Gildeuretinol in geographic atrophy: the primary endpoint was missed

Design

24-month, double-masked, multicentre randomised placebo-controlled trial (2:1)

Population

198 participants aged 60 or older with geographic atrophy

Primary outcome

Geographic atrophy growth rate from baseline to 24 months

Effect

1.62 vs 1.87 mm²/yr; difference -0.25 (95% CI -0.53 to 0.03; P = 0.075)

In this 24-month, double-masked trial, 198 people aged 60 or older with geographic atrophy were randomised 2:1 to oral gildeuretinol acetate 14 mg daily or placebo. The primary endpoint was lesion growth rate.

Growth was 1.62 mm² per year with gildeuretinol and 1.87 with placebo (difference -0.25 mm² per year, 95% CI -0.53 to 0.03; P = 0.075), a 13.4% relative reduction that did not meet significance. A prespecified analysis allowing for a 6-month delay in effect gave -0.29 mm² per year (95% CI -0.576 to -0.004; nominal P = 0.047). Low-luminance acuity fell by 3.9 letters against 8.3 with placebo (difference 4.4 letters, 95% CI 0.40 to 8.41), and best-corrected acuity declined 6.9 against 10.2 letters (difference 3.3, 95% CI -0.63 to 7.23; P = 0.099).

Only 69.2% completed the study. The authors state that the primary endpoint was not met and that the secondary results are exploratory. Gildeuretinol is not an established treatment for geographic atrophy.

• Do not start gildeuretinol, or an equivalent supplement, for geographic atrophy on this evidence.
• Continue counselling on smoking cessation and current supplementation guidance.
• Monitor lesion area with fundus autofluorescence and low-luminance acuity, which declined in this study.
• Discuss approved options and low-vision support with patients who have geographic atrophy.

Why it matters

Geographic atrophy has few options, so a near-miss invites interest, but only the pre-specified primary result counts as confirmation.

Don't overread it

The prespecified primary endpoint was not met, and the favourable findings are exploratory with nominal P values.

The statistics, in plain English

A confidence interval that includes zero (-0.53 to 0.03) means the growth-rate difference could be no effect. 'Nominal' P values for secondary outcomes are not corrected for the many tests run, so they can easily arise by chance and should be used to plan a new trial rather than to claim benefit.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.