PROGRESS-HF did not show improvement in the primary ventricular-volume comparison with placebo. Describing only the active groups' before-and-after changes would conceal that result. Primary report
What was reported
The double-blind phase 2 trial randomized 71 people with reduced-ejection-fraction heart failure to placebo or two elamipretide groups for 28 days. Cardiac MRI measured left ventricular end-systolic volume. Neither active group showed a statistically significant advantage over placebo at four weeks. The abstract reports similar rates of study-drug-related adverse events and concludes that the primary volume outcome was not improved. Primary report
Our appraisal: the comparator belongs in the headline
In a randomized trial, a decline in a measurement within one active group is not enough to attribute the change to treatment. The placebo group can also change because of background care, natural variation, measurement error, or other processes. The randomized between-group comparison is the relevant anchor.
For a hypothetical illustration, if an active group improves by four units and a comparator improves by four units, the active group's improvement is real as a description but supplies no four-unit comparative advantage. This is a teaching example, not a substitute for this trial's estimates.
The result should neither be hidden nor exaggerated. It does not establish the hoped-for four-week imaging benefit, but one bounded trial is not a test of every disease, exposure, or duration. Full appraisal should inspect precision, baseline balance, missing MRI data, and the prespecified analysis.
Any broader elamipretide overview should include this trial alongside earlier acute signals, without allowing a plausible mitochondrial mechanism to override the clinical comparison.
Frequently asked questions
Was MRI used?
Yes, for the ventricular-volume assessment.
Did either active group beat placebo on the reported primary comparison?
No statistically significant advantage was reported.
Can before-and-after change alone prove efficacy?
No. The comparator's change is essential.
Does a null result prove every possible effect is zero?
No. Interpretation remains bounded by the design and precision.
Were all adverse events necessarily absent?
No. Similar related-event rates are not absence of events.
Why retain this study in an overview?
It is an important controlled test, regardless of whether the result is favorable.
What remains to be verified?
Complete MRI methods, statistical plan, safety tables, and publication updates.