The use of a lumped parameter model to optimize biventricular defibrillator programming: results obtained by a randomized and prospective study on 60 patients

Objectives: Biventricular defibrillator (BIV) implantation is a succesfull treatment for chronic heart failure despite of 30% not responders, partly due to sub-optimal atrioventricular (AV) and interventricular (VV) intervals programming. Aim of this work was to develop a numerical model (NM) for AV and VV optimization. Methods: A lumped parameter circulatory model was updated. Atria, ventricles and septum were described by variable elastance models and their mechanical activity was related to ECG. BIV was modelled as an impulse generator driving heart chamber contraction. 60 patients were enrolled and randomized into two groups (A, B). All patients were studied by Echo, ECG, pressure measurement, six minute walking and Minnesota tests before BIV, 1, 3 and 6 months (f-u1,3,6) later. In Group A, AV and VV were programmed and modified, if necessary, at each f-u by NM optimizing hemodynamic variables. Group B was programmed by standard algorithm optimizing QRS. Results: There were no statistically significant differences between A and B at the baseline in QRS and left ventricular end systolic (LVes) and end diastolic volumes (Lved) (QRS:A:154.7±39.5ms; B:141.9±39.5ms; LVes:A:187.1±92.7mL; B:155.6±57.6mL; LVed:A: 246.1±101.2; B:226.1±67.1) and between simulated and measured data in A at the baseline and at f-u6 (p>0.3). At f-u6, comparing A to B, it was observed: no statistically significant difference in QRS reduction (p>0.7), a positive trend in Six minute and Minnesota tests (p=ns) and in ventricular remodeling (A:?LVes=-56.6mL, ?LVed=-53.7ml; B:?LVes=-28.1mL, ?LVed= -22.45mL, p<0,05). Conclusions: For each patient, the optimal AV and VV are different and can change at each follow-up. A personalized and dyamic therapy based on the developed NM could improve patients' outcome.