Effects of Genetic Variation on the Relationship between Diet and Cardiovascular Disease Risk

2008

Date Source: 

Thomas Wolever of the Department of Nutritional Sciences, University of Toronto

Organizer: 

Alison Gibbs, Department of Statistics, University of Toronto

Background

Risk Factors of Cardiovascular Disease
 

Cardiovascular disease is the most common cause of death in Canada. High levels of LDL (low-density lipoprotein) cholesterol and low levels of HDL (high-density lipoprotein) cholesterol are associated with increased risk of cardiovascular disease. In order to reduce levels of LDL cholesterol and increase levels of HDL cholesterol, dietary recommendations include modifying the type of fat consumed by reducing the intake of saturated fatty acids (SFA) and increasing the intake of polyunsaturated fatty acids (PUFA).
 

The risk of cardiovascular disease is also increased by the presence of chemicals called inflammatory cytokines, such as tumor necrosis factor-alpha (TNF-alpha). These are produced in high amounts in people who are obese or insulin resistant or have type 2 diabetes. One of the ways TNF-alpha may affect cardiovascular disease risk is by modifying the way dietary fat is related to fats (or lipids) in the blood, such as cholesterol and triglycerides.
 

Genetic Variation in Cardiovascular Disease Risk Factors
 

In a single strand of DNA, there may be many locations for which all members of the population have the same base. Assume that for one short region every individual is AAGCCTA. In this region there is no polymorphism.
 

Now suppose that some individuals vary at a particular position in the sequence, for example:
 

Joe                                          AAGCCTA
Maude                                     AAGCCTA
Lucy                                        ATGCCTA

This population is polymorphic at position 2. This is an illustration of a single-nucleotide polymorphism.
 

Because genes are found in two copies, one on each of a pair of chromosomes, individuals can have copies that differ at a polymorphic position. So Lucy could have two genes that differ:
 

Lucy1                                       ATGGCTA
Lucy2                                       AAGGCTA

Lucy is heterozygous at position 2 and she is homozygous at the other five positions. At position 2, it is possible for individuals to be heterozygous (TA), or homozygous (AA or TT).
 

The gene for TNF-alpha has two single-nucleotide polymorphisms of interest. Specifically, substitution of an adenine (A) residue for guanine (G) in position 238 ( 238G -> A) of the TNF-alpha gene reduces the rate of transcription and production of TNF-alpha, while the 308G -> A substitution has the opposite effect. These variants are present in about 20-30% of the population.
 

Related Studies
 

A recent cross-sectional analysis of observational data from subjects with type 2 diabetes showed that the 238G -> A and 308G -> A variants affected the relationships between PUFA intake and HDL cholesterol and between PUFA intake and HDL’s major structural protein, apo A-1. In subjects with the variants associated with an increased production of TNF-alpha (-238GG and -308AA or GA), PUFA intake was negatively related to HDL and apo A-1 levels, whereas in subjects with the variants associated with reduced production of TNF-alpha (-238AA or GA and -308GG), PUFA intake was positively related to HDL and apo A-1 levels. These findings are summarised in the table below.
 

  Position
Variant 238 308
AA Reduced TNF-alpha and positive relationship between PUFA intake and HDL. Increased TNF-alpha and negative relationship between PUFA intake and HDL.
GA Reduced TNF-alpha and positive relationship between PUFA intake and HDL. Increased TNF-alpha and negative relationship between PUFA intake and HDL.
GG Increased TNF-alpha and negative relationship between PUFA intake and HDL. Reduced TNF-alpha and positive relationship between PUFA intake and HDL.
 


The Case Study Data
 

In this case study, we will examine data collected during a recent clinical trial involving subjects with type 2 diabetes. Our primary objective is to examine the effect of randomly assigned dietary changes on HDL and apo A-1 and whether the diet-cholesterol relationship differs among the polymorphisms associated with the production of TNF-alpha.
 

The subjects were randomly assigned to one of three diets; two of the three study diets were high in carbohydrates and the third was high in fat. The subjects were followed on their dietary treatments for one year. Cholesterol measurements were taken at baseline and then at 4, 12, 26, 39 and 52 weeks. Dietary intake data were collected from three-day food records in which the subjects recorded everything they ate. The subjects completed the records on multiple occasions both before the trial started and then during the trial while on their dietary treatment. Components of the subjectsf dietary intake pre-trial and during the trial are derived from the food records. One of the dietary components measured is the percentage of food energy intake from PUFA.
 

Objectives
 

Primary objective
 

The primary objective is to consider how the variants of the TNF-alpha gene modify the effect of changes in PUFA intake on changes in HDL and apo A-1 concentrations in subjects with type 2 diabetes.
 

Secondary objectives
 

It is also of interest to know if the polymorphic variants of the TNF-alpha gene modify the effect of changes in PUFA intake on other lipid concentrations. That is, in addition to HDL and apo A-1, are there effects on LDL, apo B-1, triglycerides, total cholesterol, and cholesterol ratio?
 

In addition to PUFA intake, it is of interest to know which other dietary components (protein, carbohydrates, saturated fat, monounsaturated fat, and alcohol) affect the lipid measurements.

 

Research Question: 

Do variants in the TNF-alpha gene modify the effect of changes in polyunsaturated fat intake on changes in HDL and apo A-1 concentrations in subjects with type 2 diabetes?

 

Variables: 

The data are provided in three files and are available in plain text and MS Excel files. Dots indicate missing values. All missing values can be considered missing at random.
 

The files are:
 

  1. Subject information (Data files: Plain textMS Excel
    Variables:
    • ID: subject identifier
    • Diet: randomly assigned diet (hifat=high in fat; hicarb1/hicarb2=high in carbohydrates).
    • Gender
    • Age in years at the start of the clinical trial
    • BMI: body mass index, a measure of obesity
    • Centre: subjects were recruited at 5 centres in cities across Canada
    • Statins: 1 = the subject was on cholesterol lowering medication; 0 = the subject was not on cholesterol lowering medication
    • TNF238: subjectfs polymorphic variant of the TNF- alpha gene at position 238 (GG, GA, or AA)
    • TNF308: subjectfs polymorphic variant of the TNF- alpha gene at position 308 (GG, GA, or AA)
       
  2. Lipids (Data files: Plain textMS Excel
    Variables:
    • ID: subject identifier
    • Weeks: the number of weeks since the start of the trial when the measurement was taken; 0 is baseline
    • HDL
    • Apo-A1
    • LDL
    • Apo-B1: LDL’s major structural protein
    • Triglycerides
    • Total cholesterol
    • Cholesterol ratio: ratio of total cholesterol to HDL
       
  3. Dietary intake (Data files: Plain textMS Excel
    Variables:
    • ID: subject identifier
    • When: pre = the dietary intake was measured before the start of the clinical trial; during = the dietary intake was measured while the subject was on the dietary treatment
    • KCAL: food energy intake
    • Protein: percent of energy intake from protein
    • Fat: percent of energy intake from fat
    • Carbohydrates: percent of energy intake from carbohydrates
    • SFA: percent of energy intake from saturated fats
    • MUFA: percent of energy intake from monounsaturated fats
    • PUFA: percent of energy intake from polyunsaturated fats
    • Alcohol: percent of energy intake from alcohol