Characterization of retinal microvasculature and its relations to cognitive function in older people after circuit resistance training

Abstract

Purpose:

To characterize the changes of retinal microvascular density and their relations to cognitive function in the healthy older people without known cognitive impairment after an 8-week high-speed circuit resistance training program (HSCT).

Methods:

Twenty cognitively normal older people were recruited and randomly assigned to either the HSCT group or control group (CON). Twelve subjects (age 70.8 ± 5.8 yrs) in the HSCT group trained three times per week for 8 weeks. Eight subjects in the CON group (age 71.8 ± 4.8 yrs) did not perform formal training. Both eyes of each subject were imaged using optical coherence tomography angiography (OCTA) at baseline and at 8-week follow-up. The densities of the retinal vascular network (RVN), superficial vascular plexus (SVP), and deep vascular plexus (DVP) were measured. In addition, their cognitive functions were tested using the NIH toolbox.

Results:

There were significant increases in pattern comparison processing speed (PAT, P = 0.02) and fluid composite score (FCS, P = 0.005) at the follow-up in the HSCT group. Although the vessel densities did not differ between visits in either group, the variation (i.e., change) in retinal vessel density of SVP was negatively related to the changes of FCS (r = −0.54, P = 0.007) and the List Sorting Working Memory test (r = −0.43, P = 0.039) in the HSCT group.

Conclusions:

This is the first study to reveal that the individual response of the SVD was related to the improvement in the cognition in cognitively normal older people after HSCT.

Introduction

Accumulating evidence supports the positive effects of physical activity, such as high - speed circuit resistance training (HSCT), on cognitive function;^1–3 however, the underlying biological mechanisms are not fully understood. Cerebral microvascular alterations are known to be one of the major pathogenic contributors to cognitive impairment.^4–6 Aging and vascular risk factors contribute to the damages of small cerebral vessels, which present in brain magnetic resonance imaging (MRI) as lacunes, white matter hyperintensities (WMH), or microbleeds.⁷ These cerebral small vessel alterations are reported to be related to cognitive function decline.^8–10 However, it is difficult to visualize and assess the cerebral microvasculature in vivo directly. The brain and retinal vasculature share similar anatomic and physiologic features.^11,12 Alterations in retinal microvasculature have been suggested to reflect similar changes occurring in the brain and may provide a window to monitoring the vascular effect of physical activity. Indeed, retinal microvessel density decreases with aging^13,14 and in patients with cognitive function impairment.¹⁵

Optical coherence tomography angiography (OCTA) has been used to study retinal microvascular responses to the physical activities; however, these studies were conducted in healthy young adults after short periods of physical activity.^16–19 Whether the vascular responses to exercise relate to cognitive function in older people has yet to be tested. The goal of the present study was to characterize changes in the retinal microvascular density and their relationship to cognitive functions in cognitively normal older people after 8-week of HSCT.

Methods

Subjects

The study was approved by the institutional review board for human research at the University of Miami. Written informed consent was signed by each of the study participants, who were treated according to the Declaration of Helsinki. Cognitively normal healthy older people (age > 65 years) were recruited at the Department of Kinesiology and Sports Sciences at the University of Miami. Each subject had an ophthalmic examination by an ophthalmologist (HJ), which included best-corrected visual acuity, color vision, intraocular pressure (IOP) and a slit-lamp examination of anterior and posterior segments. Subjects with cardiovascular diseases or systemic diseases such as a history of stroke, coagulopathy, and uncontrolled hypertension and diabetes were excluded from the study. Individuals with a refractive error greater than −6 diopters (D) or +6 D, obvious ocular media opacity, macular degeneration, glaucoma or other ocular problems were also excluded. Individuals who had participated regularly (at least twice per week) in a resistance training program were also excluded. A total of 20 eligible subjects were randomly assigned into two groups: 12 subjects in HSCT group and 8 subjects in the control group (CON). The participants were advised to refrain from strenuous physical exercise and caffeine or alcohol on the day of imaging.

