Effect of Panretinal Photocoagulation on Macular and Disc Vasculature using Optical Coherence Tomography Angiography

Background: To evaluate the changes of macular in the (SCP) and (DCP), flow, and parapapillary flow after panretinal photocoagulation (PRP). Method: In this prospective interventional non-comparative case series, patients with very severe nonproliferative (NPDR) and early proliferative diabetic retinopathy (PDR) and no significant macular edema who were candidates for panretinal photocoagulation underwent measurement of corrected distance visual acuity (CDVA), optical coherence tomography (OCT), Optical coherence tomography angiography (OCTA) at the baseline (1–5 days before PRP), 1, and 5 to 7 months following completion of PRP treatment. Results: Thirty-nine eyes from 21 patients with diabetes were enrolled. foveal SCP (P > 0.1), foveal DCP (P > 0.1), parafoveal SCP (P > 0.1), and parafoveal DCP (P > 0.1) did not change 1 month and 6 months after PRP. The parafoveal inner retina thick slab density was significantly decreased at 6 months after PRP (p=0.015). Deep FAZ area constricted 6 months following PRP (P = 0.075). Based on calculated circularity index, the FAZ area became significantly more circular. (P=0.047). One month after PRP the inside disc vascular density was significantly increased from baseline (p=0.041); while, it was decreased to lower than baseline amount, 6 months after PRP . Conclusion: Although OCTA parameters were not significantly affected by PRP at both short- (1-month) long-term (6-month) follow-up, but FAZ area may be become more circular and regular after PRP may be due to reflow of occluded capillary plexus.

vasculature and hemodynamics.
[3] The decrease in vascular endothelial growth factor (VEGF) may result in retinal and choroidal flow redistribution [4,5]. Additionally, the choroidal vasculature and thickness changes following PRP, have been investigated [6,7]; however, while some studies have reported that PRP can slow the process of retinal capillary closure in patients with diabetic retinopathy others have noted progressive capillary closure due to VEGF reduction. [8][9][10][11]. The most important shortcoming of such studies is the utilization of fluorescein angiography (FA) for quantification of retinal ischemia. The assessment of nonperfusion in the deep retinal capillary plexus is limited by FA due to masking with leakages and haemorrhages. [12] Optical coherence tomography angiography (OCTA), as a depth-resolved non-invasive modality rendered the possibility of mapping the retinal vasculature at different capillary plexuses. [13][14][15] There are multiple studies demonstrating the capability of OCTA in the quantification of microvascular density, capillary non-perfusion, Foveal avascular area (FAZ) and choroidal flow in diabetic patients.
[ [16][17][18][19] This study aims to evaluate the short-term and long term alterations in macular vascular density, FAZ area and regularity, choroidal flow, macular thickness and parapapillary flow after PRP in patients with early proliferative diabetic retinopathy (PDR) and very severe nonproliferative (NPDR)..

