This article describes a method for measuring retinal vasculature reactivity in vivo with human subjects using a gas breathing provocation technique to deliver vasoactive stimuli while acquiring retinal images.
The vascular supply to the retina has been shown to dynamically adapt through vasoconstriction and vasodilation to accommodate the metabolic demands of the retina. This process, referred to as retinal vascular reactivity (RVR), is mediated by neurovascular coupling, which is impaired very early in retinal vascular diseases such as diabetic retinopathy. Therefore, a clinically feasible method of assessing vascular function may be of significant interest in both research and clinical settings. Recently, in vivo imaging of the retinal vasculature at the capillary level has been made possible by the FDA approval of optical coherence tomography angiography (OCTA), a noninvasive, minimal risk and dyeless angiography method with capillary level resolution. Concurrently, physiological and pathological changes in RVR have been shown by several investigators. The method shown in this manuscript is designed to investigate RVR using OCTA with no need for alterations to the clinical imaging procedures or device. It demonstrates real time imaging of the retina and retinal vasculature during exposure to hypercapnic or hyperoxic conditions. The exam is easily performed with two personnel in under 30 min with minimal subject discomfort or risk. This method is adaptable to other ophthalmic imaging devices and the applications may vary based on the composition of the gas mixture and patient population. A strength of this method is that it allows for an investigation of retinal vascular function at the capillary level in human subjects in vivo. Limitations of this method are largely those of OCTA and other retinal imaging methods including imaging artifacts and a restricted dynamic range. The results obtained from the method are OCT and OCTA images of the retina. These images are amenable to any analysis that is possible on commercially available OCT or OCTA devices. The general method, however, can be adapted to any form of ophthalmic imaging.
The metabolic demand of the retina is dependent on an adequate and constant supply of oxygen provided by a well-regulated system of arterioles, capillaries and venules1. Several studies have demonstrated that the function of larger caliber human retinal vessels can be assessed in vivo with various physiologic2,3,4,5 and pharmacologic6,7 stimuli. In addition, abnormal function of this vascular system is common in retinal vascular diseases such as diabetic retinopathy where retinal vascular reactivity (RVR) has been shown to be attenuated even in its earliest stages8,9 through both gas provocation9 and flickering light experiments5,10,11. Retinal vascular risk factors such as smoking have also been correlated with impaired RVR12 and retinal blood flow13. These findings are important since the clinical symptoms of retinal vascular disease occur relatively late in the disease process and proven early clinical markers of disease are lacking14. Thus, assessing RVR can provide useful measures of vascular integrity for the early assessment of abnormalities that can initiate or exacerbate retinal degenerative diseases.
Previous RVR experiments have usually relied upon devices such as a laser blood flowmeter9 or fundus cameras equipped with special filters15 for retinal image acquisition. However, these technologies are optimized for larger diameter vessels such as arterioles16 and venules15, which are not where gas, micronutrient and molecular exchange occur. A more recent study was able to quantify the RVR of capillaries using adaptive optics imaging17, but despite the improved spatial resolution, these images have a smaller field size and are not FDA approved for clinical use18.
The recent advent of optical coherence tomography angiography (OCTA) has provided an FDA approved, noninvasive and dyeless angiographic method of assessing capillary level changes in human patients and subjects in vivo. OCTA is widely accepted in clinical practice as an effective tool for assessing impairment in capillary perfusion in retinal vascular diseases such as diabetic retinopathy19, retinal venous occlusions20, vasculitis21 and many others22. OCTA therefore provides an excellent opportunity for the evaluation of capillary level changes, which can have significant spatial and temporal heterogeneity23 as well as pathologic changes, in a clinical setting. Our group recently demonstrated that OCTA can be used to quantify the responsiveness of retinal vessels at the capillary level2 to physiologic changes in inspired oxygen, which is a retinal vasoconstrictive stimulus16,24, and carbon dioxide, which is a retinal vasodilatory stimulus3,5.
The goal of this article is to describe a protocol that will allow the reader to assess the retinal vascular reactivity of the smaller arterioles and capillary bed using OCTA. The methods are adapted from those presented in Lu et al.25 who described the measurement of cerebrovascular reactivity with magnetic resonance imaging. Although the present methods were developed and used during OCTA imaging2, they are applicable to other retinal imaging devices with relatively simple and obvious modifications.
This study was approved by the University of Southern California Institutional Review Board and adhered to the tenets of the Declaration of Helsinki.
1. Setup of Gas Non-rebreathing Apparatus
Figure 1: Diagram of the non-rebreathing apparatus. The full setup has been broken into three separate units according to their function and the frequency with which they are dealt with independently. These include: the Air-Control Unit, the Non-rebreathing Unit, and the Subject/Imaging Device Unit Please click here to view a larger version of this figure.
