Biomarkers for Risk Assessment in Atrial Fibrillation

Abstract

Background

Atrial fibrillation (AF) is associated with an increased risk of thromboembolism, which can be significantly reduced with anticoagulant treatment. Key goals in the clinical management of AF are the identification of patients at high risk for developing AF and accurate stratification of the risk of stroke and systemic embolic events (S/SEE) as well as treatment-related major bleeding.

Content

In this review, we describe the expanding evidence regarding the use of circulating biomarkers for predicting the risks of both incident AF and its clinically important complications of S/SEE and treatment-related major bleeding. We also review emerging biomarker-based scores for assessing these risks.

Summary

Patients with AF undergo progressive cardiac structural remodeling, which may precede the onset of the arrhythmia. Abnormal concentrations of circulating biomarkers reflecting the underlying pathophysiologic mechanisms of hemodynamic stress (i.e., natriuretic peptides), inflammation (i.e., C-reactive protein), and myocardial fibrosis identify patients at higher risk of developing AF. Circulating biomarkers can also be used to identify patients with AF who are at greatest risk for developing S/SEE or major bleeding. In particular, biomarkers of hemodynamic stress, myocardial injury (i.e., cardiac troponin), and coagulation activity (i.e., D-dimer) are key indicators of thromboembolic risk, and cardiac troponin and growth-differentiation factor-15 are strongly associated with risk of anticoagulant-related major bleeding. The biomarker-based age, biomarker, clinical history (ABC)-stroke and ABC-bleeding risk scores improve risk stratification for S/SEE and major bleeding, respectively, when compared with traditional clinical risk scores like the CHA₂DS₂-VASc and HAS-BLED scores.

Introduction

The prevalence of atrial fibrillation (AF) is increasing with an estimated 1 in 3 people developing AF in their lifetime (1). This rise in prevalence has paralleled aging of the population, and the increasing burden of heart failure and metabolic risk factors, including diabetes and obesity, which have well-established associations with AF (1). Given the increasing burden of AF and its association with adverse cardiovascular outcomes, AF has emerged as a major public health problem.

The most feared complications of AF are stroke and systemic embolic events (S/SEE), which are highly morbid and often fatal (2). In epidemiologic studies conducted prior to the widespread use of anticoagulants, patients with AF had a ∼5-fold higher risk of ischemic stroke than did patients without AF (1); however, this risk is significantly reduced by treatment with oral anticoagulants. Unfortunately, ∼80% of first AF episodes are not recognized by the patient (3), and may not prompt appropriate treatment for prevention of thromboembolic complications. In addition, although efficacious for thromboprophylaxis, anticoagulants increase the risk of bleeding. As such, clinicians must systematically weigh these competing risks when making decisions about initiating oral anticoagulation (4, 5).

Accurate risk assessment is critical in the clinical management of AF. For estimating the risks of S/SEE and major bleeding, cardiovascular professional society guidelines (4, 5) currently recommend using clinical risk scores, such as the CHA₂DS₂-VASc and HAS-BLED scores, respectively; these scores, however, have only modest prognostic performance (6). Therefore, the search for enhanced tools for risk assessment remains a scientific and clinical priority. Over the last 2 decades, multiple studies have demonstrated that circulating biomarkers may improve risk prediction of S/SEE and bleeding when combined with or used in lieu of clinical risk assessment (7–11). In this review, we describe the expanding evidence regarding the use of circulating biomarkers for predicting the risks of incident AF and its clinically important complications of S/SEE and treatment-related bleeding.

Predicting Onset of Atrial Fibrillation

Pathophysiologic Mechanisms of Incident Atrial Fibrillation

The complex pathophysiologic alterations underlying AF have been well-described (12). In response to external stressors, the atria undergo progressive structural remodeling characterized by the histologic hallmarks of fibroblast activation, connective tissue deposition, and atrial myocardial fibrosis (Fig. 1) (13, 14). In turn, the electrophysiologic properties of the atria are disrupted, contributing to the initiation and self-perpetuation of AF (15). Progressive atrial fibrosis also tracks with clinical progression of AF (i.e., from paroxysmal to persistent to permanent AF) and with declining efficacy of pharmacologic and catheter-based interventions for restoring sinus rhythm (16).