Methods
This prospective interventional case series was approved by local institutional review board (IR.TUMS.FARABIH.REC.1398.021, http://ethics.research.ac.ir/) and verbal informed consents were obtained from the patients. The study adhered to the tenets of the Declaration of Helsinki.
From February 2015 to November 2018, consecutive patients with very severe nonproliferative (NPDR) and early proliferative diabetic retinopathy (PDR) and no significant macular edema who were candidates for panretinal photocoagulation were enrolled in this experiment. Patents with high risk PDR, moderate or mild NPDR, central macular thickness more than 300 micron or evidence of any centre involved cystoid maculae edema based on optical coherence tomography (OCT), presence of fibrovascular proliferation in the macular area, uncontrolled glaucoma, uveitis, eyes with visual acuity less than 20/200 and refractive error > +3 and <-3 were excluded. The eye was also excluded in case of significant medial opacity impeding high quality imaging. History of previous panretinal photocoagulation, macular photocoagulation and anti-VEGF injections were other criteria for exclusion.
Patients underwent thorough ophthalmic examination including biomicroscopy and dilated indirect ophthalmoscopy and fundus photo. A masked optometrist measured the corrected distance visual acuity (CDVA) (using snellen chart), and the result was then converted to logarithm of the minimum angle of resolution (LogMAR).
RTVue XR 100 Avanti instrument (Optovue, Inc., Fremont, CA, USA) was used to perform optical coherence tomography angiography (OCTA). Images were acquired at the baseline (few days before PRP), 1, and 5 to 7 months following completion of PRP treatment. A 3×3 mm foveal centered image was obtained consecutively for each eye. Quality score more than 5 based on OCTA software report was the minimum requirement and imaging was repeated until this goal was achieved. Eyes with significant image quality or different artifacts preventing accurate measurement of the vascular density, flow or FAZ area were excluded.
After using projection artifact removal (PAR) algorithm, the built-in module in AngioAnalytics software (version 2017.1.0.151), different layers of retina were segmented automatically. The retinal slab for superficial capillary plexus (SCP) en face image was defined at an inner boundary at 3 μm under the internal limiting membrane (ILM) and having an outer boundary at 15 lm beneath the inner plexiform layer (IPL). The deep capillary plexus (DCP) en face image started at 15 μm beneath the IPL and ended at 70 μm below the IPL. Manual correction was executed and propagated in case of erroneous determination by the built-in software. In addition, to capture all flow signals that might be affected by induced intraretinal edema following PRP, an ''inner retinal slab'' was manually modified by starting at 3 μm under the ILM and ending at 70 μm or more beneath the IPL.
The vascular density was measured and recorded by the built-in software at the mentioned slabs..
The fovea and parafovea were defined based on Early Treatment Diabetic Retinopathy Study (ETDRS) grid considering the 1mm and 3mm rings.. The FAZ area in the DCP was automaticaly measured in mm 2 and rechecked by by two experienced investigators. If the software could not detect the outlines 6 of the non-flow area, an experienced investigator manually depicted the borders.
The automatically generated Perim (perimeter of the outlined FAZ) was used to calculate the circularity index with the below formula, it is a measure of compactness of a shape relative to a circle: The circularity index of a regular circle is 1.0. Thus, a ratio closer to 0 indicates an irregular shape.
The automatically measured choroidal flow, the peripapillary, inside disk and also the whole disc density were also documented. Patients were evaluated at months 4, 8 and 12 to determine the necessity of additional PRP.
Intravitreal bevacizumab was injected after PRP if the CMT was more than 310 and vision was decreased to less than 20/25.

