Improving the Analysis of Choroidal Laser Doppler Flowmetry Measurements; Application in Patients with Retinal Diseases

Abstract

Purpose: Laser Doppler flowmetry (LDF) has been shown to be useful in measuring subfoveal choroidal blood flow (ChBF) in healthy volunteers. However, the variability of the ChBF measurements in patients with retinal disease is not well studied. We investigated the variability of LDF measurements and the applicability of established analysis protocols in subjects with diabetic maculopathy (DM) and controls.

Methods: Studies were performed in two groups of subjects utilizing a fundus camera based LDF device, in which the probing illumination beam and light detection receptor are separated, and following a test protocol developed over several iterations. Healthy volunteers underwent LDF measurement at varying direct current (DC) levels in order to investigate the relationship between ChBF and DC. A second study comprised 5 groups: DM without clinically significant macular oedema (CSMO); DM with CSMO; ischaemic DM; diabetic controls without retinopathy; healthy controls. The relationship between DC and ChBF was studied together with the extent of variability due to retinal disease. Several statistical approaches for data pre-processing and variance stabilization were investigated.

Results: Our results confirm existing knowledge about the relationship between ChBF and DC, while, uncovering several challenges. Firstly, the log-ChBF decreased with unit increase in log-DC if gain was kept constant. We found that the variability of ChBF changes with the level of DC and vice-versa (2-fold decrease and 10-fold increase). By applying a log-transformation to both data sets the variability was stabilized. Second, we found that maintaining a specific range of DC in patients with retinopathy proved difficult. We solved this by characterizing the subjects’ ChBF response as a linear function of log-DC, hence utilizing the data from a wider DC range and by summarizing the flow response by 2 parameters: intercept and slope. The mean (standard error; SE) of slope parameter was -0.87 (0.02) in DM patients without CSMO, -0.89 (0.01) in DM with CMSO, -0.93 (0.02) in ischaemic DM, -0.88 (0.04) in diabetic controls and -0.84 (0.02) in healthy controls. SEs were less than 10% of the mean which is acceptable for future comparative analyses.

Conclusions: Our improved data analysis protocol has the potential to allow for more effective comparison across groups of retinal diseases than traditional approaches.

Keywords: 452 choroid • 468 clinical research methodology • 499 diabetic retinopathy