Antiplatelet therapy with aspirin and a thienopyridine reduces major adverse cardiac events (MACE) following percutaneous coronary intervention (PCI) ((1),2). Platelet function tests, including traditional light transmission aggregometry (LTA) and contemporary point-of-care assays, have identified variable patient response to thienopyridine agents, particularly clopidogrel (3). Furthermore, in a number of studies, lower platelet inhibition at a single time point during thienopyridine therapy (also called “higher on-treatment platelet reactivity”) has been associated with a higher risk for adverse thrombotic and ischemic cardiovascular events following PCI ((4),5). However, platelet function testing does not reliably predict ischemic outcomes and supports the notion that high residual platelet reactivity is not the only factor contributing to these events (4).

Multiple prior studies have identified serum fibrinogen as a risk factor for both short- and long-term adverse cardiovascular events ((6),(7),(8),(9),(10),(11),(12),(13),(14),15). As an acute phase reactant involved in the final common pathway of the coagulation cascade and essential component of platelet crosslinking in thrombus formation, fibrinogen possesses a clear biological mechanism for its adverse cardiovascular effects (16). A relationship between serum fibrinogen and on-treatment platelet reactivity has also been demonstrated ((6),17), making it unclear whether these risk factors are associated with adverse ischemic events through independent or interactive effects. This study was undertaken to clarify the roles of fibrinogen and on-clopidogrel residual platelet reactivity on short-term ischemic cardiac outcomes following PCI.

The study protocol was approved by the Institutional Review Board of the University of California, San Diego, and written informed consent was obtained from study participants. Patients with established coronary artery disease undergoing elective PCI requiring treatment with clopidogrel were prospectively recruited and enrolled. Subjects included those undergoing PCI for stable angina or unstable angina, or elective PCI with objective evidence of myocardial ischemia. All subjects were required to have negative biomarkers of ischemic cardiac injury prior to enrollment. All subjects received either daily clopidogrel 75 mg for a minimum of 7 days or a 600-mg clopidogrel bolus loading dose at least 12 h before enrollment. Exclusion criteria included age younger than 18 years, anemia (hemoglobin <10 g/dl), thrombocytopenia (platelet count <100,000 cells/μl), diagnosis of inherited platelet disorders, serum triglycerides >400 mg/dl, chronic liver disease, intravenous glycoprotein IIb/IIIa inhibitor administration (within previous 30 days), and positive markers of myocardial injury or acute myocardial infarction (within the previous 72 h). Subjects were not excluded for diagnosis of systemic inflammatory diseases, such as rheumatoid arthritis or systemic lupus erythematosus, or renal failure.

Baseline demographic information and laboratory data including complete blood count, complete lipid panel, serum markers of inflammation (C-reactive protein [CRP] and fibrinogen), and cardiac markers of myocardial ischemia (creatine kinase-myocardial band [CK-MB], troponin I and/or troponin T) were obtained in all subjects. Standard laboratory tests were performed by the clinical laboratory at the University of California, San Diego Medical Center. Fibrinogen was measured using the Beckman Coulter ACL-TOP analyzer (Beckman Coulter Life Sciences, Indianapolis, Indiana) with Instrumentation Laboratory reagent (Instrumentation Laboratory, Lexington, Massachusetts). Cardiac markers of ischemic injury were measured every 8 h following PCI until hospital discharge or up to 24 h to assess degree of myocardial injury. Platelet function testing was performed using the bedside rapid platelet function VerifyNow P2Y12 assay (Accumetrics, San Diego, California). VerifyNow P2Y12 assay design has been previously described (18). Briefly, the VerifyNow system utilizes cartridge-based assays and optical turbidometry to rapidly assess platelet function. The P2Y12 assay features platelet P2Y12 receptor-specific activation using adenosine diphosphate (ADP) and prostaglandin E1 agonists to activate platelets following thienopyridine use. On-treatment platelet reactivity is reported as “result” P2Y12 reactivity units (PRU). High-concentration thrombin receptor-activating protein maximally activates platelets to determine a pre-treatment “base” PRU. Platelet inhibition (PI) or percent change in platelet reactivity from baseline is calculated as: (1 − [result PRU/base PRU]) × 100%.

