Effects of panretinal photocoagulation on macular vascular density and Foveal avascular zone area in patients with severe non-proliferative diabetic retinopathy by optical coherence tomography angiography


 Aim: To describe the changes of optical coherence tomography angiography (OCTA) characteristics in severe non-proliferative Diabetic Retinopathy(S-NPDR) eyes. And study the effect of Panretinal photocoagulation(PRP) on these characteristics.Methods: This is a prospective study including 31 eyes from 18 consecutive patients with S-NPDR and 31 eyes of healthy subjects. We measured macular vascular density (VD) and foveal avascular zone(FAZ) area by an OCTA device. All patients with S-NPDR underwent PRP treatment. In S-NPDR eyes, the VD and foveal avascular zone(FAZ) area were assessed at 1 week before PRP, 1 month, 3 months and 6 months following PRP.Results: Compared with the normal control group ,in superficial retinal capillary plexus (SCP), macular VD decreased in S-NPDR group expect foveal VD. In foveal ,p=0.7;In parafoveal and perifoveal, p＜0.001. In deep retinal capillary plexus (DCP), macular VD was also lower in S-NPDR group. In foveal, p=0.01.In parafoveal and perifoveal, p＜0.001. And FAZ area expanded in S-NPDR eyes(p=0.05). In S-NPDR eyes , at 3 months after PRP, foveal VD in DCP increased significantly(p=0.04). At 6 months after PRP, foveal VD increased both in SCP and DCP (P=0.01,P=0.008,respectively). At 6 months after PRP,FAZ area decreased (p=0.04).Conclusions : Patients with S-NPDR have retinal microcirculation disorder, and PRP can partially improve the macular microcirculation. VD and FAZ can be used as sensitive indicators for follow-up observation of diabetic retinopathy.


Introduction
Diabetic retinopathy(DR) is a common microvascular complication of diabetes, and remains the leading cause of blindness in working-aged people. Clinically, DR is divided into non proliferative DR(NPDR) and proliferative DR(PDR), which is marked by the appearance of retinal neovascularization (1). Panretinal photocoagulation (PRP) has been standard treatment for PDR since the advent of ophthalmic lasers .The aim of PRP is to cause regression of retinal neovascularisation in retina ,reduce the risk of vitreous hemorrhage, tractional retinal detachment and thereby prevent the progression of visual loss (2).Although PRP has been widely used in clinic, its mechanism is not very clear. The main hypothesis of the mechanism for PRP is that the PRP damages some of the highly metabolically photoreceptors, which consume lots of oxygen (3).This process may improve the oxygen supply to the inner retina, and reach a new balance between oxygen supply and demand ,to reduce the stimulus for neovascularization. Besides ,the decreased area of retinal tissue leads to improved oxygenation and a reduction in the levels of vascular endothelial growth factor(VEGF). A reduction in levels of VEGF may be important in reducing the risk of harmful new vessels forming.
Previous studies have investigated the change of ocular circulation after PRP and reported PRP reduces retinal blood ow (4). However, these studies concentrated on blood ow of whole retina rather than macula, which is more important and particular for the maintenance of visual function. According to our expectation, when the laser was placed on the peripheral parts of retina, it would cause thermal damage to outer retinal tissues, and blood will be distributed from periphery to macula region. It means although blood ow decreases in whole retina, the blood ow to macula may increase after PRP. FAZ is a capillaryfree area surrounded by separation of inner retinal layers. It is highly sensitive to ischemic events and can be an indicator of several pathologic processes. In our expectation, FAZ area may remain unchanged or decreased following effective PRP .
Optical coherence tomography angiography (OCTA) is a new, noninvasive imaging technique that generates real-time data on vascular structures of eye without intravenous injection of uorescent dyes. By OCTA, we can get information of individual layers and different parts of retina. Besides, OCTA becomes a new tool for follow-up observation of changes in FAZ (5). Using OCTA, we evaluated the changes of macular VD and FAZ area between healthy eyes and S-NPDR eyes. And then we studied the long-term effects of PRP on macular VD and FAZ areas of S-NPDR patients after PRP.

