SEARCHOTHERophthalmology, retina

other·ophthalmology, retina, neuro-ophthalmology, nutrition, bioinformatics·PMC10674956

Construction of an Exudative Age-Related Macular Degeneration Diagnostic and Therapeutic Molecular Network Using Multi-Layer Network Analysis, a Fuzzy Logic Model, and Deep Learning Techniques: Are Retinal and Brain Neurodegenerative Disorders Related?

Pharmaceuticals · 15 authors, 12 centres

AI SUMMARY

FIDELITY 100%

This summary was generated by AI from a single paper. It has not been reviewed by a clinician and is not clinical advice. Verify against the source before acting on it.

AGENT 01 · STRUCTURED SUMMARY

TRACE TO SOURCE

“

TL;DR

This study constructed a multi-layer molecular network for neovascular age-related macular degeneration (nAMD), identifying 30 key genes, 6 miRNAs, 4 lncRNAs, 3 shared metabolites with Alzheimer's disease and schizophrenia, and 9 shared SNPs with neurodegenerative disorders. The findings suggest a connection between nAMD and brain neurodegenerative diseases, and demonstrate the utility of artificial intelligence (LSTM, fuzzy logic, genetic algorithms) for identifying important metabolites. These results may inform diagnostic biomarkers and drug repurposing for nAMD.

Full summary

5,163 CHARS

**Background:** Neovascular age-related macular degeneration (nAMD) is a leading cause of irreversible vision loss in the elderly. Current treatment relies on intravitreal anti-VEGF injections, but a significant portion of patients show incomplete response due to overlapping and compensatory angiogenic pathways. A systems-level understanding of nAMD pathophysiology is needed to identify novel therapeutic targets and biomarkers. This study aimed to construct a comprehensive molecular network for nAMD using multi-layer network analysis, fuzzy logic, and deep learning, and to explore connections with neurodegenerative disorders such as Alzheimer's disease (AD), schizophrenia, multiple sclerosis (MS), and Parkinson's disease (PD).

**Methods:** The study integrated multiple data sources: (1) Disease modules for nAMD were identified using the NeDRex platform with MuST and DIAMOnD algorithms, yielding 230 genes. (2) A comprehensive nAMD protein-protein interaction network (AMD-PPIN) was built from databases (DisGeNET, STRING, etc.) and literature on 89 signaling pathways related to angiogenesis and anti-VEGF resistance, resulting in 7061 genes. (3) An anti-VEGF resistance-related network (AV-DRN) was constructed with 4340 genes. Topological analysis (degree, betweenness, etc.) identified hubs, and integration of these three lists produced 31 essential genes. A gene regulatory network (miRNA-gene) was built using NetworkAnalyst and miRTarBase, identifying 6 miRNAs (hsa-mir-124-3p, hsa-mir-335-5p, hsa-mir-661, hsa-mir-29b-3p, hsa-mir-29c-3p, hsa-mir-450a-1-3p) and 14 genes. An lncRNA-miRNA interaction network (miRNet) identified 4 lncRNAs (NEAT1, KCNQ1OT1, SNHG17, XIST). Metabolites (115) and SNPs (317) relevant to nAMD were extracted from literature. Pathway enrichment (KEGG, SMPDB) and joint pathway analysis were performed using MetaboAnalyst 5.0. Metabolite-gene-disease and SNP-gene-disease networks were constructed. A fuzzy logic model integrated four centrality parameters (degree and betweenness from MMIN and MGDIN networks) to calculate a merit for each metabolite. A second fuzzy logic model combined pathway impact (degree, betweenness, closeness) and FDR to compute an output merit for each metabolic route. A Long Short-Term Memory (LSTM) network was trained on 55 metabolites and 33 pathways to predict route output merit. A binary genetic algorithm (GA) searched a 56-dimensional space (55 metabolites + output merit) to identify the most impactful metabolites.