Whole blood samples were obtained from the radial or femoral artery via the catheter introducer sheath immediately prior to PCI or from a vein in the antecubital fossa using standard phlebotomy techniques. Blood was transferred into standard tubes for serologic assays and Bio-One Vacuette tubes (Greiner Bio-One North America, Monroe, North Carolina) containing 3.2% sodium citrate for platelet function testing.

The study was conducted using a nested case-control design. Sensitive troponin I/T–defined periprocedural myocardial infarction (PPMI) was used as the primary study endpoint, and CK-MB–defined PPMI was used as the secondary study endpoint. Separate analyses using both endpoints were planned. Sample size calculation was performed based on a previous study showing a relationship between elevated serum fibrinogen at presentation and 48-h MACE (acute myocardial infarction and/or cardiac death) in patients hospitalized for acute coronary syndrome (7). In order to detect a 51 mg/dl mean difference in fibrinogen level between cases and controls, an alpha of 0.05, and power of 80%, an estimated 25 subjects were needed in each group. Enrollment was continued until 25 troponin I/T–defined PPMI events occurred.

Study data were collected and analyzed using SPSS software (version 18.0, SPSS, Chicago, Illinois). Categorical variables are reported as percentages, whereas continuous variables are reported as mean ± SD. Normal distribution of continuous variables was confirmed using Kolmogorov-Smirnov testing. The primary PPMI endpoint was prospectively defined as either troponin I or troponin T level >3× the upper limit of normal (ULN) (troponin I reference range 0 to 0.5 ng/ml; troponin T reference range 0 to 0.03 ng/ml) (19), whereas the secondary PPMI endpoint was defined as CK-MB level >3× ULN (CK-MB reference range 0 to 4.8 ng/ml) (19). Definitions of high on-clopidogrel platelet reactivity used in this study included PRU ≥230 (20) and PRU ≥208 (21) based on results of the GRAVITAS (Gauging Responsiveness with A VerifyNow assay–Impact on Thrombosis And Safety) trial. Relationships between continuous variables and outcomes were determined using independent-samples 2-tailed t-tests. Relationships between categorical variables and outcomes were determined using chi-square tests. Variables with significant univariate relationships (p < 0.05) were entered together into a multiple variable logistic regression model for analysis of each outcome. Specific testing between serum fibrinogen, measures of platelet reactivity (PRU or PI), and markers of systemic inflammation (CRP, platelet count, white blood cell count) within a multiple variable model was prospectively planned to evaluate for interdependent effects. Significant variables after multiple variable testing (p < 0.05) were tested for interactions if necessary. To determine a cutoff point for serum fibrinogen level predicting PPMI, a receiver-operating characteristic (ROC) curve was plotted for each outcome definition. The fibrinogen cutoff level producing the maximum sum of sensitivity and specificity was determined, and a 2-tailed p value reported.

A total of 189 subjects (age 63.8 ± 11.6 years, 74.1% male) were enrolled in this study (Table 1). A high prevalence of cardiovascular risk factors and prior coronary revascularization was observed (PCI 63.0%, coronary artery bypass grafting 18.0%). Baseline fibrinogen was 318.7 ± 107.8 mg/dl, whereas only 26.3% of subjects had elevated CRP >0.5 mg/dl. Mean on-treatment PRU was 211.6 ± 91.5 (45% with PRU ≥230), and PI was 35.9 ± 24.9%. There was no angiographic evidence of an adverse event in any study subject following PCI. A ROC curve was plotted to demonstrate the relationship between PI and high on-clopidogrel platelet reactivity defined as PRU ≥230 (area under the curve [AUC]: 0.950; p < 0.001) (Figure 37_gr1). PI level of 30.9% was determined to produce the greatest sum of sensitivity and specificity in predicting PRU ≥230 (86.5% sensitivity, 90.6% specificity; p < 0.001).