2. Preparing the Subject for Imaging
3. Gas Provocation Experiment and Image Acquisition
4. Experimental Clean Up
5. OCTA Data Export and Analysis
The output from this experiment consists of the manual readings taken from the pulse oximeter, the timing noted for gas exposure or OCTA scanning and the raw OCTA imaging data. An OCTA image consists of the OCT B-scans and the decorrelation signal associated with each B-scan. The data parameters are given by the specifications of the device. A swept source laser platform OCTA machine with a central wavelength of 1040–1060 nm was used. The images provide a transverse resolution of 20 µm and optical axial resolution of 6.3 µm. Most often, the OCTA data is presented in a 2D enface format as has been shown in the representative Figure 2. Many metrics exist for quantifying this data in a way that allows for comparisons between subjects and among different conditions. A representative metric, vessel skeleton density (VSD), is shown together with full retinal angiograms in Figure 2. As the capillaries vasoconstrict and vasodilate in response to the gas exposure, the capillary density also changes. Hypercapnic conditions are expected to result in an increase in VSD and hyperoxic conditions are expected to result in a decrease in VSD when compared to room air conditions.
Figure 2: Representative results of vessel skeletal density (VSD) in hyperoxic, room air, and hypercapnic conditions. This graphic shows the 3 mm x 3 mm OCTA angiograms and vessel density findings of a healthy 76 year-old female subject. Row 1 shows a single representative horizontal OCT B-scan through the fovea with decorrelation signal above the retinal pigment epithelium represented by red for each of the gas breathing provocation conditions—100% O2, room air and 5% CO2 respectively. Row 2 consists of a single OCTA enface image constructed from 256 OCTA B-scans, one of which is shown in row 1. Row 3 consists of those same OCTA images in Row 2 after post-processing in which the vessels were binarized and skeletonized. Row 4 consists of a heat map showing VSD calculated locally from the images in row 3. Note that the total VSD and relative number of local VSD hot spots increases as one progresses in the columns from left to right. Please click here to view a larger version of this figure.
The methodology just described is the complete protocol for a gas breathing provocation experiment that allows for the measurement of a subject’s RVR in a controlled environment at specific timepoints with no modifications to the OCTA imaging device and minimal discomfort or risk to the subject. This setup is described in a way that allows for easy modifications to fit the needs of the researcher. It can accommodate additional tubing to fit different clinic rooms and certain elements such as the in-house tubing or elbow joint may be omitted or substituted with other components. Figure 1 shows how the key parts of the setup—the Air Control Unit, Non-rebreathing Unit, and Subject/Imaging Device Unit—interface with each other in one simple connection. Gas mixtures can be easily controlled using the Douglas bag as a reservoir. In addition, supplementary monitors can be added at several points in the setup. For example, the elbow joint contains an optional sampling port which may be used to measure the gases in the subject’s exhalation such as end tidal CO2 for more accurate characterization of the state of the subject’s breathing. The strength of this non-rebreathing apparatus is in its adaptability to both clinic conditions and researcher’s requirements. Though OCTA imaging is used, other imaging modalities could conceivably be implemented with this gas setup.
The order of exposure to gases during testing may be important to not bias the reactivity measures. Studies by Tayyari et al.24 have suggested that a vasoconstrictive state of retinal vessels persisted after the conclusion of a hyperoxic gas challenge and may impact hypercapnic RVR assessment. However, others have shown retinal vessel oxygenation27 and retinal vessel diameter16 both return to baseline within 2.5 min following the cessation of hyperoxic breathing. The duration of the gas provocation is also important. Previous work has shown that vasoconstriction is measurable after 1 min of hyperoxic exposure and that almost all vasoconstriction has occurred after 4–5 min of onset. Vessel diameters will then remain stable with oxygen exposure for over at least 20 min28. In the case of hypercapnic gas provocation, peak effects to the retinal arterial and venous vessel diameters were observed after 3 min of exposure to 5% carbon dioxide conditions4. The method proposed is this study begins imaging after 1 min of gas non-rebreathing because the effect of hypercapnia on cerebral vascular reactivity has been shown to be equivalent at 1 and 4 min, thereby reducing the time necessary for imaging and patient discomfort significantly29.
By using a mouthpiece with a nose clip, this setup may improve upon those experiments using a gas mask. Previous studies inducing hyperoxic conditions using a mouthpiece noted a mean increase in the blood oxygen concentration of retinal arterioles of 2%15 compared with a 5% increase30 when using a mask. However, by adding a nose clip, this method should reduce the potential for subjects to inspire any amount of air through their nose as may have occurred in this previous study. The potential for error in the setup must be balanced with the comfort of the patient and the additional complications of wearing a face mask while using an unmodified OCTA system. These include making space for the mask at the OCTA31 and the potential for gas exchange and mixing in the large space occupied by the mask itself32. One concern regarding the mouthpiece setup is the potential for compounded vasoconstrictive effects on the RVR due to changes in the partial pressure of CO2 (PCO2) during the induction of hyperoxia33. The breathing apparatus may be modified to control this confounding effect by maintaining a constant end tidal partial pressure of carbon dioxide with a sequential rebreathing circuit33,34.
