Keywords
biomarkers - deep vein thrombosis - functional status - pulmonary embolism - venous
thromboembolism
Venous thromboembolism (VTE), which encompasses pulmonary embolism (PE) and deep vein
thrombosis (DVT), is a common cardiovascular disease with an annual incidence ranging
from 1 to 2 per 1,000 individuals.[1] It poses a major burden on the population, considering that VTE associated with
hospitalization is one of the leading causes of disability-adjusted life-years lost.[2] Furthermore, VTE has a considerable financial impact on health care systems, with
total estimated costs within the European Union-28 ranging from €1.5 to €13.2 billion
per year.[3] Several strategies to reduce the burden of VTE have been proposed, such as increasing
VTE awareness, improving VTE risk assessment, providing appropriate use of thromboprophylaxis,
and ensuring accurate VTE surveillance.[4]
There are several more implications arising from VTE for affected patients. Apart
from traditional clinical outcomes such as recurrence, anticoagulation-associated
bleeding, and mortality, patients experiencing a VTE event are at risk of impaired
quality of life, physical incapacity, and psychosocial distress in the short- and
long-term periods.[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23] To optimally assess these patient-centered outcomes, a standard set of outcome measures
has been developed recently.[24] Among those measures, the post-VTE functional status (PVFS) scale has been selected
to evaluate functional limitations after VTE. The PVFS scale is intended to capture
the whole range of functional limitations in patients with VTE, including patients
with both PE and DVT.[25]
[26]
[27] When applying the PVFS scale in a prospective cohort study of patients with acute
VTE, we observed an overall improvement in functional limitations during a 12-month
follow-up.[28] However, patients did not return to pre-VTE functional status despite standard anticoagulation
treatment. Although we were able to identify several clinical characteristics associated
with functional limitations, a considerable variability in data could not be explained
by clinical characteristics alone.[28] This unexplained persistence of functional limitations might suggest that the acute
VTE event initiates an underlying process, which results in physical impairment. However,
further details of this process and the role of biomarkers have not been explored.
Therefore, we aimed to investigate the association of biomarkers, focusing on parameters
reflecting the hemostatic, inflammatory, and cardiovascular systems, with persisting
functional limitations in patients with VTE.
Materials and Methods
Study Design and Patient Population
This project was conducted within the framework of the ongoing BACH-VTE study—“A prospective
observational study to investigate predictors of Bleeding and Assess long-term outComes
on Health in patients with Venous ThromboEmbolism.” The BACH-VTE study is a prospective,
observational cohort study at the Medical University of Vienna, Austria. The detailed
study characteristics, including schedule, inclusion and exclusion criteria, and outcomes,
have been published previously.[27]
[28]
[29]
[30] Shortly, adult patients with an objectively confirmed acute DVT and/or PE are eligible
for inclusion within 21 days of diagnosis. The main exclusion criterion is therapeutic
anticoagulation for any reason in the 3 months before DVT/PE diagnosis. Outcomes of
the BACH-VTE study include incidence and risk factors of bleeding, functional limitations,
postthrombotic and post-PE syndromes, quality of life, and psychological sequelae,
among others. The follow-up visits are incorporated into routine clinical care, with
the first follow-up at 3 months and the second at 1 year. The last follow-up is scheduled
at 5 years.
For the present analysis, we excluded patients with active cancer, pregnancy, and
puerperium at the time of diagnosis. Further, we only considered patients included
between July 2020 and November 2023 who completed the first follow-up visit. We assessed
clinical characteristics and demographics in a face-to-face interview at study inclusion
and checked self-reported data with medical records. Furthermore, we performed a blood
draw by sterile venipuncture. After 3 months, we performed the follow-up visit, including
an assessment of functional limitations and another blood draw.
All included patients provided written informed consent prior to study inclusion.