Post-procedural troponin I or troponin T was obtained in 95.2% of subjects (n = 180) with troponin I/T–defined PPMI occurring in 13.9% of subjects (Table 2). Subjects with troponin I/T–defined PPMI had higher fibrinogen levels (363.1 ± 131.0 mg/dl vs. 309.1 ± 99.6 mg/dl; p = 0.017), were older (68.2 ± 10.1 years vs. 63.0 ± 11.8 years; p = 0.040), and had higher platelet counts (243.6 ± 64.1 cells × 10³/μl vs. 218.4 ± 55.5 cells × 10³/μl; p = 0.041) compared with those without a troponin I/T–defined PPMI. There was a trend towards a greater number of stents used per procedure among the PPMI group (2.08 ± 0.91 vs. 1.68 ± 1.02; p = 0.065), whereas prevalence of elevated CRP >0.5 mg/dl (24.0% vs. 25.7%; p = 0.86), on-treatment PRU (226.0 ± 81.5 vs. 209.8 ± 92.1; p = 0.41), and PI (31.1 ± 20.3% vs. 36.2 ± 25.5%; p = 0.34) were similar between those with and without a troponin I/T–defined PPMI. The prevalence of high on-clopidogrel platelet reactivity defined as PRU ≥230 (48.0% vs. 45.2%; p = 0.79), PRU ≥208 (52.0% vs. 51.6%; p = 0.97), or PI <30% (60.0% vs. 45.2%; p = 0.17) was also similar between those with and without a troponin I/T–defined PPMI.

Fibrinogen level, age, and platelet count were entered together into a multiple variable regression model and both fibrinogen (p = 0.031) and age (p = 0.035) remained independently associated with troponin I/T–defined PPMI (Table 3). Further testing did not show an interaction between fibrinogen level and age. After on-treatment PRU and PI were separately included into multiple variable regression models with significant univariate factors (fibrinogen level, age, and platelet count), only serum fibrinogen level and age remained significantly associated with troponin I/T–defined PPMI. The aforementioned multiple variable regression testing was repeated while including other markers of systemic inflammation (CRP and white blood cell count) (Table 3). When controlling for these inflammatory markers, serum fibrinogen remained significantly associated with troponin I/T–defined PPMI, whereas age and platelet function measures did not.

CK-MB measurements were obtained in all subjects. The incidence of CK-MB–defined PPMI was 5.8% (Table 4). There was no difference in baseline characteristics, medical history, medication use, or lipid profile between subject groups with or without a PPMI. Among procedural characteristics suggesting higher PCI complexity, including treatment of type C or bifurcation lesions, only the number of stents used per procedure (2.36 ± 0.92 vs. 1.68 ± 1.02; p=0.027) was associated with occurrence of PPMI. On platelet function testing, baseline PRU (312 ± 32.7 vs. 328.2 ± 57.8; p = 0.38), on-treatment PRU (224.7 ± 96.4 vs. 210.8 ± 91.4; p = 0.62), and PI (30.2 ± 20.0% vs. 36.1 ± 25.3%; p = 0.45) were not different between the 2 groups. The prevalence of high on-clopidogrel platelet reactivity defined as PRU ≥230 (45.5% vs. 44.9%; p = 1.00), PRU ≥208 (45.5% vs. 51.7%; p = 0.69), or PI <30% (63.6% vs. 45.5%; p = 0.24) was also similar between those with and without a CK-MB–defined PPMI. However, subjects with PPMI had higher fibrinogen levels (403.4 ± 128.0 mg/dl vs. 313.5 ± 104.6 mg/dl; p = 0.007) without increase in CRP (CRP >0.5 mg/dl; 36.4% vs. 25.7%; p = 0.48). Serum fibrinogen level alone remained significantly associated with CK-MB–defined PPMI after significant univariate factors were included together with either on-treatment PRU or PI within multiple variable models. Furthermore, this relationship remained significant when other markers of systemic inflammation (CRP, platelet count, and white blood cell count) were included in the models.

A ROC curve showed a fibrinogen cutoff level of 345 mg/dl to have optimal combined sensitivity and specificity for CK-MB–defined PPMI (AUC: 0.700, p = 0.026; 72.7% sensitivity, 64% specificity, positive predictive value 11.1%, negative predictive value 97.4%) (Figure 37_gr2A). The ROC curve analysis did not yield a significant fibrinogen cutoff value to predict troponin I/T–defined PPMI (AUC: 0.590; p = 0.15). However, those with a serum fibrinogen level ≥345 mg/dl had higher occurrence of PPMI by both troponin I/T >3× ULN (20.9% vs. 9.7%; p = 0.036) and CK-MB >3× ULN (11.1% vs. 2.6%; p = 0.023) (Figure 37_gr3). Compared with dichotomous markers of high on-treatment platelet reactivity, only elevated fibrinogen level ≥345 mg/dl was associated with increased risk of PPMI defined by either troponin I/T (odds ratio [OR]: 2.5; 95% confidence interval [CI]: 1.04 to 5.77) (Figure 37_gr4) or CK-MB (OR: 4.75; 95% CI: 1.21 to 18.5) (Figure 37_gr5) definition.

