McAuley et al., (2012). A whole-body mathematical model of cholesterol metabolism and its age-associated dysregulation.
October 2013, model of the month by Nick Juty
Original model: BIOMD0000000434
Cardiovascular disease is by far the most prevalent disease in ageing populations. Correlated with alterations in lipid metabolism profiles, it has estimated incidence rate of 30-40% in the UK population, over the age 85. Low Density Lipoprotein Cholesterol (LDL-C), a prominent component in lipid metabolism, stands out as a major contributory factor. Furthermore, it is apparent that neither nutritional status nor physical activity have any effect on the rising levels of LDL-C with age.
Besides its well publicized detrimental effects, cholesterol is also an important component of all cell membranes, being a hormone precursor and playing a crucial role in absorption of lipid soluble vitamins. It's absorption from the gut is documented as being inefficient, and also displays high variability between individuals (30-80%). The precise transport and enzymatic mechanisms involved, particularly pertaining to how cholesterol traverses enterocyte membranes, is not well established.
The hepatic system is central in cholesterol metabolism, with the liver able to synthesize VLDs (very low density lipoproteins), which are converted into IDLs (intermediate density lipoproteins (IDLs) through the action of lipoprotein lipase (LPL). LPLs can be taken up by the liver directly, or further hydrolysed into LDLs, the main cholesterol carrier in the blood. LDLs may also be taken up through the LDL-receptor (LDLR), which is highly expressed in the liver, and expressed in peripheral tissues. The hepatic receptor is transcriptionally regulated by intracellular cholesterol levels.
It has been demonstrated that:a) There is age-associated decline in the clearance rate of LDL-C from the blood, as well as a decrease in the number of hepatic LDLRs. b) Intestinal cholesterol absorption increases with age in some species.
In this paper, the authors take a mechanistic approach to construct a model, with these observations in mind, making extensive use of published experimental measurements over the last seventy years. The model incorporates dietary cholesterol absorption in the intestine, and hepatic LDL-C clearance from the plasma [1, BIOMD0000000434]. It consists of 6 compartments (Figure 1), and is composed of a series of coupled ODEs.
Figure 2 Simulations reflecting changes in dietary cholesterol intake, in comparison to previously published data. (A) Mean changes in dietary cholesterol intake and its effect on plasma cholesterol levels using values taken from 167 feeding studies . (B) Relationship between age and LDL-C in males . Figures are taken from .
Figure 3 (A) LDL-C plasma levels over a range of cholesterol absorption values plotted against age. (B) LDL-C plasma levels over a range of LDL receptor-mediated clearance rates plotted against age. Figures are taken from .
Quantitative data was used to define parameters in the model wherever possible, though in some instances it was either not available, was estimated, or was known to vary significantly between individuals. The latter case is exemplified by the documented variability of cholesterol absorption in the intestine, and hence is one of the parameters that was further investigated through sensitivity analysis (k6 in the model; Figure 4).
The model was able to reproduce major experimental findings, but it was acknowledged that it is not completely accurate in all cases; It has been previously demonstrated that for each 100mg/day increase in dietary cholesterol intake, there was expected to be a 1.9 mg/dL increase in LDL-C . The model however resulted in an approximate 10mg/dL increase.
The authors go on to suggest how dietary changes could be implemented to modify cholesterol absorption, and parameter modifications could be used to simulate such changes; consumption of three grams per day of fibre reduces cholesterol absorption by 15%, while two grams per day of plant sterols can reduce LDL-C concentrations in plasma by 10-15%. The authors also suggest a number of possible refinements that could be made to the model itself, and extensions to include other more detailed models to better reflect clinical situations. One such example would be to obtain specific kinetic information such as Km values from enzyme databases, for reactions, which are currently associated with nominal values.
The most interesting part of this paper is the detailed methods section, which presents clear rationale for creating the model, and in step-wise fashion details the design decisions that were made. It describes the role of extensive literature reviews in model parametrisation, and highlights the importance of re-parametrising the model at each subsequent model extension phase. In addition the authors encode the model using established standards (specifically SBGN and SBML) to facilitate its reuse and extension.
Figure 1 Network diagram: The model comprises 6 compartments: Dietary Intake (1), Intestinal Tissue (2), Excretion (3), Plasma (4), Hepatic Tissue (5) and Peripheral Tissue (6). Arrows indicate the flow of cholesterol through the system with labels indicating its form. Enzymatic activities are displayed as round arrow heads, with enzymes represented as blue spheres. Figure taken from .
To evaluate whether increased LDL-C was due to increased absorption or decreased clearance, the authors ran simulations for both conditions, and compared the results with published data (Figure 2A, 2B). Changing the cholesterol absorption efficiency from 50% (simulation for an average 20 year old male) to 80% (65 year old male), the authors demonstrated an increase of 34mg/dL in LDL-C levels (Figure 3A). An even more pronounced increase in LDL-C levels was observed in simulations where there was a 50% reduction in the number of hepatic LDL receptors (mimicking LDL receptor number for a 65 year old). This resulted in LDL-C levels increasing by over 100 mg/dL (Figure 3B). It was concluded that LDL-C clearance levels had the most significant impact on LDL-C, and hence may be the major contributor to this effect.
Figure 4 Steady-state LDL-C levels generated through modification of the k6 model parameter which specifies cholesterol absorption efficiency in the intestine. Figure taken from .
Figure 5 LDL-C levels generated through modification of the k1 model parameter which specifies cholesterol ingestion for an individual. Figure taken from .
- McAuley et al. A whole-body mathematical model of cholesterol metabolism and its age-associated dysregulation. BMC Sys. Biol. (2012), 6:130.
- McNamara, DJ. The impact of egg limitations on coronary heart disease risk: do the numbers add up?.J Am Coll. Nutr. (2000), 19:540S-548S.
- Abbott et al.Joint distribution of lipoprotein cholesterol classes. The Framingham study.Arteriosclerosis. (1983), 3:260:272.