The study was conducted according to the principles of the Declaration of Helsinki
and was approved by the local Ethics Committee of the Medical University of Vienna
(EK 1045/2020). Study data were collected and managed using REDCap electronic data
capture tools hosted at the Medical University of Vienna.[31]
Assessment of Functional Limitations
We assessed functional limitations with the PVFS scale, which is an ordinal measure
modeled after the modified Rankin scale for patients with stroke and intends to capture
the whole range of functional limitations after VTE.[25]
[26]
[27] The scale ranges from 0 to 4, with the following distinct grades: 0, no functional
limitations; 1, negligible functional limitations; 2, slight functional limitations;
3, moderate functional limitations; and 4, severe functional limitations. Thus, higher
scale grades indicate more functional limitations. In our study, we assessed the PVFS
scale through a structured interview at the first follow-up, that is, 3 months after
diagnosis, and asked patients about their functional limitations prior to the VTE
diagnosis as suggested in the scale manual.[26]
Biomarker Measurement
We intended to evaluate the association of functional limitations with readily available
biomarkers reflecting the hemostatic, inflammatory, and cardiovascular systems. Thus,
we considered the following parameters: D-dimer, fibrinogen, factor VIII (FVIII),
von Willebrand factor antigen (VWF), and C-reactive protein (CRP). Furthermore, we
evaluated troponin T and N-terminal pro-B-type natriuretic peptide (proBNP) in patients
with PE. All biomarkers were routinely measured at study inclusion and the first follow-up
visit at 3 months. However, for some hemostatic biomarkers, that is, FVIII and VWF,
this routine measurement was implemented during the conduct of the study rather than
from the beginning. Therefore, missing value rates for these biomarkers, especially
at the time of study inclusion, are higher than for other biomarkers. Units, normal
ranges, and other specifications of biomarkers are summarized in [Supplementary Table S1] (available in the online version only).
We defined two distinct aims for our study. First, we aimed to investigate the association
of biomarkers measured at study inclusion—within 21 days from VTE diagnosis—and functional
limitations after 3 months. This analysis assesses the prognostic value of these biomarkers,
potentially allowing for the early identification of patients at risk for persisting
functional limitations. Second, we examined the association between biomarkers measured
at follow-up—3 months after VTE diagnosis—and functional limitations at this time
point. This analysis reflects the diagnostic value of biomarkers for persisting functional
limitations and might, therefore, generate hypotheses regarding the underlying pathophysiological
mechanisms of long-term limitations.
Statistical Analysis
Categorical variables are summarized as absolute and relative frequencies, and continuous
variables as median and 25th and 75th percentiles, that is, interquartile range (IQR).
Missing values are reported but were not imputed. The correlation between all considered
biomarkers was evaluated with Spearman's rank correlation coefficient and displayed
graphically. To investigate the association between biomarkers and functional limitations
at follow-up, we built proportional odds logistic regression models (cumulative logit
link models). The dependent variable was the PVFS scale grade 3 months after the VTE
event, the independent variables were the biomarkers of interest. Every biomarker
was evaluated in a separate model, and we considered levels at study inclusion, that
is, within 21 days of VTE diagnosis, and at follow-up, that is, 3 months after diagnosis.
Due to right-skewed distribution, we log-transformed D-dimer, CRP, troponin T, and
proBNP. First, we performed univariable modelling, resulting in unadjusted odds ratios
(ORs) and corresponding 95% confidence intervals (CIs). Then, we adjusted for clinical
characteristics for which we previously identified an association with functional
limitations and for potential confounders, that is, sex, age, body mass index (BMI),
VTE type (i.e., PE or DVT), and history of cardiovascular or pulmonary disease.[28] Age was modeled in a nonlinear fashion, using restricted cubic splines with 3 degrees
of freedom. The knots were placed at the 5th, 35th, 65th, and 95th percentiles. To
account for functional limitations unrelated to VTE, we adjusted for PVFS scale grade
before VTE diagnosis. The resulting unadjusted and adjusted ORs indicate the likelihood
of having a higher PVFS scale grade in relation to the respective increase in each
biomarker.
To investigate the ability of the biomarkers to distinguish patients with functional
limitations from those without, we compared proportions of patients per PVFS scale
categories in patients with high (≥50th percentile) and low (<50th percentile) biomarker
levels. Furthermore, we evaluated the discriminative performance of biomarkers for
the presence of PVFS scale >1 with the area under the receiver operating characteristic
curve (AUC-ROC). Corresponding CIs were computed with 2,000 stratified bootstrap replicates.
All analyses were done in R version 4.3.3 using the rms package.[32]
[33]
Results
Our study cohort encompassed 290 patients with a median (IQR) age of 54.9 (43.1–64.2)
years, including 41.4% women and 134 (46.2%) patients with PE. Detailed clinical characteristics
are displayed in [Table 1]. The median (IQR) PVFS scale grade at 3 months was 1 (0–2).
