A risk factor is a statistical predictor of disease and its identification is usually based on associations demonstrated between the predictor and the disease in epidemiologi cal studies. As a consequence, a risk factor may or may not be causally related to
the disease. For example, it may be a secondary manifestation of a more basic underlying abnormality or an early symptom of the disease.
1.4.1 Epidemiology
The established "causative" risk factors of CHD fail to account for the total in cidence of CHD.39,148' 158 Nor do they explain the absence of clinical CHD in persons with these risk factors.148 Furthermore, the known risk factors have a limited ability to predict disease in an individual,152 and this low predictive accuracy limits their use for identifying persons with asymptomatic CAD. Indeed, the low predictive accuracy for identify individuals with CHD also applies to groups.153,159 The results of the Pooling Project defined a wide range of factors increasing the risk for CHD within different populations, and demonstrated significantly increased risk with levels of associated factors well below the 95th percentile, particularly when occur ring in combination.148 These observations have profound implications for public health and the prevention of premature CHD. For the individual, these results have illustrated major limitations in the ability to prospectively predict disease occurrence, even when the strongest risk factors are combined. Quantification of risk factors in the form of a decile risk has been used to determine the likelihood of developing CHD.160 In those with the lowest risk it was 5%, and in those with the highest risk it was 21.4% over 8 years.160 This synergistic effect of the major risk factors also remains to be explained. Use of risk factor profiles can identify 10% of the general population who will have 22% of the CHD in the population.160
A number of observations made in epidemiological studies further illustrate the incomplete foundation that known major risk factors make for explaining all CHD. Firstly, international population studies reveal a significant variation in CHD mortal ity in different countries.161 This variation is probably only partly due to population differences in blood cholesterol levels.162 It should be emphasised that cross-cultural comparisons do indicate that CAD is rare in populations with cholesterol levels less than 4.2 mmol/1.162 However, regional differences within some countries cannot be wholly explained on the basis of differences in levels of the 3 major risk factors for CHD. Secondly, the Framingham Study logistic, useful for predicting CHD in high risk population groups, over-estimates CHD incidence in "low risk countries". This indicates that the influence of risk factors in the predominantly Caucasian population of Framingham is greater than in non-Westernised countries. The reason for this observation is unknown, but may simply be due to the lack of underlying atheros clerosis in populations with low cholesterol levels leaving no substrate for the other "causative" risk factors to act upon and result in obstructive lesions causing clinical CAD. Thirdly, even though total cholesterol levels predict long term CHD mortality it is not a major factor in predicting all-cause mortality.159 The association between cholesterol and CHD at "borderline hyperlipidaemia" may reflect the correlation of cholesterol with other risk factors.159 Finally, and of particular relevance for the hypothesis in this Thesis, the incidence of CHD in France is relatively low compared to other Western Countries with equal intake of saturated fats and population cho lesterol levels.163 Moreover, other risk factors for CHD, including BP levels, body- mass index, and cigarette use for men are no lower in France than in other industri alised countries.163,164 One proposed explanation for this variation is the inhibition of platelet reactivity by alcohol.164
Even though population studies have shown that all of the variance for CHD devel opment in a population cannot be explained by the major risk factors, 148 there are
caveats to this observation. A limited number of risk factors have been included in these studies. There may also be the effects of longitudinal trends on the variables investigated, either increasing if favourable, or decreasing if unfavourable. Further, the majority of persons who have a MI are in the second, third and fourth quintiles for cholesterol level, as are the majority of persons who do not have CHD. 165,166
1.4.2 Risk Factors and Coronary Artery Pathology
The strength of association between CHD risk factors and pathologically defined atherosclerosis also has limitations. The observation that atherosclerosis is closely associated with major risk factors is only true if the correlation is estimated on a group basis. The correlation is much weaker when estimated on an individual basis. 152 For instance, correlation coefficients between cholesterol level prior to
death and quantitatively evaluated coronary atherosclerosis range from 0.15 to 0.35
for the individual analysis, but 0.76 for groups. This indicates that individuals with higher cholesterol levels are not all equally susceptible to CHD. 152 Furthermore,
there is a progressive rise in the prevalence of all major risk factors with age, in parallel with a continual reduction in the predictive accuracy. 151,167 When comparing
homogeneous age-sex-race groups without significant risk factors, the range of the extent of CAD in any single arterial segment is similar between groups, but there is up to a 5 fold difference in the mean measure for the extent of disease for all the coronary arterial tree between the groups. 153 That is, the median varies between
groups. Variability among arteries in the same individual implies that the response of the arterial wall differs, assuming that the composition of the blood reaching each artery is identical. Indeed, there is a wide variation in the extent of raised lesions in male smokers and non-smokers with an enormous overlap despite significant differ ences in mean values. 158 Even though the presence of CAD in a population corre
lates with the incidence of CHD, and CHD cases almost always have more atheros clerosis on the average than non-CHD cases in the same location, the incidence of CAD is significantly greater than the incidence of CHD, implicating other factors as well as the presence of atherosclerosis in the development of CHD.39 Furthermore,
even after classifying cases according to risk factors in a prospective autopsy study, there can be individual variation in the extent of raised atherosclerotic lesions, and the risk factors may only explain about 26% of the individual variation in raised lesions. 168
Important determinants of whether an individual develops CHD may be local factors within the artery and/or may be related to the morphology of the atherosclerotic plaque. Indeed, mechanical, hydraulic and cellular/molecular characteristics of the plaque may be important determinants in the development of clinically significant lesions. 169 For instance, although the distribution of circumferential tensile stress
across the intima is altered by the atherosclerotic plaque, and areas of high stress correlate with the site of intimal tears,33 these sites are also affected by variation in
the mechanical strength of the intimal cap.33 Sites of intimal tears appear to be weak
1.4.3 Risk Factors and Angiographic CAD
Although risk factors are clearly associated with both clinical manifestations of CAD and angiographically dem onstrated CAD, they are not accurate predictors of angio graphic C A D .149 B etw een 9% and 17% o f m ale subjects have no d efin ab le risk facto r fo r CHD in the presence o f angiographically defined C A D .149 In addition, limited gains accrue from the use of discrim inant function analysis in correctly allo cating patients into angiographically defined disease and non-disease groups accord ing to the major risk factors.149 This topic is elaborated on in Chapter 8.
