D-Dimer Predicts Long-Term Cause-Specific Mortality, Cardiovascular Events, and Cancer in Patients With Stable Coronary Heart Disease
- is corrected by
- Other version(s) of this article
You are viewing the most recent version of this article. Previous versions:
D-dimer, a degradation product of cross-linked fibrin, is a marker for hypercoagulability and thrombotic events. Moderately elevated levels of D-dimer are associated with the risk of venous and arterial events in patients with vascular disease. We assessed the role of D-dimer levels in predicting long-term vascular outcomes, cause-specific mortality, and new cancers in the LIPID trial (Long-Term Intervention with Pravastatin in Ischaemic Disease) in the context of other risk factors.
LIPID randomized patients to placebo or pravastatin 40 mg/d 5 to 38 months after myocardial infarction or unstable angina. D-dimer levels were measured at baseline and at 1 year. Median follow-up was 6.0 years during the trial and 16 years in total.
Baseline D-dimer levels for 7863 patients were grouped by quartile (≤112, 112–173, 173–273, >273 ng/mL). Higher levels were associated with older age, female sex, history of hypertension, poor renal function, and elevated levels of B-natriuretic peptide, high-sensitivity C-reactive protein, and sensitive troponin I (each P<0.001). During the first 6 years, after adjustment for up to 30 additional risk factors, higher D-dimer was associated with a significantly increased risk of a major coronary event (quartile 4 versus 1: hazard ratio [HR], 1.45; 95% confidence interval, 1.21–1.74), major cardiovascular disease (CVD) event (HR, 1.45; 95% confidence interval, 1.23–1.71) and venous thromboembolism (HR, 4.03; 95% confidence interval, 2.31–7.03; each P<0.001). During the 16 years overall, higher D-dimer was an independent predictor of all-cause mortality (HR, 1.59), CVD mortality (HR, 1.61), cancer mortality (HR, 1.54), and non-CVD noncancer mortality (HR, 1.57; each P<0.001), remaining significant for deaths resulting from each cause occurring beyond 10 years of follow-up (each P≤0.01). Higher D-dimer also independently predicted an increase in cancer incidence (HR, 1.16; P=0.02).The D-dimer level increased the net reclassification index for all-cause mortality by 4.0 and venous thromboembolism by 13.6.
D-dimer levels predict long-term risk of arterial and venous events, CVD mortality, and non-CVD noncancer mortality independent of other risk factors. D-dimer is also a significant predictor of cancer incidence and mortality. These results support an association of D-dimer with fatal events across multiple diseases and demonstrate that this link extends beyond 10 years’ follow-up.
Editorial, see p 724
What Is New?
A high D-dimer level is an important independent risk factor for a wide range of vascular events, cancer events, and multiple cause-specific mortalities in the long term (16 years).
What Are the Clinical Implications?
D-dimer level predicts risks of cardiovascular disease and all-cause mortality independently and, in addition to known traditional risk factors and biomarkers, should be considered in future risk nomograms and clinical decisions.
These results may also help guide management decisions about the likely absolute benefit of statin, antithrombosis, and other therapies.
D-dimer, a degradation product of cross-linked fibrin, is a marker of hypercoagulability and thrombosis.1 Moderately high levels of D-dimer have been associated with a risk of subsequent thrombotic events2–11 (venous12,13 and arterial6,8,9,11), particularly in patients with prior vascular disease,14–17 and have also been linked to a risk of cancer.18,19
The role of D-dimer level in predicting outcomes over ≥5 years in patients with stable coronary heart disease (CHD) is not as clearly defined, particularly in the context of the many other known risk factors for vascular events. The LIPID trial (Long-Term Intervention With Pravastatin in Ischaemic Disease) evaluated pravastatin treatment in 9014 patients with stable CHD and followed them up for 16 years. This gave us an opportunity for robust assessment of outcomes in a well-characterized cohort of patients whose D-dimer levels had been measured at least 5 months after their acute coronary syndromes, after the period of hypercoagulability.20–22
We aimed to assess how well D-dimer levels predict the rate of major vascular events, including CHD, cardiovascular disease (CVD), and venous thromboembolic events (VTE), during the 6 years of the trial, as well as cancer incidence and cause-specific mortality during 16 years of follow-up overall, after adjustment for other measured risk factors. We also evaluated the effects of pravastatin treatment within groups defined by baseline D-dimer levels.
Availability of Data
LIPID trial data are not automatically available to other researchers. Complete individual patient data have been provided to other research groups for meta-analysis, such as the international studies of the Cholesterol Treatment Trialists’ Collaboration. Proposals for analyses or collaborative studies by other researchers are welcome. The LIPID data set is held by the National Health and Medical Research Council Clinical Trials Centre, University of Sydney.
Study Design and Population
The design of the LIPID study has been reported.23 Patients 31 to 75 years of age with a myocardial infarction or hospital admission for unstable angina 5 to 38 months previously and normal total cholesterol and triglyceride levels were assigned to 40 mg pravastatin or placebo daily. The median follow-up was 6.0 years. All deaths, myocardial infarctions, and strokes were reviewed by blinded outcome-assessment committees. Myocardial infarction was defined as definite development of new pathological Q waves of at least 0.03 second in width in at least 2 electrocardiographic leads or 2 of the following: a history of typical ischemic pain lasting for 15 minutes and unresponsive to sublingual nitrates, elevation of creatine kinase myocardial fraction more than twice the upper limit of normal, or evolution of ST-T changes. A VTE (deep venous thrombosis or pulmonary embolus) was defined as hospitalization with a diagnosis of VTE identified by International Classification of Diseases, Ninth Revision code 415.1, 451.11, or 453.8. VTE events were based on hospital medical records and not centrally adjudicated.
After the LIPID trial finished, the trial cohort was followed up for an additional 10 years for cause-specific mortality and new cancers. Death, myocardial infarction, and stroke were adjudicated centrally for a further 2 years (first 8 years in total). Thereafter, data for death, cause of death, and the incidence of new cancers were obtained by direct follow-up and from national death registries and state-based cancer registries in Australia and New Zealand, as previously reported.20,24
A total of 7863 patients from the 9014 LIPID patients (87%) had baseline plasma samples available for measurement of biomarkers. Of these, 6754 of the 6889 alive at the end of the trial consented to additional long-term follow-up (Figure I in the online-only Data Supplement).
The LIPID study was approved by the ethics committee at each participating center. All patients gave written informed consent for the trial and separately for further clinic or remote follow-up.
