Genetically proxied CETP inhibition tracked with macular degeneration

Design

drug-target Mendelian randomisation with observational genetic association analyses in two cohorts

Population

457,242 UK Biobank participants without macular degeneration at baseline; 7,058 Singapore eye study participants

Primary outcome

risk of age-related macular degeneration by subtype

Effect

Europeans: advanced OR 1.45 (95% CI 1.25-1.67), early 1.29 (1.19-1.41); East Asians: exudative 3.14 (2.16-4.57)

Cholesteryl ester transfer protein inhibitors are a lipid-modifying class developed for cardiovascular benefit. This analysis asked what lifelong inhibition of that target does to the retina, using drug-target Mendelian randomisation — genetic variants that mimic the drug's effect, used as a natural experiment — in 457,242 UK Biobank participants without macular degeneration at baseline and 7,058 participants in a Singapore eye study.

The direction was consistent and the magnitude differed by ancestry. In Europeans, greater genetically proxied inhibition was associated with advanced macular degeneration (OR 1.45, 95% CI 1.25-1.67) and early disease (1.29, 1.19-1.41). In East Asians the estimates were larger: exudative disease 3.14 (2.16-4.57), typical neovascular disease 2.84 (1.69-4.79) and polypoidal choroidal vasculopathy 3.18 (2.05-4.93). Observationally, each standard deviation of greater proxied inhibition carried a hazard ratio of 1.03 (1.01-1.06) for incident disease in the UK Biobank. The Singapore genetic score showed a U-shape, with raised odds at both extremes.

What this is and is not matters. Mendelian randomisation estimates the effect of a lifetime of genetically determined target inhibition, not of a drug taken for a few years, so the magnitudes do not transfer to prescribing. It is also not a finding about statins, ezetimibe or any other lipid drug — the instrument is specific to this one protein.

The authors' framing is the right one: a potential retinal safety signal requiring validation in trials and longitudinal observation. For an ophthalmologist the practical consequence today is to take a lipid drug history where macular degeneration progresses unexpectedly, and to be able to say, if asked, that the signal is genetic rather than observed in treated patients.

• Take a lipid-modifying drug history in patients with unexplained progression of macular degeneration
• Do not extend this to statins or ezetimibe; the genetic instrument is specific to one protein target
• Explain to patients that this is a genetic signal, not an observed effect in people taking these drugs
• Note the larger estimates in East Asian populations and for polypoidal choroidal vasculopathy
• Do not advise anyone to stop a cardiovascular drug on the basis of this analysis

Why it matters

A class developed for the heart may carry a cost in the eye, and nobody is currently looking for it.

Don't overread it

Genetically proxied inhibition is not the same as taking the drug — this is a hypothesis for trials to test, not evidence of harm from treatment.

The statistics, in plain English

Mendelian randomisation uses inherited variants as a proxy for a drug's effect, which avoids the confounding of observational studies but estimates something different — a lifetime of modest target inhibition rather than a course of treatment. An odds ratio of 3.14 in East Asians therefore cannot be read as a threefold risk from taking a drug. The much smaller observational hazard ratio of 1.03 in the same study is the estimate closest to what a person's genotype does over a lifetime, and it is barely above 1.0. The U-shaped association in the Singapore cohort, with raised odds at both extremes, does not fit a simple dose-response and is a reason for caution about the whole picture.

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.