Markedly Lower Rates of Age-Related Macular Degeneration in Malta Compared to European Countries: Results from The Malta Eye Study, Indicating Possible Divergent Genetic Ancestry?

Introduction

Age-related macular degeneration (ARMD) is the most common maculopathy causing visual impairment in those aged over 75 years.^1–3 It begins with drusen (lipofuscin deposits) and may progress to central vision loss through dry (retinal pigment epithelial dysfunction, geographic atrophy) or wet (neovascularization, fluid, hemorrhage, scarring) forms.⁴

Globally, ARMD affects 8.69% of individuals aged 45–85, with early and late stages affecting 8.01% and 0.37%, respectively.⁵ In the United States (≥40 years), prevalence is 11.64% for early and 0.94% for late ARMD,⁶ while in Europe (≥60 years), it is 25.3% and 2.41%, respectively.⁷

Meta-analyses highlighted older age, smoking, prior cataract surgery, and family history as key ARMD risk factors (OR ≥ 1.5).⁸ Recent data added male sex, hypertension, cardiovascular disease, and diabetes.⁹ Age remains the strongest non-modifiable risk, with prevalence rising sharply after 75 years.⁵ Rates are highest in Northern Europeans.^10,11 Early ARMD is less common in Asians (6.8%) and Africans (7.1%) compared with Europeans (11.2%), late ARMD is less common in Africans (0.3% vs 0.5% in Europeans), and overall (“any”) ARMD is lower in Asians (7.4%) and Africans (7.5%) than in Europeans (12.3%).⁵ Modifiable behavioral risks include smoking, which triples ARMD risk,¹² low physical activity,¹³ and Western diets,¹⁴ while Mediterranean eating patterns, olive oil consumption,¹⁵ and exercise^16,17 are protective. Obesity nearly doubles the odds of late ARMD.¹⁸ Sunlight exposure has also been associated with risk in multiple studies.^19–22 Hyperopia is a possible ocular risk^23,24 while cataract surgery has been associated with ARMD,^8,25–30 though recent evidence remains conflicting.^31,32

Genetic susceptibility contributes substantially to ARMD risk at the population level. Variants in CFH and ARMS2/HTRA1, related to complement regulation and oxidative stress, were among the earliest identified.³³ A landmark genome-wide association study by the International ARMD Genomics Consortium identified 34 risk loci,³⁴ later expanded to 63 through further analyses,^35–37 implicating key pathways such as inflammation, lipid metabolism, and extracellular matrix regulation. Differences in genetic architecture between populations help explain variation in ARMD prevalence, while gene–environment interactions further shape individual and population risk.³⁸ Smoking and obesity, in particular, have been shown to markedly increase ARMD risk in individuals carrying high-risk variants, especially in CFH and ARMS2.³⁹ The development of polygenic risk scores supports early risk stratification and may guide both personalized lifestyle interventions and public health strategies targeting high-risk groups.^40–42

Malta, a small Southern European island-state with a growing, ageing population of 563,443,⁴³ has the highest overweight/obesity prevalence in the EU (69.75%; 95% CI: 68.32–71.18%), a trend rising over three decades.⁴⁴ It also ranks second in Europe for UV radiation, with 8.2 average daily sunshine hours and a mean annual temperature of 19.5°C.^45,46 Smoking rates in Malta are similar to other European countries.⁴⁷ Given that obesity, UV exposure and smoking are established ARMD risk factors, these may influence its prevalence in Malta. A 1959 Maltese blindness report noted no ARMD cases,⁴⁸ while a 1989 study found a 0.4% prevalence of senile macular degeneration in those aged 40+.⁴⁹

The Malta Eye Study (TMES) is a population-based, cross-sectional study assessing the prevalence of visual impairment and common eye diseases in Maltese adults aged 50–80. This paper focuses on ARMD, presenting its prevalence and associated risk factors in Malta, and evaluating how these patterns align with established European estimates.

Materials and Methods

TMES methodology, including sample size derivation, is detailed elsewhere.⁵⁰ The study adhered to The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist,⁵¹(Supplementary Table 1).

Ethical Clearance and Permissions

Ethical approval was granted by the University of Malta Faculty Research Ethics Committee (FRECMDS_1819_94) before data collection in 2019. The study followed the Declaration of Helsinki⁵² and General Data Protection Regulation (GDPR)⁵³ guidelines, with hospital and Data Protection Officer approvals. Informed consent was obtained, and data were pseudonymized.

Study Population

A random sample of 4006 individuals aged 50–80 years, stratified by age, sex, and locality, obtained from the Malta Electoral Register, was invited for an ophthalmic assessment between September 2021 and July 2024. Chi-square and binomial tests assessed age-sex subgroup representativity and weights (Population proportion / Sample proportion) adjusted the sample to census data for ages 50–80.

