Determinants of diastolic dysfunction: a geometric matter?