Blood was drawn after a 12-hour fast into EDTA tubes, and plasma samples were stored in freezers at −70°C until analysis. D-dimer levels (nanograms per milliliter) were measured at randomization and at 1 year by microparticle enzyme immunoassay (Abbott Diagnostics) in the MORGAM (MONICA [Multinational Monitoring of Trends and Determinants in Cardiovascular Disease], Risk, Genetics, Archiving, and Monograph) biomarker laboratory.11
Additional biomarkers in stored plasma samples were chosen to reflect the range of pathobiological processes considered important in the development of atherothrombotic disease: B-type natriuretic peptide (BNP; hemodynamic), sensitive troponin I (myocardial micronecrosis), cystatin C (renal function), high-sensitivity C-reactive protein (hs-CRP; inflammation), lipoprotein-associated phospholipase A2 activity (inflammation), midregional proadrenomedullin (humoral), and lipoprotein(a) (lipids). They were analyzed centrally in the MORGAM biomarker laboratory, as previously reported.11,25
Analyses for outcomes to 6 years were prespecified in a biomarker protocol with the composite of CHD death and nonfatal myocardial infarction (CHD events) as the primary end point.25 Other end points were major CVD events (CVD death, nonfatal myocardial infarction, and stroke), stroke, CHD death, CVD death, and total mortality. End points for additional analyses were cause-specific mortality and the incidence of patients with new cancer over 16 years. The data for D-dimer were grouped by quartile with baseline cut points of ≤112, >112 to 173, >173 to 273, and >273 ng/mL for quartiles 1 through 4, respectively. Increases in D-dimer levels between baseline and 1 year were also grouped by quartile (≤ −32, > −32 to 1, >1 to 36, and >36 ng/mL). The effect of pravastatin on change in D-dimer was examined with the Wilcoxon rank-sum test.
For analysis of continuous variables, tests for linear trend over the D-dimer quartiles used a generalized linear model, incorporating the median D-dimer value. For categorical variables, ordinal or logistic regression was used.
The relationships between D-dimer levels and outcomes for the first 6 years (main trial period) were assessed with prespecified Cox regression models with adjustment for treatment and traditional risk factors: age, sex, prior stroke, diabetes mellitus, current smoking, hypertension, fasting glucose, total cholesterol, apolipoprotein B, apolipoprotein A1, high-density lipoprotein cholesterol, triglycerides, nature of qualifying prior acute coronary syndrome, timing of coronary revascularization, systolic blood pressure, atrial fibrillation, estimated glomerular filtration rate, body mass index, dyspnea class, angina grade, white blood cell count, peripheral vascular disease, and use of aspirin at baseline. Each of these had been shown to be an independent predictor of CVD events in risk models in LIPID.26–28 All analyses were also adjusted for baseline anticoagulant treatment, which is known to modify D-dimer levels. The assumptions of proportional hazards were met for these models. Event rates were calculated as Kaplan-Meier estimates, except for cancer incidence, for which Fine and Gray estimates accounting for competing risks were used.29
In addition, backward selection techniques were used to produce models with adjustment for the significant traditional risk factors listed above, as well as significant biomarkers among BNP, hs-CRP, cystatin C, lipoprotein(a), sensitive troponin I, midregional proadrenomedullin, and lipoprotein-associated phospholipase A2 activity.25 The analysis of VTE events used a multivariate Cox regression model adjusted for risk factors related to VTE events on univariate analyses. These variables were baseline D-dimer quartiles, sex, age, nature of prior acute coronary syndrome, timing of coronary revascularization, body mass index (kilograms per meter squared), dyspnea class, angina grade, stroke, peripheral vascular disease, and aspirin. For analyses of long-term follow-up data (median, 16 years), similar Cox regression methods were applied but with the following modifications. First, the outcomes assessed were cause-specific mortality and the incidence of new cancers. Second, to meet the assumption of proportional hazards, treatment was included in the Cox model partitioned into 2 periods: the 6 years of the main trial and the additional 10 years of extended follow-up. Third, the variables selected for adjusting the effects of D-dimer levels were chosen according to other variables that were significant for non-CVD mortality, cancer mortality, or cancer incidence. Each model of significant risk factors was built with backward selection (P<0.05) but always retaining D-dimer, anticoagulant treatment, and assignment to pravastatin or placebo. The association of D-dimer and subsequent events was also examined with 5-year bands (0–5, 6–10, and >10 years), including D-dimer as a time-varying covariate, to assess whether the strength of association varied over time. In addition, a sensitivity analysis assessed the relative risk of D-dimer levels on all-cause mortality in 50-ng/mL groups and as a continuous variable using a restricted cubic spline with the R package rms.30
The relationship between change in D-dimer to 1 year and subsequent events was assessed in a landmark analysis of outcomes for 6715 patients who survived to 1 year. These analyses were also adjusted for significant risk factors found by backward selection and baseline D-dimer levels. P values for trend were calculated by use of the median D-dimer level in each quartile.
Discrimination of each risk model was assessed with the net reclassification index (NRI), a measure of correct movement into other risk categories, upward for those who have events and downward for those without events, minus incorrect reclassification. The index has a maximum of 200 and was calculated from Kaplan-Meier probabilities at 5 years continuously and with the following cut points; 0.6%, 1%, and 1.7% for VTE and 5%, 8%, and 12% for all-cause mortality. For the landmark analysis including data up to 16 years, the event probabilities were calculated at 15 years and were 22%, 34%, and 51% for all-cause mortality. The base model for each biomarker—D-dimer, BNP, and troponin I—included the significant traditional risk factors detailed above, treatment, and anticoagulation use. In addition, the NRIs for age and sex were calculated by use of a base model with these variables removed. The 95% confidence intervals (CIs) were calculated by bootstrapping with 1000 replications. No adjustments were made for multiple comparisons, and analyses used SAS 9.3 (SAS Institute Inc, Cary, NC) and R (R Core Team, Vienna, Austria). All authors had access to the study data.
Baseline Characteristics and D-Dimer Levels
Participants with higher D-dimer levels were significantly older, were more often female, and had a higher risk score for coronary events (Table 1). They were also more often taking CVD medicines and were more likely to have a history of hypertension, lower renal function, and other elevated biomarkers, including BNP, troponin I, cystatin C, and hs-CRP (each P<0.001).