During the testing, patients may feel short of breath when breathing through the tube circuit even though they are oxygenating well. This sensation is potentially due to the increased resistance to gas flow when breathing through tubing. Several steps can be taken to ensure the subject does not become disconcerted or alarmed. First, it is important to minimize the length of dead space between the subject’s mouth and the two-way non-rebreathing valve to minimize rebreathing of gas. Even with a very short segment, subjects can still “feel” like breathing is more difficult. Therefore, it is important to have the subject breathe through the gas non-rebreathing apparatus before the initiation of any data collection to familiarize the subject with the setup. The examiner should remind the subject to breathe slowly and deeply, keep a close eye on the pulse oximetry readings and inform the subject of its findings for reassurance. Also, ensure that the subject can sit comfortably and rest their head easily on the OCTA headrest while the mouthpiece is inserted. This involves directing the mouthpiece tube through and around the OCTA chinrest so that the subject need not bite down with force to keep it in their mouth. Remind the subject to maintain gaze at the fixation target and limit actions that result in eye or head movement, including talking, as these can introduce motion artifacts into the OCTA scans. The subject should be encouraged to withdraw from the experiment if the discomfort from participating in the study goes beyond the barest minimum.
Hypercapnia and hyperoxia are not expected to have a significant effect on mean arterial pressure at the magnitude and duration of gas variation seen in this study especially in hemodynamically normal subjects35,36. However, measurement of blood pressure during gas breathing provocations may be useful if the measurement procedure itself does not confound the study or increase subject anxiety during testing. If the preferred stimuli for assessing the RVR is to increase mean arterial pressure, alternative methods such as the hand-grip test37,38,39 or cold pressor test40, which can more directly and effectively increase a subject’s blood pressure, may be considered.
OCTA allows for good intravisit and intervisit reproducibility in both healthy patients and those with retinopathy with most coefficients of variation for vessel density less than 6%41,42. In a patient population of interest, such as that of diabetic patients, the intersession coefficient of variability for vessel density remained below 6% even at an interval of one month43. Thus, this method could be used to follow the longitudinal changes in RVR. During longitudinal follow-ups, however, it will be important to keep track of the potential confounders to retinal vascular reactivity assessment such as coffee intake44. There may also be a need to be sensitive to diurnal variation which can impact the reactivity depending on the condition and retinal layer being studied45,46,47.
Despite the broad applicability of the method, a few factors need to be considered during patient recruitment. Although this non-rebreathing procedure does not use a hypoxic gas mixture, the increased resistance to respiration through the tube could pose additional risks to those already with obstructive lung diseases including asthma and chronic obstructive pulmonary disease. For subjects, including those with heart conditions, in which shortness of breath is already a concern, their participation in the study should receive additional scrutiny. In the case of more common vascular diseases including hypertension and diabetes, gas challenge tests have been performed with similar gas compositions in these patient populations in several studies8,9,48, and more recently with the described method2, and there have been no reports of adverse events in these papers.
Furthermore, although OCTA images contain significant information about the function of the retina and many parameters can be computed to quantify the morphology of the capillary bed49,50, as with many other imaging technologies, limitations in interpreting OCTA scans exist. Imaging defects including displacement artifacts, motion artifacts and projection artifacts50 can affect imaging quality. OCTA relies upon flow to detect signal without visualizing the endothelium or vascular wall. As a result, OCTA metrics involve indices that are representative of the intrinsic vascular properties but may not be perfect representations of the microvasculature. Comparisons with histology have shown that the real density of retinal vasculature may be greater than assessed with OCTA51. Additionally, temporal changes in flow within microvessels less than 10–15 µm can cause variation in OCTA image intensity between scans23. This is suspected to be due to flow rates below a minimum detectable velocity.
To conclude, the convenience of the gas exchange setup, the low cost of the materials, and the ability for the method to be applied to a wide variety of ophthalmic imaging devices mean that it will remain relevant to retinal imaging, especially with OCTA systems. By stimulating both a positive and negative RVR response, this setup may also be used to probe retinal vascular disease physiology as well as the limits of the OCTA systems themselves by visualizing those vessels that evade detection using the current technology but are apparent with additional stimulation.
The authors have nothing to disclose.
This work was supported by NIH K08EY027006, R01EY030564, UH3NS100614, Research Grants from Carl Zeiss Meditec Inc (Dublin, CA) and Unrestricted Department Funding from Research to Prevent Blindness (New York, NY).
