Ophthalmology

Run-in phases and single-arm designs reduce sample sizes in GA trials

Geographic atrophy enlargement rates are stable within eyes, enabling use of run-in natural-history data to boost power. In silico power analyses using GA Minocycline Trial parameters showed 2-year RCT sample size 156 for 90% power versus 82, 40, and 26 with 3-, 6-, and 9‑month run-ins. Single-arm trials required even smaller samples (2‑year example: 14 with 9‑month run-in vs 26 for RCT with same phases). These design gains held across 1‑year trials, 80% power scenarios, and alternate outcome transformations, but are simulation-based and need prospective validation. 1

Short-term periocular or systemic corticosteroids did not raise IOP in three RCTs

Three randomized trials assessed short-term periocular, topical, or systemic corticosteroids in 84 participants with IOP as an endpoint. Periocular hydrocortisone produced a small right-eye IOP decrease (-2.5 mm Hg, 95% CI -3.9 to -1.1; P = .002) compared with untreated controls. The other two trials showed no significant IOP changes overall, supporting ocular safety for short courses. Apply usual IOP monitoring for at‑risk patients despite reassuring short-term findings and limited sample sizes. 2

Predictors of tocilizumab response in moderate-to-severe thyroid eye disease

In 234 patients with moderate-to-severe TED treated with IV tocilizumab, 66.2% (155) met the response definition of ≥2‑point CAS reduction. Mean CAS fell from 3.42 to 1.70 and TRAb declined after treatment, indicating clinical and serologic improvement. Shorter disease duration and higher baseline TRAb, fibrinogen, and HDL‑C independently predicted response (multivariate model AUC = 0.695). These predictors may help select TCZ candidates, but single-center retrospective design limits causal inference and external generalizability. 3