Small case series can reveal clinically interesting patterns while leaving treatment attribution unresolved. This report should be read with both of those features in view. Primary study
The reported cases
Nine patients with chronic neurotrophic corneal epithelial defects received sterile thymosin beta-4 eye drops through compassionate use. Six with geographic defects showed healing, whereas three with punctate defects did not show a demonstrable clinical change. All reported less irritation. Stromal thinning was observed in one patient. Primary report
Our appraisal: a pattern is not a predictive rule
The apparent difference between defect types is a hypothesis-generating observation. It does not establish that the morphology predicts a causal treatment response. Baseline severity, duration, coexisting disease, concomitant care, and chance may differ between small subgroups.
Without a concurrent randomized comparator, improvement cannot be cleanly separated from the natural course, other interventions, or selection of patients likely to respond. A history of poor healing makes the observation noteworthy but does not create a control group.
Symptoms and epithelial findings should also be reported separately. Reduced irritation does not necessarily mean the defect closed. Similarly, closure does not automatically establish restored sensation or long-term visual benefit. Each endpoint needs its own definition and follow-up.
The stromal-thinning observation should remain visible without asserting that the treatment caused it. In an uncontrolled series, adverse-event attribution is uncertain in both directions. A fair account neither dismisses the event nor labels every event drug-induced.
The most useful next evidence would come from controlled studies with predefined defect categories, standardized background care, masked assessment, and recurrence tracking. This report provides a small human feasibility and observation record, not definitive efficacy or a basis for preparing an ophthalmic product outside controlled care.
Frequently asked questions
Was this randomized evidence?
No; it was a compassionate-use case series.
Did every defect pattern show healing?
No.
Did symptom improvement always mean documented closure?
No.
Was a structural concern observed?
Yes; stromal thinning in one patient.
Does that establish drug causation?
No.
Can six responding cases define a reliable responder profile?
Not by themselves.
What should confirm the pattern?
A controlled study with prespecified morphology and outcome assessment.