Hemodynamic stressors, including structural heart disease and hypertension, may initiate the cascade leading to atrial remodeling and fibrosis (2). Systemic inflammation, often related to chronic infection, rheumatologic disease, or metabolic disease, may also be a key antecedent in this cascade. In the case of metabolic disease, increased epicardial adipose tissue, which secretes proinflammatory adipocytokines onto the adjacent atrial myocardium, may be a mechanistic link (Fig. 1) (17).

Circulating Biomarkers and Incident Atrial Fibrillation

Given the underlying pathophysiologic mechanisms implicated in the development of AF, namely hemodynamic stress, inflammation, and myocardial fibrosis, multiple studies have investigated potential associations between biomarkers reflecting these pathobiological axes and risk of incident AF (Fig. 2).

Natriuretic peptides (NPs), including B-type natriuretic peptide (BNP) and N-terminal pro-B-type natriuretic peptide (NT-proBNP), are neurohormones that are secreted from cardiomyocytes in response to increased wall tension from pressure or volume overload. Although NPs were initially embraced for diagnostic use in acute heart failure, circulating NP concentrations also forecast risk of multiple different cardiovascular diseases, including AF (18). An association between circulating NP concentration and incident AF has now been well-established in large community-based cohorts, including the Framingham Offspring Study, Cardiovascular Health Study, and the CHARGE-AF Consortium, among others (18–21). In the Framingham Offspring Study, for example, BNP identified a strong gradient of risk for new-onset AF, independent of clinical risk factors for AF such as age, sex, body mass index, hypertension, valvular heart disease, and heart failure (adjusted hazard ratio per 1 standard deviation [SD] log-transformed BNP 1.62, 95% confidence interval [CI] 1.42–1.86; P < 0.001). Moreover, BNP improved discrimination when added to clinical variables (c-index 0.80 [95% CI 0.78–0.83] from 0.78 [95% CI 0.75–0.81]), and was the strongest predictor of AF risk among a panel of circulating biomarkers reflecting other biological pathways (19).

C-reactive protein (CRP), a hepatically-derived protein, is a biomarker of systemic inflammation. Although other biomarkers reflecting specific immune pathways have been explored as research tools, CRP is the most widely used inflammatory biomarker in clinical practice because of its stability in vitro. CRP is not only a strong indicator of atherothrombotic risk, but also has important associations with prevalent and incident AF, independent of clinical risk factors and other circulating biomarkers (adjusted hazard ratio per 1 SD 1.25, 95% CI 1.07–1.46; P = 0.004) (19, 22). In addition, CRP concentration appears to correlate with AF burden (22).

Several putative biomarkers of fibrosis have been examined as potential indicators of the risk of incident AF. Although extensive histologic and radiographic evidence support an association between atrial myocardial fibrosis and the risk of incident AF, the associations between such circulating biomarkers of fibrosis and AF risk have generally been less strong. Soluble ST2, a member of the interleukin-1 receptor family, and galectin-3, a ß-galactoside-binding lectin, have both garnered interest as AF biomarkers because of their regulatory role in myocardial fibrosis and established associations with heart failure. While higher circulating concentrations of each biomarker have been associated with a higher risk of developing AF in several cohorts, these associations are more modest than the associations with NPs and CRP, and are not independent of traditional clinical risk markers (23, 24). Biomarkers of collagen deposition and extracellular matrix remodeling, including the matrix metalloproteinase family, and their endogenous inhibitors, the tissue inhibitors of matrix metalloproteinases, have also had possible epidemiological links with the onset of AF (25). In addition, specific patterns of collagen-related biomarkers (e.g., the ratio of serum carboxy-terminal telopeptide of collagen type I to serum matrix metalloproteinase−1) may improve AF risk prediction compared with the individual markers, suggesting that multi-marker approaches with fibrosis biomarkers may offer superior discrimination (25). Nevertheless, no fibrosis biomarker or combination of biomarkers has emerged as a dominant marker for assessment of AF risk.

Clinical Applications

Although circulating cardiovascular biomarkers, most notably BNP and NT-proBNP, add to clinical risk factors for estimating a patient’s risk of developing AF, their clinical application remains uncertain. AF risk assessment using biomarkers has the potential to define a population of patients who might be followed more closely with electrocardiographic monitoring with the goal of facilitating earlier detection of AF and more timely initiation of anticoagulant therapy. This is the hypothesis being tested in the ongoing STROKESTOP II trial, a randomized, controlled, population-based study of AF screening, in which the decision to use prolonged intermittent electrocardiographic monitoring is directed by NT-proBNP concentrations for patients allocated to the intervention arm (26). In addition, AF screening strategies that leverage circulating biomarkers might reinforce efforts to reverse clinical AF risk factors, such as diabetes and obesity. Nevertheless, in the absence of existing therapies for the primary prevention of AF or evidence-based, cost-effective AF screening strategies, a role for incorporating biomarker-based screening into clinical practice is, as yet, undetermined.

