But we should always question composite end-points, especially when they are the primary endpoint of a study. In this case, the primary endpoint was defined as "By intention to treat, a composite of all-cause mortality, HF-related urgent care (any unplanned hospital visit requiring intravenous heart-failure therapy), or greater than 15% increase in left ventricular end-systolic volume index by echocardiography, over a mean 36-month follow-up."
Note the word "or" between each of these measures.
What I found interesting in this trial was that most of the composite endpoint benefit was driven primarily by the echocardiographic measure of left ventricular volume index. (A portion of Table 2 from the New England Journal of Medicine is duplicated below):
| biV Pacing | RV Pacing | biV ICD | RV ICD | |
| Primary Endpoint | 108 | 127 | 52 | 63 |
| Event Related to LV Volume Index | 56 | 79 | 31 | 36 |
| Urgent Care Visit for Heart Failure | 40 | 38 | 16 | 23 |
| Death | 12 | 10 | 5 | 4 |
It is not surprising that biventricular pacing improves the LV volume index (biventricular pacing improves the coordination of contraction of the medial and lateral LV walls by its very physiology). But the as far as heart failure and death as subsets of the therapy, there really wasn't much of a difference.
To me, this is not that surprising. In the MIRACLE trial (the sentinel Medtronic trial of biV pacing in heart failure that I participated in years ago), patients were required to have left ventricular ejection fractions (LVEF) less than 35% and a QRS complex greater than 130 msec. Subsequent analysis of that trial showed that LVEF improved on average about 5-6% with biventricular pacing (a small but significant improvement). Patients with severely reduced LV functions at the outset of the MIRACLE trial, then, had the most to gain clinically from the 5-6% LV function improvement compared to patients whose LV functions were less compromised. Therefore, it is not surprising to me that biV pacing had less effect on the clinical indices of heart failure visits or death in the BLOCK-HF's cohort of patients whose LV functions were more robust.
It will be interesting to see if CMS sides with Medtronic's conclusions and agrees to pay for the extra lead placement in complete heart block patients with only mildy reduced LVEF less than 50%.
Since I implant these devices, I certainly hope so.
-Wes
As the principal investigator of BLOCK HF, I would like to make two important points about our results. First, the primary endpoint was a composite of time to FIRST event of death, heart failure-related urgent care, or a >15% increase in LVESVI. It was not uncommon for patients to experience the increase in LVESVI first, and then subsequently develop heart failure requiring treatment or die. Thus, the increase in LVESVI tended to dominate the primary endpoint. It appears that this is often the sentinel event, and then clinical deterioration follows. Second, when we looked at the clinical components of the primary endpoint (heart failure-urgent care and death, without the LVESVI), there was still a significant difference in favor of biventricular pacing. These findings reinforce our conclusion that in patients with AV block and left ventricular systolic dysfunction, biventricular pacing leads to superior outcomes compared to standard right ventricular pacing.
ReplyDeleteAnne -
ReplyDeleteThanks for the clarification.
A little further clarification. A cautionary approach is always a good place to start evaluating any study. Any study must pass two “tests.” Do the endpoints meet statistical significance and do they pass the sniff test of clinical relevance?
ReplyDeleteThe statistical analysis of the various components of the primary endpoint are shown.
1) Mortality alone: more observed deaths in RV arm but not statistically significant (AHA presentation/NEJM manuscript),
2) Mortality/HF urgent care: Significant reduction in risk among patients with BiV pacing (AHA presentation/NEJM manuscript),
4) Mortality/HF hospitalization: significant reduction in risk of an event among patients with BiV pacing (AHA presentation/NEJM manuscript),
5) HF hospitalization alone: significant reduction in risk of an event among patients with BiV pacing (AHA presentation/NEJM manuscript).
Beyond the obvious importance of death and urgent HF care, significant LV dilation (typically 15% change in LVESV in the CRT trials) should be considered a significant clinical event. In HF studies, changes in LV volumes track very reliably with mortality risk.
It is important to further study the data to see if we can tease out subgroups that are the most likely to benefit from the increased cost and risk of an LV lead. For starters, relative risk reduction in the primary endpoint was similar between the ICD (lower EF) and pacer (higher EF) groups. I hope the clinical community and the payors will recognize this study as proof of anecdotal observations with sound patho-physiologic underpinnings. It is a better way to treat certain patients than the current standard of care.
Could yourself or the study investigators clarify the difference in the selection criteria between the published paper in the NEJM and the study protcol?
ReplyDeleteThe study protocol on the NEJM supplementary material make mention that inclusion criteria is a EF < 45% however in the published article it states < 50% ...therefore does the data published correlate to the 45-50% cohort of patients?
Thank you