The HSCT

Training was performed 3 times per week for 8 weeks by the subjects in the HSCT group. The HSCT protocol used in the present study has been proven to increase oxidative capacity.^1,20 Participants performed 1 set of 10–12 repetitions on each of the ten computerized pneumatic machines in the circuit, completing one circuit during week 1, two circuits week 2, and three circuits weeks 3–8. They had minimal recovery time between exercises. The exercises for each circuit were performed in the following order when possible: leg press (LP), seated row (SR), leg curl (LC), chest press (CP), hip adduction (HAD), lat pulldown (LAT), hip abduction (HAB), triceps extension (TE), seated calf raise (CR) and biceps curl (BC) (Fig. 1). To reduce neuromuscular fatigue, upper body and lower body exercises were alternated, and a 1–2 min recovery was provided after each circuit. Prior to training, the maximum weights (1RM) that could be performed on each machine were established as described previously.²¹ Briefly, after a movement-specific warm-up, the subject performed 10 repetitions of the exercise at a low resistance. The next weights were close to the participant’s predicted maximum as agreed upon by both the tester and subject. The weight was then increased or decreased depending on the subject’s ability to perform the repetition using correct technique. All 1RM values were determined within 4 to 5 trials and a one-minute recovery was provided between attempts. The 1RM values for the LP, SR, HAB, TE and CR were ascertained on day one; whereas the 1RM values for the CP, LC, LAT and HAD and BC were established on day two.

Figure 1.

The exercise order of each circuit.

The load used for each machine during HSCT was based on the optimal loads that maximize mechanical power on that machine.²¹ Optimal loads included: 40%1RM for the LAT, 50%1RM for the LP, CP and SR, 55% 1RM for BC, 60%1RM for LC and TE, and 65% 1RM for CR, HAD and HAB. Participants were asked to move “as quickly as possible” during the concentric (shortening) phases of each exercise and take 2s for the eccentric (lengthening) phase. To reduce increases in vascular pressure that may occur during lifting, participants were instructed to exhale throughout the concentric phase of each lift and inhale during the eccentric phase. Progression was based on power plateaus shown on the machines’ digital displays. When power values plateaued within 5% of the previous day’s performances, loads were increased by 5%, and training was continued until the next power plateau. During the first 3 weeks, training volume increased gradually with one circuit during week 1, two circuits during week 2, and three circuits during weeks 3–8. The entire workout took 40–45 minutes.¹ The CON group received no intervention, and the participants were required to maintain their activity levels and dietary habits during the 8-week study period. All HSCT sessions were supervised by trained research assistants.

OCTA

All participants were imaged using OCTA at baseline and 8-week follow-up (one day after training). An Angiovue OCTA device (AngioVue, Optovue, Inc., Fremont, CA, USA) was used to image the retinal angiography and the tissue volume of the retina. The OCTA system is a spectral-domain OCT system with a scan speed of 70,000 A-scan per second and an axial resolution of 5 μm.²² In the present study, an Angio retina of the 3 × 3 mm scan protocol was used to scan the macula centered on the fovea. An image quality of ≥ 7/10 was used. Angiographic (i.e., en face view) images of the total retinal vascular network (RVN), superficial vascular plexus (SVP), and deep vascular plexus (DVP) were exported for further processing and analysis of vessel density using fractal analysis. According to the default settings in the OCTA system, the superficial layer was defined from the inner limiting membrane (ILM) to the outer boundary of the inner plexiform layer (IPL). The deep layer was demarcated from the outer boundary of the IPL to the outer boundary of the outer plexiform layer (OPL). The whole retinal layer was segmented from the ILM to the outer plexiform layer (OPL).^23,24 The superficial retinal layer is composed of the ganglion cell and inner plexiform layers, whereas the deep retinal layer is composed of the inner nuclear and outer plexiform layers. These layers encompass the entire retinal vascular network.²³

To measure the vessel density, OCTA enface images with an image size of 304 × 304 pixels were resized to 1,024 × 1,024 pixels for vessel segmentation by a custom software program in Matlab (The MathWorks, Inc., Natick, MA, USA).¹³ A series of image processing procedures, including inverting, equalizing, and removing non-vessel structures and background noise, were used to create a binary image of the vessels. In the binary image, the large vessels, defined as any vessels with a diameter of ≥ 25 μm, were extracted from the OCTA enface images. The remaining vessels in the skeletonized images were then defined as the small vessels. The small vessels of RVN, SVP, and DVP were analyzed. The foveal avascular zone (FAZ), detected based on the intensity gradient of the image, was used to determine the center of the fovea for outlining the area with a diameter of 2.5 mm.²⁵ Since the FAZ is approximate 0.6 mm in diameter,²⁶ the annulus from 0.6 to 2.5 mm was used for the analysis of the vessel density. Using the fractal analysis toolbox (TruSoft Benoit Pro 2.0, TruSoft International, Inc., St. Petersburg, FL, USA), the box-counting method was used to calculate the fractal dimension (Dbox) of the annulus. This represents the vessel density (VD). The VD measurements included VD in the RVN (RVD), in the SVP (SVD) and in the DVP (DVD).

