Opportunities, challenges, and difficulties in NMR-based metabolomics applied to neovascular age-related macular degeneration (nAMD) patient follow-up.

[en] INTRODUCTION: This study applies NMR-based metabolomics to investigate neovascular age-related macular degeneration (nAMD), addressing challenges in patient management, disease progression evaluation, and treatment response assessment. A two-year follow-up of 29 nAMD patients undergoing treatment provided 231 time points for analysis.
METHODS: Over the two-year period, 11 males and 18 females (aged 61-92 years) were monitored, yielding 231 time points. At each time point, blood samples for NMR metabolomics analysis, clinical measurements (e.g., lactate, glucose levels, HDL/LDL cholesterol, and blood pH), and optical coherence tomography (OCT) images were collected to evaluate the progression of choroidal neovascularization. 1H-NMR metabolomic analysis led to the quantification of over 60 metabolites and of the major lipoprotein fractions. Both multivariate and univariate statistical approaches tailored for longitudinal data were employed to identify biomarkers correlating metabolomic changes with ocular alterations during disease progression.
RESULTS AND DISCUSSION: Despite a rigorous analytical workflow enabling precise quantification of over 60 metabolites and the application of advanced statistical tools for longitudinal data, achieving consistent results across the cohort proved challenging. The dataset's heterogeneity, reflecting real-world clinical practice, complicated the derivation of global conclusions. Personalized analyses on a patient-by-patient basis successfully identified individual correlation models, but a universal model remained elusive. This study highlights the inherent challenges of translating findings from controlled settings into clinical practice, where factors such as visit frequency, treatment variability, and disease heterogeneity limit data uniformity. We emphasize the importance of experimental design in longitudinal studies, particularly when dealing with incomplete and variable datasets. We are therefore confident that, considering both the challenges and difficulties identified in this work and the preliminary results presented here, it is possible to develop predictive and individualized models for monitoring patients with nAMD. Such models could greatly assist clinicians in providing better care for these patients.