Limits...
Analysis of heterogeneity and epistasis in physiological mixed populations by combined structural equation modelling and latent class analysis.

Fenger M, Linneberg A, Werge T, Jørgensen T - BMC Genet. (2008)

Bottom Line: Although very few gene variants were directly associated with metabolic syndrome in the total study sample, almost one third of all possible epistatic interactions were highly significant.In addition, the genetic variance increased by an average of 35-fold when analysed in the subpopulations.The major conclusions from this study are that the likelihood of detecting true association between genetic variants and complex traits increases tremendously when studied in physiological homogenous subpopulations and on inclusion of epistasis in the analysis, whereas epistasis (i.e. genetic networks) is ubiquitous and should be the basis in modelling any biological process.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department of Clinical Biochemistry and Molecular Biology, University Hospital of Copenhagen, Denmark. mogens.fenger@hvh.regionh.dk

ABSTRACT

Background: Biological systems are interacting, molecular networks in which genetic variation contributes to phenotypic heterogeneity. This heterogeneity is traditionally modelled as a dichotomous trait (e.g. affected vs. non-affected). This is far too simplistic considering the complexity and genetic variations of such networks.

Methods: In this study on type 2 diabetes mellitus, heterogeneity was resolved in a latent class framework combined with structural equation modelling using phenotypic indicators of distinct physiological processes. We modelled the clinical condition "the metabolic syndrome", which is known to be a heterogeneous and polygenic condition with a clinical endpoint (type 2 diabetes mellitus). In the model presented here, genetic factors were not included and no genetic model is assumed except that genes operate in networks. The impact of stratification of the study population on genetic interaction was demonstrated by analysis of several genes previously associated with the metabolic syndrome and type 2 diabetes mellitus.

Results: The analysis revealed the existence of 19 distinct subpopulations with a different propensity to develop diabetes mellitus within a large healthy study population. The allocation of subjects into subpopulations was highly accurate with an entropy measure of nearly 0.9. Although very few gene variants were directly associated with metabolic syndrome in the total study sample, almost one third of all possible epistatic interactions were highly significant. In particular, the number of interactions increased after stratifying the study population, suggesting that interactions are masked in heterogenous populations. In addition, the genetic variance increased by an average of 35-fold when analysed in the subpopulations.

Conclusion: The major conclusions from this study are that the likelihood of detecting true association between genetic variants and complex traits increases tremendously when studied in physiological homogenous subpopulations and on inclusion of epistasis in the analysis, whereas epistasis (i.e. genetic networks) is ubiquitous and should be the basis in modelling any biological process.

Show MeSH

Related in: MedlinePlus

The figure illustrates the simplified model used in the present study. As explained in Methods liver symbolize a surrogate of processes in the tissues with focus on the insulin metabolism. C-peptide denotes the secretion of C-peptide and hence insulin from the pancreatic β-cell. The insulin on the right side of the liver indicates the amount of insulin actually reaching the general circulation influencing the metabolism in peripheral tissues. Most of the insulin execute its action in and is internalized by the liver. Glucose and lipids are both metabolised in the liver, but also in many other tissues, in addition to influencing insulin production and secretion in the pancreas. All these processes can of course be modelled, but at the moment it will impose severe computer challenges. Nevertheless, this highly simplistic model turns out to be very efficient in the latent class analysis (see text).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2483291&req=5

Figure 1: The figure illustrates the simplified model used in the present study. As explained in Methods liver symbolize a surrogate of processes in the tissues with focus on the insulin metabolism. C-peptide denotes the secretion of C-peptide and hence insulin from the pancreatic β-cell. The insulin on the right side of the liver indicates the amount of insulin actually reaching the general circulation influencing the metabolism in peripheral tissues. Most of the insulin execute its action in and is internalized by the liver. Glucose and lipids are both metabolised in the liver, but also in many other tissues, in addition to influencing insulin production and secretion in the pancreas. All these processes can of course be modelled, but at the moment it will impose severe computer challenges. Nevertheless, this highly simplistic model turns out to be very efficient in the latent class analysis (see text).

