Introduction

Stroke is, according to the World Health Organization, the second leading cause of death worldwide [1, 2]. In 2017, 8% of the deaths in Europe were attributed to acute stroke. Moreover, stroke survivors often suffer from severe disability, which in turn generates an elevated use of health and social care resources, up to 8% [2]. Improving the current methods of stroke risk stratification may aid clinical decision-making when considering stroke prevention therapies, thereby reducing the burden of stroke on patients and resources.

Atrial cardiopathy has recently emerged as a significant precursor for stroke, which has led to the suggestion of a new model for the mechanism of stroke pathogenesis in atrial fibrillation (AF): aging and systemic vascular risk factors cause abnormal atrial tissue substrate (atrial cardiopathy), which results in both AF and thromboembolism [3]. Of note, it is well established that the risk of stroke is strongly associated with AF [4].

The total atrial conduction time (TACT), measured noninvasively as PA-TDI interval (the time interval from initiation of the P wave to the peak of the A’ wave of the atrial tissue Doppler tracing) [5, 6], has been shown to be correlated with the degree of atrial fibrosis and predict the occurrence of postoperative AF after cardiac surgery[5] as well as predict stroke risk in patients with AF, independent of the CHA2DS2-VASc score [6].

In the general population, stroke risk can be calculated using the revised Framingham stroke risk profile (rFSRP), which has been recently described and validated in two external cohorts (including patients without a history of AF) [7]. However, the rFSRP may not directly reflect atrial cardiopathy, as this scoring system does not include any echocardiographic or electrocardiographic parameters [8,9,10].

Therefore, we hypothesized that the TACT, as a surrogate parameter for atrial cardiopathy, improves the efficacy of stroke prediction when added to the rFSRP in patients without known AF.

Materials and methods

Study ethics

The study protocol was approved by the local ethics committee of Heinrich Heine University Düsseldorf (Study Number 5228R). Informed written consent was obtained from all study participants.

Study population

We prospectively enrolled consecutive patients in our echocardiography laboratory. Patients were considered eligible if they underwent transthoracic echocardiography, were older than 18 years of age, and provided informed consent before inclusion. Patients were excluded if they had a history of stroke (ischemic or hemorrhagic) or AF/atrial flutter (at least one episode lasting longer than 30 s). Patients with a history of hospital admission due to decompensated heart failure within 6 months of the date of recruitment were also excluded.

Comorbidities, medications, and epidemiological data were recorded. In addition to the echocardiographic assessment, we calculated the rFSRP to determine stroke risk at 2 years [7]. The follow-up period was planned to be 24 months.

Echocardiography

A Vivid 7 machine (GE Healthcare, Chicago, IL, USA) was used to perform echocardiography. Two-dimensional transthoracic echocardiography was performed in all patients, and echocardiographic parameters were assessed according to the guidelines of the American Society of Echocardiography and were supervised by experienced cardiologists certified in adult transthoracic echocardiography by the European Association of Cardiovascular Imaging [11]. Essentially, left atrial (LA) diameter was measured at the end-systolic time point in the parasternal long axis view, and the left ventricular ejection fraction was measured using the bi-plane method of disks (modified Simpson’s rule, apical four-chamber and two-chamber views, respectively).

The TACT was estimated using the PA-TDI interval. This was measured twice for every patient, and the mean value was then calculated. In the apical four-chamber view, the pulsed-wave tissue Doppler sample was placed on the lateral wall of the LA above the mitral annulus. The time interval from initiation of the P wave (lead II as recorded by the echocardiography machine) to the peak of the A’ wave of the atrial tissue Doppler tracing, as shown in Online Resource 1, was defined as the PA-TDI interval, as previously described [12].

Follow-up

The primary outcome was defined as the occurrence of ischemic stroke (both major and minor [13]) during the 2 years of follow-up. The secondary outcome was defined as newly diagnosed AF (at least one episode lasting longer than 30 s) within 2 years.

Major stroke was defined as an episode of neurological dysfunction caused by focal cerebral, spinal, or retinal infarction, with symptoms persisting for longer than 24 h [14]. Minor stroke was diagnosed in patients with a National Institutes of Health Stroke Scale score of ≤ 3 points. AF was diagnosed by 12-lead electrocardiogram (ECG) and Holter monitor.

A detailed overview of the inclusion process and follow-up is presented in Online Resource 2 (panel A).

The necessary sample size was calculated prior to patient recruitment. Specifically, we hypothesized that patients with prolonged TACT will have a higher incidence of stroke than patients without TACT prolongation during the 2 years of follow-up. We estimated the annual incidence of patients with and without TACT prolongation, and without known AF, as 2.5% and 0.25%, respectively, based on previous work from other groups [6, 15, 16]. For a 2 years follow-up period, with a dropout rate of 15% and one-tailed testing, a total of 367 participants were calculated to be necessary (statistical power of 0.8).

