Retinal microvascular function is associated with the cerebral microcirculation as determined by intravoxel incoherent motion MRI
1. Introduction
]. Through the application of an advanced MRI technique, intravoxel incoherent motion (IVIM), microcirculatory-related parameters can be measured [
]. IVIM is a diffusion-weighted imaging technique which is able to separate the fast (blood flow-enhanced) diffusivity of water particles in the microvasculature from the slow (restricted) water diffusion within the parenchyma. Next to the parenchymal diffusion, IVIM can provide two microcirculatory-related measures: the perfusion volume fraction f, representing the volume of blood flowing through the capillaries, and the pseudo-diffusion coefficient D*, reflecting the fast directional changes in the cerebral microcirculatory blood stream [
]. In earlier studies, we demonstrated differences in IVIM parameters between healthy controls and patients with cerebral small vessel disease, a common microvascular pathology [
,
]. Furthermore, we showed associations between IVIM measures and WMH volumes [
]. Altered IVIM measures have been proposed to be a sign of early microvascular dysfunction, possibly preceding macrostructural MRI lesions [
].
]. These include the presence of a blood-tissue-barrier and auto-regulation of blood flow. Associations between retinal vessel geometry and diameters, and macrostructural MRI markers of microvascular dysfunction have previously been shown [
,
,
]. Furthermore, functional properties of the retinal vessels can be assessed by using a dynamic vessel analyzer (DVA), which measures the vasoreactivity of small retinal vessels in response to a flicker-light stimulus. This response can be regarded as a measure of neurovascular coupling, which is a mechanism to adjust blood flow in neural tissue in response to neural activity or an increased metabolic demand [
]. A previous study showed that patients with WMH had impaired retinal vasoreactivity compared to healthy controls [
]. Moreover, it has been demonstrated that DVA is capable of assessing subtle, early stage changes in vascular function. For example, DVA could already detect microvascular dysfunction in pre-diabetic patients, which further worsens as type 2 diabetes develops [
]. Furthermore, healthy individuals who developed hypertension later in life had an impaired vascular response to flicker light several years before the manifestation of disease []. This way, the retina could be a window to the brain in a very early stage of microvascular disease, when macrostructural abnormalities are still absent. At this stage, IVIM measures might already be altered as well, however dedicated microvascular MRI methods are not widely available and MRI in general is much more expensive than retinal imaging. By exploring the association between retinal vasoreactivity and cerebral microcirculatory properties as assessed through IVIM, we aim to investigate whether early alterations in retinal microvascular function can exhibit corresponding microcirculatory effects in the brain.
4. Discussion
The current study examined retinal microvascular function in relation to cerebral microcirculatory properties measured by IVIM. Our results revealed an association between retinal arteriolar vasoreactivity in response to flicker light stimulation (i.e. neurovascular coupling) and the IVIM-derived microcirculatory diffusivity D*, which depends on the microvascular blood velocity and the architecture of the microvascular bed, in the NAWM and CGM.
]. A previous study demonstrated that patients with WMH had impaired retinal arteriolar and venular vasoreactivity compared to healthy controls [
]. In our study, we showed a modest correlation between arteriolar vasoreactivity in the retina and microcirculatory properties in the brain. These findings suggest that changes in retinal microvascular function may parallel similar functional changes in the cerebral microvasculature. However, a considerable amount of the variance in D* is explained by other factors than the retinal arteriolar vasoreactivity, which makes it difficult to accurately predict cerebral microcirculatory derangement using retinal imaging, especially for individual patients.
]. Nonetheless, retinal arterioles and venules may behave differently during the complex pathophysiological changes that are part of the development of cerebral small vessel disease. This could explain why we found an association between retinal arteriolar, but not venular, dilation response and D*. However, the exact pathophysiological mechanisms differing between arterioles and venules remain elusive. Two previous studies have examined the relation between retinal vasoreactivity and cerebral blood vessel function [
,
]. In one of these studies, patients were included based on the presence of WMH on brain MRI, and in the other one, patients were included based on the presence of diabetes mellitus. They showed an association between retinal venular, but not arteriolar, reactivity and cerebral vasoreactivity as measured by transcranial Doppler. The discrepancy with the current study can be explained by differences in methods: these studies measured an increase in cerebral blood flow following a hyperventilation/breath hold maneuver, whereas we measured the cerebral microcirculation in resting state. Furthermore, transcranial Doppler measures blood flow in the larger cerebral vessels, while IVIM measures properties of the microvasculature. These previous studies did also show an inversed association between retinal arteriolar and venular dilation response and the pulsatility index as well as the resistency index in the middle cerebral artery [
,
], which both are considered to be indirect measures of microvascular compliance. The latter is more consistent with the findings in our study.