Table 1
Patient demographics and clinical characteristics
|
Full cohort (n = 290)
|
|
Female, n (%)
|
120 (41.4)
|
|
Age, median (IQR)
|
54.9 (43.1–64.2)
|
|
BMI, median (IQR)[a]
|
27.7 (24.7–31.6)
|
|
Type of VTE, n (%)
|
–
|
|
Pulmonary embolism[b]
|
134 (46.2)
|
|
Deep vein thrombosis
|
156 (53.8)
|
|
Unprovoked VTE, n (%)
|
174 (60.0)
|
|
Provoked VTE, n (%)[c]
|
116 (40.0)
|
|
Major persisting risk factor
|
25 (8.6)
|
|
Major transient risk factor
|
38 (13.1)
|
|
Minor transient risk factor
|
66 (22.8)
|
|
History of VTE, n (%)
|
85 (29.3)
|
|
History of cardiovascular or pulmonary disease, n (%)[d]
|
67 (23.1)
|
|
Smoking, n (%)
|
–
|
|
Current
|
71 (24.5)
|
|
Former
|
63 (21.7)
|
|
Never
|
156 (53.8)
|
Abbreviations: BMI, body mass index; IQR, interquartile range; VTE, venous thromboembolism.
a Data missing for one patient.
b With or without deep vein thrombosis.
c Some patients had more than one risk factor; patients with cancer, patients with
pregnancy, and patients in the postpartum period were excluded.
d Including coronary artery disease, chronic heart failure, arrhythmia, peripheral
artery disease, cerebrovascular disease, and chronic pulmonary disease.
Biomarkers at Study Inclusion
The median (IQR) time from VTE diagnosis to study inclusion and blood draw was 3 (1–7)
days. Median (IQR) biomarker levels at study inclusion are shown in [Supplementary Fig. S1] (available in the online version only), and a correlation matrix between all considered
biomarkers is shown in [Supplementary Fig. S2] (available in the online version only). D-dimer was missing in 20 patients, fibrinogen
in 15, FVIII in 75, VWF in 62, and CRP in 15. Detailed clinical characteristics of
patients with and without missing biomarker values are shown in [Supplementary Table S2] (available in the online version only). In 134 patients with PE, troponin T was
missing in 24 and proBNP in 23 patients.
All hemostatic and inflammatory biomarkers measured at study inclusion showed a significant
and independent association with functional limitations after 3 months ([Table 2]). A 50-unit increase in FVIII and VWF was associated with 1.21-fold increased odds
of having a higher PVFS scale grade after 3 months (OR: 1.21, 95% CI: 1.01–1.45 for
both). In patients with PE, cardiac biomarkers were not significantly associated with
functional limitations. When stratifying patients by biomarker levels into a high-level
and a low-level group (≥50th percentile and <50th percentile, respectively), hemostatic
and inflammatory biomarkers measured at study inclusion showed a moderate performance
in differentiating patients with high and low PVFS scale grades ([Fig. 1]). In patients with PE, cardiac biomarkers showed a poor to moderate performance
in differentiating patients ([Supplementary Fig. S3] [available in the online version only]). The absolute and relative frequencies of
patients within the PVFS scale grade according to biomarker levels are shown in [Supplementary Table S3] (available in the online version only). Less than 1% of patients with a study inclusion
D-dimer or VWF level below the cohort median had a PVFS scale grade of 4, and nearly
70% of those with a study inclusion fibrinogen level below the cohort median had a
PVFS scale grade of 0 or 1. The discriminatory performance of hemostatic and inflammatory
biomarkers for the presence of PVFS scale grade >1 was moderate, with AUC-ROCs (95%
CIs) ranging from 0.57 (0.51–0.64) for CRP to 0.62 (0.55–0.69) for VWF ([Supplementary Table S4] [available in the online version only]).