**Key Results:** The final molecular network comprised 30 key genes (including VEGFA, CFH, APOE, TIMP3, COL1A1, FN1, SQSTM1, TLR4, SLC16A8, NMNAT1, ERCC6), 6 miRNAs, and 4 lncRNAs. Enrichment analysis of the 31 genes revealed significant pathways: Staphylococcus aureus infection (FDR=0.000152), protein digestion and absorption (FDR=0.00031), ECM-receptor interaction (FDR=0.00427), complement and coagulation cascades (FDR=0.0401), and AGE-RAGE signaling (FDR=0.00492). miRNA enrichment identified pathways including terpenoid backbone biosynthesis (FDR=0.0022855) and arachidonic acid metabolism (FDR=0.0280283). Metabolite enrichment (KEGG) highlighted aminoacyl-tRNA biosynthesis (FDR=7.02×10^-12), glyoxylate and dicarboxylate metabolism (FDR=0.000176), and arginine biosynthesis (FDR=0.00562). Joint pathway analysis identified six critical metabolic pathways: alanine, aspartate, and glutamate metabolism (FDR=0.0005232); glycine, serine, and threonine metabolism (FDR=0.000031504); arginine biosynthesis (FDR=0.00083089); sphingolipid metabolism (FDR=0.0072172); cysteine and methionine metabolism (FDR=0.000016837); and arginine biosynthesis (FDR=0.000831). Three key metabolites (L-glycine, L-arginine, L-lysine) were common between AMD, AD, and schizophrenia. Nine key SNPs were shared between nAMD and neurodegenerative disorders: rs1061170 (CFH) common in nAMD, AD, MS, and schizophrenia; rs699947 (VEGFA) common in nAMD, AD, MS, PD, and schizophrenia; rs429358 (APOE) common in nAMD, AD, MS, and PD. The binary-GA optimization identified 25 most valuable metabolites (listed in Table 1 of the paper) that significantly influenced metabolic route output merit. The LSTM model achieved an average R-squared of approximately 0.64.

**Clinical Implications:** This study provides a comprehensive molecular framework for nAMD, highlighting the interplay of angiogenesis, inflammation, ECM remodeling, metabolism, and complement pathways. The identified genes (e.g., CFH, VEGFA, APOE, TIMP3, SLC16A8) and non-coding RNAs (miR-29 family, miR-124, NEAT1, XIST) represent potential diagnostic biomarkers and therapeutic targets. The shared metabolites and SNPs between nAMD and neurodegenerative disorders (AD, schizophrenia, MS, PD) suggest common pathogenic mechanisms, which could inform screening for neurodegenerative risk in AMD patients and families. The use of AI (fuzzy logic, LSTM, genetic algorithms) to prioritize metabolites demonstrates a novel approach for drug repurposing and precision medicine. The 25 key metabolites identified may guide development of metabolic interventions or combination therapies to overcome anti-VEGF resistance.

STUDY TYPE

other

SPECIALTY

ophthalmology

SUMMARISED BY

AI pipeline

FIDELITY CHECK

100% · A

ORIGINAL PAPER

PMC10674956

Construction of an Exudative Age-Related Macular Degeneration Diagnostic and Therapeutic Molecular Network Using Multi-Layer Network Analysis, a Fuzzy Logic Model, and Deep Learning Techniques: Are Retinal and Brain Neurodegenerative Disorders Related?