Ophthalmic Assessment

The individuals were examined at Mater Dei Hospital between September 2021 and May 2024 and at Gozo General Hospital between June 2024 and July 2024.

The assessment involved visual acuity tests and autorefraction using the Visionix device (Visionix),⁵⁴ subjective refraction, anthropometric (weight, height) and clinical (blood pressure and capillary blood glucose) measurements, Goldmann tonometry, a slit lamp based anterior segment and dilated posterior segment indirect (90D) examination, fundus photography and swept-source optical coherence tomography (OCT) scanning of the macula and disc. The questionnaire included themes related to sociodemographic status, medical and ocular history.

Definitions Used

The presence and severity of ARMD were classified according to the Age-Related Eye Disease Study (AREDS) criteria.⁵⁵ Category 1 indicated no drusen or <15 small drusen (<63 µm) in both eyes. Category 2 included ≥15 small drusen (<63 µm), <20 intermediate drusen (63–124 µm), or pigmentary abnormalities in at least one eye. Category 3 was defined by the presence of large drusen (≥125 µm), ≥20 intermediate drusen (63–124 µm), or non-central geographic atrophy in at least one eye. Category 4 indicated advanced ARMD, or visual acuity worse than 20/32 attributable to lesions of non-advanced ARMD (eg, central large drusen) in one eye only.

For analysis, ARMD status was grouped as: no ARMD (Category 1), early ARMD (Categories 2 and 3), and late ARMD (Category 4), based on the more severely affected eye. Grading was performed by a single observer (DA) using both direct fundus examination and retinal images. As a quality control measure, a single-blinded validation was conducted by an independent reviewer (FC), who graded fundus images post hoc for a subsample of 74 participants (148 eyes). The validation sample was selected to detect an expected agreement of κ = 0.9 with a precision of ±0.1, with a 50% prevalence of ARMD in the subset (sample size based on Shoukri et al’s formula).⁵⁶ The resulting inter-rater agreement for the presence or absence of ARMD was substantial (κ = 0.7).

Statistical Analysis

Statistical analyses were done using IBM SPSS Statistics v23 (IBM Corp.).⁵⁷ Demographic variables (age, sex, district) were compared with national census data⁵⁸ for representativeness. Descriptive statistics were used to provide prevalence estimates with 95% confidence intervals. Categorical predictors of ARMD were tested with Chi-square or Fisher’s exact tests. Significant univariate predictors (p<0.05) were included in backward stepwise binary logistic regression models on the basis of the Wald test, with absence of ARMD as reference. The models’ final steps were presented. Missing data, being minimal, were handled by listwise deletion, assumed missing completely at random, with no imputation.

Results

A total of 1794 individuals were assessed between September 2021 and July 2024, yielding a turnout of 44.8%. Sample representativity was evaluated against the national census,⁵⁸ showing good alignment for age, sex, and district when considered separately. However, there was an underrepresentation of males (p<0.001) and females (p=0.009) aged 50–59 in Northern Harbor, and of females aged 50–59 in Gozo (p=0.004). To correct this, weighting based on national census⁵⁸ data was applied. Missing data were minimal: 0.4% for most questionnaire items, 0.2% for VI classification, 8.1% for education (post-assessment), and 4.8% for autorefraction.

Prevalence of Age-Related Macular Degeneration

The prevalence of any ARMD in either eye was 6.5% (95% CI 5.4–7.8%) in TMES, and when taking into consideration the worst stage in the affected eye, category 2 ARMD was present in 5.6% (95% CI 4.6–6.7%), category 3 ARMD occurred in 0.6% (95% CI 0.3–1.1%) while category 4 ARMD affected 0.4% (95% CI 0.2–0.8%) (Table 1).

Table 1 Prevalence of Age-Related Macular Degeneration (ARMD) Among the Census-Adjusted TMES Population

A comparison of the age-standardized prevalence of ARMD in individuals aged 60+ between TMES and the latest European ARMD meta-analysis⁷ showed significantly lower prevalence rates of both early and late ARMD in each age group within TMES, with the differences increasing with age (Figures 1 and 2).

Figure 1 Comparison of Age-Standardized Prevalence Rates (with 95% confidence interval bars) of Early (Categories 2 and 3) Age-Related Macular Degeneration (ARMD) between European data7 and that in TMES.

Figure 2 Comparison of Age-Standardized Prevalence Rates (with 95% confidence interval bars) of Late (Category 4) Age-Related Macular Degeneration (ARMD) between European data7 and that in TMES.