]. When there is less vasoreactivity in the retina, one would expect a lower volume of blood flowing through the capillaries in the brain (lower f), however in people with microvascular dysfunction, enlarged perivascular spaces and vessel tortuosity are also more common, both leading to a higher f. This counterbalance could explain why we did not found any associations with f. Furthermore, we measured the cerebral microcirculation in rest, whilst DVA measures an increase in retinal perfusion due to an increased metabolic demand. The different nature of these measures could also contribute to the absence of an association with f.
]. However, the fact that we did not find an association might be due to population bias, as most participants in the present study were adequately treated for their cardiovascular risk factors, and had relatively healthy brain tissue, i.e. rather low WMH volumes. Furthermore, CRAE and CRVE are structural retinal vessel measures while the flicker light-induced dilation response is a functional measure, which may indicate different aspects.
We did not find any associations between retinal arteriolar nor venular dilation response and WMH volume. The fact that we did not find an association with WMH volume, but we did find an association with the IVIM derived D* supports the theory that both DVA and IVIM can detect alterations in microvascular function at an early stage (before macrostructural brain tissue damage appears). However, not finding an association between retinal vasodilation response and WMH volume, could also have been due to a lack of statistical power in our relatively healthy population.
]. Furthermore, technical factors can also contribute to this discrepancy, since D* is sensitive to image noise [
]. Where the effect of image noise on D* values of larger regions of interest, such as CGM and NAWM, is averaged out, D* values calculated over a small region of interest, such as the WMH, are less reliable.
,
]. Furthermore, it has been demonstrated that D* is decreased in acute stroke lesions compared to the contralateral side [
]. These data confirmed that D* decreases in case of hypoperfusion. On the contrary, the interpretation of D* seemed to be less straight-forward in ageing [
] and in disorders associated with cerebral microvascular dysfunction such as cerebral small vessel disease [
] and Alzheimer’s disease [
]. A decrease in D* was expected in the aged or diseased group compared to controls, but this could not be shown. Our study population consists of middle-aged participants with a relatively high prevalence of cardiovascular risk factors, who are at risk, but do not yet have obvious cerebrovascular disease. By showing an association between retinal microvascular function and D* in this study population, our research also mainly validates the physiological concepts underlying the pseudo-diffusion measure D*.
]. Furthermore, IVIM, analogous to ASL, does not use exogenous contrast agent and is therefore safe and patient friendly. However, this study also has a few limitations. Firstly, despite the relatively high prevalence of cardiovascular risk factors in our participants, their microvascular retinal and cerebral status seemed to be quite normal (i.e. % flicker-light response comparable to previous reported values in healthy control groups [
,
] and low prevalence of macrostructural MRI markers of cerebral microvascular dysfunction). This could have imposed limitations on our statistical power to show associations and might have led to an underestimation of the observed associations. Secondly, the time passed between enrollment in The Maastricht Study (time at which baseline characteristics were recorded and retinal measurements were performed) and the IVIM MRI assessment was on average 16 months. This might have limited somewhat the validity of the retinal measures at the time of MRI evaluation, nonetheless we adjusted for this in all analyses and we do not expect a substantial change in the retinal measures in this relatively short period of time [
]. Finally, we investigated associations between a measure of neurovascular coupling in the retina and cerebral microvascular perfusion in resting state, which are obviously two different types of measures of different vascular conditions. However, both have been proposed to be altered at early stages of microvascular disease [
,
,].
In conclusion, we demonstrated that retinal microvascular function and microcirculatory properties in the brain are related. This study provides support for parallel early changes in retinal and cerebral microvascular function. This observation could offer opportunities for early screening (identification of cerebral small vessel disease at subclinical disease stages), and for monitoring the effect of therapeutic interventions including modification of traditional cardiovascular risk factors. Furthermore, retinal vessel analysis could also be used for studies in the early etiology stages of small vessel disease. Future longitudinal studies are needed to validate the observed associations.
Acknowledgements
JJ was funded by VENI research grant 916.11.059 from The Netherlands Organization for Scientific Research (NWO) and The Netherlands Organization for Health Research and Development (ZonMw) .
This work was also supported by the European Union’s Horizon 202 research and innovation programme ‘CRUCIAL’ (grant number 848109 ), European Regional Development Fund via OP-Zuid, the Province of Limburg, the Dutch Ministry of Economic Affairs (grant 31O.041 ), Stichting De Weijerhorst (Maastricht, the Netherlands), the Pearl String Initiative Diabetes (Amsterdam, the Netherlands) , CARIM , School for Cardiovascular Diseases , Universiteit Maastricht , School CAPHRI , Care and Public Health Research Institute (Maastricht, the Netherlands), NUTRIM , School of Nutrition and Translational Research in Metabolism (Maastricht, the Netherlands) , Stichting Annadal (Maastricht, the Netherlands), Health Foundation Limburg (Maastricht, the Netherlands) and by unrestricted grants from Janssen-Cilag B.V. (Tilburg, the Netherlands), Novo Nordisk Farma B.V. (Alphen aan den Rijn, the Netherlands), and Sanofi-Aventis Netherlands B.V. (Gouda, the Netherlands).