Effects of Atrial Fibrillation and Chronic Kidney Disease on Major Adverse Cardiovascular Events

Residual risk prediction in anticoagulated patients with atrial fibrillation using machine learning: A report from the GLORIA-AF registry phase II/III

BACKGROUND

Although oral anticoagulation decreases the risk of thromboembolism in patients with atrial fibrillation (AF), a residual risk of thrombotic events still exists. This study aimed to construct machine learning (ML) models to predict the residual risk in these patients.

METHODS

Patients with newly diagnosed non-valvular AF were collected from the Global Registry on Long-Term Oral Anti-Thrombotic Treatment in Patients with Atrial Fibrillation (GLORIA-AF) registry. To predict the residual risk of the composite outcome of thrombotic events (defined as ischemic stroke, systemic embolism, transient ischemic attack and myocardial infarction), we constructed four prediction models using the logistic regression (LR), random forest, light gradient boosting machine and extreme gradient boosting machine ML algorithms. Performance was mainly evaluated by area under the receiver-operating characteristic curve (AUC), g-means and F1 scores. Feature importance was evaluated by SHapley Additive exPlanations.

RESULTS

15,829 AF patients (70.33 ± 9.94 years old, 55% male) taking oral anticoagulation were included in our study, and 641 (4.0%) had residual risk, sustaining thrombotic events. In the test set, LR had the best performance with higher AUC trend of 0.712. RF has highest g-means of 0.295 and F1 score of 0.249. This was superior when compared with the CHA2DS2-VA score (AUC 0.698) and 2MACE score (AUC 0.696). Age, history of TE or MI, OAC discontinuation, eGFR and sex were identified as the top five factors associated with residual risk.

CONCLUSION

ML algorithms can improve the prediction of residual risk of anticoagulated AF patients compared to clinical risk factor-based scores.

Validating the predictive ability of the 2MACE score for major adverse cardiovascular events in patients with atrial fibrillation: results from phase II/III of the GLORIA-AF registry

The 2MACE score was specifically developed as a risk-stratification tool in atrial fibrillation (AF) to predict cardiovascular outcomes. We evaluated the predictive ability of the 2MACE score in the GLORIA-AF registry. All eligible patients from phase II/III of the prospective global GLORIA-AF registry were included. Major adverse cardiac events (MACEs) were defined as the composite outcome of stroke, myocardial infarction and cardiovascular death. Cox proportional hazards were used to examine the relationship between the 2MACE score and study outcomes. Predictive capability of the 2MACE score was investigated using receiver-operating characteristic curves. A total of 25,696 patients were included (mean age 71 years, female 44.9%). Over 3 years, 1583 MACEs were recorded. Patients who had MACE were older, with more cardiovascular risk factors and were less likely to be managed using a rhythm-control strategy. The median 2MACE score in the MACE and non-MACE groups were 2 (IQR 1-3) and 1 (IQR 0-2), respectively (p < 0.001). The 2MACE score was positively associated with an increase in the risk of MACE, with a score of ≥ 2 providing the best combination of sensitivity (69.6%) and specificity (51.6%), HR 2.47 (95% CI, 2.21-2.77). The 2MACE score had modest predictive performance for MACE in patients with AF (AUC 0.655 (95% CI, 0.641-0.669)). Our analysis in this prospective global registry demonstrates that the 2MACE score can adequately predict the risk of MACE (defined as myocardial infarction, CV death and stroke) in patients with AF. Clinical trial registration: http://www.clinicaltrials.gov . Unique identifiers: NCT01468701, NCT01671007 and NCT01937377.

Incidence and Determinants of Spontaneous Cardioversion of Early Onset Symptomatic Atrial Fibrillation

Atrial fibrillation (AF) is the most frequent chronic arrhythmia worldwide, and it is associated with significant morbidity and mortality, making it a considerable burden both to patients and the healthcare system. Nowadays, an early attempt to restore sinus rhythm in acute symptomatic AF through electrical or pharmacological cardioversion is the most common approach in the Emergency Department (ED). However, considering the high percentage of spontaneous cardioversion of paroxysmal AF reported by many studies, this approach may not be the ideal choice for all patients. In this manuscript we performed a review of the most relevant studies found in literature with the aim of identifying the main determinants of spontaneous cardioversion, focusing on those easy to detect in the ED. We have found that the most relevant predictors of spontaneous cardioversion are the absence of Heart Failure (HF), a small atrial size, recent-onset AF, rapid Atrial Fibrillatory Rate and the relationship between a previous AF episode and Heart Rate/Blood Pressure. A number of those are utilized, along with other easily determined parameters, in the recently developed "ReSinus" score which predicts the likelihood of AF spontaneous cardioversion. Such identification may help the physician decide whether immediate cardioversion is necessary, or whether to adopt a "watch-and-wait" strategy in the presence of spontaneous cardioversion determinants.

Relationship between temporal rhythm-based classification of atrial fibrillation and stroke: real-world vs. clinical trial

Background

The risk of stroke according to clinical classification of atrial fibrillation (AF) remains poorly defined. Here, we assessed the impact of AF type on stroke risk in vitamin K antagonist-treated patients with AF in ‘real-world’ and ‘clinical trial’ cohorts.

