All are now agreed that the plasma fibrinogen level is a strong and independent predictor of the risk of major episodes of coronary heart disease (CHD), i.e. coronary death and myocardial infarction. Those with fibrinogen levels in the highest third of the distribution of fibrinogen run about twice the risk of an event compared with those in the lowest third . Another way of expressing the risk is that it increases by perhaps 50% or more for every standard deviation increase in fibrinogen , which varies from about 0.5 to 0.8 g L−1. The association with stroke risk is less clear , but it is probable that fibrinogen also predicts this [4–7]. Peripheral arterial disease and re-occlusion of coronary arteries or veins after interventions such as angioplasty and bypass surgery [8–10] are other events of which the fibrinogen level does or may indicate the risk. The fibrinogen level may also be related to other conditions and causes of death, notably cancer. So it is not specific for arterial disease unless the explanation for its association with these other conditions is that fibrinogen affects the circulation in some way that hastens their onset or progression. However, this non-specificity does not invalidate consideration of fibrinogen in arterial disease, particularly in view of its effects on pathways, such as viscosity, known to affect CHD risk.
The ‘fibrinogen story’ started and has developed at least two decades after work on lipids and effective lipid-lowering treatments, so the case for deciding when and in whom fibrinogen levels should be measured should be considered accordingly—there is a great deal of catching up to be done. Interest in studying fibrinogen and CHD developed in the early 1970s following the increasing awareness of the thrombotic contribution to major events  (although much later after Morris's landmark publication than it should have done) and the need to take it into account as the immediate cause of most episodes [12–15]. Until then—surprising as this now seems—the pathogenesis of CHD had been considered by most as due mainly or entirely to the consequence of the long-term contribution of blood cholesterol to atherogenesis. Indeed, the widely used term ‘atherosclerosis’ to refer both to clinical episodes and to the atheromatous element of their pathology has been confusing, and has continued to limit full recognition of the thrombotic process.
Assessing components of the haemostatic system and their possible contribution to the thrombotic component of CHD has been technically much more demanding than for lipids. There is still no satisfactory way of measuring platelet activity in individuals in large-scale studies and most attention has hitherto been paid to the coagulation system. In this, the essential role of fibrinogen needs no further emphasis.
Observational studies of CHD (as distinct from randomized controlled trials) have two main objectives. One is to help identify those at high risk of CHD. The other is to learn more about the causes and processes involved in its onset, in which case the implications for prevention and treatment are ultimately the chief aim. Of course, one approach overlaps with the other. If—to consider the extremes to make the point—the objective is simply to identify those at risk, it is entirely appropriate to place most emphasis on the fully independent contribution of a risk factor. If, on the other hand, the processes involved in CHD are the main interest, multivariate analyses should be the method of last, rather than first, resort. Putting fibrinogen and smoking straight into a multivariate analysis will underestimate the full contribution of fibrinogen to CHD, because smoking is a cause of high levels (and similarly for other determinants of fibrinogen).
How do these considerations lead to rational decisions about measuring the fibrinogen level in individuals? Some have questioned the value of screening for the better-known CHD risk factors , but this begs the question of whether newer risk factors may help. Fibrinogen probably does, as we now have additional clues about hitherto less well-known pathways that may lead to CHD.
