Age-related macular degeneration (AMD) remains a leading cause of irreversible vision loss worldwide. While genetic susceptibility and aging are well-established non-modifiable drivers of disease, the contribution of environmental and metabolic exposures has historically been examined through single-exposure, hypothesis-driven studies. A recent groundbreaking publication in the British Journal of Ophthalmology (Zhou et al., 2026) shifts this paradigm by utilizing an Exposure-Wide Association Study (EWAS) framework to systematically screen hundreds of modifiable factors, evaluate their joint effects, and quantify the population-level preventability of AMD.Below is an in-depth clinical and epidemiological breakdown tailored for retina specialists and eye care practitioners.
Why the EWAS Approach Matters in AMD
Traditional epidemiological studies evaluating AMD risk factors often focus on isolated variables—such as cigarette smoking, specific dietary carotenoids, or physical activity. While valuable, single-exposure designs suffer from inherent limitations:
- Publication Bias: Favoring preconceived hypotheses.
- Confounding & Overestimation: Unmeasured or unadjusted exposures can inflate effect sizes and yield Type I errors.
- Overlooked Risk Factors: Novel or non-obvious environmental factors go unscreened.
By applying a hypothesis-free EWAS strategy to 478,867 UK Biobank participants followed over a median of 13.63 years (during which 10,903 participants developed incident AMD), the researchers established a systematic blueprint of modifiable risk factors.
Study Design & Multi-Tiered Screening Pipeline
The investigation followed a rigorous three-step analytical workflow:
- Univariate Screening (Exposure-Wide Scan):
- 331 modifiable factors across 9 broad categories were evaluated using Cox proportional hazard models, adjusting for age, sex, and AMD Polygenic Risk Score (PRS).
- After Bonferroni correction ($p < 9.62 \times 10^{-5}$), 177 factors across 7 categories passed initial screening.
- Mutually Adjusted Validation:
- Significant factors were entered into mutually adjusted multivariable Cox regression models to account for collinearity and shared variance.
- 34 independent exposures across 5 categories retained statistical significance:
- Sociodemographic
- Medical Conditions
- Lifestyle
- Physical Measures
- Biomarkers
- Joint Effects & Population Attributable Fraction (PAF):
- Combined joint effect scores were constructed to assess risk strata (favourable, intermediate, unfavourable) across categories.
- Principal component analysis was used to account for inter-variable correlation, yielding conservative weighted PAF estimates.
Key Clinical Findings & Category-Specific Drivers
331 Modifiable Factors Screened
│
▼
177 Significant in EWAS Scan (7 Categories)
│
▼
34 Mutually-Adjusted Validated Factors (5 Categories)
┌──────────────────────┬──────────────────────┐
▼ ▼ ▼
Medical Conditions Physical Measures Biomarkers
• Diabetes / HbA1c • BMI / Fat Mass • Cystatin C
• Hypertension • Hand Grip Strength • Glycated Hb / Glucose
• Gastritis • Forced Vital Capacity • Lipoproteins
1. Medical Conditions & Multimorbidity (Largest PAF Contributor)
Comorbid conditions accounted for the single largest proportion of preventable AMD risk.
- Glycemic Control & Diabetes: Diabetes mellitus, elevated blood glucose, and higher HbA1c were significantly associated with incident AMD.
- Systemic Inflammatory Conditions: Gastritis and duodenitis emerged as novel risk factors. Chronic mucosal inflammation may influence retinal health through systemic inflammatory pathways or altered nutrient absorption.
- Cardiovascular & Respiratory Factors: Hypertension, systemic vascular conditions, and chronic respiratory disorders were positively linked to disease onset.
2. Biomarkers
- Renal & Inflammatory Markers: Elevated Cystatin C and glycated hemoglobin (HbA1c) were among the strongest biomarker hits.
- Lipid Pathways: Apolipoprotein B and direct LDL markers demonstrated distinct relationships, corroborating previous Mendelian randomization studies on lipid transport in drusen formation.
3. Physical Measures & Frailty Metrics
- Grip Strength: Lower hand grip strength—a marker of systemic frailty and sarcopenia—was associated with higher AMD risk.
- Body Composition vs. BMI Paradox: While higher Body Mass Index (BMI) showed a slight protective hazard ratio in this specific multivariable model, the authors highlight that BMI is a crude metric. Detailed body composition parameters (fat mass distribution vs. muscle mass) are critical for accurate risk estimation.