Mentions: The structural model (SEM) of the metabolism of glucose and hence the metabolic syndrome is schematically shown in Figure 1. The metabolic syndrome is conceived as diminished glucose utilisation (uptake and processing of glucose) in peripheral tissues caused by increasingly inefficient action of insulin, i.e., insulin resistance evolves in the tissues. "Increasingly" should be conceived both as differences in insulin response between homogeneous subpopulations determined by the subpopulation genotype and as a modulation of insulin resistance by non-genetic factors. The pancreatic β-cell is the sole physiological source of insulin for which synthesis and secretion are influenced by numerous substances, including glucose. However, most of the insulin does not reach the general circulation, as it is metabolised in its first passage through the liver. Only 15–30% of the insulin secreted from the pancreas actually reaches the general circulation. The actual secretion by pancreatic β-cells can be estimated by measuring C-peptide, which is cleaved from proinsulin when insulin is secreted from the pancreas. C-peptide is not cleared by the liver and therefore reflects the β-cell activity. However, we are interested in the general metabolic state of the organism and therefore the biological activity of insulin (rather than in its production), and thus we define insulin as the principal indicator of this metabolic state. Nonetheless, the actual production of insulin is still an integrated circuit in the metabolic pathways, and C-peptide is therefore included into the model as a covariant (Figure 1). Several metabolites other than glucose influence the metabolic process that in turn also directly influences the secretion of insulin from pancreatic β-cells, but complete modelling of these pathways is complicated and computer- intensive and will be the subject of future studies. Nevertheless, the simple model used here is very efficient (see below), and therefore all latent variables are modelled in one complex variable, which we name "liver" as a surrogate for all tissues involved in glucose metabolism and insulin resistance, recognising the over-simplification this implicates. Note that no covariates are allowed to directly affect insulin, because, although correlated to insulin, they are not directly explanatory, but act through cellular processes.


Analysis of heterogeneity and epistasis in physiological mixed populations by combined structural equation modelling and latent class analysis.

Fenger M, Linneberg A, Werge T, Jørgensen T - BMC Genet. (2008)

The figure illustrates the simplified model used in the present study. As explained in Methods liver symbolize a surrogate of processes in the tissues with focus on the insulin metabolism. C-peptide denotes the secretion of C-peptide and hence insulin from the pancreatic β-cell. The insulin on the right side of the liver indicates the amount of insulin actually reaching the general circulation influencing the metabolism in peripheral tissues. Most of the insulin execute its action in and is internalized by the liver. Glucose and lipids are both metabolised in the liver, but also in many other tissues, in addition to influencing insulin production and secretion in the pancreas. All these processes can of course be modelled, but at the moment it will impose severe computer challenges. Nevertheless, this highly simplistic model turns out to be very efficient in the latent class analysis (see text).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2483291&req=5

