Non-invasive infrared spectroscopy of aqueous humour samples in glaucoma compared to controls

Abstract

Purpose : Metabolic profiling has gained increasing interest in recent years to determine potential biomarkers for glaucoma. This study aims to understand the use of vibrational spectroscopy for the evaluation of aqueous samples and to determine differences between glaucoma and controls for downstream development of cheaper, portable technology for metabolomics evaluation.

Methods : Aqueous samples were acquired from patients undergoing glaucoma surgery (n=10) or patients having cataract surgery in whom cataract had been excluded (n=10) and stored at -80°C by the Liverpool Research Eye Biobank. A high throughput Fourier transform infrared (FTIR) spectrometer was used to obtain spectra and identify specific molecular signatures from the aqueous samples and then analysed at the Centre for Metabolomics Research. Briefly, aqueous samples were transferred to a silicon substrate (Bruker Ltd, Coventry, UK) and dried in an oven (50 °C) prior to data collection. FTIR data were acquired in the mid-IR range (4000–600 cm^-1), with 64 spectral co-adds and 4 cm^-1 resolution, using a Bruker Invenio FTIR spectrometer. A total number of 4 infrared spectra were collected from each sample/patient.

Results : Infrared spectral data collected from diseased and control samples (Fig. 1a) were used as inputs for partial least squares discriminant analysis (PLS-DA) in order to differentiate diseased from control samples. The classification model obtained for the PLS-DA using infrared data is shown in (Fig. 1b). Bootstrapping was performed 10000 times to determine whether the infrared spectrum was disease or control. The output shows two distributions; a ‘real’ (in blue) and ‘null/random’ (in red) model, these models displayed a correct classification rate of 75%.

Conclusions : Vibrational spectroscopy, a differing modality of metabolomics, provides evidence of an ability to determine spectral data differences between glaucoma patients compared to controls. This highlights the potential of the use of vibrational spectroscopy as a cost-effective metabolomics technique for the analysis ocular samples and cells.

This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.

 

A). Average infrared spectra acquired from control and glaucoma samples; spectra are offset for clarity. B). Classification model obtained for the PLS-DA using infrared data where blue are the observed CCR (correct classification rate) from the bootstrap models along with permutation testing in red which represents the null models.