Design and population
The study followed the tenets of the Declaration of Helsinki, and was approved by the Ethics Committee of First A liated Hospital of Nanjing Medical University (2020-SR-269).This study prospectively analyzed patients with S-NPDR who had recently received PRP in the First A liated Hospital of Nanjing Medical University. The S-NPDR patients who were ready to receive PRP, and age-matched healthy subjects were included in the study. All participants or guardians were informed consent. And all participants were not involved in the design,or conduct, or reporting, or dissemination plans of our research.

Panretinal photocoagulation treatment
According to the recommendations of the Early Treatment Diabetic Retinopathy Study Group, PRP was applied in 4 peripheral quadrants in 2 to 4 sessions by an experienced ophthalmologist who was unaware of the study design. All photocoagulation sessions were performed with an interval of 1 week. A total of 1200 to 2000 spots were applied per eye with the spot size of 200-500 µm and pulse duration of 0.2 seconds.

Optical coherence tomography angiography
Optical coherence tomography angiography scans were obtained using AngioVue (version 2017.1.0.151); Optovue, Fremont, CA, USA) with the angio retina mode (6 × 6 mm). Automatic segmentation was performed by the viewing software to generate en face projection images of the super cial retinal capillary plexus (SCP), deep retinal capillary plexus (DCP). The SCP is segmented from inner internal limiting membrane to 9 µm above lower boundary of inner plexiform layer. The DCP is segmented from 9 µm above inner plexiform layer to 9 µm below lower boundary of outer plexiform layer (Fig. 1). To reduce the effect of artifacts and segmentation errors, we excluded images with severe artifacts(≤ 6/10) and performed manual correction of the segmentation of retina in every image by a well-trained ophthalmologist. Then, images were checked by another two ophthalmologists, only when both of them agreed the segmentation, can an image be used in analysis.Angiography information calculated by embedded software in the machine comprises the area of the foveal avascular zone (FAZ), and the VD of SCP, DCP. VD was de ned as the percentage area occupied by vessels. All OCTA examinations were performed by the well-trained studier.

Statistical analysis
Statistical analysis was performed using an SPSS software package (SPSS 22.0). Independent samples t-test was used to compare the change of VD and FAZ area. P < 0.05 is considered as statistically signi cant.

Patient demographics
A total of 80 eyes were recruited in our study, and 18 of these eyes were excluded because of poor image quality and wrong segmentation of retina. Demographic and anatomic characteristics of patients included in this study are summarized in Table 1. In S-NPDR eyes which received PRP, mean log MAR BCVA at baseline (0.4 ± 0.2) showed no signi cant change( p = 0.6) with the last visit after PRP(0.4 ± 0.1).
Mean HbA1c at baseline was 5.6 ± 1.2(%),which also had no signi cant change (p = 0.5) to 5.4 ± 1.4(%) at the last visit. 2. Macular VD between normal eyes and S-NPDR eyes before PRP Table 2 shows the difference between healthy eyes and S-NPDR eyes before PRP. In SCP, macular VD was signi cantly lower in S-NPDR group than that in normal subjects expect fovea VD. (p<0.001,p = 0.7). In DCP, macular VD was signi cantly lower in S-NPDR group than that in normal eyes. In fovea, p = 0.01, and in parafovea and perifovea, p<0.001.  Table 3 shows the changes of VD in S-NPDR eye after PRP. Three months after PRP, the VD of DCP fovea increased rst (p = 0.04). Six months after PRP, the foveal VD of SCP and DCP increased signi cantly (p = 0.01,p = 0.008,respectively). There was no signi cant change in blood vessel density in other parts of the macula. Figure 2 illustrates the trend of VD in normal eyes, S-NPDR eyes before and after PRP.  Tables 2 and 3 show that, compared with healthy eyes, the FAZ area of S-NPDR eyes increased (p = 0.05).

