Caffeine is an adenosine antagonist present in coffee and other beverages. It has an impact on several ocular indices, including ocular blood flow,1 retinal vessel diameter,2 and ocular accommodation.3 Its effect on intraocular pressure (IOP) has been widely studied. Consumption of caffeine caused an acute increase in IOP in glaucoma patients, but there was no significant effect in healthy individuals.4 This finding informed eye care practitioners to remind glaucoma patients not to consume caffeine prior to IOP measurement.

The choroid is responsible for the vascular supply of the outer retina, including photoreceptors at the macula. Previous studies using enhanced depth imaging (EDI) from optical coherence tomography (OCT) revealed thinning of the choroid at the fovea after caffeine intake. Vural et al.5 reported a reduction in choroidal thickness after taking one cup of Turkish coffee (with 57 mg caffeine). The choroid was thinnest two hours after intake and returned back to baseline four hours later. Altinkaynak et al.6 found a similar thinning and rebound pattern occurred after taking a 200 mg caffeine capsule. Other studies confirmed that caffeine intake could reduce choroidal thickness for a period of 2 to 4 h.78

OCT angiography (OCTA) is an advanced non‐invasive ophthalmic imaging modality. It can capture microvasculature structures of both the inner and outer retina and the choroid without the use of fluorescent dye, as used in fluorescein angiography.910 High resolution OCTA can aid detection of proliferative eye problems, including myopic choroidal neovascularization,11 diabetic retinal neovascularization,12 and microaneurysms.13 Fan et al.14 applied OCTA and found that high myopes had reduced macular vessel length density, making them more susceptible to pathological changes. Other investigators have observed reduced retinal vasculature in high myopes.15‐18 A recent study evaluated the effect of caffeine intake on retinal vessel length density utilizing OCTA19. Both the superficial and deep retinal vasculature was found reduced. It was a placebo‐controlled study involving two groups of subjects, and both eyes were included in their analysis.

High myopia is frequently associated with retinal vascular alterations, such as narrower retinal vessels or increased vessel resistance.2021 OCTA has been used to show reduced retinal vasculature of high myopes.14‐18 Further reduction due to consumption of caffeine, if any, could have negative impact to physiology of the macula. It is worth of testing the effects of caffeine intake in high myopes. The current study used a placebo‐controlled balanced crossover study to monitor effects of caffeine on the vasculature of the inner retina of healthy high myopes.

Eighteen subjects were recruited, 6 male and 12 female. Their mean age was 24.3 years (SD 3.1). The average spherical equivalent was ‐8.05 ± 1.71D (‐6.00 to ‐11.75D) in the examined eyes. Analyses included 11 right eyes and 7 left eyes.

Table 1 shows BMI, blood pressure, NCT and GAT IOP, axial length, and subfoveal choroidal thickness results. There was no significant difference in BMI at the two visits (Wilcoxon test, p = 0.17). Blood pressure had significant interaction (p < 0.01). Pairwise comparisons found significant increase in systolic and diastolic blood pressure when taking caffeine. Although a significant interaction was found in NCT IOP (p < 0.01), no measurement points had IOP significantly different from baseline. The GAT IOP and axial length were similar throughout the study (p > 0.05). Subfoveal choroidal thickness demonstrated a significant effect with Time (p < 0.01) and interaction (p < 0.01). It showed a trend of reduction when taking caffeine. However, pairwise comparisons did not find any measurement points with significant difference as compared with baseline.

Vessel length density at the central 1‐mm, and 1‐3 mm annulus showed significant effect with Time (p < 0.01). Only VD at the central 1‐mm was significantly reduced compared with baseline after 180 min of caffeine intake. Although a trend of reduced VD at the 1‐3 mm annulus was apparent, no pairwise comparisons with baseline could reach statistical significance (Table 2). Fig. 1 shows VD at four quadrants in the 1‐3 mm annulus. No significant variation with Time or interaction was found from taking either caffeine or placebo capsules (all p > 0.05).

Fig. 1.