Cognitive function assessments

Cognitive assessments, including the Mini-Mental State Examination (MMSE) and the NIH Toolbox Cognitive Battery, were performed at baseline and the follow-up visits. The MMSE is one of the most commonly administered screening measures of general cognitive ability.²⁷ For this study, it was used as a screening tool and to examine changes in attention, memory, ability to understand instructions, and orientation. The maximum score for the MMSE is 30, and a higher score suggests higher cognitive ability. The NIH Toolbox Cognition Battery is a well-validated measurement tool for assessing cognitive function, which provides a computer-administered assessment of a comprehensive range of cognitive functions.^28,29 For the purposes of this study, five tests from the battery were selected: Flanker Inhibitory Control and Attention Test (FLNK, executive function, and attention), Dimensional Card Sort Test (DCCS, executive function, and task shifting), Picture Sequence Memory Test (PSM, episodic memory), List Sorting Working Memory Test (LSWM, working memory), and Pattern Comparison Processing Speed Test (PAT, processing speed). Age-adjusted scale scores were used for comparative analyses. The normative mean for these scores is 100, and the standard deviation is 15. Therefore, a score near 100 indicates average, over 115 above-average, and under 85 below-average for fluid cognitive ability. To assess overall cognition, a Fluid Cognition Composite Score (FCS) was computed by averaging the normalized scores of each measure and then deriving scale scores based on this new distribution.

Statistical Analyses

Descriptive statistics and data analyses were conducted using a statistical software package (SPSS for Windows 25.0; SPSS Inc., Chicago, Illinois, USA). T-tests were used to determine if differences existed in the demographic variables between baseline and follow-up. To count inter-eye correlation and visits within the group and between groups, generalized estimating equation (GEE) models were used to analyze the changes of these variables. Age, sex, and eye were covariates. A chi-square test was used to test for differences due to sex between groups. Pearson correlations were used to determine the relationships between cognition tests and ocular variables. P < 0.05 was considered significant.

Results

Demographic characteristics of the subjects by group are presented in Table 1. There were no significant differences in any variables, including cognitive function tests, between groups at baseline or follow-up in CON group (all P > 0.05).

Table 1.

Demographic Characteristics and cognitive function of Study Subjects

	HSCT			CON
	Baseline	Follow-up	P Value	Baseline	Follow-up	P Value
Subjects, n		12		8
Age, y		70.8 ± 5.8		71.8 ± 4.8		0.71
Sex, M:F		5 : 7		4 : 4		0.71
HTN, n		6		4		0.85
Diabetes, n		0		0		n.a
Dyslipidemia, n		7		5		0.73
Smoke		0		0		n.a
Education, y		15.9 ± 1.7		16.6 ± 1.5		0.40
SBP (mmHg)	137 ± 14	127 ± 13	0.08	130 ± 27	133 ± 20	0.75
DBP (mmHg)	81 ± 11	76 ± 9	0.28	75 ± 8	79 ± 12	0.33
HR (beat / min)	63 ± 13	64 ± 10	0.91	62 ± 11	66 ± 7	0.27
MAP	100 ± 9	93 ± 8	0.09	93 ± 13	97 ± 13	0.50
MOPP	55 ± 9	51 ± 5	0.06	52 ± 7	54 ± 9	0.56
IOP (mmHg)	17 ± 3	17 ± 4	0.98	16 ± 3	17 ± 3	0.31
MMSE	30 ± 1	30 ± 1	0.59	30 ± 0	29 ± 1	0.08
FLNK	103 ± 12	109 ± 14	0.07	101 ± 8	101 ± 14	0.95
DCCS	115 ± 17	112 ± 17	0.60	115 ± 7	110 ± 12	0.26
PS	100 ± 12	106 ± 16	0.11	102 ± 10	109 ± 8	0.09
LS	103 ± 13	105 ± 12	0.63	104 ± 12	113 ± 12	0.11
PAT	98 ± 25	107 ± 26	0.02	97 ± 15	103 ± 15	0.49
FCS	105 ± 15	113 ± 16	0.005	105 ± 50	110 ± 7	0.30