|Characteristic||D-Dimer Quartile||P for Trend†|
|≤112 ng/mL (n=1992)||112–173 ng/mL (n=1968)||173–273 ng/mL (n=1941)||>273 ng/mL (n=1962)|
|D-dimer level, ng/mL||80.7 (20.6)||141.7 (17.5)||217.4 (27.4)||613.8 (402.9)|
|Age at randomization, y||58.0 (51.0–64.0)||62.0 (56.0–67.0)||64.0 (58.0–68.0)||65.0 (59.0–69.0)||<0.001|
|Female, n (%)||226 (11)||355 (18)||379 (20)||373 (19)||<0.001|
|Time from qualifying event, mo||14.3 (8.0–25.2)||14.5 (8.0–25.1)||14.1 (8.0–25.0)||13.0 (7.7–24.6)||0.08|
|Current smoker, n (%)||180 (9)||193 (10)||165 (9)||197 (10)||0.36|
|Hypertension, n (%)||777 (39)||797 (40)||826 (43)||891 (45)||<0.001|
|Diabetes mellitus, n (%)||147 (7)||179 (9)||172 (9)||178 (9)||0.17|
|Obesity (BMI ≥30 kg/m2), n (%)||325 (16)||351 (18)||361 (19)||360 (18)||0.17|
|BMI, median (IQR), kg/m2||26.3 (24.2–28.7)||26.5 (24.4–28.9)||26.5 (24.3–29.1)||26.3 (24.0–29.0)||0.73|
|Previous stroke, n (%)||54 (3)||73 (4)||85 (4)||110 (6)||<0.001|
|Systolic blood pressure, mm Hg||132 (19)||134 (19)||134 (19)||136 (20)||<0.001|
|Diastolic blood pressure, mm Hg||81 (11)||80 (11)||80 (11)||81 (11)||0.22|
|History of atrial fibrillation, n (%)||21 (1)||28 (1)||24 (1)||37 (2)||0.03|
|Baseline lipids, mmol/L|
|Total cholesterol||5.6 (0.8)||5.7 (0.8)||5.7 (0.8)||5.6 (0.8)||0.11|
|LDL cholesterol||3.9 (0.7)||3.9 (0.8)||3.9 (0.7)||3.9 (0.7)||0.92|
|HDL cholesterol||0.96 (0.23)||0.96 (0.24)||0.95 (0.24)||0.95 (0.24)||0.08|
|Triglycerides||1.6 (1.1–2.2)||1.6 (1.2–2.2)||1.6 (1.2–2.2)||1.5 (1.2–2.1)||0.02|
|Previous coronary revascularization, n (%)|
|No revascularization||1158 (58)||1165 (59)||1164 (60)||1123 (57)||0.17|
|PCI only||256 (13)||219 (11)||184 (9)||211 (11)|
|CABG only||512 (26)||522 (27)||542 (28)||565 (29)|
|PCI and CABG||66 (3)||62 (3)||51 (3)||63 (3)|
|Qualifying event, n (%)|
|No myocardial infarction||708 (36)||705 (36)||712 (37)||718 (37)||0.82|
|Single myocardial infarction||1093 (55)||1025 (52)||995 (51)||1002 (51)|
|Multiple myocardial infarction||191 (10)||238 (12)||234 (12)||242 (12)|
|Medication, n (%)|
|Aspirin||1661 (83)||1637 (83)||1605 (83)||1598 (81)||0.08|
|Angiotensin-converting enzyme inhibitors||244 (12)||287 (15)||331 (17)||392 (20)||<0.001|
|β-Blocker||975 (49)||939 (48)||921 (47)||856 (44)||<0.001|
|Calcium antagonist||596 (30)||658 (33)||698 (36)||736 (38)||<0.001|
|Anticoagulant||73 (4)||34 (2)||28 (1)||43 (2)||0.07|
|Warfarin||73 (4)||34 (2)||28 (1)||42 (2)||0.05|
|LIPID risk score26|
|Mean risk score||5.2 (3.2)||5.8 (3.5)||6.1 (3.5)||6.3 (3.6)||<0.001|
|Baseline biomarker concentrations*|
|Estimated glomerular filtration rate, mL/min||73 (64–83)||70 (61–80)||68 (58–79)||67 (57–79)||<0.001|
|White blood cell count, ×109/L||6.8 (5.9–8.1)||7.1 (6.0–8.2)||7.0 (6.0–8.3)||7.1 (6.1–8.3)||<0.001|
|hs-CRP||1.8 (0.9–3.4)||2.3 (1.2–4.2)||2.8 (1.4–5.1)||3.3 (1.6–7.0)||<0.001|
|Sensitive troponin I, mg/L||0.009 (0.006–0.020)||0.011 (0.006–0.020)||0.010 (0.006–0.022)||0.012 (0.006–0.023)||0.004|
|Sensitive troponin not detectable, n (%)||818 (41)||747 (38)||733 (38)||669 (34)||<0.001|
|Lipoprotein(a), nmol/L||12.8 (6.2–41.5)||12.9 (6.5–41.2)||13.9 (6.9–45.6)||15.8 (7.1–48.6)||<0.001|
|Lp-PLA2 activity, nmol/min/mL||259 (227–291)||259 (229–29.21)||264 (230–297)||263 (231–296)||<0.001|
|BNP, pg/mL||18.3 (8.2–38.6)||23.4 (10.1–47.4)||26.3 (10.9–54.0)||28.4 (10.8–63.5)||<0.001|
|Midregional proadrenomedullin, nmol/L||0.44 (0.37–0.52)||0.48 (0.39–0.57)||0.49 (0.40–0.61)||0.50 (0.37–0.62)||<0.001|
D-Dimer Levels and Clinical Outcomes During 6 Years of the LIPID Trial
A higher baseline D-dimer level was strongly associated with a risk of subsequent CVD events (Table 2). Five-year rates are shown for comparison with other trials and clinical prediction models. After adjustment for all significant traditional CVD risk factors previously identified in LIPID25–28 and novel biomarkers, a level of D-dimer in quartile 4 compared with quartile 1 was associated with a higher risk of CHD events, major CVD events, total CVD events, and death resulting from any cause (all P<0.001).
