The index of this PhD thesis is available here.
This PhD thesis contains a report of
research aimed at improving the diagnosis of cardiovascular disease in an early
stage by investigating lipoprotein biology. Lipoproteins are the particles that
carry cholesterol and triglycerides through the bloodstream. Especially low-density
lipoprotein (LDL) cholesterol and LDL particle concentrations are known to be
positively associated with cardiovascular disease risk. Technological advances
allow the full size spectrum of
lipoproteins to be measured in increasing detail. Although such a
detailed measurement contains much information, it does not lead to diagnostic
values that both use all relevant information and that doctors can easily use.
We have
developed a mathematical model to aid the interpretation of detailed lipoprotein
measurements. This model is able to derive information on the status of
lipoprotein metabolism from a single lipoprotein measurement. This PhD thesis
reports on the development of the model and model-based diagnostics, their
application in a nutritional and a pharmacological human intervention study,
and their application as predictors for cardiovascular disease.
Part I: Model
development
In
chapter 2, we introduced the novel model framework (Particle Profiler) and
evaluated its feasibility. The framework was tested using existing stable
isotope flux data. The model framework implementation we presented reproduced
the flux data and derived lipoprotein size pattern changes that corresponded to
measured changes. It also sensitively indicated changes in lipoprotein
metabolism between patient groups that are biologically plausible. Finally, the
model was able to reproduce the cholesterol and triglyceride phenotype of known
genetic diseases like familial hypercholesterolemia and familial
hyperchylomicronemia.
In chapter 3, we further developed and
calibrated Particle Profiler using subjects with specific genetic conditions.
We subsequently performed technical validation and worked at an initial
indication of clinical usefulness, starting from available data on lipoprotein
concentrations and metabolic fluxes. Since the model outcomes cannot be
measured directly, the only available technical validation was corroboration.
For an initial indication of clinical usefulness, pooled lipoprotein metabolic
flux data was available from subjects with various types of dyslipidemia. Therefore,
we investigated how well lipoprotein metabolic ratios derived from Particle
Profiler distinguished reported dyslipidemic from normolipidemic subjects. The
VLDL metabolic ratios outperformed each of the classical diagnostics
separately, and also added power of distinction when included in a multivariate
logistic regression model on top of the classical diagnostics.
Part II:
Applications
In
chapter 4, we investigated how dietary MCFA and linoleic acid (C18:2n-6)
supplementation and body fat distribution affect the fasting lipoprotein
subclass profile, lipoprotein kinetics, and postprandial fatty acid kinetics.
Dietary MCFA supplementation led to a less favorable lipoprotein profile than
C18:2n-6 supplementation. Surprisingly, these differences were not due to
elevated VLDL production, but rather to lower lipolysis and uptake rates. The
expected higher VLDL production after MCFA intervention was diminished due to
chain elongation and metabolic control in the liver. Observed changes in
lipolysis and uptake rates are consistent with the mechanism of PPAR activation
by linoleic acid. Differences in body fat distribution only correlated with
differences in LDL cholesterol values.
In chapter 5, we investigated the
response of lipoprotein profiles to fibrate therapy in the Genetics of Lipid
Lowering Drugs and Diet Network (GOLDN) study. Fibrates but show heterogeneous
treatment response. Using k-means clustering, we divided the baseline NMR lipoprotein
profiles of 775 participants in the GOLDN study into three subgroups. The
subjects in each subgroup showed differences in conventional lipid
characteristics and in presence/absence of cardiovascular risk factors at
baseline. Modeling analysis using Particle Profiler revealed differences in
VLDL performance between the subgroups. This parameter was high in the first
‘healthy’ subgroup, lower in the second subgroup, and lowest in the third
subgroup, with a progressively larger influence of extrahepatic lipolysis
dysfunction versus liver uptake dysfunction. The second and third subgroups
showed a positive, yet distinct lipid response to fenofibrate treatment.
Therefore, using lipoprotein profile clustering, we identified two subgroups of
responders to fenofibrate therapy with different underlying disturbances in
lipoprotein metabolism. The total responder subgroup is larger than responder
subgroups identified with baseline triglyceride and HDL cholesterol cutoffs.
In chapter 6, we used Particle Profiler
to calculate lipoprotein metabolic ratios from NMR lipoprotein profiles
measured in the Framingham Heart Study. We tested whether lipoprotein metabolic
ratios can improve cardiovascular risk prediction. Two ratios, that reflect
VLDL metabolism, significantly improve cardiovascular risk prediction above the
classical Framingham Risk Score. The first ratio is called the ‘extrahepatic
lipolysis ratio’. This ratio proves to be especially important when estimating
the risk for people who take blood pressure medication: when study participants
take this medication and the lipolysis of VLDL particles is low, the risk of
cardiovascular disease dramatically rises. The second ratio is called the
‘hepatic turnover ratio’. This ratio is especially important for people that
are on the border between low and medium risk, judging from their LDL particle
number. If in that case, the liver turnover of VLDL particles is good, risk is
low; if the liver turnover of VLDL particles is impaired, the risk is in the
medium category. So in this chapter we identified two lipoprotein metabolic
ratios that significantly improve cardiovascular risk prediction. For clinical
application, further validation is required.
Part III:
Reflections
In chapter7, we included methodological and technical considerations useful for
researchers starting to develop computational model-based diagnostics using
clinical chemistry data. We discussed the issues that a modeler has to take
into account during the design, construction and evaluation phases of model
development. We use the example of Particle Profiler development as a case
study to illustrate our considerations. The case study also offers techniques
for efficient model formulation, model calculation, workflow structuring, and
quality control.
The general discussion includes additional
methodological considerations related to Particle Profiler, a comparison of
Particle Profiler with other modeling studies, future research recommendations
and a general conclusion.