Statistical Analysis
To assess the relation of the variables in different times, we used correlation coefficient. Also, we Nine eyes were excluded due to poor imaging qualities at 1 or 6 months after PRP or loss to follow up.
PDR was regressed after 1 month in all the 39 eyes and there was no need for additional PRP. The Table 1 shows visual acuity, superficial and deep retinal capillary densities, deep FAZ area, perim, circularity index and retinal thickness measurements before and after PRP. Foveal vessel density in the SCP and DCP were increased after PRP, although they weren't statistically significant (all P > 0.1). Parafoveal vessel density in the SCP and DCP statistically unchanged before and after PRP (all P > 0.1). Although the foveal inner retina thick slab density wasn't significantly different from the baseline at 1 st and 6 th month following PRP (P = 0.9 for both) but the parafoveal inner retina thick slab density was significantly decreased at 6 month after PRP (p = 0.015). The Central macular thickness (CMT) was less than 310, 1 month after PRP, in all cases. It was increased to more than 320 in 4 eyes at 6 months after PRP, that anti-VEGF was injected in 3 of them due to the CDVA less than 20/25.
Although choroidal flow was decreased after PRP but it was statically significant neither at 1 month nor at 6 months (P = 0.31 and 0.23, respectively).
Peripapapillary vascular density was decreased from 48.33 ± 3.63 at baseline to 46.37 ± 3.81 at 1 month after PRP (p = 0.082) and then increased to 47.04 ± 3.58. (p = 0.423 from baseline) One month after PRP the inside disc vascular density was significantly increased from 48.768 ± 5.879 at baseline to 55.603 ± 5.962 (p = 0.041); while, it was decreased to lower than baseline amount, 6 months after PRP (45.8 ± 5.176) (p = 0.106 from the baseline). (Table2) Mean vascular density of the superficial and deep capillaries in the fovea and parafoveal area, remained statistically unchanged short term and long term after PRP. Although, parafoveal "inner retina" vascular density decline started at 1 month after PRP, its reduction became statically significant 6 months after PRP. (Table1) It appears that this decrease was mostly due to superficial vascular density reduction, as deep vascular density at parafoveal area remained stable either 1 month or 6 months from the baseline. We found small parafoveal cysts in 16 eyes of 39 enrolled eyes in the study, which may induce the density reduction due to imaging faults. Also the rearrangement of vascular plexus after PRP and inward migration of the parafoveal vessels due to these cysts may be the other explanation for parafoveal inner retinal vascular density reduction. [24,25] In this investigation, foveal superficial and deep vascular density slightly increased following PRP; while not statistically significant. Redistribution of the foveal superficial and deep vascular circulation after PRP might be a reason for this phenomena. In addition, nitric oxide(NO) overproduction due to PRP-induced inflammation, may play an important role in vasodilation of the retinal capillaries. It could be a trigger for reperfusion of the occluded vessels, make them more detectable with OCTA.
[26, 27] This increase in vascular density was significantly associated with FAZ constriction (p = 0.05) We detected a mild but statisticaly significant increase in the CMT 1 month following laser treatment that remained significant until 6 months. But, there wasn't any significant change in CDVA after PRP  [26] Using different measurement techniques; OCTA vs. Doppler flowmetry or ICGA may be the main reason for this discrepancy between the studies.
OCTA provides an effective way to visualize the vessels in the peripapillary and inside disk region.
This allows us to evaluate their changes in the density after PRP. Peripapapillary vascular density was decreased from baseline to 1 month after PRP (p = 0.082) and then increased after 6 months to the amount less than baseline (p = 0.423). One hypothesis for this reduction is peripapillary neovascularization reduction. Inside disc vascular density was increased at 1 month PRP and then the decrease to lower than baseline after 6 months may be due to optic disc tiny vessels dilatation immediately after PRP that return to normal diameter along with neovascularization regression in the following months. [44] Our study had several limitations. The major drawback is the relatively small sample size impeding us from drawing a conclusion regarding the association between alterations of vessel density and PRP treatment. As this investigation was done on the eyes without significant macular edema and the eyes with high risk proliferative retinopathy were also excluded, therefore the results of this study should not be extrapolated to the treatment of the eyes with more severe retinopathies or macular edema. During the imaging process, several artifacts may interfere with OCTA measurements, especially in eyes with macular edema. Unfortunately, despite recent advances in the OCTA software, many of these artifacts cannot be readily fixed with the present-day technologies. Although the images with low quality were excluded, other forms of artifacts including projection, motion and segmentation artifact might still negatively affect our measurements. Another drawback is the lack of data regarding metabolic control in our patient, however the 5 to 7 months follow-up period deems it less necessary. Confounding factors such as the need for anti-VEGF injections and loss to follow up also precluded us from data gathering longer than 6 months.

Conclusion
Although OCTA parameters were not significantly affected by PRP at both short-(1-month) long-term (6-month) follow-up, but FAZ area may be become more circular and regular after PRP may be due to reflow of occluded capillary plexus.

Declarations
Ethics approval and consent to participate Availability of data and materials: The datasets used and /or analysed during the current study are available upon reasonable request.   The FAZ also became more circular during the study as evident by the images (A to C and D to F).