For several years there has been a largely sterile debate as to whether fibrinogen is a cause or simply a marker of CHD risk. Very probably, the answer is both. Figure 1 shows various characteristics influencing the fibrinogen level and several pathways that are either known to contribute to CHD or may do so . One of these pathways is the effect of the fibrinogen level on the development of atheroma in the arterial wall, presumably in the long term, while the other mechanisms are likely to promote the much more acute, short-term complication of thrombosis. Atheroma is an inflammatory process which may lead to an increase in fibrinogen as a chronic or acute-phase protein, so in this sense a high level would indeed be a marker of underlying vessel wall pathology. There may even be a sequence in which fibrinogen contributes to atheroma, which leads in turn to a further increase in fibrinogen and thus to a higher level of the activity of the pathways through which fibrinogen increases the risk of a major CHD event. For fibrinogen, this is a more helpful way of thinking about the ‘marker or cause’ explanation than ‘reverse causality’, an ambiguous term which may or may not imply direct causality and which many do not understand properly anyway. It seems probable that high fibrinogen levels, whatever their explanation to begin with, are likely to be involved in the causation of the acute thrombotic event. (Whether C-reactive protein or fibrinogen is associated more closely with CHD is debated, but at present at least as much, if not more, is known about the ways through which fibrinogen contributes to CHD. It may, of course, be that both are valuable, but it is fibrinogen that is under consideration here.)
However, another argument against a causal role for fibrinogen has been advanced. The theory of ‘Mendelian randomization’ is that a genetic marker associated with the fibrinogen level should also be associated with the occurrence of clinical CHD events. It has been used to question the role of fibrinogen in causation, as in one study there was no clear association with CHD of one of the polymorphisms associated with the fibrinogen level . For a variety of statistical and other technical reasons, the question remains open. Very recently, however, the failure of the Mendelian approach to show clear associations with clinical events of polymorphisms for lipid fractions contributing to CHD shows that there are problems with the Mendelian approach that remain to be explained . The balance of evidence does seem to favour a causal role (as well as that of a marker). Therefore, as with the cholesterol debate several years ago, what would be the effects of lowering the fibrinogen level?
It is a desirable condition for screening that there should be an effective treatment for the risk factor involved . At present, there are no selective oral agents for lowering fibrinogen. Fibrates such as bezafibrate do so by an amount that would be expected to lead to a worthwhile reduction in CHD, particularly as they also have beneficial lipid-modifying properties. However, trials of bezafibrate and similar fibrates show mostly only modest reductions of around 10% in major events , and it may be that the rise in homocyst(e)ine caused by these fibrates offsets much of their potentially beneficial effect . Is it justifiable still to measure fibrinogen? (Gemfibrozil, which does reduce CHD [23,24], differs biochemically from other fibrates and tends to raise fibrinogen .)
Although formal research remains to be conducted, clinical experience shows that high fibrinogen is sometimes the only risk factor in someone who has previously not had an event but who may be at risk of one. (Patients who have already had events are at high risk for that reason alone, and should be taking statins and aspirin and perhaps other agents.) Consider a 45-year-old man who has an uncertain family history but a fibrinogen level in the top 10% of the distribution. Low-dose aspirin might well be indicated (provided his annual risk of a major event is more than 1% because, at lower risk, hazards may well outweigh benefit —an important consideration not generally recognized) even though (pace Wilson and Jungner) there is at present no effective agent for lowering fibrinogen itself. This scenario draws on several only partly satisfied requirements—the number of people with high fibrinogen levels but no other obvious risk factors, having the resources to measure levels more than once (as with any characteristic such as cholesterol on which clinical decisions may be based), being able to use the fibrinogen level in a reliable risk scoring system, and laboratories with good quality control that can say what their reference ranges are, for example. It will also be particularly valuable to be able to allow for the regression dilution effect , which depends on having more than a single value for individuals in order to avoid underestimating the strength of the fibrinogen/CHD association. The case for answering all these questions is a strong one, and closely resembles the questions about cholesterol 20 or 30 years ago. They will be clarified only by measuring the fibrinogen level in many people in different settings and for different reasons, and with increasingly helpful clinical decisions in mind.
In fact, there is, as already suggested, the single individual who should be considered. Ask yourself whether, not so far having had a myocardial infarct but concerned about marginal or uncertain ‘conventional’ risk factors (perhaps your cholesterol level and your family history), you would completely forgo measurement of your fibrinogen level, even if you regard it only as a marker, to help you make a rational decision about taking low-dose aspirin or some other prophylactic measure?