Pharmaceuticals

READ THE FULL PAPER

JUMP TO SECTION

ABSTRACT

Neovascular age-related macular degeneration (nAMD) is a leading cause of irreversible visual impairment in the elderly. The current management of nAMD is limited and involves regular intravitreal administration of anti-vascular endothelial growth factor (anti-VEGF). However, the effectiveness of these treatments is limited by overlapping and compensatory pathways leading to unresponsiveness to anti-VEGF treatments in a significant portion of nAMD patients. Therefore, a system view of pathways involved in pathophysiology of nAMD will have significant clinical value. The aim of this study was to identify proteins, miRNAs, long non-coding RNAs (lncRNAs), various metabolites, and single-nucleotide polymorphisms (SNPs) with a significant role in the pathogenesis of nAMD. To accomplish this goal, we conducted a multi-layer network analysis, which identified 30 key genes, six miRNAs, and four lncRNAs. We also found three key metabolites that are common with AMD, Alzheimer’s disease (AD) and schizophrenia. Moreover, we identified nine key SNPs and their related genes that are common among AMD, AD, schizophrenia, multiple sclerosis (MS), and Parkinson’s disease (PD). Thus, our findings suggest that there exists a connection between nAMD and the aforementioned neurodegenerative disorders. In addition, our study also demonstrates the effectiveness of using artificial intelligence, specifically the LSTM network, a fuzzy logic model, and genetic algorithms, to identify important metabolites in complex metabolic pathways to open new avenues for the design and/or repurposing of drugs for nAMD treatment.

CITATION EXPORT

Hamid Latifi-Navid, Amir Barzegar Behrooz, Saleh Jamehdor et al.. Construction of an Exudative Age-Related Macular Degeneration Diagnostic and Therapeutic Molecular Network Using Multi-Layer Network Analysis, a Fuzzy Logic Model, and Deep Learning Techniques: Are Retinal and Brain Neurodegenerative Disorders Related?. Pharmaceuticals. (2023). https://doi.org/10.3390/ph16111555

AUTHORS

15 · 12 CENTRES

Hamid Latifi-Navid, Amir Barzegar Behrooz, Saleh Jamehdor, Maliheh Davari, Masoud Latifinavid, Narges Zolfaghari, Somayeh Piroozmand, Sepideh Taghizadeh, Mahsa Bourbour, Golnaz Shemshaki, Saeid Latifi-Navid, Seyed Shahriar Arab, Zahra-Soheila Soheili, Hamid Ahmadieh, Nader Sheibani

1Department of Molecular Medicine, National Institute of Genetic Engineering and Biotechnology, Tehran 1497716316, Iran; [email protected] (H.L.-N.); [email protected] (M.D.); [email protected] (N.Z.); [email protected] (S.P.); [email protected] (S.T.); [email protected] (Z.-S.S.)
2Departments of Ophthalmology and Visual Sciences and Cell and Regenerative Biology, University of Wisconsin School of Medicine and Public Health, Madison, WI 53705, USA
3Department of Human Anatomy and Cell Science, University of Manitoba College of Medicine, Winnipeg, MB R3T 2N2, Canada; [email protected]
4Electrophysiology Research Center, Neuroscience Institute, Tehran University of Medical Sciences, Tehran 1416634793, Iran
5Department of Virology, Faculty of Medicine, Hamadan University of Medical Sciences, Hamadan 6517838636, Iran; [email protected]
6Department of Mechatronic Engineering, University of Turkish Aeronautical Association, 06790 Ankara, Turkey; [email protected]
7Department of Physiology and Pharmacology, Schulich School of Medicine & Dentistry, Western University, London, ON N6A 5C1, Canada
8Department of Biotechnology, Alzahra University, Tehran 1993893973, Iran; [email protected]
9Department of Studies in Zoology, University of Mysore, Manasagangothri, Mysore 570005, India; [email protected]
10Department of Biology, Faculty of Sciences, University of Mohaghegh Ardabili, Ardabil 5619911367, Iran; [email protected]
11Biophysics Department, Faculty of Biological Sciences, Tarbiat Modares University, Tehran 1411713116, Iran; [email protected]
12Ophthalmic Research Center, Research Institute for Ophthalmology and Vision Science, Shahid Beheshti University of Medical Sciences, Tehran 1666673111, Iran; [email protected]

EDITOR / REVIEWER

Marino F. A. Santos, Luís Passarinha