Age Related Macular Degeneration by Age and Sex

The prevalence of any ARMD and bilateral ARMD increased by age group and was higher among males overall, especially from 60 years and over (Figures 3–5). Similar trends were observed for early and intermediate ARMD in the worse affected eye. Advanced ARMD, including both geographic atrophy and neovascular ARMD, increased with age but had no sex preference.

Figure 3 The prevalence of age-related macular degeneration (ARMD) laterality (with 95% confidence interval bars) by age group and sex among the census-adjusted TMES population. ARMD OU: Bilateral ARMD.

Figure 4 The prevalence of age-related macular degeneration (ARMD) severity in the worse affected eye (with 95% confidence interval bars) by age group and sex among the census-adjusted TMES population.

Figure 5 The prevalence of geographic atrophy and neovascular age-related macular degeneration (ARMD) (with 95% confidence interval bars) by age group and sex among the census-adjusted TMES population.

Predictors of Age-Related Macular Degeneration

Due to the relatively small number of ARMD cases (n = 117), only predictors of any ARMD were analyzed. Stepwise multivariate logistic regression modelling was employed, which led to the exclusion of several predictors that were initially significant in univariate analysis (Supplementary Table 2), including education level, employment status, sunglasses wear, antiplatelet use, astigmatism, and corneal arcus. Sex was included in the model, especially in view of its theoretical relevance. After adjusting for age and sex, increasing age remained significantly associated with ARMD (OR per year 1.08 95% CI 1.05–1.11, p<0.001) and male sex was independently associated with greater risk (OR: 1.57 95% CI 1.01–2.44, p=0.043).

Discussion

Comparison with Previous Local Rates

Since the previous study did not provide defining criteria for “senile macular degeneration”, the earlier reported prevalence of 0.4% cannot be meaningfully compared with current data.⁴⁹ This is really the first comprehensive, cross sectional prevalence study, on a large sample of the population, using a random sample of 1% of the population, stratified for age, sex and locality.

Comparison with European Data

ARMD prevalence was notably lower among the TMES cohort (Figures 1 and 2). Although age groups and ARMD categories were matched to those used in the literature, comparisons should be interpreted with caution. TMES had an upper age limit of 80 years, unlike the European meta-analysis,⁷ which included older participants. Additionally, the meta-analysis combined studies with varying grading methods, and its reported categories, early/intermediate and late ARMD, correspond to AREDS stages 2/3 and 4, which TMES used.

One possible explanation for the lower prevalence observed in Malta is a difference in genetic susceptibility. The Maltese population may have a lower frequency of high-risk alleles, such as CFH Y402H and ARMS2/HTRA1 A69S, or a higher frequency of protective variants. This genetic background could also contribute to the lack of significant associations between ARMD and modifiable risk factors such as smoking (p = 0.430) or obesity in the TMES cohort. Notably, even when high-risk variants are present, recent evidence suggests that their effect sizes vary by ancestry and these risk variants have been shown to confer weaker effects in individuals with non-European ancestry.⁵⁹

The genetic structure of the Maltese population is distinct from that of Central and Northern Europe. Studies have shown close genetic affinity with Sicilians and Southern Italians,⁶⁰ along with traces of Near Eastern ancestry.^61,62 This distinct ancestry may modulate both the expression of risk alleles and their interaction with environmental exposures, potentially attenuating the risk conferred by smoking or obesity.

In Salandra, Basilicata (Southern Italy), the prevalence of ARMD among individuals aged 60–89 years was notably lower than in other European populations, with advanced ARMD observed in only 1.1% and early changes, defined as drusen larger than 50 μm and more than 10 in number, reported in 4.5%.⁶³ In contrast, a citation reported ARMD prevalence of 22.3% among adults aged 40+ in the Sicilian town of Casteldaccia.⁶⁴ Although the original source is not readily accessible, the findings highlight the need for caution when comparing regional prevalence estimates, as factors such as local environmental exposures, dietary habits, and methodological differences between studies may significantly influence outcomes, regardless of underlying genetic similarities.

Sardinians, unlike Sicilians, have a distinct genetic ancestry closely aligned with Early European Farmers.⁶¹ They had a relatively low ARMD prevalence, 6.9% for early and 0.7% for late ARMD among individuals aged 50+, and their genetic predisposition by the CFH Y402H allele and ARMS2/HTRA1 polymorphisms were low.⁶⁵ In contrast, Northeast Italy reported a much higher prevalence of 62.7% (95% CI: 59.4–66.0%) among individuals aged 60–80 years.⁶⁶ Northeast Italy differs genetically from Sardinia and Malta, reflecting more Central European ancestry, and is also less traditionally Mediterranean in climate and diet. These combined genetic and lifestyle differences likely contribute to the higher ARMD prevalence observed there. These examples imply complex interplay between genetics, ancestry and environmental factors in determining ARMD.

