A trial can miss its primary endpoints while identifying a promising imaging signal. ReCLAIM-2 requires both parts of that story. Primary report
Study snapshot
The double-masked phase 2 study randomized 176 patients with noncentral geographic atrophy to elamipretide or placebo. Neither primary endpoint, low-luminance visual acuity change or transformed geographic-atrophy area change, met statistical significance. Additional analyses favored treatment on ellipsoid-zone preservation and a categorical vision outcome, with nominal P values. Injection-site reactions were common adverse events. Primary report
Our appraisal: an imaging signal does not rewrite the trial objective
The primary endpoint hierarchy defines the intended confirmatory question. Favorable additional analyses can justify future testing without turning the original primary comparisons into successes.
The term “nominal” deserves attention. It should prompt readers to inspect whether and how the analysis controlled for multiple comparisons. A small P value among several tested outcomes does not carry the same evidentiary role as a prespecified, multiplicity-protected primary result.
An imaging measure and a visual-function measure also answer different questions. Even when a structural feature is biologically relevant, its treatment-related change does not automatically quantify the patient's experienced benefit. The relationship must be established rather than assumed.
The sensible next step is a prospectively designed test of the proposed endpoint, with justified clinical interpretation and appropriate safety follow-up. Review should inspect missing images, grading procedures, endpoint definitions, and effect estimates. This historical study summary does not establish a current approved indication, nor does it validate nontrial products marketed with the SS-31 name.
Frequently asked questions
Did ReCLAIM-2 meet its primary endpoints?
No.
Were all additional findings negative?
No. Some imaging and categorical analyses favored treatment.
Does nominal significance resolve multiple testing?
Not by itself.
Is retinal structure identical to experienced visual function?
No. Their relationship needs evidence.
Can a new endpoint be tested in a later trial?
Yes, prospectively and with an appropriate design.
Were adverse events absent?
No. Injection-site reactions were common.
What should a final appraisal preserve?
The primary failures, additional signals, uncertainty, and trial-specific formulation.