Results are presented as the mean ± standard deviation. Abbreviations: M = male; F = female; SBP = systolic blood pressure; DBP = diastolic blood pressure; HR = heart rate; MAP = mean arterial pressure; MOPP = mean ocular perfusion pressure; IOP = intraocular pressure; MMSE = Mini Mental State Test; FLNK = Flanker; DCCS = Dimensional Change Card Sort; PS = Picture sequence; LSWM = List sorting working memory; PAT = Pattern comparison processing speed; FCS = Fluid composite score; n.a = not applicable; Bold font denotes significant change at P < 0.05.

There were, however, significant increases of PAT (P = 0.02) and FCS (P = 0.005) from baseline to the follow-up in HSCT group. (Table 1, Fig. 2).

Figure 2. Cognitive function tests in HSCT and CON groups.

In the HSCT group, there were significant increases of PAT (P = 0.02) and FCS (P = 0.005) from baseline to the follow-up (A). In the CON group, there were no significant differences in all cognitive function tests between visits (all P > 0.05, B). HSCT: high-speed circuit training group; CON: control group; MMSE: mini-mental state test; FLNK: NIH Flanker; DCCS: Dimensional Change Card Sort; PS: picture sequence; LSWM: list sorting working memory; PAT: comparison processing speed; FCS: fluid composite score. Bars = standard errors.

The vessel densities in RVN, SVP and DVP did not differ at baseline and 8-week follow-up in either group (P > 0.05, Fig. 2); however, variations (i.e., changes) in retinal vessel density of SVP were negatively correlated to changes in FCS (r = −0.54, P = 0.007, Fig. 3) and LSWM (r = −0.43, P = 0.04) in the HSCT group.

Figure 3. VD in HSCT and CON groups at baseline and 8-week follow-up.

There were no significant differences in RVD, SVD, and DVD in both groups between visits (all P > 0.05). HSCT: high-speed circuit training group; CON: control group; VD: vessel density; RVD: VD in the total retinal vascular network; SVD: VD in the superficial vessel plexus; DVD: VD in the deep vessel plexus. Bars = standard errors.

Discussion

To the best of our knowledge, this is the first prospective study to characterize the changes of retinal microvasculature and its relationship to cognitive function in older people without known cognitive impairment after circuit resistance training. The key finding is that the decreased density of SVD was related to the improvements in cognitive function (i.e., FCS and LSWM) in the training group, as well as significant improvements in PAT and FCS following HSCT. Since OCTA does not measure blood flow, vessels visualized using OCTA are those with moving blood clusters. In other words, the vessel density measured using OCTA indicates the recruited capillaries which are needed for transporting blood.³⁰ Therefore, the decrease in vessel density may imply that a lower level of the capillary recruitment is sufficient to transport enough blood at a higher flow rate for tissue perfusion. In the present study, while average vessel densities did not change after training, the negative correlation between the SVD and cognitive function indicated that the improved cognitive function after training could be mediated by the vascular effects.

