Idiopathic atrial fibrillation patients rapidly outgrow their low thromboembolic risk: a 10-year follow-up study

The studied population has been described previously [3]. In brief, we studied 41 consecutive idiopathic paroxysmal AF patients and 45 healthy sinus rhythm (SR) control patients who were referred to the outpatient clinic of one of the authors between 2004 and 2007 for a standard transthoracic echocardiographic examination (in the work-up for AF or for cardiovascular screening purposes in the SR control population). Individual AF treatment and follow-up/discharge was left to the discretion of the treating physician. The study was approved by the local Ethics Committee and complied with the Declaration of Helsinki. Informed consent of all patients was obtained. In the current report we extended the total follow-up duration and focused on the anticoagulation strategy.

Idiopathic AF and healthy SR were strictly defined as the absence of any CVD including hypertension, diabetes, or hypercholesterolaemia. In addition, no coronary artery disease (from absence of typical exercise-related angina pectoris and beyond), no peripheral vascular disease, no congestive heart failure (clinical signs or symptoms, ejection fraction > 50%, normal diastolic function), no previous stroke, no thyroid or pulmonary disease, preserved renal function (MDRD > 60 ml/min/1.73m²), no malignancy, and no evidence for structural CVD on echocardiography. Female gender was not included in CHA₂DS₂-VASc in the absence of other stroke risk factors [5, 6].

The combined endpoint was defined as all emerging CVDs, including all-cause death, thromboembolic complications, congestive heart failure (left ventricular ejection fraction, LVEF, < 45%, admission), significant coronary artery disease, diabetes mellitus, and new onset hypertension (need for antihypertensive drugs, or development of hypertrophy > 10 mm). Major adverse cardiac and cerebrovascular events (MACCE) were defined as cardiovascular death, stroke, myocardial infarction or repeated revascularisation, and the occurrence of acute heart failure.

Patient characteristics at the time of echocardiography and during follow-up were obtained. At the end of follow-up, general practitioners and anticoagulation clinics were contacted in order to obtain additional follow-up data.

Statistical analysis was performed using SPSS statistical software (IBM SPSS statistics version 23.0, IBM Corporation, Armonk, NY, USA). Categorical variables are reported as number of patients and percentage and were compared using chi-square testing. Continuous variables are presented as mean ± standard deviation or median (interquartile range) and were compared with an independent t-test (two-tailed) for normally distributed continuous variables and Mann–Whitney test (two-tailed) for all not normally distributed variables. All parameters showing a significant univariable relation with the occurrence of CVD during follow-up using Cox regression were included as covariates in a multivariate Cox regression model. Results were checked for colinearity and interaction among covariates. Proportional hazards were checked. Manual backwards selection was used to construct the final models (retention level set at 0.10). Hazard ratios (HR) and 95% confidence intervals (CI) were calculated. A p-value of 0.05 was considered statistically significant.

Baseline characteristics were the same in AF cases and controls (Tab. 1; [3]).

Tab. 2 shows follow-up characteristics. Follow-up was incomplete in 4 patients (mean follow-up in these patients was 4.8 ± 1.1 years). During a mean period of 10.7 ± 1.6 years from baseline, patients with AF died or developed CVD significantly more often than SR controls (63% vs 31%, log rank p < 0.001). Even after the initial 5 years of follow-up, survival curves show divergent patterns (log rank p = 0.006). Log rank p-values for true MACCE (all-cause death excluded) are 0.005 for complete follow-up and 0.034 after the initial 5 years of follow-up. The Kaplan-Meier curve with cumulative incidence of CVD during the first 5 years of follow-up was published previously [3].

Multivariable Cox regression analysis revealed that baseline posterior wall width (8.3 ± 0.7 vs 8.7 ± 0.8 mm, p = 0.013. HR 1.9; 1.2–2.9, p = 0.003), a history of AF (38% vs 64%, p = 0.019. HR 2.6; 1.3–4.9, p = 0.005), and age (50 ± 13 vs 57 ± 8 years, p = 0.012. HR 1.0; 1.0–1.1, p = 0.018) were independent significant predictors of CVD development.

During follow-up a total of 4 SR patients developed AF. As a result, there were 45 patients with a diagnosis of AF to study the prescription of antithrombotic therapy at the end of follow-up. Two female patients (aged 65 and 67 years) had a CHA₂DS₂-VASc score of 2 at baseline; these patients were excluded from calculation of duration to reach a CHA₂DS₂-VASc score of 2.

Among 45 AF patients, a total of 35 (78%) developed a CHA₂DS₂-VASc score of 1 or higher, of which 24 (53%) developed a CHA₂DS₂-VASc of 2 or higher. A total of 10 (22%) patients retained their low thromboembolic risk (CHA₂DS₂-VASc 0). Fig. 1 shows characteristics regarding the development and distribution of increased CHA₂DS₂-VASc score during follow-up, as well as the prescription of (novel) oral anticoagulants, (N)OAC.