Table 2
Association between biomarker levels at study inclusion and functional limitations
at 3 months
|
Unadjusted OR (95% CI)
|
Adjusted OR (95% CI)[a]
|
|
D-dimer (per double, μg/mL)
|
1.20 (1.05–1.37)
|
1.17 (1.01–1.35)
|
|
Fibrinogen (per 50 mg/dL increase)
|
1.11 (1.02–1.22)
|
1.09 (1.00–1.20)
|
|
FVIII (per 50% increase[b])
|
1.27 (1.07–1.50)
|
1.21 (1.01–1.45)
|
|
VWF (per 50% increase[b])
|
1.34 (1.13–1.59)
|
1.21 (1.01–1.45)
|
|
CRP (per double, mg/dL)
|
1.12 (1.02–1.23)
|
1.11 (1.00–1.23)
|
|
Troponin T[c] (per double, ng/L)
|
1.16 (0.93–1.44)
|
1.10 (0.84–1.46)
|
|
proBNP[c] (per double, pg/mL)
|
1.13 (0.97–1.31)
|
1.05 (0.87–1.27)
|
Abbreviations: CI, confidence interval; CRP, C-reactive protein; FVIII, factor VIII;
OR, odds ratio; proBNP, N-terminal pro-B-type natriuretic peptide; VWF, von Willebrand
factor antigen.
a Adjusted for sex, age, body mass index, venous thromboembolism (VTE) type (i.e.,
pulmonary embolism or deep vein thrombosis), history of cardiovascular or pulmonary
disease, and post-VTE functional status scale before VTE diagnosis. Age was modelled
as a continuous variable using restricted cubic splines with four knots at the 5th,
35th, 65th, and 95th percentiles.
b Unit of FVIII and VWF is a percentage.
c Troponin T and proBNP were only evaluated in patients with pulmonary embolism, therefore,
the multivariable model for these variables did not include VTE type.
Fig. 1 Proportions of patients per post-VTE functional status (PVFS) scale category at 3
months with high (≥50th percentile) and low (<50th percentile) biomarker levels measured
at study inclusion. Dark green refers to PVFS scale of 0 (no functional limitations),
light green to 1 (negligible functional limitations), yellow to 2 (slight functional
limitations), orange to 3 (moderate functional limitations), and red to 4 (severe
functional limitations). CRP, C-reactive protein; FVIII, factor VIII; perc, percentile;
VTE, venous thromboembolism; VWF, von Willebrand factor antigen.
Biomarkers at 3 Months
The median (IQR) time from VTE diagnosis to follow-up blood draw was 98 (93–115) days.
All biomarker values decreased compared to study inclusion ([Supplementary Fig. S4] [available in the online version only]). A correlation matrix of all considered
biomarkers is shown in [Supplementary Fig. S5] (available in the online version only). D-dimer was missing in 33 patients, fibrinogen
in 28, FVIII in 39, VWF in 33, and CRP in 15. Detailed clinical characteristics stratified
by missing biomarker levels are shown in [Supplementary Table S5] (available in the online version only). In 134 patients with PE, troponin T was
missing in 35 and proBNP in 30 patients.
After adjusting for confounders, only D-dimer remained significantly associated with
functional limitations (adjusted OR [95% CI], 1.22 [1.00–1.49]; [Table 3]). Similar to biomarkers measured at study inclusion, biomarkers measured after 3
months showed a moderate performance in differentiating patients with high and low
PVFS scale grades ([Fig. 2]). In patients with PE, cardiac biomarkers showed a poor performance ([Supplementary Fig. S6] [available in the online version only]). The absolute and relative frequencies of
patients within the PVFS scale grade according to biomarker levels measured at 3 months
are shown in [Supplementary Table S6] (available in the online version only). The discriminatory performance of biomarkers
measured at 3 months for the presence of PVFS scale grade >1 was similar to the performance
of biomarkers measured at study inclusion, with the highest AUC-ROC (95% CI) displayed
by D-dimer and fibrinogen (0.62 [0.55–0.69] and 0.63 [0.55–0.69], respectively; [Supplementary Table S7] [available in the online version only]).
