Abstract
Purpose :
TPLR is an objective outcome measure of visual function especially relevant in patients with severe vision loss and unsteady fixation. Our previous work has supported the hypothesis that early constriction is dominated by rod and cone photoreceptor function, and TPLR latency to a carefully chosen stimulus could be a simple and useful endpoint for some clinical trials. Here we compare two approaches for estimating TPLR latency.
Methods :
Dark adapted pupillometry (Roland Consult) with red and blue stimuli was performed in subjects with normal vision (age range: 22-55 years). TPLR latency was measured for 1 s long stimuli over a large range of luminances (-2.3 to 2.7 log phot-cd.m-2). Latency was defined as the time to reach a criterion constriction. The criterion was set as either the fixed value of 0.3 mm (absolute) or varied as 4% of pre-stimulus baseline diameter for each trace (relative).
Results :
The baseline diameters varied from 5.2 to 8.4 (mean±SD=7.0±1.06) mm across subjects and stimuli. For the dimmest red stimuli that elicited response in all subjects (-2.3 log phot-cd.m-2), latency to absolute or relative criteria showed qualitatively similar variation (A). TPLRs to photopically matched (but scotopically mismatched) blue stimuli had faster latencies. Overall variation of latencies was similar with absolute and relative measures (B). Mean latencies for the absolute criterion ranged from 450 ms for the dimmest red to 289 ms for the brightest blue (C and D). Mean latencies for the relative criterion ranged from 443 to 283 ms (C and D). The variances for either criterion choice were not statistically different.
Conclusions :
Latencies estimated using an invariant (absolute) criterion were nearly identical to those estimated using a normalized (relative) criterion across a wide range of luminance. Over the range of baseline pupil diameters considered, normalization does not appear to reduce variability. Considering the loss of information by normalization, we suggest use of invariant criterion to measure latency.
This is a 2020 ARVO Annual Meeting abstract.