|End Point and D-Dimer Level, ng/mL||Events, n/N||5-y Event Rate (95% CI), %||Adjusted I* HR (95% CI)||P forTrend||Adjusted II† HR (95% CI)||P forTrend|
|112–173||266/1968||11.2 (9.9–12.7)||1.41 (1.17–1.69)||1.18 (0.98–1.42)|
|173–273||280/1941||12.6 (11.2–14.1)||1.55 (1.29–1.86)||1.20 (0.99–1.44)|
|>273||351/1962||15.0 (13.5–16.7)||1.97 (1.66–2.35)||1.45 (1.21–1.74)|
|112–173||58/1968||2.6 (2.0–3.4)||1.17 (0.80–1.70)||0.88 (0.60–1.28)|
|173–273||83/1941||3.3 (2.6–4.2)||1.73 (1.22–2.45)||1.08 (0.75–1.56)|
|>273||117/1962||5.3 (4.3–6.4)||2.54 (1.83–3.52)||1.37 (0.96–1.95)|
|Major CVD events§|
|112–173||311/1968||13.1 (11.7–14.7)||1.32 (1.12–1.56)||1.08 (0.91–1.28)|
|173–273||352/1941||15.4 (13.9–17.1)||1.57 (1.33–1.84)||1.14 (0.96–1.36)|
|>273||471/1962||20.4 (18.6–22.3)||2.16 (1.85–2.51)||1.45 (1.23–1.71)|
|Total CVD events‖|
|112–173||742/1968||32.6 (30.6–34.7)||1.17 (1.05–1.30)||1.07 (0.96–1.18)|
|173–273||771/1941||34.8 (32.7–37.0)||1.27 (1.15–1.41)||1.09 (0.98–1.21)|
|>273||858/1962||39.4 (37.2–41.6)||1.45 (1.31–1.61)||1.20 (1.08–1.33)|
|112–173||152/1968||6.2 (5.2–7.3)||2.20 (1.67–2.90)||1.59 (1.20–2.10)|
|173–273||161/1941||7.0 (6.0–8.3)||2.43 (1.85–3.20)||1.46 (1.10–1.94)|
|>273||268/1962||11.0 (9.6–12.5)||4.12 (3.19–5.32)||2.23 (1.70–2.93)|
|112–173||214/1968||8.2 (7.1–9.5)||1.83 (1.47–2.28)||1.34 (1.08–1.68)|
|173–273||244/1941||9.9 (8.6–11.3)||2.17 (1.75–2.69)||1.35 (1.08–1.69)|
|>273||387/1962||15.5 (14.0–17.2)||3.52 (2.88–4.30)||1.99 (1.61–2.46)|
|112–173||25/1968||0.8 (0.5–1.4)||1.67 (0.89–3.13)||1.60 (0.85–3.00)|
|173–273||30/1941||1.3 (0.9–2.0)||2.06 (1.12–3.79)||1.91 (1.04–3.52)|
|>273||62/1962||2.9 (2.2–3.8)||4.36 (2.51–7.57)||4.03 (2.31–7.03)|
Baseline D-dimer in quartile 4 compared with quartile 1 was associated with a higher risk of VTE, overall and separately for deep venous thrombosis (quartile 4 versus 1: hazard ratio [HR], 3.35; P for trend=0.003) and pulmonary embolism (quartile 4 versus 1: HR, 5.31; P fortrend <0.001). The higher risk of VTE remained after adjustment for other risk factors (Table 2). These risk factors were dyspnea grade >1, anticoagulation use, a history of peripheral vascular disease, and no coronary revascularization after the acute coronary event. In a sensitivity analysis using restricted cubic splines, the association of D-dimer and all-cause mortality appeared to be nonlinear: higher baseline D-dimer levels were associated with higher mortality up to ≈400 ng/mL but with a less strong relationship above this level (Figure II in the online-only Data Supplement).
The NRI over the main trial period, when D-dimer was added to the risk model, was 4.0 for all-cause mortality and 13.6 for VTE events. In comparison, the NRIs for all-cause mortality were 8.20 for BNP and 3.56 for sensitive troponin I and for VTE were 0.01 for BNP and 3.57 for sensitive troponin I. Changes in the C statistic were relatively small for most variables21 with the exception of D-dimer for VTE (from 0.716 1 to 0.742) and age for long-term all-cause mortality (from 0.649–0.701) (Table I in the online-only Data Supplement).
Pravastatin Treatment, D-Dimer, and Clinical Outcomes
Pravastatin treatment compared with placebo was associated with a small but significant reduction in D-dimer levels from baseline to 1 year (P<0.001), as reported previously.25 However, this change did not explain any of the effect of pravastatin in reducing CHD events, which remained highly significant after adjustment for changes in D-dimer (P<0.01).
As previously reported, pravastatin significantly reduced CVD end points.21 It did not affect the rate of VTE overall or within each quartile (Table II in the online-only Data Supplement). The relative reduction in CVD events was similar in each quartile of baseline D-dimer levels, but absolute benefits were larger among patients in the higher D-dimer quartiles.
D-Dimer Levels, Cause-Specific Mortality, and Cancer During 16 Years
Higher D-dimer levels were associated with a significant and independent higher risk of cause-specific mortality (death from CHD, CVD, cancer, or non-CVD noncancer) after adjustment for all significant traditional CVD risk factors and novel biomarkers (each P<0.001; Table 3).
|D-Dimer Level, ng/mL||Events, n/N||15-y Event Rate, % (95% CI)||HR (95% CI)*||P for Trend*|
|Non-CVD noncancer mortality|
The risk prediction of D-dimer levels over time is illustrated in the Figure, in which mortality rates for quartile groups continue to diverge well after 10 years of follow-up. This is illustrated further in Table III in the online-only Data Supplement, which shows that a higher D-dimer level was associated with a higher risk of cause-specific mortality for those deaths occurring beyond 10 years within each major category: quartile 4 versus 1, for CVD death (HR, 1.59; 95% CI, 1.29–1.96; P=0.004), for cancer death (HR, 1.58; 95% CI, 1.15–2.15; P=0.01), for non-CVD noncancer death (HR, 1.70; 95% CI, 1.28–2.26; P=0.004), and for all-cause mortality (HR, 1.65; 95% CI, 1.43–1.92; P<0.001).
Higher baseline levels of D-dimer (quartile 4 versus 1) were associated with a significantly higher rate of new cancers (Table 4) and remained a moderate risk factor for new cancers after adjustment for other significant cancer risk factors: age, sex, smoking, hs-CRP, white blood cell count, and aspirin use.
|End Point and D-Dimer Level (ng/mL)||Events, n/N||15-y Event Rate‡||Adjusted I* HR (95% CI)||P for Trend||Adjusted II† HR (95% CI)||P for Trend|
|112–173||471/1968||23.1 (21.2–25.8)||1.33 (1.18–1.49)||1.09 (0.97–1.23)|
|173–273||492/1941||23.9 (22.0–25.8)||1.44 (1.28–1.62)||1.08 (0.96–1.22)|
|>273||502/1962||24.5 (22.6–26.4)||1.59 (1.42–1.79)||1.16 (1.03–1.31)|
|112–173||181/1968||9.9 (8.6–11.5)||1.37 (1.11–1.70)||1.07 (0.86–1.33)|
|173–273||244/1941||13.7 (12.0–15.5)||1.98 (1.62–2.43)||1.39 (1.13–1.72)|
|>273||254/1962||15.2 (13.4–17.1)||2.28 (1.86–2.79)||1.54 (1.25–1.91)|
Change in D-Dimer Levels and Mortality
An increase in D-dimer between baseline and 1 year was associated with an increase in the risk of death, death resulting from CVD, and death from non-CVD noncancer causes over 16 years, but not in the risk of CHD mortality, when adjusted for baseline D-dimer and other significant risk factors (Table IV in the online-only Data Supplement). Conversely, an increase in D-dimer from baseline was associated with a reduction in the risk of death when adjusted by the D-dimer level at 1 year (data not shown). Hence, the prognostic value of the increase in D-dimer from baseline appears to be related to providing a slightly better estimate of the long-term average D-dimer level. This is also reflected in the higher NRI for all-cause mortality over 16 years, which was 4.4 on the basis of 2 measurements of D-dimer compared with 3.1 for 1 measurement.