Predicting Thromboembolism in Patients with Atrial Fibrillation

Pathophysiologic Mechanisms of Complications of Atrial Fibrillation

Stratifying the risk of S/SEE to inform decision-making regarding the initiation of anticoagulation therapy is the cornerstone of clinical management of AF patients. Analogous to risk screening for incident AF, blood-based biomarkers may provide insight into the pathophysiologic mechanisms implicated in AF-related thrombosis, and may, in turn, be applied clinically to improve risk stratification.

The traditional pathophysiologic model of thromboembolism in patients with AF is centered around the notion that thrombus forms in the fibrillating left atrium due to stasis. However, this model has been challenged over the last 2 decades by the observations that a rhythm control strategy, as compared with a rate control strategy, does not mitigate stroke risk, and that the timing of S/SEE often does not correlate with the timing of AF episodes (27). Moreover, empiric observations from large clinical datasets indicate that cardiovascular comorbidities (e.g., diabetes and heart failure) rather than AF parameters (e.g., AF frequency and duration) are stronger clinical indicators of thromboembolic risk in patients with AF; hence, their inclusion in the CHA₂DS₂-VASc score. Given this shift in paradigm, a more appropriate mechanistic framework for AF-related thrombosis, is grounded in the classical triad proposed by Virchow of 1) alterations in blood flow (i.e., stasis), 2) vascular endothelial injury, and 3) hypercoagulability.

Within this framework, some of the biologic pathways associated with AF initiation and self-perpetuation also relate to thromboembolic risk. For example, in addition to their associations with progressive fibrotic changes underpinning AF onset, cardiomyocyte stress pathways also reflect the downstream atrial dysfunction seen in patients with AF. This so-called “atrial myopathy,” in which the normal transit of blood through the heart is perturbed, independently correlates with thromboembolic risk (28). Other important biomarkers of thrombotic risk include those reflecting myocardial injury and coagulation activity (i.e., thrombosis and fibrinolysis), which are more germane to Virchow’s axes of vascular endothelial injury and hypercoagulability.

Circulating Biomarkers to Predict Stroke or Systemic Embolism

The association of NPs with thromboembolic risk was first demonstrated in the biomarker substudy (n = 6189) of the RE-LY trial, a multinational study comparing dabigatran vs. warfarin in patients with AF. In that analysis, the investigators observed that NT-proBNP concentrations measured in blood samples obtained at the time of randomization were independently associated with risk of S/SEE, with a 2-fold higher risk in the top vs. bottom quartile (8). In addition, NT-proBNP concentration contributed incremental prognostic information to the CHADS₂ and CHA₂DS₂-VASc clinical risk scores (8). Similarly, NT-proBNP performed well for S/SEE risk prediction in the biomarker substudies of the ARISTOTLE (apixaban vs. warfarin) and ENGAGE AF-TIMI 48 (edoxaban vs. warfarin) trials (Fig. 3), and manifested a particularly strong association with ischemic stroke (11, 30). The relationship between NT-proBNP concentrations and S/SEE risk in patients with AF offered a compelling epidemiologic correlate supporting the pathophysiologic model that atrial myocyte stress could lead to atrial dysfunction, and ultimately to thromboembolism.

The biomarker studies from the RE-LY, ARISTOTLE, and ENGAGE AF-TIMI 48 trials also established the prognostic significance of cardiac troponin (cTn) in patients with AF. In patients with AF enrolled in RE-LY, ARISTOTLE, and ENGAGE AF-TIMI 48, cTnI and/or cTnT concentrations measured at the time of randomization identified a strong gradient of risk of S/SEE during trial follow-up that was independent of clinical risk factors (Fig. 3) (8, 9, 30, 31). The pathophysiologic basis for increased thrombogenicity in patients with increased cardiac cTn remains uncertain. Analogous to NPs, increases in circulating cTn may reflect chronic myocardial ischemia secondary to elevated left atrial pressures and hemodynamic stress, although it is noteworthy that cTn provides independent prognostic information from NPs. Chronic increases in cTn may also capture chronic myocardial ischemia due to changes in microvascular blood flow, and thus provide insight into vascular endothelial integrity within the heart, where thrombus formation occurs (32).