Statistics

Data were analyzed using SPSS statistical software version 23 (SPSS Inc., Chicago, IL, USA). Continuous variables are expressed as mean ± standard deviation (or median and interquartile range [IQR]), and categorical parameters are expressed as counts and percentages. Normal distribution was assessed using the Shapiro–Wilk test. Study participants were divided by outcome, primarily whether they had suffered a stroke in the follow-up period of 24 months. We compared the demographics, clinical characteristics, and echocardiography data of both groups using the t test for continuous variables and Fisher’s exact test for categorical variables. Univariate and multivariate Cox regression analyses were performed to assess predictive values for ischemic stroke. Receiver operating characteristic (ROC) curve analysis was performed for rFSRP, TACT (via the PA-TDI method) alone and in combination, as well as for the CHA2DS2-VASc score. In order to achieve this, we combined the rFSRP and TACT into a single composite score using logistic regression model and evaluate the predictability of markers by the one-dimensional composite score [17]. Statistical significance was set at p < 0.05.

Results

Patients included

Based on the power calculation, we screened 408 patients with informed consent obtained from 393 patients. As shown in Online Resource 2 (panel A), follow-up was completed in 376 patients. We included only 376 patients who completed the follow-up in our analysis and the subsequent description of the results. Seventeen patients who were lost to follow-up (4.3%) demonstrated no significant differences in terms of clinical characteristics compared with those with complete follow-up. Baseline clinical characteristics are summarized in Online Resource 3. The mean age of our cohort was 58.5 years (median, 61 years; IQR, 23), and 54% (n = 203) were male.

Cardiovascular events during follow-up

During the follow-up period of 2 years, 10 patients (2.6%) experienced an event: 6 minor strokes and 4 major strokes, all of which were ischemic. Twenty-two patients were newly diagnosed with AF. Of note, both a diagnosis of stroke and new AF were noted in two patients.

The clinical characteristics of the patients without subsequent stroke (n = 366 [97.4%]) compared to those with stroke (n = 10 [2.6%]) demonstrated significant differences with regard to age (61 vs. 75, p = 0.003), peripheral arterial disease (10.9 vs. 40%, p = 0.005), beta-blocker medication use (36.9 vs. 70%, p = 0.03), and virtual CHA2DS2-VASc score (Table 1). The rFSRP at 2 years was significantly higher in patients with subsequent strokes than in those without (1.93 vs. 0.33%, p = 0.002), and the TACT measured via PA-TDI was 169 and 143 ms, respectively (p = 0.001).

Table 1 Follow-up stroke
Full size table

The clinical characteristics of the patients with newly diagnosed AF (n = 22) during follow-up and those without AF (n = 354) are presented in Table 2. Patients with AF were older and suffered significantly more frequently from diabetes mellitus and coronary artery disease, which is reflected by a higher virtual CHA2DS2-VASc score.

Table 2 Follow-up atrial fibrillation
Full size table

The predictors of stroke

The univariate regression analysis revealed that both rFSRP and TACT as well as the CHA2DS2-VASc score were statistically significant predictors of stroke incidence (Table 3).

Table 3 Cox regression analysis
Full size table

The relationship between the number of patients with primary/secondary outcomes and the TACT in ms is shown in Online Resource 2 (panel B). After exclusion of patients with newly diagnosed AF, the regression analysis again yielded a significant result (Table 4). Interestingly, the CHA2DS2-VASc score failed to reach statistical significance after excluding these patients.

Table 4 Cox regression analysis after excluding newly diagnosed patients with atrial fibrillation
Full size table

The predictive values of TACT for stroke

The ROC curve analysis is shown in Fig. 1 (panels A–D). Accordingly, the rFSRP and TACT had an area under the curve (AUC) of 0.79 and 0.84, respectively, whereas the combination of both variables showed an AUC of 0.85 for subsequent stroke incidence. The CHA2DS2-VASc score showed a modest AUC of 0.71 for subsequent stroke incidence.

Fig. 1
figure 1

ROC analysis for TACT (panel A), rFSRP (panel B), rFSRP + TACT (panel C), CHA2DS2VASc score (panel D) ROC receiver operating characteristic, TACT total atrial conduction time, rFSRP revised Framingham stroke risk profile

Full size image

A complete analysis of the specificity, sensitivity, positive predictive value and negative predictive value as well as the accuracy for the TACT, rFSRP and the combination of the two (using the appropriate cut-offs) is presented in Table 5.

Table 5 Sensitivity, specificity, positive predictive value, negative predictive value and accuracy for the studied predictors
Full size table

The difference in stroke incidence by TACT value

The optimal TACT cutoff identified using the Youden index was 161.43 ms. The Kaplan–Meier curve using this cutoff demonstrated that patients with a TACT value higher than 161.43 ms had a significantly increased risk of suffering a stroke during 2 years of follow-up (log-rank Mantel–Cox, 22.35; p < 0.001; Fig. 2). The baseline characteristics of the cohort adjusted for this PA-TDI cutoff are presented in the Online Resource 4.