3.FAZ area
In S-NPDR eyes receiving PRP, the area of FAZ decreased signi cantly 6 months after surgery (p = 0.04).

Discussion
In this prospective study, we found that S-NPDR eyes showed a signi cant decrease macular VD except fovea VD of SCP before PRP. Besides, FAZ area increased signi cantly. In S-NPDR eyes treated with PRP, VD only changed in fovea. Three months after PRP, the VD in fovea of DCP rst increased signi cantly.
Six months after PRP, the VD in fovea of SCP and DCP both increased. And FAZ area had a signi cant decrease at 6 months following PRP.
Consistent with previous studies (6, 7), our results showed whole en face VD decreased in SCP and DCP(p<0.001) in S-NPDR eyes. It should be noted, however, this change was less remarkable in fovea. In parafovea and perifovea, VD decreased both in SCP and DCP(p<0.001). In fovea, VD of SCP showed a non-signi cant decrease (17.6%-16.99%, p = 0.7) and VD of DCP decreased from 32.45-27.49% (P = 0.01). Perhaps it was because DR affects the peripheral retina rst, and fovea is damaged in the last stage. In addition, fovea is a region which overlies photoreceptors and has a high acuity vision.
(8)Another possibility is that relatively preserved fovea VD helps to maintain visual acuity and it may be a self-protective mechanism of retina. Of course, there is another possibility that fovea has fewer capillary than parafovea and perifovea, so the change of VD is not obvious. Further studies are needed to clarify the mechanism. Besides, we had an unexpected nding that fovea VD of SCP did not decreased signi cantly in S-NPDR eyes(p = 0.7). But fovea VD of DCP decreased signi cantly (P = 0.01). So far OCTA examinations have been performed on various layers of the retina of DR patients. In the early stage of DR, SCP showed high ow and DCP presented a steep decline of blood ow with increasing DR severity (6,9,10). Compared with SCP, blood ow of DCP decreased more steeply with the aggravation of DR. These ndings coincide with our result of fovea VD decreased only in DCP. One possible explanation is the self-regulation of SCP. With the progress of DR, SCP retains a certain degree of self-regulation function compared with DCP. Another possible explanation is that the dilated capillaries of SCP, which have increased blood ow and decreased vascular resistance, may contribute a steal phenomenon at DCP. (11) In S-NPDR eyes which received PRP, fovea VD increased. At 3 months following PRP, fovea VD in DCP increased rst(p = 0.03). At 6 months following PRP, fovea VD increased both in SCP and DCP(p = 0.01,p = 0.008, respectively).The changes of VD in S-NPDR eyes which received PRP were consistent to our expectation. However, the mechanism of this observation is not fully clari ed. Therefore ,we proposed several hypotheses. The rst assumption we put forward is that PRP may redistribute the circulation of ocular. It is generally accepted that laser energy is mainly absorbed by retinal pigment epithelium (RPE), destroys RPE and photoreceptor cells, and injures surrounding retina and choriocapillaris. Both in retina and choroid, capillary dropout in the areas of laser scars (12). There are some studies have suggested that choroidal vessels in the untreated macular area increased after retinal peripheral photocoagulation (13,14). For example, Takahashi reported PRP may increase choroidal blood ow in fovea in eyes with severe diabetic retinopathy (14). So, it is also possible that fovea VD in retina increased after PRP. Histological studies should be performed to prove this assumption. The second assumption is the recanalisation occurred in fovea vascular after PRP. As we all know, DR is a form of microangiopathy which accompanied by obstruction of capillaries. (15) Several ophthalmologists have demonstrated the occurrence of capillary revascularization (16,17). Rema Mohan and Evam Kohner reported a case which demonstrated the occurrence of capillary revascularization in a previously avascular area of the retina in a patient with advanced diabetic eye disease after extensive photocoagulation. Meanwhile, the vessels are in the plane of the retina and do not have the appearance of new vessels (16). By optics scanning light ophthalmoscopy, Chui, T.Y. observed capillary recanalized and remained perfused in fovea (17). These examples illustrated macular VD is reversible and it is possible that there are partially reversible components in DR. Thus ,we hypothesized that PRP may promote the recanalization of fovea microvascular. More studies are needed to verify these assumptions.