In addition to indicators of possible cardiac structural changes, biomarkers that directly reflect activity of the coagulation cascade and hemostasis have been investigated in both community-based and clinical trial cohorts of patients with AF (7, 30, 33). In particular, D-dimer, a fibrin degradation product that is most often used clinically for venous thromboembolic risk stratification, is a direct indicator of risk of thrombogenesis in patients with AF as well. In the ENGAGE AF-TIMI 48 trial, for example, baseline D-dimer predicted S/SEE risk during trial follow-up and improved the prognostic performance of the risk model when added to the clinical variables in the CHA₂DS₂-VASc score (30).

In addition to biomarkers of hemodynamic stress, myocardial injury, and coagulation activity, biomarkers of renal dysfunction, such as eGFR and cystatin C, have also been associated with thromboembolic risk. On the other hand, they are even stronger predictors of bleeding risk in patients with AF, thus limiting their clinical utility for S/SEE risk prediction (Fig. 2). Similarly, biomarkers of inflammation (e.g., CRP and interleukin-6) have been investigated in this clinical context, but the strength of their associations with S/SEE risk are more modest when compared with BNP/NT-proBNP, cTn, and D-dimer (34).

Clinical Applications

Although S/SEE risk scores that use clinical variables only (e.g., CHA₂DS₂-VASc) are endorsed by professional society guidelines and are widely used in clinical practice, their ability to accurately assign risk is limited (31, 35). Consequently, incorporating circulating biomarkers into clinical risk tools holds the promise of improving S/SEE risk prediction, perhaps by capturing relevant pathophysiologic mechanisms more effectively than do dichotomous clinical variables alone. Moreover, since biomarkers representing distinct pathobiological axes appear to provide complementary information, the notion of multi-marker risk stratification has particular appeal for prediction of S/SEE in patients with AF.

In a subset of patients from the ENGAGE AF-TIMI 48 trial (n = 4880), a multi-marker risk score incorporating a conventional assay for cTnI, NT-proBNP, and D-dimer was developed to estimate the probability of S/SEE or death. When these 3 biomarkers were added to the CHA₂DS₂-VASc score, the combined clinical and biomarker risk model had significantly better discrimination than did the CHA₂DS₂-VASc model alone, with an improvement in the c-index from 0.59 (95% CI 0.57–0.61) to 0.71 (95% CI 0.69–0.73;P < 0.001). In a sensitivity analysis using the endpoint of S/SEE alone (i.e., excluding all-cause mortality), the model incorporating all 3 biomarkers remained significantly better than the model with only clinical variables (0.66 [95% CI 0.63–0.70] vs. 0.58 [95% CI 0.55-0.62]) (30).

Given concerns about the stability of D-dimer measurements over even the short-term, and the observation that, in a multivariable model, D-dimer had the weakest association with S/SEE of the 3 biomarkers, other biomarker-based risk scores have restricted their focus to cTn and NPs alone. For example, the age, biomarker, clinical history (ABC)-stroke score, derived in a cohort of 14 701 anticoagulated patients from the ARISTOTLE trial, includes just 4 variables: age, high-sensitivity cTn (I or T), NT-proBNP, and prior history of stroke or transient ischemic attack. This parsimonious risk score significantly outperformed the CHA₂DS₂-VASc in that cohort (c-index 0.68 vs. 0.62; P < 0.001) as well as in a smaller validation cohort from the STABILITY trial (n = 1400) (35). Moreover, in an independent external validation of the ABC-stroke score in 8705 anticoagulated patients from the ENGAGE AF-TIMI 48 trial, the score was well-calibrated and again outperformed the CHA₂DS₂-VASc score (c-index 0.67 [95% CI 0.65–0.70] vs. 0.59 [95% CI 0.57–0.62]; P < 0.001) (Fig. 4) (31). The scores can be conveniently applied using online calculators (https://www.ucr.uu.se/en/services/abc-risk-calculators) and have now been proposed for routine use (36), but several key questions remain. First, since all studies evaluating the performance of the ABC-stroke score have been conducted in anticoagulated clinical trial cohorts, it is unknown how the score would perform in a population of patients with AF not receiving anticoagulation. Additional validation studies are needed in low-risk patients (i.e., CHA₂DS₂-VASc 0–1) not on anticoagulation, in whom improved assessment of S/SEE risk may be particularly clinically actionable. An ongoing prospective randomized trial is testing whether personalized treatment decision-support using the ABC-stroke and ABC-bleeding (see next section) scores improves outcomes in patients with AF, including in non-anticoagulated patients with newly discovered AF (NCT03753490). Second, the optimal timing and frequency for remeasuring biomarkers to refine risk prediction over time is unknown and deserves further study. For example, while there appears to be minimal fluctuation in biomarker concentrations over two months (37), there is more variability at one year (31).