4. Lifestyle & Environmental Factors
- Tobacco Dependence: While general smoking status attenuated under mutual adjustment, mental and behavioral disorders due to tobacco use remained strongly associated with AMD, reflecting heavy dependence and pack-year accumulation.
- Screen Time & Sleep Behaviors: Novel findings included daytime dozing/somnolence and frequent computer game playing as risk factors. These suggest potential roles for circadian rhythm disruption and long-term screen exposure/sedentary behavior.
Daytime Somnolence & Age-Related Macular Degeneration (AMD)
Disrupted Circadian Rhythms
Frequent daytime dozing typically signals underlying sleep-wake cycle fragmentation and poor sleep quality:
Systemic Health & Vascular Comorbidities
Excessive daytime sleepiness serves as a marker for systemic disease processes that compromise ocular perfusion:
- Obstructive Sleep Apnea (OSA): Induces intermittent nocturnal hypoxia.
- Cardiovascular Health: Impairs systemic and choroidal vascular regulation.
- Systemic Inflammation: Promotes oxidative stress contributing to AMD progression.
Clinical Pearl: Holistic Screening
In patients presenting with AMD, screening for excessive daytime sleepiness can uncover treatable systemic comorbidities (like OSA or sleep disorders) that negatively influence retinal vascular health and disease trajectory.
Modifiable Risk Override Genetic Susceptibility
A vital takeaway for retina clinical practice is the interaction between genetic risk and lifestyle/environmental modification:
- Polygenic Risk Stratification: When participants were stratified by AMD Polygenic Risk Score (PRS), those in the high genetic risk tier who maintained unfavourable modifiable factor scores experienced significantly higher incidence of AMD compared to those with favourable factor profiles.
- Preventability Estimate: Shifting the population from the most unfavourable risk tier (worst 1/3) to favourable levels could reduce overall AMD cases by 30.4%. Under a broader public health intervention strategy (eliminating the worst 2/3 of risk factor profiles), up to 45.1% of incident AMD cases could be prevented.
Summary of Category-Specific PAF Contributions
| Factor Category | PAF (Eliminating Worst 1/3) | PAF (Eliminating Worst 2/3) |
|---|---|---|
| Medical Condition | 10.9% | 14.3% |
| Physical Measure | 6.5% | 9.5% |
| Biomarker | 7.2% | 8.0% |
| Socio-demographic | 4.2% | 8.4% |
| Lifestyle | 1.6% | 4.7% |
| Overall Combined | 30.4% | 45.1% |
Translational Implications for Retina Clinics
- Beyond AREDS Formulations: Clinical management of high-risk dry AMD patients has long centered on antioxidant supplementation. This study reinforces that systemic metabolic optimization (glycemic control, hypertension management, renal monitoring) plays an equally vital role in delaying disease onset.
- Cross-Disciplinary Care: The prominent contribution of cardiometabolic and multimorbidity factors indicates that collaborating with primary care physicians, endocrinologists, and cardiologists is essential for holistic AMD risk reduction.
- Patient Counseling with High PRS: Patients with a family history or known genetic predisposition to AMD can be reassured that 30% to 45% of disease burden remains modifiable through proactive lifestyle and health interventions.
Study Limitations to Consider
- Cohort Homogeneity: The UK Biobank consists predominantly of individuals of white European ancestry, requiring cautious generalization to diverse ethnic populations.
- Outcome Capture: AMD diagnosis was determined via medical and inpatient coding rather than protocolized fundus photography/OCT screening, potentially missing asymptomatic early-stage non-exudative cases.
- Observational Design: Despite 5-year lag sensitivity analyses to reduce reverse causality, residual confounding remains possible in observational exposures.
Conclusion
This exposure-wide association study demonstrates that age-related macular degeneration is heavily influenced by a broad spectrum of modifiable systemic factors. Integrating systemic health awareness into retinal practice offers a powerful lever to attenuate disease incidence on a population scale.
Reference
Zhou J, He Y, Wang J, Wu H, Huang X, Cao J, Tham YC, Zhang C, Cheng CY, Ye J. Exposure-wide approaches identifying modifiable factors for age-related macular degeneration. Br J Ophthalmol. 2026;110(7):907-914. doi:10.1136/bjo-2025-328446
Daytime somnolence/dozing and high screen/gaming time were identified as novel risk factors, highlighting potential mechanisms involving circadian rhythm disruption, poor sleep quality, or obstructive sleep apnea (OSA).