Table 3
Association between biomarker levels at 3 months and functional limitations at 3 months
|
Unadjusted OR (95% CI)
|
Adjusted OR (95% CI)[a]
|
|
D-dimer (per double, μg/mL)
|
1.35 (1.13–1.62)
|
1.22 (1.00–1.49)
|
|
Fibrinogen (per 50 mg/dL increase)
|
1.28 (1.12–1.48)
|
1.09 (0.92–1.29)
|
|
FVIII (per 50% increase[b])
|
1.31 (1.07–1.60)
|
1.12 (0.90–1.39)
|
|
VWF (per 50% increase[b])
|
1.28 (1.06–1.56)
|
1.06 (0.86–1.32)
|
|
CRP (per double, mg/dL)
|
1.19 (0.79–1.79)
|
1.04 (0.90–1.21)
|
|
Troponin T[c] (per double, ng/L)
|
1.16 (0.79–1.71)
|
1.31 (0.76–2.26)
|
|
proBNP[c] (per double, pg/mL)
|
1.21 (0.96–1.53)
|
0.95 (0.69–1.32)
|
Abbreviations: CI, confidence interval; CRP, C-reactive protein; FVIII, factor VIII;
OR, odds ratio; proBNP, N-terminal pro-B-type natriuretic peptide; VWF, von Willebrand
factor antigen.
a Adjusted for sex, age, body mass index, venous thromboembolism (VTE) type (i.e.,
pulmonary embolism or deep vein thrombosis), history of cardiovascular or pulmonary
disease, and post-VTE functional status scale before VTE diagnosis. Age was modelled
as a continuous variable using restricted cubic splines with four knots at the 5th,
35th, 65th, and 95th percentiles.
b Unit of FVIII and VWF is a percentage.
c Troponin T and proBNP were only evaluated in patients with pulmonary embolism, therefore,
the multivariable model for these variables did not include VTE type.
Fig. 2 Proportions of patients per post-VTE functional status (PVFS) scale category at 3
months with high (≥50th percentile) and low (<50th percentile) biomarker levels measured
at 3 months. Dark green refers to PVFS scale of 0 (no functional limitations), light
green to 1 (negligible functional limitations), yellow to 2 (slight functional limitations),
orange to 3 (moderate functional limitations), and red to 4 (severe functional limitations).
FVIII, factor VIII; CRP, C-reactive protein; perc, percentile; VTE, venous thromboembolism;
VWF, von Willebrand factor antigen.
Discussion
In our prospective cohort study, we observed an independent association of hemostatic
and inflammatory biomarkers at VTE diagnosis with functional limitations 3 months
after VTE. Furthermore, higher D-dimer levels 3 months after VTE were significantly
associated with persisting functional limitations.
The association between biomarkers and persisting functional limitations in patients
with VTE has not been investigated previously. Since we considered biomarker measurements
both at the time of diagnosis and after 3 months, we can draw several conclusions
from our work. First, the independent association of biomarkers at the time of VTE
diagnosis with functional limitations 3 months afterward underlines their prognostic
value. Thereby, patients at a higher risk of functional limitations could be identified
early in their disease process, enabling a comprehensive and close follow-up procedure
with an early possibility of intervention, for example, with rehabilitation programs.
The efficacy and cost-effectiveness of such interventions remain to be evaluated.
Second, the independent association between D-dimer measured 3 months after diagnosis
and functional limitations at the same time can be considered hypothesis-generating,
potentially reflecting a mechanistic relationship. While details of a potential mechanism
need to be evaluated in basic and translational research studies, the association
may be explained by clot persistence, endothelial dysfunction or damage, a general
state of hypercoagulability, or others.
The hemostatic and inflammatory biomarkers at hand have been investigated in various
patient cohorts and clinical scenarios. High levels of VWF and FVIII have been associated
with an increased risk of thrombosis.[34]
[35] At the same time, the release of VWF from Weibel Palade bodies can be triggered
by cellular injury.[36] Thus, increased levels might be a sign of endothelial cell dysfunction, damage,
or activation, presumably caused by VTE or at least more pronounced in patients with
persisting functional limitations.[37]
[38] We observed a comparable association of VWF and FVIII with functional limitations.
This is plausible due to their tight connection since VWF functions as a carrier for
FVIII.[39] Apart from the general population, VWF and FVIII have been studied in patients with
cancer, showing an association with a higher risk of thrombosis, poorer overall survival,
and the presence of metastases.[40]
[41]
[42] Furthermore, FVIII has been associated with recurrent VTE and postthrombotic syndrome
(PTS) in children and young adults.[43] Considering inflammatory biomarkers, a systematic review evaluated the association
between biomarkers of inflammation and the development of PTS, suggesting an association
of CRP and PTS.[44] This is in-line with the association between CRP and functional limitations observed
in our study. Apart from CRP, intracellular adhesion molecule-1 showed promising results
regarding the association with PTS, with statistical significance at several time
points after proximal DVT and a dose–response association.[44]
[45] However, we cannot compare this result to our study since we only evaluated CRP.