Predicting Major Bleeding in Patients with Atrial Fibrillation

Pathophysiologic Mechanisms and Circulating Biomarkers of Bleeding

In parallel with assessing S/SEE risk, clinicians must also evaluate bleeding risk when making decisions about initiating or continuing oral anticoagulation (4, 5). Complicating this risk-benefit calculation is the fact that many of the clinical risk factors for major bleeding are also risk factors for stroke (e.g., older age, prior stroke, renal dysfunction) (38). Similarly, some cardiovascular biomarkers, such as cTn, also appear to associate with both S/SEE and bleeding risk. For example, baseline cTnI concentration was strongly associated with International Society of Thrombosis and Hemostasis major bleeding among patients enrolled in the ARISTOTLE trial, irrespective of treatment group assignment (apixaban vs. warfarin) (9). Likewise, in the ENGAGE AF-TIMI 48 biomarker cohort, baseline cTnT concentration identified a gradient of risk for International Society of Thrombosis and Hemostasis major bleeding that remained significant after adjusting for the components of the HAS-BLED score (age, history of hypertension, history of abnormal renal or liver function, history of stroke or transient ischemic attack, history of major bleeding, medication use predisposing to bleeding, and alcohol use) (Fig. 3) (31). Mechanistically, the association of cTn with both S/SEE and major bleeding risk may reflect its role as a biomarker of cardiovascular endothelial integrity. As a practical matter, though, this dual association has motivated interest in identifying other biomarkers that more specifically reflect bleeding risk.

The marker that has earned the most attention as a risk indicator of major bleeding in patients receiving antithrombotic therapy is growth differentiation factor-15 (GDF-15), a member of the transforming growth factor (TGF)-ß family that is secreted by a broad range of cell types in response to tissue hypoxia and oxidative stress (29, 31, 39). In both the ARISTOTLE and ENGAGE AF-TIMI 48 biomarker substudies, GDF-15 was significantly associated with major bleeding even after adjusting for the clinical components of the HAS-BLED score, cTn, and NT-proBNP (Fig. 3) (29, 31).

Clinical Applications

As with the development of the ABC-stroke score, the recognition that circulating biomarkers could improve assessment of major bleeding risk in patients receiving anticoagulation for AF motivated the development of an analogous risk score combining clinical and biomarker variables. Derived in 14 537 patients in the ARISTOTLE trial, the ABC-bleeding score incorporates age, history of bleeding, and 3 biomarkers (hemoglobin, GDF-15, and high-sensitivity cTn) (40). Compared with the HAS-BLED score and a clinical bleeding risk model called the ORBIT score, the ABC-bleeding score improved discrimination for predicting major bleeding in this clinical trial cohort (c-index 0.68 [95% CI 0.66–0.70] for ABC-bleeding vs. 0.61 [95% CI 0.59–0.63] for HAS-BLED vs. 0.65 [95% CI 0.62–0.67] for ORBIT; P < 0.001 for each comparison). When assessed in external validation cohorts from the RE-LY (n = 8468) and ENGAGE AF-TIMI 48 (n = 8705) trials, the ABC-bleeding score also significantly outperformed the HAS-BLED score (c-index 0.71 [95% CI 0.68–0.73] vs. 0.62 [95% CI 0.59–0.64] and 0.69 [95% CI 0.66–0.71] vs. 0.62 [95% CI 0.60–0.64], respectively; P < 0.001 for both) (Fig. 4) (31, 40).

