CAPITULO II: DESARROLLO DEL SISTEMA
2.2 DISEÑO DEL SISTEMA
2.2.1 PRIMER SPRINT
2.8.1 Tests used for analysis
Only white Caucasian individuals of European ancestry were included in the analysis. Normally distributed data were expressed as mean and standard deviation (SD) or standard error o f the mean (SEM). Concentrations o f NOx, neopterin, serum triglyceride and fibrinogen were log transformed, prior to analysis, to normalise their distributions. Geometric means and approximate SD are reported for these variables. Non-normally distributed data were expressed as median and range. Univariate analysis of continuous variables was carried out by one-way analysis of variance (ANOVA). Welch's test was used where there was evidence o f unequal variances between groups.
As the distribution of plasma NOx and neopterin were highly skewed, Kruskal-Wallis (KW) non-parametric tests were used to determine the relationship with genotype in men who developed an IHD event and those who remained free. Medians with interquartile range are presented. Geometric means and appropriate SD are reported for these variables. The KW test was used to assess the effect of genotype on NOx and
multiple logistic regression models. The effect of possible confounding factors such as age and BMI were assessed using the Spearman Correlation coefficient.
The student's T-test (two group comparisons) or ANOVA (three group comparisons) were used to compare normally distributed variables. Non-normally distributed variables were compared by the Mann-Whitney test (two group comparisons) or the KW test (three group comparisons). Differences in clinical and biochemical characteristics according to smoking status were analysed using one-way ANOVA. Non- and ex-smokers were combined for analysis and compared with smokers.
For association studies, candidate gene approach was used It tests the hypothesis that a specific allele occurs more frequently in diseased subjects (those who suffered an IHD event) than in those without the disease (i.e. no IHD event). Frequency of a particular marker allele in the group of unrelated 'cases' were compared to a group of unrelated 'controls', who were closely matched for biochemical factors such as age and BMI so as to avoid differences between allele frequencies bearing no causal relationship at all with the disease.
An increase in allele frequency in the cases compare to the controls suggests the probability o f a cause/effect relationship between the 'associated' alleles and the disease. Either the gene itself or a nearby gene has a functional genetic variant, which might contribute to the disease process. A significant association between a genetic marker and a disease implies that either the genetic marker either resides in the disease locus itself or within 500 kb of the disease DNA locus (Kurtz et a l, 1992).
The relative risk (RR) gives a convenient summary o f the outcome of an epidemiological study and represents a quantitative value o f an approximate estimate o f the RR for the disease associated with the factor. For all the polymorphisms, risk was estimated in rare allele carriers and homozygotes (i.e. 'exposed' individuals) combined and were adjusted for age, BMI, baseline levels o f plasma cholesterol, triglycerides, fibrinogen and sBP, using Cox's proportional hazard model. Survival analysis was carried out using Cox's model, thereby allowing for varying follow-up intervals and failure time was taken as time to the first IHD event. Significance was assessed using the Likelihood Ratio (LR) test and 95% confidence intervals (Cl) for the parameters were calculated from the standard errors by assuming normality. Results are exponentiated and are presented as the RR representing the hazard ratios (HR) and their 95% CL Binary variables were analysed using the chi-squared (%^) test for difference in proportions, test, used for unpaired sample data, determines whether there is a hypothesised statistical difference between observed expected values o f the data.
2.8.2 Hardy-Weinberg (H-W) equilibrium
Hardy-Weinberg (H-W) law states that the genotype frequencies in any generation can be predicted by knowing only the allele frequencies in the parental generation, assuming random mating o f parents and Mendelian segregation ratios for each mating type in a large population. If p is the frequency of an allele (X) and q is the frequency of a second allele (Y) in the parental generation, then q= \-p. The H-W law states that the genotype frequencies expected in the offspring are given by the binomial
• Relative frequency of common homozygous genotype, XX = • Relative frequency of heterozygous genotype, XY = 2pq • Relative frequency o f rare homozygous genotype, YY = • By definition, p^ + 2pq + q^ = \
A population is said to be in H-W equilibrium if the observed frequencies o f genotypes XX, XY and YY are equal to the expected H-W frequencies of 2pq and q^. Deviation from the equilibrium in a large random mating population occurs when there is mutation, change, selection or migration. The observed genotypes frequency for a 2-allele polymorphism were counted and all data were checked to be in H-W equilibrium. analysis was used to compare categorical data between two groups and to assess whether genotype frequencies deviated from H-W equilibrium and values and the number o f degrees of freedom were calculated. A p value > 0.05 was taken to be not significant and to conform to H-W law.
2.8.3 Allelic association and linkage disequilibrium coefficients
Allelic association and linkage disequilibrium between alleles at two different polymorphisms in the same gene cluster was estimated as described elsewhere (Chakravarti et al, 1984). It was determined as follows, with a, b, c and d are the frequencies o f the haplotypes A l B l , A1B2, A2B1 and A2B2 respectively:
Number o f A IB 1 chromosomes = a Number of A1B2 chromosomes = b
Number of A2B1 chromosomes = c Number of A2B2 chromosomes = d
The haplotype o f an individual heterozygous for two alleles (1 and 2) at each of two loci A and B could be A1B1/A2B2 or A1B2/A2B1. A correlation coefficient delta (A) was estimated, where:
A = ad-bc
V(a + b) (c + d) (a + c) (b + d)
A p value of <0.05 was taken as statistically significant. A can range between 0 and 1, with the higher the value of A the greater the degree of linkage between the 2 alleles. A negative A value means that the rare form o f one allele is in complete linkage disequilibrium with the common form o f the other allele.
The haplotypes of both heterozygous individuals, A1B2 and A2B1, were estimated by maximum likelihood analysis (Thompson et a l, 1988). The probability of the double heterozygous being distributed as A1B1/A2B2 is:
P = number o f A1B1/A2B2 haplotvpes total number of haplotypes (n)
P = a/n X d/n