CpG value of the allostatic load then the method steps to determine the CpG value of allostatic load is conducted with different starting positions and choices. This can comprise a new distribution of the first and second groups as shown with the arrow going back to box 300, but it is also possible to maintain the same groups and make different choices for the steps as shown in boxes 100 or 200.

LIST OF REFERENCE SIGNS biomarker database cells

AL concept score generation 300 separate the population into AL score on risk threshold two distinct groups ' AL risk categorization 310 calculate AL score of a risk threshold database member of the second group

AL score on Z-score with Cp levels of the first

Z-Score of reference group population 320 calculate phenotype-based

AL score on latent variable value of allostatic load of the concept system biomarker member of the second group execute an EWAS identifying 325 reiterate for all second group CpGs members decision for creating a loop 330 determine validation result for further phenotypes comparing phenotype with compiling all identified CpGs CpG-based values of the into a single list second group members calculate AL score of a 335 if negative result go back human being based on 340 validation successful measured CpG levels of his

Claims

1 . A computer implemented phenotype-based method determining a value of allostatic load of a human being based on a sample of cells of this human being, wherein the method comprises the steps of: a.) providing a biomarker database (10) of a reference population comprising biomarker values of a plurality of biomarkers, bO.) calculating one or more allostatic load scores for each member of the reference population based on one or more AL phenotype variable concepts (100) from the following list: b1.) wherein a risk threshold value is provided for each biomarker and an allostatic load score for each member is based on the sum of risk values attributed for each biomarker depending on the distance of each biomarker value of said member beyond the corresponding threshold value of said biomarker, b2.) wherein an allostatic load score for each member is based on the sum of risk values attributed for each biomarker depending on the absolute z-score of each biomarker value of said member compared to the corresponding average value of said biomarker for the reference population, b3.) wherein, independently for a predetermined number of AL concept systems, an allostatic load score for each member is based on the sum of latent variable values for such an AL concept system with biomarkers attributed to this AL concept system (100), c.) run (200), for at least one of the above chosen and applied AL phenotype variable concepts, an epigenome-wide association study identifying nbf associated CpGs(i) with i = 1 to nbf , wherein all identified CpGs are meeting Bonferonni-level significance independently for each of the above AL phenotype variable; d.) compile (210) all nbf CpGs identified in step c.) into a single list, e.) calculate (220) the allostatic load of the above mentioned human being based on his sample of cells through determining the methylation levels (CpG(i, 1), ... .,CpG(i,n_rp) for i = 1 to nbf of the predetermined methylation sites (CpG(1)... .CpG(nbf) ) of human cells related to the allostatic load for the n_rp members of said reference population using elastic net regression and general population cut-off values of individual biomarkers, wherein the calculated coefficients for each CpG are used to calculate the allostatic load of the human being providing his sample of cells.

2. The method of claim 1 , wherein the latent variables are chosen from at least one up to all following groups encompassing the cardiovascular, neuroendocrine, metabolic and/or inflammatory system, wherein biomarkers are associated with one or more of these latent variables as follows.

3. The method of claim 1 , wherein the biomarkers are chosen from the list as follows:

Body surface area (DuBois & DuBois)

4. A computer implemented method for validating the phenotype-based method determining a value of allostatic load of a human being based on a sample of cells of this human being according to any of claims 1 to 3, wherein the validation method comprises: a.) separate the reference population into a first group being a randomised sample of n_ri members of the reference population and a distinct second group of the remaining (n_rp - n_ri) members of the reference population; b.) providing a biomarker database (10) of the reference population comprising biomarker values of a plurality of biomarkers, cO.) calculating one or more allostatic load scores for each member of the reference population based on one or more AL phenotype variable concepts (100) from the following list: c1.) wherein a risk threshold value is provided for each biomarker and an allostatic load score for each member is based on the sum of risk values attributed for each biomarker depending on the distance of each biomarker value of said member beyond the corresponding threshold value of said biomarker, c2.) wherein an allostatic load score for each member is based on the sum of risk values attributed for each biomarker depending on the absolute z-score of each biomarker value of said member compared to the corresponding average value of said biomarker for the reference population, c3.) wherein, independently for a predetermined number of AL concept systems, an allostatic load score for each member is based on the sum of latent variable values for such an AL concept system with biomarkers attributed to this concept system (100), d.) run (200), for at least one of the above chosen and applied AL phenotype variable concepts, an epigenome-wide association study identifying nbf associated CpGs(i) with i = 1 to nbf , wherein all identified CpGs are meeting bonferonni-level significance independently for each of the above AL phenotype variable; e.) compile (210) all nbf CpGs identified in step c.) into a single list, f.) calculate (310) the allostatic load of a member of the second group being based on his sample of cells through determining the methylation levels (CpG(i, 1), ... .,CpG(i,n_ri) for i = 1 to nbf of the predetermined methylation sites (CpG(1)... .CpG(nbf) ) of human cells related to the allostatic load for the n_ri members of the randomized sample using elastic net regression and general population cut-off values of individual biomarkers, wherein the calculated coefficients for each CpG are used to calculate the CPG-based allostatic load of the member of the second group providing his sample of cells; g.) calculate (320) the phenotype-based value of allostatic load of the member of the second group and compare this phenotype-based value of allostatic load with the CpG-based allostatic load of the member of the second group to determine the difference between CpG-based allostatic load value and phenotype-based value of allostatic load; h.) repeat (325) step f.) and g.) for a predetermined number of members of the second group; and i.) determine (330) a negative validation result, if the determined differences between the CpG-based allostatic load value and the phenotype-based value of allostatic load for the different members of the second group exceed a predetermined threshold level for a number of at least one of the predetermined number of members of the second group.

5. A computer implemented phenotype-based method determining a value of allostatic load of a human being based on a sample of cells of this human being, wherein the method comprises the steps of: a.) providing a biomarker database (10) of a reference population comprising biomarker values of a plurality of biomarkers, bO.) calculating one or more allostatic load scores for each member of the reference population based on one or more AL phenotype variable concepts (100) from the following list: b1.) wherein a risk threshold value is provided for each biomarker and an allostatic load score for each member is based on the sum of risk values attributed for each biomarker depending on the distance of each biomarker value of said member beyond the corresponding threshold value of said biomarker, b2.) wherein an allostatic load score for each member is based on the sum of risk values attributed for each biomarker depending on the absolute z-score of each biomarker value of said member compared to the corresponding average value of said biomarker for the reference population, b3.) wherein, independently for a predetermined number of AL concept systems, an allostatic load score for each member is based on the sum of latent variable values for the predetermined number of AL concept system with biomarkers attributed to this or these AL concept system(s) (100) from the list of phenotype AL concept systems, respectively, c.) provide (300) a list of nbf CpGs according to the following list:

d.) calculate (220) the allostatic load of the above mentioned human being based on his sample of cells through determining the methylation levels (CpG(i, 1), ... .,CpG(i,n_rp) for i = 1 to nbf of the predetermined methylation sites (CpG(1)... .CpG(nbf) ) of human cells related to the allostatic load for the n_rp members of said reference population using elastic net regression and general population cut-off values of individual biomarkers, wherein the calculated coefficients for each CpG are used to calculate the allostatic load of the human being providing his sample of cells.