One of the most important findings from the external validation study in the ENGAGE AF-TIMI 48 trial was that use of the ABC-bleeding score, as compared with the HAS-BLED score, resulted in predominantly correct downward reclassification of bleeding risk. Since underuse of anticoagulation—often related to overestimation of patient bleeding risk by physicians—is a major driver of morbidity and mortality in patients with AF, this finding suggests that clinical use of the ABC-bleeding score may be particularly helpful for mitigating this treatment gap by providing reassurance to clinicians when the risk of major bleeding is lower than predicted by clinical risk factors alone.

Precision Medicine in Atrial Fibrillation

When clinicians counsel patients with AF about whether to initiate anticoagulation therapy for thromboprophylaxis, they must ultimately integrate a complex set of factors that includes assessment of the patient’s S/SEE risk, bleeding risk, and of course, patient preference. In an analysis of the performance of the ABC scores in ENGAGE AF-TIMI 48, simultaneous application of the ABC-stroke and ABC-bleeding risk scores had the additional advantage of identifying patients who are most likely to derive a benefit from treatment with non-vitamin K antagonist oral anticoagulants compared with warfarin. Specifically, among patients at low risk for both stroke and bleeding (as predicted by the ABC scores), there was no meaningful difference in the rates of the net clinical outcome of S/SEE, major bleeding, or all-cause mortality when they were treated with a high-dose edoxaban regimen, low-dose edoxaban regimen, or warfarin (31). By contrast, among patients with either high ABC-stroke scores or high ABC-bleeding scores, the edoxaban regimens were clearly superior to warfarin (31). As an additional nuance, patients in whom the ABC scores predicted a high risk of both stroke and bleeding had the best outcomes with the higher-dose non-vitamin K antagonist oral anticoagulant regimen, and those in whom the ABC scores predicted a low risk of stroke but a high risk of bleeding had the best outcomes with the lower-dose regimen. Thus, by enhancing the precision of estimated S/SEE and bleeding risks and directly informing treatment decisions, biomarker-based risk scores like the ABC-stroke and ABC-bleeding scores have the potential to support the movement towards precision medicine in the care of patients with AF.

Conclusions

Multiple studies of community-based and clinical trial cohorts have clearly demonstrated that circulating biomarkers improve risk prediction of AF incidence and AF-related complications of S/SEE and major bleeding. Furthermore, multi-marker approaches that incorporate biomarkers reflecting distinct pathobiological axes appear to offer even better discrimination. The ABC-stroke and ABC-bleeding risk scores are parsimonious risk tools that improve risk stratification of S/SEE and major bleeding, respectively, when compared with clinical risk scores like the CHA₂DS₂-VASc and HAS-BLED scores. Although current clinical practice guidelines continue to embrace these traditional clinical risk scores, the emerging evidence base supporting biomarker-based risk tools may lead to a shift in the clinical management of patients with AF.

Author Contributions

All authors confirmed they have contributed to the intellectual content of this paper and have met the following 4 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved.

Authors’ Disclosures or Potential Conflicts of Interest

Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:

Employment or Leadership

None declared.

Consultant or Advisory Role

C.T . Ruff, Roche; D. A. Morrow, AstraZeneca, Bayer Pharmaceuticals, InCarda, Merck and Co., Novartis, Roche Diagnostics.

Stock Ownership

None declared.

Honoraria

None declared.

Research Funding

D.D. Berg, C.T. Ruff, and D.A. Morrow are members of the TIMI Study Group which has received institutional research grant support through Brigham and Women’s Hospital from: Abbott Diagnostics, Amgen, Anthos Therapeutics, AstraZeneca, Bayer HealthCare Pharmaceuticals, Inc., BRAHMS, Daiichi-Sankyo, Eisai, GlaxoSmithKline, Intarcia, Janssen, Merck, Novartis, Pfizer, Poxel, Regeneron, Roche Diagnostics, Siemens, and Takeda.

Expert Testimony

None declared.

Patents

None declared.

References

Nonstandard Abbreviations:

 
• AF

  atrial fibrillation

 
• BNP

  B-type natriuretic peptide

 
• CI

  confidence interval

 
• CRP

  C-reactive protein

 
• cTn

  cardiac troponin

 
• GDF-15

  growth differentiation factor-15

 
• NP

  natriuretic peptides

 
• NT-proBNP

  N-terminal-pro-B-type natriuretic peptide

 
• S/SEE

  stroke and systemic embolic events

 
• TGF-ß

  transforming growth factor-beta