Age-related macular degeneration (AMD) is a common threat to vision. While classification of disease stages is critical to understanding disease risk and progression, several systems based on color fundus photographs are known. Most of these require in-depth and time-consuming analysis of fundus images. Herein, we present an automated computer-based classification algorithm. Algorithm development for AMD classification based on a large collection of color fundus images. Validation is performed on a cross-sectional, population-based study. We included 120 656 manually graded color fundus images from 3654 Age-Related Eye Disease Study (AREDS) participants. AREDS participants were >55 years of age, and non-AMD sight-threatening diseases were excluded at recruitment. In addition, performance of our algorithm was evaluated in 5555 fundus images from the population-based Kooperative Gesundheitsforschung in der Region Augsburg (KORA; Cooperative Health Research in the Region of Augsburg) study. We defined 13 classes (9 AREDS steps, 3 late AMD stages, and 1 for ungradable images) and trained several convolution deep learning architectures. An ensemble of network architectures improved prediction accuracy. An independent dataset was used to evaluate the performance of our algorithm in a population-based study. κ Statistics and accuracy to evaluate the concordance between predicted and expert human grader classification. A network ensemble of 6 different neural net architectures predicted the 13 classes in the AREDS test set with a quadratic weighted κ of 92% (95% confidence interval, 89%-92%) and an overall accuracy of 63.3%. In the independent KORA dataset, images wrongly classified as AMD were mainly the result of a macular reflex observed in young individuals. By restricting the KORA analysis to individuals >55 years of age and prior exclusion of other retinopathies, the weighted and unweighted κ increased to 50% and 63%, respectively. Importantly, the algorithm detected 84.2% of all fundus images with definite signs of early or late AMD. Overall, 94.3% of healthy fundus images were classified correctly. Our deep learning algoritm revealed a weighted κ outperforming human graders in the AREDS study and is suitable to classify AMD fundus images in other datasets using individuals >55 years of age.

Few genome-wide association studies (GWAS) account for environmental exposures, like smoking, potentially impacting the overall trait variance when investigating the genetic contribution to obesity-related traits. Here, we use GWAS data from 51,080 current smokers and 190,178 nonsmokers (87% European descent) to identify loci influencing BMI and central adiposity, measured as waist circumference and waist-to-hip ratio both adjusted for BMI. We identify 23 novel genetic loci, and 9 loci with convincing evidence of gene-smoking interaction (GxSMK) on obesity-related traits. We show consistent direction of effect for all identified loci and significance for 18 novel and for 5 interaction loci in an independent study sample. These loci highlight novel biological functions, including response to oxidative stress, addictive behaviour, and regulatory functions emphasizing the importance of accounting for environment in genetic analyses. Our results suggest that tobacco smoking may alter the genetic susceptibility to overall adiposity and body fat distribution.

Purpose: To systematically analyze thicknesses of retinal layers in an older population and their link to early age-related macular degeneration (AMD). Methods: In the AugUR baseline survey from a population aged ≥70 years, we conducted multimodal retinal imaging, including spectral-domain optical coherence tomography. Autosegmentation of eight distinct retinal layers was followed by manual correction of segmentation errors. AMD status was graded on color fundus images according to the Three Continent AMD Consortium Severity Scale. We tested the association of early AMD on retinal layer thicknesses by using linear mixed models and replicated significant results in independent data also from the AugUR platform. Results: When comparing layer thicknesses between early AMD and no AMD (822 eyes, 449 participants), the retinal pigment epithelium/Bruch's membrane complex demonstrated a statistically significant thickening (e.g., P = 6.41 × 10⁻⁹² for severe early versus no AMD) and photoreceptor layers showed a significant thinning. Autosegmented retinal layer thicknesses revealed similar associations as manually corrected values but underestimated some effects. Independent replication analysis in 1026 eyes (546 participants) confirmed associations (e.g., P = 9.38 × 10⁻³⁶ for retinal pigment epithelium/Bruch's membrane complex, severe early versus no AMD). Conclusions: This first population-based study on spectral-domain optical coherence tomography-derived retinal layer thicknesses in a total of ∼1000 individuals provides insights into the reliability of autosegmentation and layer-specific reference values for an older population. Our findings show a difference in thicknesses between early AMD and no AMD for some retinal layers, suggesting these as potential imaging biomarkers. The thinning of photoreceptor layers substantiates a photoreceptor cell loss/damage already occurring in early AMD.