A secondary analysis was performed to further evaluate the relationship between fibrinogen level, systemic inflammation, and the risk of PPMI. In the absence of systemic inflammation (CRP ≤0.5 mg/dl, 73.7% of subjects), there was a higher incidence of PPMI among those with fibrinogen ≥345 mg/dl (troponin I/T–defined: 27.8% vs. 9.4%, p = 0.007; CK-MB–defined: 16.2% vs. 1.0%, p = 0.002) (Figure 37_gr3). A ROC curve analysis in subjects without elevated CRP also showed a fibrinogen level ≥345 mg/dl had greatest sensitivity and specificity for predicting a CK-MB–defined PPMI (AUC: 0.821, p = 0.004; sensitivity 85.7%, specificity 76.1%, positive predictive value 16.2%, and negative predictive value 99.0%) (Figure 37_gr2B).

An interaction variable including elevated fibrinogen ≥345 mg/dl and CRP ≤0.5 mg/dl was entered together with fibrinogen and CRP level into multiple variable regression models to predict the occurrence of PPMI. This interaction term remained predictive of CK-MB–defined PPMI (OR: 11.6; 95% CI: 1.13 to 119) (whereas fibrinogen or CRP level alone did not), but was not predictive of troponin I/T–defined PPMI when either including or excluding age and platelet count covariates. An interaction variable between lower fibrinogen <345 mg/dl and CRP ≤0.5 mg/dl was also entered together with fibrinogen and CRP level into a multiple variable regression model to predict the absence of a CK-MB–defined PPMI, and was found to be an independent negative predictor of CK-MB–defined PPMI (OR: 0.086; 95% CI: 0.008 to 0.882) (whereas fibrinogen or CRP level alone was not). In the overall cohort, subjects with fibrinogen ≥345 mg/dl in the absence of systemic inflammation (CRP ≤0.5 mg/dl) had the highest incidence of CK-MB–defined PPMI (Figure 37_gr6), whereas those with fibrinogen <345 mg/dl in the absence of systemic inflammation had the lowest incidence of CK-MB–defined PPMI (Figure 37_gr7) compared with other variables known to affect the risk of PPMI.

This study was undertaken to determine the roles of serum fibrinogen and residual platelet reactivity after clopidogrel pre-treatment on ischemic events after elective PCI. The key findings of this study are: 1) elevated serum fibrinogen (≥345 mg/dl) is independently associated with periprocedural myocardial infarction; 2) fibrinogen level is a significant positive and negative predictor of PPMI in the absence of baseline systemic inflammation (CRP ≤0.5 mg/dl); 3) no association between on-clopidogrel residual platelet reactivity and PPMI is observed; and 4) a PRU >230 as measured with the VerifyNow assay corresponds to a platelet inhibition of <30.9%.

Using universal definitions of PPMI according to troponin I/T and CK-MB enzyme levels, elevated serum fibrinogen level was associated with higher rate of PPMI regardless of on-clopidogrel platelet reactivity using the point-of-care VerifyNow P2Y12 assay. By ROC curve analysis, fibrinogen level ≥345 mg/dl significantly predicted PPMI and identified individuals undergoing elective PCI with nearly a 5-fold increased odds of experiencing a CK-MB–defined PPMI. A secondary analysis demonstrated that a fibrinogen level of 345 mg/dl was an even stronger positive and negative predictor of PPMI among those without baseline systemic inflammation. No relationships between measures of platelet reactivity using the VerifyNow P2Y12 assay and PPMI were detected in this study, but the significance of these negative findings may be limited due to inadequate study power. Measurement of sensitive periprocedural ischemic markers of myocardial injury (CK-MB and troponin I/T) within a controlled environment, during which medication noncompliance was not an issue, add to the reliability of these findings. Furthermore, patients with acute myocardial infarction were excluded to minimize confounding variables, and no patients had an angiographic etiology for a PPMI. These findings are highly pertinent from both a mechanistic and clinical standpoint.