A total of 24 of 35 patients with CHA₂DS₂-VASc ≥ 1 received (N)OAC. This treatment was started within 12 months prior to or after an increased thromboembolic risk score in only 8 patients, and 5 of these patients had a CHA₂DS₂-VASc score of 1. In 11 (31%) of 35 patients, (N)OAC were started while CHA₂DS₂-VASc was still 0 at that timepoint. Overall the mean duration of over- or undertreatment with oral anticoagulation was 5.0 years (±3.0) when CHA₂DS₂-VASc > 1 was used as treatment cut-off and 4.2 years (±3.0) when CHA₂DS₂-VASc = 1 was used. Fig. 2 shows the duration of increased thromboembolic risk score, start of anticoagulation therapy and the occurrence of MACCE for individual patients with CHA₂DS₂-VASc ≥ 1.

More than half of the ‘healthy’ AF patients developed a cardiovascular condition during 10 years of follow-up. In fact, the average time to develop a CHA₂DS₂-VASc score ≥ 1 was roughly only 4 years.

Although patients with hypertension or left ventricular hypertrophy were excluded, posterior wall width at baseline (8.3 vs 8.7 mm) was a predictor of CVD. This suggests the presence of masked or prehypertension in patients participating in this study, and that even normal wall sizes show a variation that potentially represents preclinical disease once wall width is found to be at the high end of normal. This emphasises the importance of regular 24-h ambulatory blood pressure monitoring and adequate treatment of early hypertension in patients with idiopathic AF, since it relates to their vascular prognosis [7].

The patients were treated according to common daily practice (not specialised AF care). Patient dismissal, follow-up and (anticoagulant) treatment were left to the discretion of the treating physician. Interestingly, vitamin K antagonists (OAC) or NOAC were often (34%) prescribed when a patient still had CHA₂DS₂-VASc 0 (i. e. preceding elective electrical cardioversion) and not stopped subsequently. Further, 5 (21%) patients with CHA₂DS₂-VASc ≥ 2 were not receiving (N)OAC at the end of the follow-up period. An explanation might be that some patients did not have an indication for anticoagulation at baseline according to previous risk schemes (Birmingham risk schema and CHADS₂) and that patients were discharged from regular follow-up due to their healthy profile. These numbers are, however, comparable with the recent results of GARFIELD-AF and GLORIA-AF and confirm that there is still much to be gained regarding guideline-adherent AF management in common daily practice [8‐11].

Anticoagulation therapy reduces morbidity and mortality in patients with AF who are at risk for thromboembolism. Since NOAC are associated with low bleeding risk and thromboembolic risk increases within a short time frame in recently diagnosed AF patients with a low CHA₂DS₂-VASc score, it is worth considering starting anticoagulation in these patients irrespective of their thromboembolic risk score. Based on currently available data, the net clinical benefit of such a treatment approach is negative due to increased bleeding rates and low baseline thromboembolic rates (stroke risk 0.3–0.7% per year in CHA₂DS₂-VASc score 0 vs NOAC major bleeding rates: 1.4–3.6%) [12‐15]. It must be stated that the average age in these CHA₂DS₂-VASc score 0 populations is somewhat lower than the average age in our study. A randomised trial exploring the potential benefit of (low-dose?)—NOAC in CHA₂DS₂-VASc score 0 AF patients aged 50–64 years would certainly add to this consideration but would require a large study population [16].

A recently proposed novel score predicting net clinical outcome may improve risk prediction in this (especially low-risk) population, and could help guide the selection of anticoagulants [17]. Moreover, there are currently many efforts to redefine AF types into biologically plausible subtypes and to improve identification of those patients who currently have only ‘subclinical’ stroke risk factors (prehypertension, prediabetes, age 50–64 years), yet harbour a considerable stroke risk—for example through imaging or biomarker studies [18, 19].

Physicians should be encouraged to give high priority to detecting early stages of underlying CVD and to set up a solid individual follow-up plan. Review by a specialised AF service is necessary and follow-up may very well be undertaken in primary care or ideally in a nurse-led AF clinic, as these have been shown to improve survival and quality of life and are a cost-effective management strategy for patients with AF [20, 21]. The implementation of E‑health may actively involve patients in their own care and improve guideline-adherent AF management. At least the individual low-risk AF patient should be carefully instructed to contact their AF specialist in case of disease progression, age > 65 years, or the emergence of hypertension or diabetes mellitus in order to adjust antithrombotic therapy in a timely fashion.

In current practice, many of the patients described in this study are discharged from follow-up. We would like to emphasise the need to continue assessing their stroke risk, as CVD may soon emerge. Annual visits may be too large a burden on healthcare systems, with increasingly limited resources. Therefore, according to the findings in the present study we chose a 2- to 3‑year interval, leaving more frequent visits to the treating physicians’ discretion.

The study results were obtained in a relatively small and—because of our research focus—highly selected population. Nevertheless, it seems reasonable to state that a substantial number of AF patients originally diagnosed with a low thromboembolic risk develop CVD during short-term follow-up. Since SR patients were referred for cardiovascular screening purposes rather than AF, referral bias may have played a role but would—if anything—have led to a higher incidence of events in the control population rather than the reverse. Although we meticulously tried to rule out hypertension, specific cases of masked hypertension may have been missed, since 24-h ambulatory blood pressure monitoring was not performed.