Atrial fibrillation (AF) remains a common phenomenon in patients undergoing cardiac surgery, and is associated with increased morbidity and longer hospital stays[1–3]. Among multifactorial causes, interatrial conduction disturbances (IACD) have been repeatedly shown to be a trigger of paroxysmal AF in invasive electrophysiologic and non-invasive studies in non-surgical patients[4, 5]. Furthermore IACD have been proposed as one of the contributing mechanisms of postoperative AF[6–8]. One of the most frequently reported methods to detect IACD non-invasively is P-wave dispersion in surface multichannel electrocardiograms (ECG)[6, 8, 9]. It represents a sensitive marker for IACD, but does not reveal changes and differences in electromechanical activation of the respective atrium. In modern ultrasound machines, Doppler myocardial imaging (DMI) has been demonstrated as a useful, non-invasive tool to indirectly reflect the brief moments of atrial electromechanical activation due to superb temporal resolution. Findings of DMI-derived atrial electromechanical activation in healthy individuals correlate well with invasive conduction dispersion[10, 11].

The aim of this investigation was to assess the use of DMI for atrial electromechanical activation before and after cardiac surgery, and to evaluate its use as a possible predictor of postoperative AF.

DMI presented a useful tool for the detection of atrial electromechanical coupling and hence indirect assessment of atrial electrical activation and interatrial conduction times. These DMI-derived values have been shown to reveal interatrial conduction disturbances (IACD) in patients presenting postoperative AF as shown by P-wave dispersion in surgical and invasive electrophysiology in non-surgical AF[4, 8].

Proper tracking of the respective atrial wall with small regions of interest (ROI) as used in this study is very time-consuming taking up to 30 minutes and, after a learning curve of several exams, approximately 15 minutes. The chosen region of interest (e.g. base or mid portion) within the atrial wall did alter our measurements significantly. This coincides with the findings of Quintana et al., which showed that electromechanical intervals were longer near the atrio-ventricular ring than in the mid regions of the atria[13]. On the other hand, the qualitative implication of our measurements was unaltered by the numerical changes between base and mid atrial regions. Thus, in our opinion, the future use of DMI-derived measurements of atrial electromechanical coupling will not require small regions of interest, such as we used in our study. A size extension along the longitudinal axis of the ROIs was unproblematic (i.e. 6 mm), whereas ROIs should not be set closely to the AV ring and ROI-width should not exceed 2 mm in order to solely track atrial wall deformation. The measurement of all described values was very time-consuming as well, taking further 15 minutes per exam. For detecting IACD, the mere measurement of the respective P to A´start intervals, representing atrial electromechanical coupling, is sufficient enough. Therefore we used one large ROI and measured the intervals of electromechanical coupling only. This reduced each patient study to less than ten minutes making this technique useful for clinical application (data not shown).

Our findings of interatrial electromechanical coupling by DMI correlate well with findings of Quintana and Zhang, where electromechanical coupling was the earliest in the RA, followed by the IAS and the latest in the lateral LA[13, 14]. There was no significant increase in postoperative interatrial electromechanical activation times in the entire cohort, but IACD were detected in 11 patients postoperatively who developed transient postoperative AF. An M-mode based DMI study by Omi and co-authors revealed that non-surgical patients with paroxysmal AF showed prolonged interatrial conduction times measured from the onset of P-wave to the beginning of tricuspid and mitral annulus backward motion[15]. Postoperative IACD was also detected by P-wave dispersion. In these patients, P-wave dispersion was significantly increased after surgery. Our findings compare well to the findings of Tsikouris and Koide who have shown that increased P-wave dispersion was associated with higher incidence of (postoperative) AF[8, 9]. Changes in P-wave dispersion correlated well with differences in conduction time between the RA and LA derived by DMI. The fact that atrial mono- and/or dual-site pacing may prevent AF supports the concept of IACD[16–19]. In our study, IACD does not apply to every cardiac surgery patient, which seems to be one of the reasons why postoperative bi-atrial pacing is not effective in the prevention of postoperative AF in all patients[6, 16]. In previous echocardiographic studies intraoperative transoesophageal echocardiography was used to assess IACD as measured by onset of P-wave to onset of left atrial appendage (LAA) systolic ejection as measured by pulsed-waved Doppler. The group found significant differences between pre- and immediate intraoperative measurements indirectly suggesting that open-heart surgery might lead to IACD. The measurement of the time interval between onset P-wave and LAA ejection flow does not offer the comparison of right and left atrial electromechanical activation. This method therefore cannot measure interatrial electromechanical coupling as comparing DMI-derived atrial activation with much higher temporal resolution, but represents a very easy and reproductive assessment tool for indirect detection of IACD and risk prediction of postoperative AF. Our data and the data from the Turkish group would strengthen the hypothesis of IACD being one of the major contributing factors of postoperative AF in a certain percentage of patients. The clinical implication of intra- and/or early postoperatively detected IACD would be bi-atrial pacing above intrinsic heart rate as recommended by the American College of Chest Physicians[20] and/or early commencement of anti-arrhythmic medication such as amiodarone[21]. In order to determine whether IACD is a homogenous disturbance of atrial conductive function or a localized abnormality by alteration of atrial conductive properties, further electrophysiologic and echocardiographic studies with larger numbers of patients and long-term follow-up are necessary.

