Abstract
P
Pupil dynamics were measured in both eyes using a handheld ForSite digital pupillometer, (Neuroptics, Irvine, USA) at 75m (London), within 2–3 days post arrival, during the morning, at 3500 m (Namche Bazaar) and 5300 m (Everest Base Camp). Of 222 possible subjects, confounding medical conditions or missing measurements left 174 subjects' data being eligible for analysis. No subjects took medication for high altitude illness at the time of, or prior to, pupillometry. Data from one was excluded, being an outlier (with values falling over 6 standard deviations below the mean for the rest of the sample) at all three altitudes. Pupil responses to altitude were symmetrical; mean (±standard deviation) difference between left and right eye in PAC was 0.1±5.16 % (t = −0.43, df=461, p=0.666), and CV 0.02±0.4175 % (t = −1.15, df=460, p=0.2499). Given this symmetry, the average change/velocity were used as the dependent variables for purposes of data analysis. The relationships between peripheral arterial oxygen saturation (SaO2) and both PAC and CV were examined through linear mixed effects models, including inter-individual variability as a random effect. Analysis was carried out using lme4 package (Bates et al. 2013) of R for Mac version 3.0.2 (R Core Team, 2013). Models were fitted by maximum likelihood with and without oxygen saturation as a fixed effect and then compared with likelihood-ratio tests to obtain p-values; the final adequate models were then refitted using Restricted Maximum Likelihood.
SaO2 fell with ascent, and with it PAC (averaged across both eyes) (chi-squared=24.09, d.f.=4, p<0.0001) and CV (chi-squared=26.67, d.f.=4, p<0.0001), albeit with a small effect size (Pearson's r=0.12 and 0.15, p=0.009 and 0.010 respectively; Fig. 1a and 1b respectively).
Scatterplots showing the relationship between pupil oxygen saturation and a) percentage pupil aperture change and b) constriction velocity. Data are separated by location measured in color: red (sea level, London–75 m), orange (Namche– 3500 m), and blue (Everest Base Camp–5200 m). Black lines represent fitted mixed effects models.
These findings are consistent with those we previously reported, the modest effect size perhaps relating to the slower arrival at 3450/3500 m (in the past study, arrival was by air), and the 2–3 days of acclimatization allowed in comparison to the immediate measurements made in our past study. While the association of hypoxia with these pupil dynamics may be due to a direct hypoxic effect, other causal factors cannot be excluded. For instance, rising intracranial pressure (for which retinal venous congestion might prove a helpful index c.f (Wilson et al., 2013)) can cause compression of parasympathetic oculomotor nerve fibers. The symmetrical nature of our results across both eyes may support this hypothesis. Notably, our findings contrast with Schultheiss and colleagues (Schultheiss et al., 2014), who found that both CV and PAC increased as SaO2 decreased. Although Schultheiss and colleagues argue that differences in methodology and pupillometer model are important variables, our results are uniquely demonstrate consistent effects over successive trials with a consistent methodology. Furthermore, the results we now report are based on a cohort significantly larger than studied before (Cymerman et al. (2005), n=18; Wilson et al. (2008), n=17; Schultheiss et al. (2014), n=14). Given these differences, further studies relating to the time-course and cause of pupillary changes at altitude are advocated. Given the use of pupil dynamics as a measure of neurological function as a part of the Glasgow Coma Scale, further work is of particular clinical relevance.