Page 71 - Biomarkers for risk stratification and guidance in heart failure
P. 71

                                Chapter 3
it is clear that even small changes in NT-proBNP levels during serial testing are predictive of outcome.23 In this context, application of an absolute BNP or NT- proBNP level seems appropriate. How best to determine the target level? Data from earlier biomarker-guided studies and observational studies may be helpful in this regard. An approach similar to that used in the TIME-CHF study may have greatest merit— choosing levels associated with increased risk and using simple stratification based on age.1,7
Third, like earlier biomarker-guided studies, PRIMA suffered from inadequate power to test the effect of biomarker-guided treatment on some hard clinical outcomes. The initial power calculation optimistically estimated a 50% reduction in cardiovascular events by NT-proBNP-guided treatment. The primary end point was changed prior to the initiation of recruitment to “days alive and out of hospital”—arguably a less robust end point for heart failure trials that may be skewed by the potentially large number of subjects who do not suffer a hospitalization.24 Mortality and hospitalization are more robust and unbiased outcome measures that were routinely used as primary end points in other biomarker-guided studies. It is interesting to note that in PRIMA, there was a nonsignificant 21% lowering of mortality rates in the NT-proBNP-guided arm. However, PRIMA was not powered to test mortality, and the modified target sample size further reduced the ability to detect clinically significant reductions in this end point. For the observed mortality rate and effect size described in PRIMA, a study of about 2,000 patients would have been needed to provide adequate power. The trend to lower mortality in PRIMA mirrors the trends seen in trials such as TIME-CHF and BATTLESCARRED, where there were reductions in mortality, particularly in younger patients. In an attempt to address the issue of inadequate power from individual studies, 2 recently published literature-based meta- analyses combining recent biomarker-guided studies, including PRIMA, suggest that treatment guided by BNP or NT-proBNP may in fact be associated with up to a 30% mortality reduction compared with usual clinical care.25,26 These meta- analyses, however, cannot be seen as definitive since they were based only on the available summary data extracted from reports.
The PRIMA study is an important addition to the series of biomarker-guided heart failure studies and provides valuable new insights. How should we therefore interpret PRIMA and other recent studies of biomarker-guided care? First, an
70






























































































   69   70   71   72   73