Fig. 2
figure 2

Kaplan–Meier analysis

Full size image

Discussion

We investigated TACT as a potential new tool in the prediction of stroke in 376 consecutive patients in our echocardiography laboratory without clinically apparent AF. The main findings of this study are as follows:

  1. 1.

    Patients with a TACT higher than 161.43 ms have a significantly higher risk of stroke after 2 years despite having no prior history of AF or stroke.

  2. 2.

    The TACT improved the stroke prediction of the rFSRP (AUC of 0.79 for rFSRP to AUC of 0.85 for rFSRP and TACT).

TACT is a well-studied echocardiographic parameter in patients with AF. TACT has been reported to predict new-onset AF [12], stroke after catheter ablation of AF [18], postoperative AF after cardiac surgery [5], and occult AF after embolic stroke of unknown source (ESUS) [16]. Moreover, Leung et al. demonstrated in a well-powered recent study that the assessment of LA reservoir strain and PA-TDI on echocardiography after initial CHA2DS2-VASc scoring provides additional risk stratification for stroke in patients with AF [6].

Our study investigated the TACT measured noninvasively as the PA-TDI interval to predict the occurrence of stroke in patients without known AF or stroke. To our knowledge, this is the first study to investigate TACT in this subset of patients. The TACT further supports the existence of atrial cardiopathy as a substrate for thromboembolisms even in the absence of AF.

Currently, there is no clear definition of atrial cardiopathy in the literature. This concept is a complex interplay of electric, echocardiographic, and biochemical abnormalities. The ARCADIA trial [10] is currently assessing whether apixaban is superior to aspirin for the prevention of recurrent stroke in patients with ESUS and atrial cardiopathy. In this trial, atrial cardiopathy was defined as a P-wave terminal force of > 5000 μV*ms in ECG lead V1, serum NT-proBNP of > 250 pg/mL, and LA diameter index of ≥ 3 cm/m2 on echocardiogram. However, this definition does not include any echocardiographic markers correlated with atrial fibrosis, including TACT, LA reservoir strain, and mitral regurgitation.

The TACT cutoff for stroke (161.43 ms) in our model was higher than that for AF (145 ms) after ESUS [16]. Moreover, only 2 of the 10 patients with stroke had AF diagnosed during follow-up. This result suggests that ischemic stroke in patients with prolonged TACT was not necessarily caused by AF alone. The prolongation of TACT may reflect atrial cardiopathy, which has been reported to be a possible cause of both AF and ischemic stroke.

In this study, the LA diameter was not associated with the risk of subsequent stroke incidence (p = 0.85 in the univariate Cox regression analysis). In our cohort, as shown in Online Resource 3, most of the enrolled patients did not have coronary or peripheral artery disease. Perhaps because of this, the LA diameter was almost normal. The LA diameter and TACT in our study showed a statistically significant correlation. However, this was a very small positive correlation (correlation coefficient = 0.34, as shown in Online Resource 5). The key differences between our cohort and the patient population in the ARCADIA trial are age ≥ 45 years and history of ESUS, both of which increase the pretest probability of having a dilated LA.

In patients with ESUS, both dabigatran and rivaroxaban failed to show superiority over aspirin in preventing recurrent strokes [19, 20]. The choice of anticoagulant for patients with ESUS seems to represent an unbalanced risk–benefit ratio. Theoretically, patients with subclinical AF and ESUS should be anticoagulated. In this context, our data suggest that the TACT is used as a tool to improve risk stratification in patients without AF in need of anticoagulation.

This hypothesis should be further confirmed in larger prospective studies with continuous rhythm monitoring to evaluate the benefit of TACT as a possible risk index.

Our study has several limitations. First, this was a single-center, prospective cohort study. In addition, rhythm monitoring was routinely conducted through 12-lead and Holter ECGs, but the secondary endpoint (AF detection) may have missed asymptomatic paroxysmal AF. A further limitation is that we performed a sample size calculation with one-tailed testing under the alternate hypothesis that patients with prolonged TACT have a higher stroke incidence than those without. A detailed inter- and intra-observer analysis was not possible, however, the PA-TDI was measured by experienced cardiologists certified in adult transthoracic echocardiography by the European Association of Cardiovascular Imaging.

Taking into account the low number of patients with stroke and atrial fibrillation during the follow-up period, we refrained from performing a multivariate regression analysis, as the multivariate model was underpowered.

Finally, patients with more comorbidities may be underrepresented because of the enrollment criteria.

Conclusion

This study showed PA-TDI improves the predictive power when used in addition to the rFSRP. Our findings support that of atrial cardiopathy, and the PA-TDI may represent an important tool to risk stratify patients who may need closer rhythm monitoring and distinct indications for anticoagulation. However, these findings must be evaluated in larger cohorts.