Mechanistically, both serum fibrinogen and activated platelets are key components of arterial thrombus formation (22). Circulating platelets are activated by soluble platelet agonists (such as ADP, epinephrine, thrombin, and collagen) causing conformational shifts of glycoprotein IIb/IIIa receptors and externalization of these receptors to the platelet cell surface. Activated receptors then bind to circulating fibrinogen and fibrin molecules, and facilitate platelet crosslinking, thrombosis, and clot formation. Platelet activation also facilitates release of platelet alpha-granules, containing additional fibrinogen and other coagulation factors, further promoting localized thrombus formation ((22),23). Due to the interplay between fibrinogen and ADP-mediated platelet activation during thrombus formation, uncertainty exists regarding the independent effects of these factors on adverse thrombotic and ischemic cardiac events. However, the results of the current study suggest that an elevated fibrinogen level ≥345 mg/dl is related to significant platelet cross-linking and thrombus formation independent of residual P2Y12 receptor-mediated platelet activity during clopidogrel therapy, and in the absence of general inflammation.

Clinically, both elevated serum fibrinogen and high on-treatment platelet reactivity signify increased risk of adverse ischemic cardiovascular events. Large epidemiological studies have shown fibrinogen to be an independent long-term risk factor for primary MACE, including death, nonfatal myocardial infarction, and stroke ((12),(13),(14),15). Other smaller studies with shorter follow-up periods have also shown fibrinogen to be an independent risk factor for repeat MACE in those with existing coronary artery disease ((6),(7),(8),(9),(10),(11),). Similarly, the GRAVITAS trial and a number of smaller studies have shown that higher platelet reactivity with clopidogrel therapy is independently associated with an increased risk of adverse thrombotic and ischemic cardiovascular events following PCI ((20),21). However, the positive predictive value of platelet reactivity measures for such events is low, and there is limited consensus on the definition of suboptimal platelet inhibition ((4),(5),(20),(21),25).

Our group ((17),26) and others ((6),27) have previously shown an association between elevated fibrinogen level and lower platelet reactivity during antiplatelet therapy. Using the VerifyNow IIb/IIIa assay, we have shown an association between fibrinogen level ≥375 mg/dl and inadequate inhibition of platelet aggregation during PCI despite single- or double-bolus administration of the potent glycoprotein IIb/IIIa inhibitor eptifibatide (26). In another study, using the VerifyNow P2Y12 assay, we demonstrated that fibrinogen level ≥375 mg/dl was associated with higher residual platelet reactivity following clopidogrel pre-treatment before PCI (17). Althoff et al. (6) used multiple electrode aggregometry (Multiplate analyzer, Dynabyte, Munich, Germany) and identified both serum fibrinogen and recent myocardial infarction as independent risk factors for subsequent elevated platelet reactivity during dual-antiplatelet therapy with aspirin and clopidogrel. Bernlochner et al. (27) used the Multiplate analyzer to demonstrate a relationship between a fibrinogen level ≥350 mg/dl and higher platelet aggregation following PCI and dual antiplatelet treatment using aspirin and clopidogrel. Recognizing the association between increased fibrinogen and platelet reactivity, the current study was designed to prospectively evaluate the roles of baseline serum fibrinogen and platelet reactivity following dual antiplatelet therapy (aspirin and clopidogrel) in predicting a short-term, adverse ischemic cardiac event following elective PCI.

In this study, a fibrinogen level ≥345 mg/dl had the greatest sensitivity and specificity after ROC curve analysis for predicting CK-MB–defined PPMI, and this level also remained significantly associated with troponin I/T–defined PPMI. Other groups, including Arnau Vives et al. (7), Toss et al. (8), and Kaski et al. (24), studied higher-risk individuals with non–ST-segment elevation acute coronary syndromes and reported fibrinogen cutoff levels of 375 mg/dl (upper tertile/ROC curve), 400 mg/dl (upper tertile), and 447 mg/dl (ROC curve) to be predictive of repeat occurrence of MACE. The ROC curve analyses from studies by Toss et al. (8), Kaski et al. (24), and the current study show similar results, with AUC ranges of 0.70 to 0.73 (sensitivity 60% to 73%, specificity 64% to 70%) and fibrinogen cutoffs possessing low positive predictive values (6.2% to 14.9%) and high negative predictive values (94.4% to 98.8%). The variation observed between our cutoff value of 345 mg/dl and the others is likely attributed to differences in study populations (lower-risk elective PCI subjects in our study) and outcomes measured. The overall predictive ability of these cutoffs suggests that depressed fibrinogen level accurately signifies a lower risk of secondary thrombotic/ischemic cardiac events, whereas elevated fibrinogen level alone does not reliably predict such events. Of note, we demonstrated an incrementally increased AUC (0.821) and predictive values (sensitivity 85.7%, specificity 76.1%, positive predictive value 16.2%, and negative predictive value 99.0%) after performing ROC curve analysis among subjects with CRP ≤0.5 mg/dl, suggesting an improved positive and negative predictor role for fibrinogen in the absence of systemic inflammation.