Methods

Post-hoc analysis of patient-level data from the Murcia AF Project and AMADEUS trial. Clinical classification of AF was based on contemporary recommendations from international guidelines. Study endpoint was the incidence rate of ischaemic stroke. Stroke risk was determined using CHA₂DS₂-VASc score and CARS. A modified CHA₂DS₂-VAS‘c’ score that applied one additional point for a ‘c’ criterion of continuous AF (i.e. non-paroxysmal AF) was calculated.

Results

We included 5,917 patients: 1,361 (23.0%) real-world and 4,556 (77.0%) clinical trial. Baseline demographics were balanced in the real-world cohort but clinical trial participants with non-pAF (vs. pAF) were older, male-predominant and had more comorbidities. Crude stroke rates were comparable between the groups in real-world patients (IRR 0.72 [95% CI,0.37-1.28], p = 0.259) though clinical trial participants with non-pAF had a significantly higher crude rate of stroke events (IRR 4.66 [95%,CI,2.41-9.48], p < 0.001). Using multivariable analysis, AF type was not independently associated with stroke risk in the real-world (adjusted HR 1.41 [95% CI,0.80-2.50], p = 0.239) and clinical trial (adjusted HR 1.16 [95% CI,0.62-2.20], p = 0.646) cohorts, after accounting for other risk factors. There was no significant improvement in the CHA₂DS₂-VAS‘c’ compared to CHA₂DS₂-VASc score in either cohorts (p > 0.05).

Conclusions

Overall, our results support the need for anticoagulation based on thromboembolic risk profile rather than AF type.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11239-022-02638-0.

Real world data of anticoagulant treatment in non-valvular atrial fibrillation across renal function status

The objective is to assess the impact of anticoagulant treatment in non-valvular atrial fibrillation (AF) and different categories of renal dysfunction in real world. Electronic Health recordings of patients with diagnosis of AF and renal function collected throughout 5 years and classified according to KDIGO categories. Stroke, transitory ischemic attack (TIA), intracranial hemorrhage and all-cause mortality were identified. Anticoagulant treatments during the study period were classified in untreated (never received therapy), VKA, NOAC and Aspirin. The risk of events was calculated by Cox-proportional hazard models adjusted by confounders. A total of 65,734 patients with AF, mean age 73.3 ± 10.49 years old and 47% females and follow-up of 3.2 years were included. KDIGO classification were: G1 33,903 (51.6%), G2 17,456 (26.6%), G3 8024 (12.2%) and G4 6351 (9.7%). There were 8592 cases of stroke and TIA, 437 intracranial hemorrhage, and 9603 all-cause deaths (incidence 36, 2 and 38 per 103 person/year, respectively). 4.1% of patients with CHA2DS2-VASc Score 2 or higher did not receive anticoagulant therapy. Risk of stroke, TIA, and all-cause mortality increased from G1 to G4 groups. Anticoagulant treatments reduced the risk of events in the four categories, but NOAC seemed to offer significantly better protection. Renal dysfunction increases the risk of events in AF and anticoagulant treatments reduced the risk of stroke and all-cause mortality, although NOAC were better than VKA. Efforts should be done to reduce the variability in the use of anticoagulants even in this high risk group.

Chronic kidney disease and risks of adverse clinical events in patients with atrial fibrillation

BACKGROUND

Chronic kidney disease (CKD) is highly prevalent in patients with atrial fibrillation (AF). However, the association between CKD and clinical consequences in AF patients is still under debate.

METHODS

We included 19,079 nonvalvular AF patients with available estimated glomerular filtration rate (eGFR) values in the Chinese Atrial Fibrillation Registry from 2011 to 2018. Patients were classified into no CKD (eGFR ≥ 90 mL/min per 1.73 m²), mild CKD (60 ≤ eGFR < 90 mL/min per 1.73 m ²), moderate CKD (30 ≤ eGFR < 60 mL/min per 1.73 m ²), and severe CKD (eGFR < 30 mL/min per 1.73 m ²) groups. The risks of thromboembolism, major bleeding, and cardiovascular mortality were estimated with Fine-Gray regression analysis according to CKD status. Cox regression was performed to assess the risk of all-cause mortality associated with CKD.

RESULTS

Over a mean follow-up of 4.1 ± 1.9 years, there were 985 thromboembolic events, 414 major bleeding events, 956 cardiovascular deaths, and 1,786 all-cause deaths. After multivariate adjustment, CKD was not an independent risk factor of thromboembolic events. As compared to patients with no CKD, those with mild CKD, moderate CKD, and severe CKD had a 45%, 47%, and 133% higher risk of major bleeding, respectively. There was a graded increased risk of cardiovascular mortality associated with CKD status compared with no CKD group: adjusted hazard ratio [HR] was 1.34 (95% CI: 1.07-1.68,P = 0.011) for mild CKD group, 2.17 (95% CI: 1.67-2.81,P < 0.0001) for moderate CKD group, and 2.95 (95% CI: 1.97-4.41, P < 0.0001) for severe CKD group, respectively. Risk of all-cause mortality also increased among patients with moderate or severe CKD.

CONCLUSIONS

CKD status was independently associated with progressively higher risks of major bleeding and mortality, but didn't seem to be an independent predictor of thromboembolism in AF patients.