Page 167 - Simplicity is Key in CRT

analyses in chapters 4 and 5 as to whether all proposed LBBB definitions are associated with outcomes to CRT, can be explained by the differences in number of patients included (‘statistical power’). For instance, in chapter 4 only 90 patients classified as LBBB according to the AHA/ACC/HRS LBBB definition , with only 19 events occurring during follow-up. In chapter 5 213 patients were classified as LBBB according to the AHA/ACC/HRS definition, experiencing 49 events.
Another explanation for the different results may be the use of different observers in the two analyses. This may, as mentioned in chapter 3, induce significant inter-observer variability. Although in both studies the AHA/ACC/ HRS definition provided the highest specificity, there was a considerable difference in specificity between the two studies. In the analysis in chapter 4 29.0%, and in the analysis in chapter 5, 18.5% of patients were classified as LBBB according to AHA/ACC/HRS definition.
If inter-observer variability would indeed be the cause of the difference in specificity, this would add to the concern about heterogeneity of LBBB classification. A secondary analysis in the MADIT-CRT trial strengthens this hypothesis by showing that when ‘LBBB-like IVCD patterns’ (n=191, 62% of IVCDs) were added to the ‘LBBB group’, association with the primary endpoint remained significant with odds ratio increasing only from 0.47 to 0.55 [13]. In contrast, overall results in the IVCD patient group (n=308) treated with CRT showed a trend towards worsening outcomes as compared to the ICD treated group.
The studies in chapters 4 and 5 have a limitation which precludes extrapolation of their results in an absolute way to all patients referred for CRT. The retrospective nature results in bias of excluding patients deemed not suitable by treating physicians. This has probably lead to a lower overall proportion of non-LBBB patients (as it seems less likely that CRT is denied to an LBBB patient). Furthermore, the absence of a non-CRT treated “control group” does not allow to completely contribute the association of outcomes to the effect of therapy, as this may be an effect of the presence of LBBB or associated patient characteristics itself. The logical next step would be to analyse the different LBBB definitions in a landmark CRT-trial and evaluate whether outcome of CRT differs between LBBB definitions.
In conclusion, the results from chapter 4 and chapter 5 reassure the value of LBBB as a positive predictor of CRT benefit, regardless of the definition of LBBB. However, considering that in the CRT patient cohort the event rate of all- cause mortality, cardiac transplantation, or LVAD implantation was the same (33.3%) in non-LBBB patients according to AHA/ACC/HRS definition and in LBBB patients according to the MADIT definition, one should give considerable thought to the preferred definition. When applying the numbers found in the real life cohort presented in this thesis; using AHA/ACC/HRS definition in 100 CRT patients would lead to 47 (out of 71) non-LBBB patients with good clinical outcome in CRT being denied therapy. On the other hand, using MADIT definition in the same 100 CRT patients would lead to 25 (out of 75) LBBB patients) being treated although clinical outcome is not good. Using AHA/ ACC/HRS definition a likely responder is denied appropriate therapy, possibly increasing health care costs as well, since CRT has been shown to be cost-effective, certainly in responders. In the latter case the patient is inadvertently exposed to the risk of procedure related complication and possibly even to adverse effects of CRT, as shown by the Echo-CRT trial [15]. The device implantation without clinical benefit will definitely increase health care costs.
One major limitation in the patient data used in this thesis and therefore, the abovementioned translation into practice, is the absence of a untreated control group (no CRT device implanted or CRT-off) in the studies. Because of this lack of controls, we cannot attribute worse outcome to the absence of effect of CRT. The bad outcome could be caused by baseline worse HF condition, in which an untreated patients would have reached the endpoint even sooner.
167