In patients with stable CHD, higher D-dimer levels at baseline are significantly related to the risk of vascular events, both arterial and venous, over 6 years. This association is independent of traditional risk factors for CVD and for venous thromboembolic disease, as well as more recently identified biomarkers, including BNP, hs-CRP, troponin I, cystatin C, and midregional proadrenomedullin. Furthermore, D-dimer remains a highly significant and sustained predictor of CVD and all-cause mortality over 16 years, making this an additional biomarker to consider in risk stratification and clinical decisions.
In our previous analysis, D-dimer was one of several biomarkers to predict CVD events over the 6 years of the main trial.25 In that setting, other factors such as BNP and troponin I contributed even more to net reclassification than D-dimer. However, for long-term fatal outcomes, D-dimer was particularly important. The most striking aspect of our new results is the impact of D-dimer in predicting diverse cause-specific deaths—from CVD, cancer, and other causes—over a long period. These associations remain strong >10 years after the initial D-dimer reading and may relate to a long-term risk of thrombotic events in association with many pathogenic pathways, including atherosclerosis, cancer progression, and other inflammatory and infectious disease processes. Many studies have described associations of D-dimer with some of these outcomes2–19 but not with the range of outcomes, duration of effect, and level of independence seen here. A particular strength is that these findings remained significant after the inclusion of most other important biomarkers, including BNP, CRP, cystatin C, midregional proadrenomedullin, and sensitive troponin I (albeit with a non–high- sensitivity troponin I assay).25
The association between VTE and arterial disease events has been well described, raising the possibility of at least some common pathogenic pathways or common underlying risk factors.31–34 Common risk factors associated with arterial and venous disease include age, obesity, diabetes mellitus, and metabolic syndrome. For our population of patients with stable coronary disease, common independent risk factors identified for both arterial and venous events included a history of peripheral artery disease and dyspnea, but the most important common risk factor was D-dimer level. This may reflect a prothrombotic state that increases susceptibility to a major thrombotic event when there are additional underlying risk factors. There is also evidence that high D-dimer levels reflect an inflammatory state: in the LIPID study, D-dimer correlated with levels of hs-CRP but remained a highly significant predictor even after adjustment for hs-CRP.
The use of anticoagulation at baseline and during the trial is unlikely to have confounded these results. Because anticoagulation may lower D-dimer levels, all analyses were adjusted for such use at baseline. Higher baseline D-dimer levels were associated with a greater uptake of anticoagulant therapy during the trial (related mainly to subsequent thrombotic events), which may have slightly attenuated the strength of association with fatal events had such treatment not been used.
D-dimer levels were associated with risk not only of CVD death but also separately of cancer death and non-CVD noncancer death. The link between D-dimer levels, thrombosis, and cancer events has been well described.35 Studies have found a higher rate of occult cancers among patients with a VTE event and high D-dimer levels, and a higher D-dimer level has been associated with poorer prognosis and death in patients with cancer.19 This may relate to the hypothesis that hypercoagulability is associated with cancer and is consistent with some observations that anticoagulant treatment slows cancer progression.36
D-dimer may be a marker of coagulation activation, thrombin generation, and fibrin formation, which have been implicated in angiogenesis, tumor cell invasion, tumor progression, and metastatic spread.18 Furthermore, patients with many cancers are at increased risk of VTE: D-dimer may be linked to worse outcomes in cancer, both its progression and subsequent thrombotic adverse and fatal complications. In LIPID, higher D-dimer levels were associated with an increase in cancer incidence (possibly linked to occult cancers) and to cancer deaths over 16 years. These results appear across multiple cancers and are not limited to 1 or few cancer types.
The interrelation between thrombosis, inflammation, and cancer is well recognized but is very complex.37 There appear to be several shared molecular pathways in common in the pathogenesis of atherosclerosis and cancer, including inflammation, oxidative stress, and uncontrolled cell proliferation,38,39 as well as common risk factors for CVD and cancer.40 D-dimer as a marker of thrombosis may be reflecting increased activity in a variety of these complex processes, which may be a result of, rather than a cause of, such common pathways.
The link between D-dimer levels and non-CVD noncancer mortality over 16 years is novel and intriguing. It is presumably related to thrombosis and related pathogenic pathways being common in many disease processes beyond CVD and cancer. Alternatively, it could relate to patients with other diseases being misclassified as dying of a disease when death is in fact the result of an unclassified thrombosis event (eg, an unrecognized fatal pulmonary embolus). The causes of death in the long-term follow-up of LIPID beyond 8 years relied mainly on death registry classification, which has >90% accuracy, as shown by our earlier validation study linking death registry data to centrally adjudicated review.24 However, there was some lessening in the strength of association of D-dimer with deaths resulting from CVD that was not seen with deaths resulting from non-CVD noncancer causes, making this at least part of the explanation.
The effects of pravastatin on reducing CVD events appear to be related primarily to reductions in low-density lipoprotein cholesterol, although other effects such as reduced inflammation (eg, reflected by lipoprotein-associated phospholipase A2 activity) also appear to play a role.41 Pravastatin was shown to moderately reduce D-dimer levels, but D-dimer does not account for any of the clinical benefit of pravastatin.
The relative effect of pravastatin on clinical event reduction was similar for each baseline quartile of D-dimer level, but absolute benefits were appreciably greater in the top quartile; in other words, the number needed to treat to prevent an event was lower in the top quartile. This points to a potentially useful clinical role for D-dimer in prognosis and in guiding treatment choices.
Patients with high D-dimer levels may be most likely to benefit from new anticoagulant treatments that act on this pathophysiological pathway. Trial evidence shows that compared with placebo, these agents reduce ischemic events42 and venous thromboembolism,43 but they increase the risk of bleeding,42,43 indicating a need to target patients most likely to benefit. Such a strategy is worth considering in relation to a broad range of patients beyond those with prior CHD such as in the ADVICE trial (Attenuation of D-Dimer Using Vorapaxar to Target Inflammatory and Coagulation Endpoints; NCT02394730), which is testing this question in patients with HIV. However, the association of D-dimer with these outcomes is not necessarily part of a causal pathway, and it will be important that such trials are undertaken to confirm or refute potential benefit.