Results are converted to left‐eye configuration. Mean vessel length density (mm-1) of the superior (top), temporal (right), inferior (bottom) and nasal (left) quadrants at different time points. Upper figures: caffeine visit; Lower figures: placebo visit.

(0.19MB).

Perfusion area density results in each visit are shown in Table 3. There was significant effect with Time at the central 1 mm, 1‐3 mm annulus, and the 3‐mm full circle (p < 0.01). Pairwise comparisons revealed a significantly lower PD at the central 1‐mm circle after 180 min of caffeine intake. Fig. 2 shows PD at four quadrants in the 1‐3 mm annulus. No significant variation with Time or interaction was found from taking either caffeine or placebo capsules (all p > 0.05).

Fig. 2.

Results are converted to left‐eye configuration. Mean perfusion area density (%) of the superior (top), temporal (right), inferior (bottom) and nasal (left) quadrants at different time points. Upper figures: caffeine visit; Lower figures: placebo visit.

(0.2MB).

To our knowledge, the present study is the first to evaluate the effect of caffeine on retinal vasculature of the macula in high myopes using OTCA. Apparently 200 mg caffeine consumption did not change superficial retinal vasculature very much. Caffeine consumption causes IOP elevation in patients with glaucoma or ocular hypertension.2728 Although this elevation may not be clinically significant,29,30 glaucoma patients and glaucoma suspects are reminded not to consume caffeine before follow‐ up eye examination. Recent studies using a higher dose of caffeine intake revealed IOP rise in healthy adults.3132 Nakano et al.33reported that habitual coffee consumption was associated with a slightly lower IOP in healthy subjects. Kim et al.34 did not find causal effect of coffee drinking on IOP. They further concluded that effect of habitual caffeine consumption depended on genetic predisposition.

We could not identify any measurement points with IOP significantly different from baseline. This is important because acute IOP rise could result in reduced retinal vasculature. Ma et al.35 found reduced vessel density at both the macula and optic nerve head during IOP spike after laser peripheral iridotomies. Wang et al.36 had similar findings. The reduction of vessel density was more significant in radial peripapillary capillary than that of the entire retina when IOP rise was moderate, from 5 to 10 mmHg. Zhang et al.37 did not find significant change in vessel density using dark room prone provocative test to induce IOP rise. It might be due to IOP rise not great enough to induce changes in retinal vasculature.35 Our subjects had similar IOP level throughout the study. Any changes in retinal vasculature should not be due to IOP variation.

Terai et al.2 analysed retinal vessel diameter from fundus photography and found constriction of retinal vessels one hour after consuming 200 mg caffeine capsule. Retinal vasculature can be evaluated non‐invasively using OCTA. Our placebo‐controlled balanced crossover study revealed some effects of 200 mg caffeine on superficial retinal vessel length density/perfusion area density in healthy high myopes. Although there was a trend of reduced vessel length density (Table 2) and perfusion area density (Table 3), compared with baseline, only the central 1‐mm circle had significant reductions 3 h after caffeine consumption. When macular regions were segregated into four quadrants, all quadrants had similar vessel length and perfusion area densities compared with baseline. Karti et al.19 found significant reduction in vessel length density one hour after caffeine intake, which involved all four quadrants of the macula. Liu et al.25 found that inferior nasal sector was less susceptible from axial elongation. They hypothesized that it could be related to larger blood flow area in the nasal region.38

It is not surprising to find a trend of reduced retinal blood vessel area (perfusion area density) following caffeine consumption. Caffeine exerts its antagonist effect at adenosine A2A receptors39 leading to vasoconstriction.40 Changes in superficial retinal vasculature could be due to vasoconstriction. Such decreasing trend was not obvious from taking placebo. However, the reduced vasculature was minimal (vessel length density: reduced by 1mm-1, Table 2; perfusion area density: reduced by 1%, Table 3) 3 h after caffeine consumption. This may be due to limited sensitivity of OCTA to detect the small physiological changes in capillaries related to caffeine intake, which could differ from pathological changes, such as those observed in diabetic retinopathy,4142 high myopic choroidal neovascularization,11 or age‐related macular degeneration.4344

Jung et al.45 reported that amount of astigmatism could influence OCTA vessel length density and perfusion area density at different quadrants. It could be due to optical blur across different meridians because current OCTA systems incorporate only spherical lens to compensate refractive errors. However, our subjects did not have much astigmatism. The mean astigmatism was ‐1.00D (SD0.64D).