1.4.4 Genetic and Hereditary Factors
Fam ily studies consistently dem onstrate a prom inent fam ilial grouping o f CHD. In m ales, there is likely a genetic predisposition to both prem ature C H D ,154,155,170' 175 and the major risk factors for CH D .154,155,165,173,176 The fam ilial occurrence o f CAD is well docum ented by case-control,177' 180 prospective,181 and twin studies.182,183 No discrete genetic m ode o f inheritance has em erged for the m ajority o f family aggre gates with CAD and only occasionally have single gene disorders accounted for the development of premature CAD in a general population.184 A family history of MI is an independent risk factor for MI, and the num ber of relatives and the age at which they are affected is related to the strength of the association.180
The tendency for first degree relatives of CAD patients to have more adverse risk factors than the general population has been well docum ented,176,178,185 particularly so when CAD is m anifest at a young a g e.176 The m ajor risk factors per se have a substantial fam ilial com ponent, w hether of genetic or environm ental origin. It may be that the co n trib u tio n o f genetic factors is g reater than can be attrib u ted to the inheritance o f the m ajor risk factors alone.186 For most of the im portant risk factors, familial associations arise from both a shared genotype and a shared environment.187 The fam ilial nature o f CAD may be explained in part by com m on environm ental factors shared by family members. Nonetheless, the classic risk factors do not entire ly predict fam ilial occurrence o f CHD, even if the genetic risk factors are elim inat ed .165,173 It is even possible that the risk ratio for a positive family history o f CHD is greater than for the highest quintile of cholesterol levels,165 and that the contribution of heredity may exceed that of environment.173
The exact nature o f the genetic factors, which are independent of the known risk fac tors and which im part a predisposition to CAD, are obscure. In addition, for young m ales, the heredity association is particularly im portant,170 and it may even be re stricted to younger m en.188 The relationship in fem ales is less certain .175,188 U nfor tunately specific in h eritan ce p attern s, genetic m echanism s or gen e-en v iro n m en t interactions for CAD have not been well defined and require m ore intense evalua tion.154,155,170,173 D iseases attributable to multifactorial inheritance are caused by the small effects of many genes, and only in a small number o f cases are determ ined by single genes of large effect.189,190
It is assumed that there are multiple predisposing hereditary factors that interact with environm ental triggers to produce CAD, rather than a single gene mode o f inherit-
ance. Evidence exists, however, that in patients susceptible to CHD there are single m ajor gene influences, and, although research has concentrated on lipid disorders, there may be m ajor gene influences other than through lipoprotein abnorm alities.165 For instance, ’abnorm al’ platelet response to ADP have been found m ore often in the close relatives o f young men with AMI than in a general population.191 Indeed, other heritable risk factors appear to contribute strongly to the occurrence of prem a ture CA D .192
Risk factors other than those currently known must mediate the effects o f the family a g g re g atio n o f C A D .193 The p resen ce o f an en v iro n m en tally sen sitiv e in h e rite d abnormality of platelet reactivity is one possibility that requires evaluation.
1.4.5 Sex Differences in CHD
Differences in known risk factors fail to explain the male sex predominance in either atherosclerosis or CHD incidence.152,155,194' 196 H ow ever, there is an absence o f d e tailed studies o f the variability in CHD susceptibility betw een men and w om en.196 T h e re are a lso u n e x p la in e d d iffe re n c e s in the p ro g n o s is re la te d to C A D risk fa c to rs197 and the presence o f C H D .194,195,197 For exam ple, w om en appear to have m uch low er rates of CHD than men at the same values o f cholesterol and increased risk at higher levels of cholesterol.198 Furtherm ore, there are im portant differences in the natural h isto ry ,195 the risk facto r p ro file 194,195,198,197 and p ath o lo g y 199,200 o f CHD in women com pared to males. However, most young women with overt CHD have a detectable risk factor, but also have a high rate of non-atherosclerotic MI.200
1.4.6 Conclusion
D espite the concerns discussed above, epidem iological evidence is persuasive that the 3 m ajor risk factors (tobacco use, hypertension and hypercholesterolaem ia) are causative in their association with CHD,161 in particular, lipid abnormalities.201 Even though a calculated attributable risk for these factors may be as high as 90% ,202 their lack o f diagnostic specificity and low predictive accuracy for the developm ent o f CHD rem ains to be explained. Many individuals remain free o f the clinical m anifes tations o f CAD even when m ultiple risk factors are p resen t.148,160,166,203,204 F urther m ore, th e ir poor predictive accuracy persists into old age 151.160.167-205 and, w hile certain factors are significantly correlated with angiographic findings, their value for predicting the severity o f CAD disease is also very lim ited.149,206,207 Indeed, in the younger p atient the progression o f coronary atherosclerosis into distinct coronary artery stenosis m ay largely be determ ined by m echanism s other than the presently know n risk fa c to rs.207 B ased on p resen t evidence, there seem s to be at least two processes inv o lv ed in the developm ent o f atherosclerotic vascular disease, one in re sp o n se to lo c a lise d lip id accu m u latio n w ith in the v essel w all and the o th e r a throm botic response leading to more rapid developm ent o f stenotic coronary artery lesions.9