Interestingly, both fovea VD of SCP and DCP increased ,but the enhancement of DCP was more notable and earlier. In our investigation between S-NPDR eyes and normal eyes, S-NPDR eyes showed a signi cant decrease in fovea VD of DCP(p = 0.01),while fovea VD of SCP had no signi cant change(p = 0.7). Besides, as we have mentioned, many previous studies have suggested microvascular changes occur at DCP earlier than SCP in DR (6,9,10). These above two points may help explaining the phenomenon that DCP is more susceptible to PRP. Previous studies together with our observations suggest a role of DCP in evaluating therapeutic effect and monitoring DR progression.
Furthermore, we found FAZ area in S-NPDR eyes increased (p = 0.05) before PRP. At 6 months following PRP, FAZ area decreased (p = 0.04). In DR, an increase in FAZ area caused by capillary alteration has already been shown and it is in line with our nding (18). The FAZ is a specialized region of the human retina which is capillary-free and approximates the region of highest cone photoreceptor density and oxygen consumption. (19) It is well known that diabetes induces microvascular diseases including not only retinopathy but also choroidopathy ,and the disrupted choroidal circulation also has clinical relevance and contributes to the pathogenesis. Cells in FAZ are principally nourished by the choroid and during physiologic conditions the choroidal circulation is able to meet the metabolic demands of the FAZ (20). And the severity of DR is associated with enlarged FAZ area (21). Our nding that FAZ area decreased after PRP suggests PRP could reverse the progression of DR. Consequently, FAZ area may be a marker for monitoring the therapeutic effect in DR.
Before our study, Fawzi et.al studied 10 DR eyes and found VD decreased only in middle capillary plexuses which is located between SCP and DCP after PRP (22). But they only studied whole en face VD of retina and ignore different change in different region of macula and FAZ area.
Of course, there are several limitations in this study. Firstly, it is impossible for OCTA machine to avoid all of artifacts and segmentation errors of retina, particularly in eyes with poor vision. To reduce the effect of artifacts and segmentation errors, manual adjustments were performed to segment retina and edge FAZ in every image. Secondly, during our follow-up period, ve eyes appeared cystoid macular edema. Some previous studies reported a decrease of macular VD in patients with macular edema.(23, 24) So we supposed that macular edema will not make VD decreased. In addition, anti-VEGF therapy has been widely applied in recent years and has achieved short-term returns. However, anti-VEGF therapy needs high costs, regular follow-up and it is still to be studied whether it will relapse after anti-VEGF withdrawal.
If we can combine anti-VEGF and PRP, the macular edema after PRP may be inhibited. We are also carrying out relevant research. Last but not least, the patient numbers in our study was still limited. Future studies with larger samples are needed to verify our results.
In summary, by compared with healthy eyes, we found macular VD decreased in S-NPDR eyes, and the decrease of fovea VD is not as signi cant as parafovea and perifovea. After PRP, only fovea VD increase, especially in DCP. It suggests us fovea VD could be a possible predictor of the effectiveness of PRP. In addition, FAZ area decreased at 6months after PRP. It suggested FAZ area may also help in the assessment of treatment e cacy. Our study provided a new angle to the study of mechanisms of PRP and further work is needed to understand the effect of PRP better.   The trend of VD and FAZ area in normal eyes, S-NPDR eyes before PRP and S-NPDR eyes after PRP Figure 2 The trend of VD and FAZ area in normal eyes, S-NPDR eyes before PRP and S-NPDR eyes after PRP