Background: The PR-interval reflects atrial and atrioventricular activation. A prolonged PR has been linked to heart failure (HF), atrial fibrillation and mortality. This might at least in part be the consequence of compromised cardiac performance (e.g. left ventricular filling). On the other hand, very long PR may just be a marker of advanced heart disease. Alternative measures that account for atrial and ventricular activation might better predict adverse hemodynamic consequences of delayed atrioventricular conduction. The DEEPER-PR mechanistic pilot study suggested that various novel measures of AV conduction are associated with Doppler-derived cardiac performance. However, the clinical significance of these novel parameters is unknown.

Study objective: To examine the association of novel measures of AV conduction with clinical outcomes in a HFrEF population with non-LBBB and a wide QRS (non-LBBB ICD-only MADIT-CRT subset).

Methods: Time intervals from anonymized scanned baseline 12-lead ECGs from the Multicenter Automatic Defibrillator Implantation Trial – Cardiac Resynchronization Therapy (MADIT-CRT), non-LBBB patients (n=537) were assessed in detail using ImageJ (public domain software provided by NIH). The primary endpoint of the study was the combined endpoint of heart failure (HF) or death events. The earliest onset of atrial activation (P wave onset), zero crossing of the P wave in lead V1, earliest onset of the QRS complex, time interval to the first peak of the R wave in V1 and V6, and QRS width were determined. The PR interval and 8 novel parameters (see figure 1) were measured and calculated accordingly. Bivariate non-parametric correlations were calculated to ascertain the associations between these covariates. In addition, Kaplan-Meier analyses by quintiles of all AV delay measures, and multivariable Cox regression analyses were performed among ICD-only patients (n=209) to determine the associated risk regarding of the combined MADIT-CRT clinical outcome of HF events or death.

Results: Weak correlations were found between the AV delay variables (figure 2). Four of the variables (P0PV1, P0R, PePV6, PR) were significantly associated to the risk of HF or death with the upper quintile defining a high-risk HF population, as compared to quintiles 1 to 4 (see figure 3). This indicated a threshold behavior of the associations. P0PV1 was the most powerful risk predictor (Chi-square=16.31, HR for Q5 vs. Q1-4=3.12, CI=1.80-5.41, p<0.001), followed by PRmax (Chi-square=13.19, HR for Q5 vs. Q1-4=2.81, CI=1.61-4.92, p<0.001), P0R (Chi-square=10.42, HR=2.54, CI=1.44-4.48, p=0.001), and PePV6 (Chi-square=7.96, HR for Q5 vs. Q1-4=2.2, CI=1.29-4.07, p=0.005).

Conclusion: This post-hoc ECG-based study identified in MADIT-CRT non-LBBB patients the left-to-right atrioventricular activation delay (P0PV1) as the most powerful risk predictor of HF or death. Prospective evaluation of this measure is warranted.