Postoperative reduction of P to peak A’-wave intervals coincides with postoperative reduction of peak diastolic longitudinal deformation velocities, but not with TEMA. We noticed a significant postoperative reduction of right atrial TEMA while left atrial TEMA stayed fairly unchanged. Due to the fact that postoperative heart rates were significantly increased, we adjusted our measurements accordingly. This postoperative reduction of right atrial TEMA lead to a significantly increased difference between postoperative right and left atrial TEMA in the entire cohort. These findings might indicate decreased right atrial refractoriness as described in the literature by electrophysiologic studies[22].

Three recent echocardiographic studies assessed the predictive value of pulsed-wave (PW)- DMI in patients prior to cardiac surgery[23–25]. Here, the time interval between onset of P-wave and peak of A’-wave derived from PW-DMI of the left lateral atrial wall was measured. This so called PA-TDI interval, if prolonged, correlated well to postoperative occurrence of AF as well as to fibrosis of the right atrial tissue[24]. Besides this, further methodical differences in comparison to our study have to be discussed. The scope of all three studies was the predictive value of the PA-TDI interval. Patients were therefore only assessed prior to surgery, not after, thus missing postoperative changes in electromechanical activation. The time interval from onset of P-wave to peak of A’-wave was regarded as representing atrial electromechanical coupling. There was no further explanation to this definition. As described by Quintana and in our opinion atrial electromechanical coupling is represented by the time interval of onset of P-wave to onset of A’-wave, when atrial deformation begins[13].

In all three studies frame-rates of PW-DMI were not described: Pre-settings in PW-DMI of the leading echocardiography machine vendors only show very low frame rates of approximately 35 frames per second, if the whole four chamber sector is used. Using 35 frames per second at a heart rate of 80/min means that 21 frames per cycle are recorded.

To increase frame rates the scanning sector must be reduced to the region of interest and frame rates per se increased on the respective machine setting in order to achieve frame rates higher than 90 frames per second. Looking at the figures of these studies, where in DMI-Mode the full apical four chamber view was used for pulsed-wave Doppler assessment, one must assume that frame rates were too low for measuring these brief electromechanical time intervals.

Nevertheless, PW-DMI with adequate frames rates measuring respective atrial electromechanical coupling (onset P-wave to onset A’-wave) should allow a more clinical application to detect interatrial conductance disturbances rather than carrying out the measurement off-line in quantitative analysis.

DMI was able to detect disturbances in atrial electromechanical coupling in concordance to findings of prolonged postoperative P-wave dispersion. This simple, easy to obtain, reproducible, and clinically useful tool might help quantifying atrial electromechanical function, thus extending the informative value of routine postoperative echocardiography. Equally effective to P-wave dispersion, DMI derived atrial electromechanical activation may help to early identify the risk for postoperative AF.

DMI can be a helpful non-invasive tool to detect IACD. DMI might be a good predictor for postoperative AF. However, further investigations with larger numbers of patients are required to evaluate its value as a prediction tool.