An encouraging numerical difference is not automatically a statistically established primary result. This small ophthalmic thymosin beta-4 study illustrates why timing matters. Primary report
Study snapshot
Eighteen patients with neurotrophic keratopathy received RGN-259 ophthalmic solution or placebo. At four weeks, complete healing occurred in six of ten treated participants and one of eight controls; the comparison had P = .0656. Later healing and symptom analyses included favorable findings. The report disclosed developer involvement and did not identify significant adverse effects in this small sample. Primary report
Our appraisal: keep the original endpoint visible
A later positive observation can be clinically interesting without replacing the original time-specific comparison. The protocol should establish which endpoint and visit carried the confirmatory question, and how additional analyses were handled.
Small denominators make the percentages appear dramatic while leaving substantial uncertainty. Counts should accompany percentages so readers can see how few events determine the result. A single additional healed participant would materially change a group proportion in this study.
The formulation boundary is equally important. An ophthalmic preparation tested for a particular corneal disorder does not validate systemic use, sports-injury claims, or products labeled TB-500. Exact identity and delivery cannot be inferred from a related name.
Finally, absence of a detected safety problem in 18 participants cannot establish broad safety. Full review should inspect baseline defect characteristics, masking, assessment methods, missing visits, recurrence definitions, and multiplicity. The study supports further investigation of this specific clinical question, not a generalized wound-healing promise.
Frequently asked questions
Was the four-week healing comparison significant at .05?
No. Its reported P value was .0656.
Were later findings uniformly negative?
No. Some later and symptom analyses favored treatment.
Why report counts beside percentages?
The groups were very small.
Can a later result replace an earlier primary comparison?
Not without changing the evidentiary claim.
Does this establish systemic TB-500 efficacy?
No.
Does the small safety sample exclude uncommon harms?
No.
What should release review prioritize?
Endpoint hierarchy, visit timing, recurrence, exact formulation, and safety uncertainty.