Pravastatin did not reduce the rate of VTE. This is in contrast to observational studies of statins and to the JUPITER trial (Justification for the Use of Statins in Prevention: An Intervention Trial Evaluating Rosuvastatin) of rosuvastatin44,45 but is consistent with the more definitive meta-analysis of >150 000 patients randomized to statins46,47 and more recent studies.48,49
Despite the striking and sustained prediction over 16 years in a well-characterized and carefully assessed cohort, some caution is needed in generalizing these results more widely, and validation in other long-term cohorts would be of value. We did not adjust for other biomarkers of hypercoagulability such as fibrinogen or prothrombin fragment 1+2 because they were not measured. Factors associated with VTE other than D-dimer in our model may represent overfit. In addition, on-study VTE data were not adjudicated. Nevertheless, compared with results of observational cohorts,50 data on biomarkers such as D-dimer from randomized trials like this may be conservative and less likely to overestimate associations than smaller studies.51
In patients with stable coronary disease, a high D-dimer level is an important independent and sustained risk factor for a wide range of vascular events, cancer events, and multiple cause-specific mortalities in the long term. It predicts risk independently and in addition to known traditional risk factors and biomarkers and could be considered in future risk nomograms for a variety of outcomes, with the potential to guide management decisions.
The authors thank Dr Tim Brighton, Haematology South Eastern Area Laboratory Services, (SEALS), for hematology review and comments. Rhana Pike from the NHMRC Clinical Trials Centre assisted with the manuscript.
Sources of Funding
The LIPID study was supported by a research grant from Bristol-Myers Squibb and was conducted under the auspices of the National Heart Foundation of Australia. This biomarker study was supported by a grant from the National Health and Medical Research Council (1037786).
Dr White reports grants from Sanofi Aventis, Eli Lilly, the National Institutes of Health, Omthera Pharmaceuticals, Elsai Inc, and DalGen Products and Services; grants and personal fees from Pfizer; personal fees and nonfinancial support from AstraZeneca; and personal fees from Sirtex and Acetelion. Dr Sullivan reports grants and other support from Amgen and Merck and grants from Sanofi, Eli Lily, and Pfizer. Dr Keech reports travel grants or honoraria from Abbott, Amgen, AstraZeneca, Bristol-Myers Squibb, Eli Lilly, Merck, Novartis, Pfizer, Roche Diagnostics, and Solvay. Dr Tonkin reports research funding support from Bayer and honoraria for advisory board participation or speaker fees from Aventis, Boehringer Ingelheim, Pfizer, and Sanofi. The other authors report no conflicts.
Meade TW, Mellows S, Brozovic M, Miller GJ, Chakrabarti RR, North WR, Haines AP, Stirling Y, Imeson JD, Thompson SG. Haemostatic function and ischaemic heart disease: principal results of the Northwick Park Heart Study.Lancet. 1986; 2:533–537.CrossrefMedlineGoogle Scholar
Cushman M, Lemaitre RN, Kuller LH, Psaty BM, Macy EM, Sharrett AR, Tracy RP. Fibrinolytic activation markers predict myocardial infarction in the elderly: the Cardiovascular Health Study.Arterioscler Thromb Vasc Biol. 1999; 19:493–498.LinkGoogle Scholar
Danesh J, Whincup P, Walker M, Lennon L, Thomson A, Appleby P, Rumley A, Lowe GD. Fibrin D-dimer and coronary heart disease: prospective study and meta-analysis.Circulation. 2001; 103:2323–2327.LinkGoogle Scholar
Di Castelnuovo A, de Curtis A, Costanzo S, Persichillo M, Olivieri M, Zito F, Donati MB, de Gaetano G, Iacoviello L; MOLI-SANI Project Investigators. Association of D-dimer levels with all-cause mortality in a healthy adult population: findings from the MOLI-SANI study.Haematologica. 2013; 98:1476–1480. doi: 10.3324/haematol.2012.083410.CrossrefMedlineGoogle Scholar
Folsom AR, Delaney JA, Lutsey PL, Zakai NA, Jenny NS, Polak JF, Cushman M; Multiethnic Study of Atherosclerosis Investigators. Associations of factor VIIIc, D-dimer, and plasmin-antiplasmin with incident cardiovascular disease and all-cause mortality.Am J Hematol. 2009; 84:349–353. doi: 10.1002/ajh.21429.CrossrefMedlineGoogle Scholar
Lowe GD, Rumley A. Use of fibrinogen and fibrin D-dimer in prediction of arterial thrombotic events.Thromb Haemost. 1999; 82:667–672.CrossrefMedlineGoogle Scholar
Lowe GD, Yarnell JW, Rumley A, Bainton D, Sweetnam PM. C-reactive protein, fibrin D-dimer, and incident ischemic heart disease in the Speedwell study: are inflammation and fibrin turnover linked in pathogenesis?Arterioscler Thromb Vasc Biol. 2001; 21:603–610.LinkGoogle Scholar
Ridker PM, Hennekens CH, Cerskus A, Stampfer MJ. Plasma concentration of cross-linked fibrin degradation product (D-dimer) and the risk of future myocardial infarction among apparently healthy men.Circulation. 1994; 90:2236–2240.LinkGoogle Scholar
Wannamethee SG, Whincup PH, Shaper AG, Rumley A, Lennon L, Lowe GD. Circulating inflammatory and hemostatic biomarkers are associated with risk of myocardial infarction and coronary death, but not angina pectoris, in older men.J Thromb Haemost. 2009; 7:1605–1611. doi: 10.1111/j.1538-7836.2009.03574.x.CrossrefMedlineGoogle Scholar
Zakai NA, Katz R, Jenny NS, Psaty BM, Reiner AP, Schwartz SM, Cushman M. Inflammation and hemostasis biomarkers and cardiovascular risk in the elderly: the Cardiovascular Health Study.J Thromb Haemost. 2007; 5:1128–1135. doi: 10.1111/j.1538-7836.2007.02528.x.CrossrefMedlineGoogle Scholar