Overall, the association of hemostatic and inflammatory biomarkers measured at VTE
diagnosis with functional limitations 3 months afterward complement the known prognostic
role of these biomarkers for adverse outcomes in different patient cohorts. While
we can only speculate on the details, possible mechanisms include thromboinflammation,
endothelial cell damage or activation, hypercoagulability, or an increase in biomarkers
as a surrogate for the extent and severity of the initial thrombotic event, reflecting
an acute phase reaction.
When evaluating biomarkers 3 months after VTE, only D-dimer remained independently
associated with functional limitations. Apart from its role in the diagnosis of VTE,
the utility of D-dimer as a biomarker has been shown in several other clinical scenarios.
Those include the prediction of VTE in ambulatory patients with solid cancer, the
identification of patients at low risk of recurrent VTE, and the prediction of long-term
risk of arterial and venous events, cancer, and mortality in patients with stable
coronary heart disease.[46]
[47]
[48] Importantly, the measurement of D-dimer in these scenarios is performed when patients
are not receiving anticoagulation, which is in contrast to our study. Regarding the
long-term consequences of VTE, an association of D-dimer with recurrent VTE and PTS
in children and young adults has been suggested.[43] Other explanations for an elevation in D-dimer values include the presence of cancer,
advanced age, pregnancy, and puerperium, among others.[49] However, these factors are unlikely to explain the association with persisting functional
limitations in our study due to the exclusion of patients with active cancer, pregnancy,
and puerperium and the adjustment for important confounders, including age, sex, and
comorbidities. D-dimer is a degradation product of fibrin.[50] Thus, higher levels might indicate the presence of residual thrombosis and/or increased
hemostatic activity and hypercoagulability, which could explain the persistence of
functional limitation. The hypercoagulability might be associated with endothelial
cell function, platelet activation, inflammation, or others. Overall, this observation
must be considered hypothesis-generating and should be investigated in more detail.
Surprisingly, we could not identify an association of troponin T and proBNP with persisting
functional limitations in patients with PE. Troponins have been shown to be associated
with short-term mortality and adverse outcomes in patients with PE.[51] A similar association has been observed for elevated BNP values.[52] Consequently, troponin levels have been incorporated into the risk assessment strategy
of acute PE in current guidelines, and the role of proBNP to provide additional prognostic
information has been acknowledged.[53] However, both troponins and proBNP have low specificity and positive predictive
value while displaying high sensitivity and negative predictive value.[54]
[55]
[56] This could partially explain why we did not observe an association of elevated levels
with persisting functional limitations. More importantly, the number of patients was
very limited in this subgroup of our study, which most likely resulted in an underpowered
analysis.
While we performed multivariable analyses per biomarker adjusted for potential confounders,
we did not consider multiple biomarkers in a single model. The aim of this study was
to generate hypotheses rather than predictive or causal models. Further, some biomarkers
we considered certainly inherit collinearity due to their strong interplay, for example,
VWF and FVIII, as can also be seen from the correlation matrices. Thus, incorporating
all considered biomarkers in a single multivariable model and trying to interpret
the respective ORs might be misleading and would not have added further value.
Our study has several limitations. First, the number of patients was limited, resulting
in a considerable degree of uncertainty, especially in the subgroup analysis of patients
with PE. Second, there was a considerable number of missing values in biomarker measurements,
which is a result of the implementation of routine laboratory measurement during the
conduct of the study rather than from the beginning. However, demographics and clinical
characteristics were similar between patients with and without missing values. Third,
our results must be considered hypothesis-generating as they neither represent a predictive
nor a causal association between biomarkers and persisting functional limitations.
Fourth, we only evaluated readily available biomarkers. Thus, future studies might
take a broader approach and evaluate further biomarkers such as prothrombin fragment
1.2, thrombin–antithrombin complexes, or sP-selectin, among others. Lastly, the monocentric
design of the study and the recruitment of patients in a tertiary care center might
lead to a selection bias and limit the generalizability of the results.
Conclusion
In conclusion, hemostatic and inflammatory biomarkers measured at VTE diagnosis and
D-dimer measured after 3 months were independently associated with persisting functional
limitations after VTE. This underlines their prognostic role for persisting functional
limitations and suggests an underlying pathophysiological mechanism.