Figure 1: The figure illustrates the simplified model used in the present study. As explained in Methods liver symbolize a surrogate of processes in the tissues with focus on the insulin metabolism. C-peptide denotes the secretion of C-peptide and hence insulin from the pancreatic β-cell. The insulin on the right side of the liver indicates the amount of insulin actually reaching the general circulation influencing the metabolism in peripheral tissues. Most of the insulin execute its action in and is internalized by the liver. Glucose and lipids are both metabolised in the liver, but also in many other tissues, in addition to influencing insulin production and secretion in the pancreas. All these processes can of course be modelled, but at the moment it will impose severe computer challenges. Nevertheless, this highly simplistic model turns out to be very efficient in the latent class analysis (see text).
Mentions: The structural model (SEM) of the metabolism of glucose and hence the metabolic syndrome is schematically shown in Figure 1. The metabolic syndrome is conceived as diminished glucose utilisation (uptake and processing of glucose) in peripheral tissues caused by increasingly inefficient action of insulin, i.e., insulin resistance evolves in the tissues. "Increasingly" should be conceived both as differences in insulin response between homogeneous subpopulations determined by the subpopulation genotype and as a modulation of insulin resistance by non-genetic factors. The pancreatic β-cell is the sole physiological source of insulin for which synthesis and secretion are influenced by numerous substances, including glucose. However, most of the insulin does not reach the general circulation, as it is metabolised in its first passage through the liver. Only 15–30% of the insulin secreted from the pancreas actually reaches the general circulation. The actual secretion by pancreatic β-cells can be estimated by measuring C-peptide, which is cleaved from proinsulin when insulin is secreted from the pancreas. C-peptide is not cleared by the liver and therefore reflects the β-cell activity. However, we are interested in the general metabolic state of the organism and therefore the biological activity of insulin (rather than in its production), and thus we define insulin as the principal indicator of this metabolic state. Nonetheless, the actual production of insulin is still an integrated circuit in the metabolic pathways, and C-peptide is therefore included into the model as a covariant (Figure 1). Several metabolites other than glucose influence the metabolic process that in turn also directly influences the secretion of insulin from pancreatic β-cells, but complete modelling of these pathways is complicated and computer- intensive and will be the subject of future studies. Nevertheless, the simple model used here is very efficient (see below), and therefore all latent variables are modelled in one complex variable, which we name "liver" as a surrogate for all tissues involved in glucose metabolism and insulin resistance, recognising the over-simplification this implicates. Note that no covariates are allowed to directly affect insulin, because, although correlated to insulin, they are not directly explanatory, but act through cellular processes.

Bottom Line: Although very few gene variants were directly associated with metabolic syndrome in the total study sample, almost one third of all possible epistatic interactions were highly significant.In addition, the genetic variance increased by an average of 35-fold when analysed in the subpopulations.The major conclusions from this study are that the likelihood of detecting true association between genetic variants and complex traits increases tremendously when studied in physiological homogenous subpopulations and on inclusion of epistasis in the analysis, whereas epistasis (i.e. genetic networks) is ubiquitous and should be the basis in modelling any biological process.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department of Clinical Biochemistry and Molecular Biology, University Hospital of Copenhagen, Denmark. mogens.fenger@hvh.regionh.dk

ABSTRACT

Background: Biological systems are interacting, molecular networks in which genetic variation contributes to phenotypic heterogeneity. This heterogeneity is traditionally modelled as a dichotomous trait (e.g. affected vs. non-affected). This is far too simplistic considering the complexity and genetic variations of such networks.

Methods: In this study on type 2 diabetes mellitus, heterogeneity was resolved in a latent class framework combined with structural equation modelling using phenotypic indicators of distinct physiological processes. We modelled the clinical condition "the metabolic syndrome", which is known to be a heterogeneous and polygenic condition with a clinical endpoint (type 2 diabetes mellitus). In the model presented here, genetic factors were not included and no genetic model is assumed except that genes operate in networks. The impact of stratification of the study population on genetic interaction was demonstrated by analysis of several genes previously associated with the metabolic syndrome and type 2 diabetes mellitus.

Results: The analysis revealed the existence of 19 distinct subpopulations with a different propensity to develop diabetes mellitus within a large healthy study population. The allocation of subjects into subpopulations was highly accurate with an entropy measure of nearly 0.9. Although very few gene variants were directly associated with metabolic syndrome in the total study sample, almost one third of all possible epistatic interactions were highly significant. In particular, the number of interactions increased after stratifying the study population, suggesting that interactions are masked in heterogenous populations. In addition, the genetic variance increased by an average of 35-fold when analysed in the subpopulations.

Conclusion: The major conclusions from this study are that the likelihood of detecting true association between genetic variants and complex traits increases tremendously when studied in physiological homogenous subpopulations and on inclusion of epistasis in the analysis, whereas epistasis (i.e. genetic networks) is ubiquitous and should be the basis in modelling any biological process.

Show MeSH
Related in: MedlinePlus