5% CO2 gas [5% CO2, 21% O2, 74% N2] (Compressed) | Institution Dependent (Praxair) | ||
Bacdown Disinfectant Detergent | Decon Labs | 8001 | https://deconlabs.com/products/disinfectant-bdd/ |
Clean-Bor Tubes (35 mm Inner Diameter) | Vacumed | 1011-108 | http://www.vacumed.com/zcom/product/Product.do?compid=27&skuid=1197 |
Cuff adapter for Douglas bag filling | Vacumed | 22254 | http://www.vacumed.com/zcom/product/Product.do?compid=27&prodid=343 |
Douglas bag (200-liter capacity) | Harvard Apparatus | 500942 | https://www.harvardapparatus.com/douglas-bag.html |
Elbow Joint (Inner Diameter 19mm/ Outer Diameter 22 mm), Modified in House | |||
Fingertip Pulse Oximeter (Pro-Series) | CMS | CMS 500DL | https://www.walmart.com/ip/Pro-Series-CMS-500DL-Fingertip-Pulse-Oximeter-Blood-Oxygen-Saturation-Monitor-with-silicon-cover-batteries-and-lanyard/479049154 |
Gas Delivery Tube (22 mm Inner Diameter) Modified in House | |||
Gas filling tube (1/8" for compressed gas) | |||
Hydrogen Peroxide Cleaner Disinfectant Wipes | Clorox Healthcare | 30824 | https://www.cloroxpro.com/products/clorox-healthcare/hydrogen-peroxide-cleaner-disinfectants/?gclid=EAIaIQobChMIk-KG4vi15QIVcRh9Ch0NNwLPEAAYASAAEgJIa_D_BwE&gclsrc=aw.ds |
Lubricant Eye Drops | Refresh | Refresh Plus | https://www.refreshbrand.com/Products/refresh-plus |
Manual Directional Control Valves: Three-Way T-Shape Stopcock Type (Inner Diameter 28.6 mm, Outer Diameter 35 mm) | Hans Rudolph | 2100C Series | www.rudolphkc.com |
Medical O2 (Compressed) | Institution Dependent | ||
Mouth piece (Silicone, Model #9061) | Hans Rudolph | 602076 | www.rudolphkc.com |
OCTA Imaging Device (PLEX Elite 9000) | Carl Zeiss Meditec, Dublin, CA, USA | https://www.zeiss.com/meditec/int/product-portfolio/optical-coherence-tomography-devices/plex-elite-9000-swept-source-oct.html | |
Phenylephrine Hydrochloride Ophthalmic Solution, USP 2.5% | Paragon Bioteck, Inc | NDC 42702-102-15 | https://paragonbioteck.com/products/diagnostics/phenylephrine-hydrochloride-ophthalmic-solution-usp-2-5/ |
Plastic Nose Clip Sterile Foam CS100 | Sklar Sterile | 96-2951 | https://www.sklarcorp.com/disposables/plastic/plastic-nose-clip-sterile-foam-box-of-100.html |
Proparacaine Hydrochloride Ophthalmic Solution, USP .5% | Bausch + Lomb | NDC 24208-730-06 | https://www.bausch.com/ecp/our-products/rx-pharmaceuticals/generics |
Regulator (tank dependent- 5% CO2: Fisherbrand Mulitstage Gas Cylinder Regulators) | Genstar Technologies Company | 10575150 | https://www.fishersci.com/shop/products/fisherbrand-multistage-cylinder-regulators-22/10575150?keyword=true |
Regulator (tank dependent- Oxygen: Fisherbrand Multistage Gas Cylinder Regulators) | Genstar Technologies Company | 10575145 | https://www.fishersci.com/shop/products/fisherbrand-multistage-cylinder-regulators-22/10575145?keyword=true |
Rubber Tubing (Inner diameter 19 mm, Outer diameter 27 mm), Made in House | |||
Sealing tape- Parafilm Wrap (2" Wide) | Cole Parmer | PM992 | https://www.coleparmer.com/i/parafilm-pm992-wrap-2-wide-250-ft-roll/0672050?PubID=VV&persist=True&ip=no&gclid=EAIaIQobChMInY3vqomz5QIVfyCtBh1VSg64EAAYASAAEgJ9n_D_BwE |
Sterile Alcohol Prep Pads | Medline | MDS090670 | https://www.medline.com/product/Sterile-Alcohol-Prep-Pads/Swab-Pads/Z05-PF03816 |
Tropicamide Ophthalmic Solution, USP 1% | Akorn | NDC 17478-102-12 | http://www.akorn.com/prod_detail.php?ndc=17478-102-12 |
Tubing Adapter, Made in House | |||
Two-way non-rebreathing valve (2600 Series- Inner Diameter 28.6 mm, Outer Diameter 35 mm) | Hans Rudolph | 2600 Series, UM-112078 | www.rudolphkc.com |