Platelet function testing after clopidogrel pre-treatment is similarly limited in its ability to positively predict secondary ischemic cardiac events. The current study did not show an association between measures of platelet reactivity and PPMI, though it may have been inadequately powered to evaluate these relationships. This finding is inconsistent with the results of several smaller studies that used the VerifyNow P2Y12 assay to predict adverse thrombotic cardiac outcomes following PCI for both acute coronary syndromes ((28),29) and elective indications ((30),31). The ROC curve analyses used in each of these studies resulted in PRU cutoffs of 235, 236, and 240 to predict adverse cardiac events. These analyses as a group produced ranges of AUC (0.62 to 0.71), sensitivity (60% to 81%), specificity (53% to 70%), positive predictive value (12% to 81%), and negative predictive value (91% to 99%) that indicate, similar to serum fibrinogen concentration, lower VerifyNow PRU predicts superior clinical outcomes, but higher PRU alone does not reliably predict adverse ischemic events.

In the current study, we also report a strong association between VerifyNow PRU ≥230 and PI <30.9%. This finding can be helpful in interpreting the results of studies identifying clinically significant VerifyNow PRU cutoffs and those using traditional LTA to identify clinically significant levels of PI. A number of previous studies have used LTA with a 10 μmol/l ADP agonist and ROC curve analyses to identify PI of 30% as a significant cutoff, below which individuals are at greater risk of adverse cardiac events ((32),(33),(34),35). In addition, the current study shows that although 45% of subjects have high on-clopidogrel platelet reactivity defined as PRU ≥230, a slightly larger percentage (53%) of subjects attained PI <30%, and a small portion of these individuals would be considered as appropriate clopidogrel responders using a PRU-based definition alone. Most importantly, it might not be appropriate to define cutoff values that are based on a single assay or level of platelet reactivity, but consideration should be given to reporting outcomes in future studies based on a range of inhibition of platelet reactivity as it is more generalizable and better understood by the majority of clinicians.

The current study was not designed to re-examine the relationship between fibrinogen and platelet reactivity, but rather to assess for interactions between these variables in predicting a short-term adverse cardiac event. The results of these analyses showed a relationship between fibrinogen and PPMI independent of VerifyNow P2Y12 platelet reactivity measures. A relationship between platelet reactivity and PPMI was not found in this study, but conclusions drawn from this result may be limited due to inadequate study power. Results may also be limited to the VerifyNow P2Y12 assay methodology, and other tests of platelet reactivity may yield different results. The findings of this study do not provide insight into whether the relationship between high platelet reactivity and ischemic cardiovascular events demonstrated in previous studies is a direct one or mediated through the effect of serum fibrinogen. Future studies evaluating platelet reactivity and adverse cardiovascular events should include simultaneously measured serum fibrinogen to clarify this relationship. Furthermore, though relationships between fibrinogen, low CRP, and PPMI were observed in a secondary analysis, limited conclusions should be drawn regarding lack of similar relationships in those with higher CRP. The significance of the reported fibrinogen cutoff may vary when related to different populations or outcome measurements.

In elective PCI patients, elevated serum fibrinogen is independently associated with PPMI regardless of on-treatment platelet reactivity or platelet inhibition using the VerifyNow P2Y12 assay. Fibrinogen level ≥345 mg/dl signifies an increased risk of a PPMI, especially in the absence of systemic inflammation (CRP ≤0.5 mg/dl), whereas no relationship between high platelet reactivity and PPMI is present in this elective PCI study population. Future studies relating platelet reactivity and adverse cardiac events should measure baseline fibrinogen to clarify whether such a relationship is a direct one or mediated through independent fibrinogen effects.