Blankenberg S, Zeller T, Saarela O, Havulinna AS, Kee F, Tunstall-Pedoe H, Kuulasmaa K, Yarnell J, Schnabel RB, Wild PS, Münzel TF, Lackner KJ, Tiret L, Evans A, Salomaa V; MORGAM Project. Contribution of 30 biomarkers to 10-year cardiovascular risk estimation in 2 population cohorts: the MONICA, Risk, Genetics, Archiving, and Monograph (MORGAM) biomarker project.Circulation. 2010; 121:2388–2397. doi: 10.1161/CIRCULATIONAHA.109.901413.LinkGoogle Scholar
Cushman M, Folsom AR, Wang L, Aleksic N, Rosamond WD, Tracy RP, Heckbert SR. Fibrin fragment D-dimer and the risk of future venous thrombosis.Blood. 2003; 101:1243–1248. doi: 10.1182/blood-2002-05-1416.CrossrefMedlineGoogle Scholar
Halaby R, Popma CJ, Cohen A, Chi G, Zacarkim MR, Romero G, Goldhaber SZ, Hull R, Hernandez A, Mentz R, Harrington R, Lip G, Peacock F, Welker J, Martin-Loeches I, Daaboul Y, Korjian S, Gibson CM. D-dimer elevation and adverse outcomes.J Thromb Thrombolysis. 2015; 39:55–59. doi: 10.1007/s11239-014-1101-6.CrossrefMedlineGoogle Scholar
Kleinegris MC, ten Cate H, ten Cate-Hoek AJ. D-dimer as a marker for cardiovascular and arterial thrombotic events in patients with peripheral arterial disease: a systematic review.Thromb Haemost. 2013; 110:233–243. doi: 10.1160/TH13-01-0032.CrossrefMedlineGoogle Scholar
Komarov A, Panchenko E, Dobrovolsky A, Karpov Y, Deev A, Titaeva E, Davletov K, Eshkeeva A, Markova L. D-dimer and platelet aggregability are related to thrombotic events in patients with peripheral arterial occlusive disease.Eur Heart J. 2002; 23:1309–1316.CrossrefMedlineGoogle Scholar
Morange PE, Bickel C, Nicaud V, Schnabel R, Rupprecht HJ, Peetz D, Lackner KJ, Cambien F, Blankenberg S, Tiret L; AtheroGene Investigators. Haemostatic factors and the risk of cardiovascular death in patients with coronary artery disease: the AtheroGene study.Arterioscler Thromb Vasc Biol. 2006; 26:2793–2799. doi: 10.1161/01.ATV.0000249406.92992.0d.LinkGoogle Scholar
Moss AJ, Goldstein RE, Marder VJ, Sparks CE, Oakes D, Greenberg H, Weiss HJ, Zareba W, Brown MW, Liang CS, Lichstein E, Little WC, Gillespie JA, Van Voorhees L, Krone RJ, Bodenheimer MM, Hochman J, Dwyer EM, Arora R, Marcus FI, Watelet LF, Case RB. Thrombogenic factors and recurrent coronary events.Circulation. 1999; 99:2517–2522.LinkGoogle Scholar
Ay C, Dunkler D, Pirker R, Thaler J, Quehenberger P, Wagner O, Zielinski C, Pabinger I. High D-dimer levels are associated with poor prognosis in cancer patients.Haematologica. 2012; 97:1158–1164. doi: 10.3324/haematol.2011.054718.CrossrefMedlineGoogle Scholar
Han D, Hartaigh BO, Lee JH, Cho IJ, Shim CY, Chang HJ, Hong GR, Ha JW, Chung N. Impact of D-dimer for prediction of incident occult cancer in patients with unprovoked venous thromboembolism.PloS One. 2016; 11:10. doi: 10.1371/journal.pone.0153514.Google Scholar
Hague WE, Simes J, Kirby A, Keech AC, White HD, Hunt D, Nestel PJ, Colquhoun DM, Pater H, Stewart RA, Sullivan DR, Thompson PL, West M, Glasziou PP, Tonkin AM; LIPID Study Investigators. Long-term effectiveness and safety of pravastatin in patients with coronary heart disease: sixteen years of follow-up of the LIPID study.Circulation. 2016; 133:1851–1860. doi: 10.1161/CIRCULATIONAHA.115.018580.LinkGoogle Scholar
- 21. LIPID Study Group,
Tonkin A, Simes RJ. Prevention of cardiovascular events and death with pravastatin in patients with coronary heart disease and a broad range of initial cholesterol levels.N Engl J Med. 1998; 339:1347–1357. doi: 10.1056/NEJM199811053391902.Google Scholar
- 22. LIPID Study Investigators,
Simes R, Hunt D, Kirby A, Tonkin A, Keech A, Aylward P, Colquhoun D, Glasziou P, Hague W, MacMahon S, Thompson P, West M, White H. Long-term effectiveness and safety of pravastatin in 9014 patients with coronary heart disease and average cholesterol concentrations: the LIPID trial follow-up.Lancet. 2002; 359:1379–1387. doi: 10.1016/S0140-6736(02)08351-4.CrossrefMedlineGoogle Scholar
- 23. LIPID Study Group. Design features and baseline characteristics of the LIPID (Long-term Intervention With Pravastatin in Ischaemic Disease) study: a randomised trial in patients with previous acute myocardial infarction and/or unstable angina pectoris.Am J Cardiol. 1995; 76:474–479.CrossrefMedlineGoogle Scholar
Magliano D, Liew D, Pater H, Kirby A, Hunt D, Simes J, Sundararajan V, Tonkin A. Accuracy of the Australian National Death Index: comparison with adjudicated fatal outcomes among Australian participants in the Long-term Intervention with Pravastatin in Ischaemic Disease (LIPID) study.Aust N Z J Public Health. 2003; 27:649–653.CrossrefMedlineGoogle Scholar
Tonkin AM, Blankenberg S, Kirby A, Zeller T, Colquhoun DM, Funke-Kaiser A, Hague W, Hunt D, Keech AC, Nestel P, Stewart R, Sullivan DR, Thompson PL, West M, White HD, Simes J; LIPID Study Investigators. Biomarkers in stable coronary heart disease, their modulation and cardiovascular risk: the LIPID Biomarker Study.Int J Cardiol. 2015; 201:499–507. doi: 10.1016/j.ijcard.2015.07.080.CrossrefMedlineGoogle Scholar
Marschner I, Colquhoun D, Simes R, Glasziou P, Harris P, Singh B, Friedlander D, White H, Thompson P, Tonkin A; LIPID Study Investigators. Long-term risk stratification for survivors of acute coronary syndromes: results from the Long-term Intervention with Pravastatin in Ischemic Disease (LIPID) Study.J Am Coll Cardiol. 2001; 38:56–63. doi: 10.1016/S0735-1097(01)01360-2.CrossrefMedlineGoogle Scholar
Stewart RW, HD, Kirby AC, Heritier SR, Simes RJ, Nestel PJ, West MJ, Colquhoun DM, Tonkin AM; Long-Term Intervention With Pravastatin in Ischemic Disease Study Investigators. White blood cell count predicts reduction in coronary heart disease mortality with pravastatin.Circulation. 2005; 111:1756–1762. doi: 10.1161/01.cir.0000160924.73417.26.LinkGoogle Scholar
West MJ, White HD, Simes RJ, Kirby A, Watson JD, Anderson NE, Hankey GJ, Wonders S, Hunt D, Tonkin AM. Risk factors for non-haemorrhagic stroke in patients with coronary heart disease and the effect of lipid-modifying therapy with pravastatin.J Hypertens. 2002; 20:2513–2517. doi: 10.1097/01.hjh.0000042890.24999.00.CrossrefMedlineGoogle Scholar
Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk.J Am Stat Assoc. 1999; 94:496–509. doi: 10.2307/2670170.CrossrefGoogle Scholar
Harrell F. Regression Modelling Strategies: With Applications to Linear Models, Logistic and Ordinal Regression, and Survival Analysis. 2nd ed. New York, NY: Springer International Publishing; 2015.CrossrefGoogle Scholar
Franchini M, Mannucci PM. Association between venous and arterial thrombosis: clinical implications.Eur J Intern Med. 2012; 23:333–337. doi: 10.1016/j.ejim.2012.02.008.CrossrefMedlineGoogle Scholar
Lowe GD. Coagulation factors, activation markers and risk of coronary heart disease: the Northwick Park Heart Studies.J Thromb Haemost. 2008; 6:256–258. doi: 10.1111/j.1538-7836.2007.02842.x.CrossrefMedlineGoogle Scholar
Prandoni P. Links between arterial and venous disease.J Intern Med. 2007; 262:341–350. doi: 10.1111/j.1365-2796.2007.01815.x.CrossrefMedlineGoogle Scholar
Sørensen HT, Horvath-Puho E, Søgaard KK, Christensen S, Johnsen SP, Thomsen RW, Prandoni P, Baron JA. Arterial cardiovascular events, statins, low-dose aspirin and subsequent risk of venous thromboembolism: a population-based case-control study.J Thromb Haemost. 2009; 7:521–528. doi: 10.1111/j.1538-7836.2009.03279.x.CrossrefMedlineGoogle Scholar
Donati MB. Thrombosis and cancer: a personal view.Thromb Haemost. 2007; 98:126–128.CrossrefMedlineGoogle Scholar
Zacharski LR. Anticoagulants in cancer treatment: malignancy as a solid phase coagulopathy.Cancer Lett. 2002; 186:1–9.CrossrefMedlineGoogle Scholar
Franco AT, Corken A, Ware J. Platelets at the interface of thrombosis, inflammation, and cancer.Blood. 2015; 126:582–588. doi: 10.1182/blood-2014-08-531582.CrossrefMedlineGoogle Scholar
Ross JS, Stagliano NE, Donovan MJ, Breitbart RE, Ginsburg GS. Atherosclerosis and cancer: common molecular pathways of disease development and progression.Ann N Y Acad Sci. 2001; 947:271–292.CrossrefMedlineGoogle Scholar
Tapia-Vieyra JV, Delgado-Coello B, Mas-Oliva J. Atherosclerosis and cancer: a resemblance with far-reaching implications.Arch Med Res. 2017; 48:12–26. doi: 10.1016/j.arcmed.2017.03.005.CrossrefMedlineGoogle Scholar
Koene RJ, Prizment AE, Blaes A, Konety SH. Shared risk factors in cardiovascular disease and cancer.Circulation. 2016; 133:1104–1114. doi: 10.1161/CIRCULATIONAHA.115.020406.LinkGoogle Scholar
White HD, Simes J, Stewart RA, Blankenberg S, Barnes EH, Marschner IC, Thompson P, West M, Zeller T, Colquhoun DM, Nestel P, Keech AC, Sullivan DR, Hunt D, Tonkin A; LIPID Study Investigators. Changes in lipoprotein-associated phospholipase A2 activity predict coronary events and partly account for the treatment effect of pravastatin: results from the Long-Term Intervention with Pravastatin in Ischemic Disease study.J Am Heart Assoc. 2013; 2:e000360. doi: 10.1161/JAHA.113.000360.LinkGoogle Scholar
Oldgren J, Wallentin L, Alexander JH, James S, Jönelid B, Steg G, Sundström J. New oral anticoagulants in addition to single or dual antiplatelet therapy after an acute coronary syndrome: a systematic review and meta-analysis.Eur Heart J. 2013; 34:1670–1680. doi: 10.1093/eurheartj/eht049.CrossrefMedlineGoogle Scholar
Castellucci LA, Cameron C, Le Gal G, Rodger MA, Coyle D, Wells PS, Clifford T, Gandara E, Wells G, Carrier M. Efficacy and safety outcomes of oral anticoagulants and antiplatelet drugs in the secondary prevention of venous thromboembolism: systematic review and network meta-analysis.BMJ. 2013; 347:f5133. doi: 10.1136/bmj.f5133.CrossrefMedlineGoogle Scholar
Glynn RJ, Danielson E, Fonseca FA, Genest J, Gotto AM, Kastelein JJ, Koenig W, Libby P, Lorenzatti AJ, MacFadyen JG, Nordestgaard BG, Shepherd J, Willerson JT, Ridker PM. A randomized trial of rosuvastatin in the prevention of venous thromboembolism.N Engl J Med. 2009; 360:1851–1861. doi: 10.1056/NEJMoa0900241.CrossrefMedlineGoogle Scholar
Squizzato A, Romualdi E, Ageno W. Why should statins prevent venous thromboembolism? A systematic literature search and a call for action.J Thromb Haemost. 2006; 4:1925–1927. doi: 10.1111/j.1538-7836.2006.02030.x.CrossrefMedlineGoogle Scholar
Rahimi K, Bhala N, Kamphuisen P, Emberson J, Biere-Rafi S, Krane V, Robertson M, Wikstrand J, McMurray J. Effect of statins on venous thromboembolic events: a meta-analysis of published and unpublished evidence from randomised controlled trials.PLoS Med. 2012; 9:e1001310. doi: 10.1371/journal.pmed.1001310.CrossrefMedlineGoogle Scholar
Ray JG, Mamdani M, Tsuyuki RT, Anderson DR, Yeo EL, Laupacis A. Use of statins and the subsequent development of deep vein thrombosis.Arch Intern Med. 2001; 161:1405–1410.CrossrefMedlineGoogle Scholar
Biere-Rafi S, Hutten BA, Squizzato A, Ageno W, Souverein PC, de Boer A, Gerdes VE, Büller HR, Kamphuisen PW. Statin treatment and the risk of recurrent pulmonary embolism.Eur Heart J. 2013; 34:1800–1806. doi: 10.1093/eurheartj/eht046.CrossrefMedlineGoogle Scholar
Nguyen CD, Andersson C, Jensen TB, Gjesing A, Schjerning Olsen AM, Malta Hansen C, Büller H, Torp-Pedersen C, Gislason GH. Statin treatment and risk of recurrent venous thromboembolism: a nationwide cohort study.BMJ Open. 2013; 3:e003135. doi: 10.1136/bmjopen-2013-003135.CrossrefMedlineGoogle Scholar
Tzoulaki I, Siontis KC, Ioannidis JP. Prognostic effect size of cardiovascular biomarkers in datasets from observational studies versus randomised trials: meta-epidemiology study.BMJ. 2011; 343:d6829. doi: 10.1136/bmj.d6829.CrossrefMedlineGoogle Scholar
Tzoulaki I, Siontis KC, Evangelou E, Ioannidis JP. Bias in associations of emerging biomarkers with cardiovascular disease.JAMA Intern Med. 2013; 173:664–671. doi: 10.1001/jamainternmed.2013.3018.CrossrefMedlineGoogle Scholar