Even though caffeine causes vessel constriction, it requires vessel diameter to be reduced to an extent quantifiable using perfusion area density or to an extent at which the capillaries are no longer detectable (vessel length density). OCTA uses different technologies to detect flow of red blood cells against static retinal tissues in order to reconstruct the retinal vasculature.9 In the current study, microvasculature of the superficial retina was captured using an optical micro‐angiography (OMAG) algorithm from the Cirrus AngioPlex. The OMAG algorithm appeared to perform better than other OCTA algorithms, such as split‐spectrum amplitude decorrelation angiography algorithm (SSADA) or OCT angiography ratio analysis (OCTARA) .46 The OMAG algorithm has fewer motion artefacts than the SSADA.4748

Although OCTA is now commonly used in ophthalmic practice, this technology is limited by severe artefacts, including shadow, defocus, and movement artefacts during acquisition.49 Averaging multiple OCTA acquisitions could enhance image quality5051 and be more sensitive in detecting differences,52 especially when the scan size is large (such as 6 × 6 mm) with low resolution in pixel/mm.53 Previous studies have documented the advantage of image averaging in examining retinal vein occlusion54 and diabetic retinopathy.55 Only one high quality OCTA image was captured in the current study, which could be due to the instrument being incapable of detecting small changes.

Another explanation for the small change in retinal vasculature could be related to the amount of caffeine used. Our subjects took a 200 mg caffeine capsule which was commonly used in previous studies.26756 Recent studies about the effect of caffeine on IOP or accommodation adopted 4 mg/kg.31325758 Other studies used 5 mg/kg5960 or even 7 mg/kg.61 Based on the weight of our subjects, the dose of caffeine intake was between 2.5 to 4.7 mg/kg. This amount could be considered low. Caffeine intake based on body weight should be used in future studies. We did not record menstrual cycle phases of female subjects. However, Kamimori et al.62 found that menstrual cycle did not alter pharmacokinetic of caffeine. We believe that gender should not be a determining factor in the results.

Some limitations were identified in the current study. The analyses were restricted to the superficial retinal vasculature due to the available OCTA metric provided by the proprietary software. Other research groups used customized algorithms to derive middle and deep capillary plexuses.63 Another limitation was not differentiating low and high caffeine consumers. Vera et al.31 found that caffeine induced greater IOP change in low‐caffeine consumers. Therefore, the ocular effect could be subject to caffeine tolerance. We did not exclude regular coffee drinkers who might also have greater caffeine tolerance. We monitored subjects for 3 h only. Future studies should monitor caffeine effects for a longer period. Effect of caffeine on subfoveal choroidal thickness could last for 4 h and it took 6 h for it to return to baseline level.56 The caffeine intake used in the current study was relatively small based on subjects’ body weights. Finally, the sample size was small, with only 6 male and 12 female, and limited to healthy young high myopes. Effect of caffeine in other refractive groups and also patients with ocular pathologies warrants further studies.

Llanas et al.64 highlighted the importance of considering ocular magnification from axial length in OCTA. Axial length of the subjects’ eyes ranged from 25.2 to 29.0 mm. Actual vessel length density could be 5% to 20% smaller than the reported values.65 However, our study included within‐subject comparison and effects between taking caffeine and placebo were still valid.

To conclude, consuming 200 mg caffeine in high myopes reduced subfoveal choroidal thickness and constricted retinal vasculature of the superficial capillary plexus. There was a trend of reduced vessel length density and perfusion area density, which, although small, lasted for a few hours.

Written informed consent was obtained before any ophthalmic examination.

None.

V.K.M.L. and A.K.C.L. designed the study; V.K.M.L. collected the data; V.K.M.L. and A.K.C.L. analysed and interpreted the data; A.K.C.L. wrote the manuscript; V.K.M.L. and A.K.C.L. involved in final approval of the article.