Gordon D. O. Lowe, Division of Cardiovascular and Medical Sciences, University of Glasgow, 10 Alexandra Parade, Glasgow G31 2ER, UK. Tel.: 0044 141211 5412; fax: 0044 141211 0414; e-mail: firstname.lastname@example.org
Since the first report by Professor Tom Meade and his colleagues , many other prospective studies have confirmed that plasma fibrinogen levels are associated with risk of coronary heart disease (CHD), whether or not there is baseline evidence of cardiovascular disease [2,3], and possibly also with increased risks of stroke or peripheral arterial disease (PAD) . In addition to a potential role in risk prediction, it is possible that higher fibrinogen levels (however they arise) may play a causal role in ischaemic events (for example, by increasing atherogenesis; increasing platelet–fibrin thrombogenesis; or increasing plasma and whole-blood viscosity which aggravates ischaemia distal to atherothrombotic stenoses ). If so, fibrinogen measurement might be justified if therapeutic fibrinogen reduction may thereby be indicated.
Such potential clinical utility of plasma fibrinogen, as with any other potential risk marker/factor for thrombotic disease, should be evaluated critically . We need to ask:
•Is the association between fibrinogen and risk of arterial thrombosis cause, consequence or coincidence? (See Fig. 1.)
•Is the proposed addition of fibrinogen to current risk prediction scores useful and practical?
Cause, consequence or coincidence?
Increased fibrinogen may be partly a consequence of atherothrombosis. Atherosclerosis is an inflammatory disease , associated with a systemic inflammatory response including increased circulating levels of acute phase proteins [e.g. fibrinogen, C-reactive protein (CRP)], white cell count and proinflammatory cytokines . Fibrinogen correlates with the extent of coronary artery disease [8,9], PAD [10,11] and carotid atherosclerosis . Part of the association between fibrinogen and risk of arterial events may therefore reflect a reactant plasma protein response to atherosclerosis: ‘reverse causality’.
Alternatively, this association may also arise because both fibrinogen and CHD are associated with established cardiovascular risk factors, confounding their relationship, which is partly coincidence  (Fig. 1). In addition to age and ‘classical’ cardiovascular risk factors , fibrinogen is related to insulin resistance and diabetes , psychosocial variables [16,17], regular exercise  and alcohol consumption .
The hypothesis that fibrinogen may play a causal role in arterial thrombosis is as yet unproven (Fig. 1). Randomized controlled trials of chronic fibrinogen reduction by certain fibrates (clofibrate, bezafibrate), which lower fibrinogen levels by 10–15%, have not shown reductions in CHD risk [20,21], due possibly to increased homocysteine levels . ‘Mendelian randomized trials’ have shown that common functional polymorphisms in the fibrinogen gene, which result in lifelong lower usual plasma fibrinogen levels (and possibly also lesser acute-phase increases ) are not associated with increased risk of CHD [13,24]. The causality of fibrinogen for arterial thrombosis is also questioned by its lack of specificity for cardiovascular mortality: fibrinogen shows least as strong a relationship to total mortality [25–27]. An alternative hypothesis to explain the relationship of fibrinogen levels (as well as CRP levels) to non-cardiovascular, as well as cardiovascular mortality, is that people with the highest inflammatory reactions to lifetime environmental stresses (stress burden and/or stress response) are more likely to develop more severe disease (vascular or non-vascular), progressing to clinical outcomes, including death (Fig. 2).
Is the proposed addition of fibrinogen to risk prediction scores useful and practical?
The proposed addition of fibrinogen to current cardiovascular risk prediction scores  is only useful if it yields clinically worthwhile additional risk stratification. It is important to establish the incremental value of fibrinogen to that of established risk predictors in large prospective studies, as was performed recently for CRP, whose additional contribution was marginal , and in meta-analyses. The Fibrinogen Studies Collaboration  will report soon on the incremental value of fibrinogen for risks of CHD, stroke and mortality, at different ages, in women as well as men, and by different types of fibrinogen assay. Until this analysis is available, it is premature to use fibrinogen for risk assessment.
Even if this analysis suggests that fibrinogen adds significantly to risk stratification by current risk scores, and might influence clinical practice (e.g. by influencing decisions about reduction in blood pressure or cholesterol, or advising aspirin, in primary CHD prevention), several practical issues remain:
•Choice of assay? The three most commonly used assays give different results for risk stratification by thirds .
•How many assays? As with blood pressure and cholesterol, several assays are required to estimate the ‘usual’ level of fibrinogen .
•Additional cost to healthcare screening?
•What should the patient be told? In contrast to causal risk factors measured routinely (e.g. smoking, blood pressure and cholesterol), which form the basis of a discussion with patients about how they might be altered, patients may be concerned when informed that their fibrinogen level is high, when there is no good evidence that it is causal, or that it should be lowered. Fibrinogen levels are lowered by standard lifestyle advice, including reducing tobacco smoke exposure and obesity, moderate alcohol consumption and increasing regular activity [18,19]. Knowing the patient's fibrinogen level will not change such advice; and prescribing fibrates which lower fibrinogen will not reduce risk [20,21]. Patients may worry needlessly, and their life insurance premium may increase.
For these reasons, I do not—yet—advocate fibrinogen measurement to assess the risk of arterial thrombosis in individual patients, with or without clinical arterial disease. However, there is much active ongoing research, so the question remains open.