The effects of physical activities on the retinal microvasculature have been previously studied using OCTA on young adults.^16–19 Immediately after performing a training program, which included sit-ups, push-ups, squats, lunges, and rope skipping, decreased peripapillary and parafoveal vessel densitiy was observed in 13 healthy young adults, and correlated with elevated systolic blood pressure.¹⁷ The authors commented that the decreased retinal perfusion during physical strain was potentially due to the redistribution of blood supply from the eye to skeletal muscle or other organs (vascular steal) during the exercises. Similar results were reported by Vo Kim et al., who evaluated the retinal vessel density in 32 healthy young adults before and after continuously cycling for 20 min.¹⁶ Because the reduction of retinal microvascular density was related to the increased systolic blood pressure, a rest period was recommended before OCTA measurement to avoid the acute impacts from systemic cardiovascular parameter modification.^16,17 In contrast, no changes of retinal vessel density were observed following a Wingate test, a moderately short (30 sec) anaerobic exercise, despite the increased blood pressure.¹⁸ Moreover, Schmitz et al. reported a decreased SVD measured one day after a 4-week high-intensity interval training in 52 healthy young adults.¹⁹ No systemic changes in cardiovascular parameters, such as blood pressure and heart rate, were reported in the study, although they may be expected to be within normal ranges in these healthy young adults given the study’s short duration. The persistent decrease in SVD one day after the training may be a precursor of long term benefits that can be derived from physical activity. Indeed, decreased oxidative stress and inflammation have been reported after physical activity, and have been proposed to contribute to a decreased incidence of diabetic retinopathy.³¹ In the present study, the OCTA measurement of the retinal microvascular density was done one day after the HSCT training program in these older participants. HSCT was chosen as an intervention due to its effectiveness in improving both cardiovascular, neuromuscular performances over more standardize training modalities³² and its proven impact on cognition in older persons.³³ There was no significant difference in systemic cardiovascular parameters and mean ocular perfusion pressure compared to the baseline. Hence, the reduced SVD that correlated with improved cognitive function suggest the long-term beneficial effect from resistance training may be mediated through vascular effects.

The improvement of the cognitive test scores (NIH PAT and FCS) found in the current study refelcted those reported in previous studies.^1,2,33,34 Although a previous study using a similar training protocol with a smaller cohort of 7 subjects did not demonstrate improvements in cognition using the NIH toolbox,¹ improvements in the Walking Response and Inhibition Test (WRIT), used to assess executive function), were demonstrated.¹ Ricardo et al. reported the improvement in cognition in 62 older subjects after 24-week resistance training.³⁴ Strassnig et al. also found the improvement in the cognitive function in a cohort of 12 patients with mental diseases after 8-week of HSCT.³³ More interestingly, in a 30-year follow-up study, the Caerphilly Cohort Study, healthy behaviors, including physical activities, and cognitive functions were accessed in a cohort of 2,235 male adults.² Physical activities were defined as walking two or more mles to work each day, cycling ten or more miles to work each day, or “rigorous” exercise described as a regular habit. The study found that the adults who participated in regular exercise during the follow-up period, had low risk (odds ratio of 0.36) of cognitive impairment at the end of the study.

There are several limitations to the present study. First, the sample size is relatively small, which may reduce the generalizability of our results; however in this novel study that attempted to establish the relationship between the retinal microvasculature and cognitive function, the small sample appeared to provide sufficient power to determine the relationship. Second, we only included elder subjects and did not recruit young subjects. Aging plays a role in the decline of retinal microvasculature,^13,14 cognition,⁶ and CBF,⁷ focusing on these vulnerable populations may provide a chance to detect the relation between retinal microvasculature and cognition as shown in the present study. Third, the training period was relatively short. In future studies a larger sample size of cognitively normal older people and patients with cognitive function decline and a longer period of training is needed to further characterize the effect of physical activiy on retinal microvascular changes. Lastly, we did not measure cerebral blood flow, which prevented direct establishment of the link between the eye and brain in vascular function. This would be a logical next step in this research process.

In summary, this is the first prospective study to characterize the change of retinal microvasculature and its relation with cognitive function in the cognitively normal older people after HSCT. The individual response of the SVD was found to be negatively related to the improvement of the cognitive function in the training group. This has the potential to be used as a diagnostic marker for monitoring the effect of the HSCT and declines due to inactivity with aging.

Figure 4. Relations between the changes of retinal VD and the changes in cognitive function tests in both HSCT and CON groups.

The changes of the FCS score were significantly correlated to the changes of SVD (r = −0.54, P = 0.007, A) in the HSCT group. The changes of the LS score were significantly correlated to the changes of RVD (r = −0.43, P = 0.04, B) in the HSCT group. HSCT: high-speed circuit training group; CON: control group; SVD: vessel density in the superficial vessel plexus; FCS: fluid composite score; LSWM: list sorting working memory.

Highlights.

• The densities of retinal vessels were measured in older people.

• Cognitive function improved after high-speed circuit resistance training.

• The vessel density was correlated with the cognitive function after training.