The Mediterranean diet and olive oil consumption are known for their protective effects against ARMD.^67,68 Although dietary habits were not assessed in this cohort, it is possible that protective dietary factors, such as adherence to a Mediterranean diet, were present within the TMES population.

Geographic location, including proximity to the equator and higher solar insolation, has been linked to lower ARMD prevalence, potentially due to the role of vitamin D. Higher early ARMD rates were found in predominantly white populations while higher late ARMD rates were found in areas with lower insolation, with whites being more susceptible in these regions.³⁸ Higher insolation may help explain Malta’s lower ARMD rates compared to Northern Europe. In Crete, a prevalence of 6.1% (95% CI: 4.4–8.2%) was reported in individuals aged 50–79, with higher rates associated with increased altitude.⁶⁴ Malta’s relatively low elevation may be another factor contributing to its lower ARMD prevalence.

Since Malta is a relatively small country, easy access to ophthalmic care and timely AREDS supplementation, may help reduce progression to late-stage ARMD, especially among older adults.

The lower ARMD prevalence among the Maltese population as opposed to Europeans warrants further investigation on specific genetic variants, and their interaction with environmental exposures. Although genetic analysis was beyond the scope of this project, DNA samples were collected from over 1400 TMES participants, and genetic data obtained from TMES controls and individuals with known ARMD, particularly those in the older age groups, may help explain these striking differences. This remains a promising area for investigation.

Associations of ARMD

Interestingly, data from TMES indicate that several well-established risk factors for AMD, beyond age and sex, such as smoking, UV exposure, obesity, hypertension, cardiovascular disease, diabetes, dementia, and pseudophakia,^{9,12,18,21,26,69,70} were not significantly associated with ARMD in this population. Upon stratifying systemic diseases such as diabetes and hypertension by their level of control, as determined by clinic blood pressure readings, the type of medications used to treat diabetes, and by clinic capillary blood glucose readings, no significant differences in ARMD rates were observed between individuals without these conditions, those with poorly controlled disease, and those with well-controlled disease.

There may be protective genetic or environmental factors specific to the Maltese population that mitigate the impact of traditional risk factors. Nonetheless, the lack of observed associations may also reflect the relatively low prevalence of ARMD in the cohort, potentially limiting the statistical power to detect significant relationships. Future studies with larger sample sizes and longitudinal or case–control designs are needed to better explore these factors.

While pseudophakia has been linked to ARMD in earlier studies,^26,30 the higher ARMD rates observed among pseudophakic individuals in this cohort were not statistically significant. In this context, pseudophakia may instead reflect access to cataract surgery and overall better eye care, possibly acting as a protective confounder rather than a true risk factor, aligning with current evidence.^31,32,71

Strengths and Limitations

The strengths of this study include its large, representative, population-based and clinician-led design, with all examinations performed by the same validated specialist. This approach eliminated inter-observer variability and ensured intra-observer consistency, thereby enhancing the validity and robustness of the findings. Furthermore, the use of pupil-dilated indirect fundoscopy and swept-source OCT enabled comprehensive and high-quality ARMD assessment. Turnout matched similar local studies,¹⁸ but the 44.8% response rate may have introduced nonresponse ad selection bias. Participants were invited to clinics rather than examined through domiciliary visits, which may have potentially discouraged attendance of older individuals who might have had visual problems, including ARMD, potentially leading to underestimation of its prevalence. While domiciliary assessments could have reduced this bias, they were not feasible in this study due to resource limitations. Additionally, we did not assess dietary patterns or AREDS supplementation, which may influence ARMD risk, and thus these factors could not be evaluated in our analyses. The low prevalence of ARMD limited the power of risk factor analyses, including those for severity subgroups. Finally, the cross-sectional design restricted causal inference due to potential biases, confounding factors, and the inability to establish temporal relationships.

Conclusion

In this study, we report a significantly lower prevalence of ARMD compared to other European populations, with most traditional risk factors, except age and male sex, not showing significant associations. While our findings suggest a lower prevalence of ARMD in the Maltese population, further research into genetic and environmental factors may help explain these patterns, considering the cross-sectional design, modest response rate, and potential selection bias. These differences may be partly explained by underlying genetic variation or distinct gene–environment interactions. As DNA samples were collected, future analyses incorporating genetic data are recommended to explore these hypotheses further and enhance understanding of ARMD risk in this population.