Topographical Distribution of Retinal and Optic Disc Neovascularization in Early Stages of Proliferative Diabetic Retinopathy

Abstract

Purpose.: We analyzed the topography of proliferative diabetic retinopathy (PDR), and visualized the distribution of neovascularization of the optic disc (NVD) and elsewhere in the retina (NVE).

Methods.: The study included 174 eyes of 106 patients with early PDR. Data on the size and location of 391 NVE and 73 NVD were converted into a database of two-dimensional retinal and optic disc charts. The geometric centers of the neovascular lesions were plotted into corresponding areas of the charts, and the topographic distributions of the NVE and NVD were visualized by merging the charts and displaying the number of overlapping lesions on color-coded contour maps.

Results.: A total of 141 (36%) NVE was located in the temporal and 250 (64%) in the nasal hemisphere (P < 0.0001). The distribution of the NVD in the temporal and nasal half of the optic disc was 46 (63%) and 27 (37%), respectively (P = 0.03). NVE in type 1 diabetes were located significantly farther from the fovea and optic disc, and were more numerous and larger than in type 2 diabetes. The number and diameter of the NVE were also significantly higher when the time from the last examination to the appearance of PDR exceeded 12 months.

Conclusions.: The majority of NVE lesions are located inferonasal to the optic disc and along the superior vascular arcades, while NVD have a predilection for the upper temporal disc rim. More extensive PDR is found in patients with type 1 diabetes and those with examination intervals longer than one year.

Introduction

Diabetic retinopathy is characterized by structural and functional alterations in the retinal microvasculature. Previous studies have shown that the microvascular disease has a predilection for certain parts of the retina. In the early stages of non-proliferative retinopathy, the vascular lesions are located most commonly in the posterior pole lateral to the macula. ¹ As the disease progresses, the vasculopathy and areas of capillary nonperfusion tend to involve the midperipheral temporal retina, and the larger vascular arcades. ^2–5 The vascular occlusions, retinal ischemia, and subsequent release of vasoproliferative factors may then lead to proliferative diabetic retinopathy (PDR), ⁶ which becomes clinically apparent as neovascularization of the optic disc (NVD) and elsewhere in the retina (NVE). Accurate data on the distribution of the neovascular lesions in PDR are scarce. Taylor and Dobree found that NVE arises most commonly from the superior temporal veins. ⁷ This observation was also reported by Feman et al., who showed that the superotemporal quadrant was the most frequent site of new vessel formation. ⁸ Except for a few studies about the origin and vascular supply of the NVD, ^9,10 we are not aware of previous reports regarding the topography of diabetic neovascularization confined to the optic disc. 

Regular screening is a widely accepted strategy to detect sight-threatening diabetic retinopathy in time to allow effective treatment. ^11,12 One of the main reasons for such screening procedures is the established efficacy of laser photocoagulation in the treatment of PDR. ^13,14 Knowledge about the pattern of new vessel formation in the diabetic fundus may facilitate early detection of PDR and, therefore, is of great importance for patient care and follow-up. The aims of our study were to analyze the topographic distribution of the early stages of NVE and NVD, and to visualize their extent in the ocular fundus. 

Methods

Patients

Haukeland University Hospital is the regional hospital for the three western counties of Norway (Hordaland, Rogaland, and Sogn og Fjordane). The population in Western Norway is homogeneous, with less than 4% of non-European origin. ¹⁵ At the Department of Ophthalmology, diabetic patients are referred routinely by ophthalmologists, endocrinologists, and general practitioners. A search in the institutional database was performed to find all patients who were diagnosed with PDR in our clinic between April 2000 and December 2011. The patients were identified using a combination of the International Classification of Diseases (ICD) diagnosis codes for diabetes mellitus and diabetic retinopathy, as well as the procedure codes for transpupillary photocoagulation and pars plana vitrectomy. A manual search in the patients' fundus image database then was performed by two experienced ophthalmologists (RWJ, JK) to find the photographs showing the very initial signs of NVE and/or NVD. Only eyes with high quality photographs, making it possible to determine clearly the presence and exact location of the neovascular lesions, were included in the study. Eyes with retinal vascular disorders other than diabetic retinopathy, and eyes that had undergone retinal laser treatment other than focal or grid laser to reduce macular edema were excluded. The study was registered and approved by The Regional Committee for Medical and Health Research Ethics, Western Norway, and followed the official ethical regulations for clinical research and the Declaration of Helsinki. 

Fundus Photography

For patients examined between April 2000 and June 2007, a digital fundus camera system Zeiss FF 450 IR (Carl Zeiss, Jena, Germany), equipped with Kodak Megaplus 1.6i/1.4i CCD cameras (Eastman Kodak, San Diego, CA), was used for routine photography with 50° fields of view. From July 2007, the Zeiss camera was replaced by the digital fundus camera system Canon CF-60 Dsi, combined with a Canon EOS-1 mark II camera (Canon, Inc., Tokyo, Japan), for photography with 60° fields of view. Routine examinations were performed by taking a total of six mydriatic photographs of each eye: one color photograph centered between the optic disc and macula, and five red-free photographs centered on the macula, temporal to the macula, nasal to the optic disc, superior to the optic disc, and inferior to the optic disc. In some patients, additional fundus photographs and fluorescein angiographic images were obtained to secure the ophthalmoscopic diagnosis. 

Background Parameters

Clinical information about the patients with eligible fundus photographs was obtained from their hospital files. The recorded data included the patient's age at the time the PDR was first diagnosed, sex, type of diabetes, age at diabetes debut, duration of diabetes, type of treatment, and time since the last examination (ophthalmoscopy and/or fundus photography) before the first appearance of PDR. 

NVE Parameters

Data regarding specific features of the neovascular lesions were obtained by careful examination of all the available fundus photographs. Recorded parameters included the total number of NVE in each eye, their largest diameter, and the distances of the nearest margin of the lesions to the fovea and optic disc margin. The metric parameters were calculated by using the horizontal diameter of the optic disc as a reference of 1.75 mm. ^16,17 The location of the NVE was determined according to their geometric center. The lesions then were categorized according to their meridional location in quadrants and hemispheres defined by a horizontal and vertical axis through the fovea. In addition, the presence of retinal hemorrhages, hard exudates, and fibrous tissue adjacent to the NVE was documented. 

NVD Parameters

Photographs of the optic discs were examined for NVD, that is new vessels on or within one disc diameter of the disc. The parameters evaluated were the number, largest diameter, and location of the NVD in each eye. The size of the NVD was calculated by using the horizontal diameter of the optic disc as a reference standard of 1.75 mm, and the location was determined according to the geometric center of the lesions in meridional sectors defined by a horizontal and vertical axis through the center of the optic disc. The extent of the NVD was expressed in three different concentric zones: within the optic cup, within the optic disc margin, and peripapillary extension. The presence of any hemorrhages, exudates, and fibrous tissue adjacent to the new vessels was also recorded. 

Fundus and Optic Disc Mapping

Some of the mapping procedures have been described previously. ^18–20 Based on the fundus photographs, all visible NVE lesions in each eye were drawn with azimuth equidistant projections on a standardized retinal drawing chart with a macular center surrounded by circles representing the equator, ora serrata, and limbus. ¹⁶ Special care was taken to correct for peripheral distortions when calculating retinal diagrams from the NVE parameters. ^21,22 Similarly, for each eye with photographically documented NVD, the lesions were drawn on a schematic chart of the optic disc designed according to the parameters described by Jonas et al. ¹⁷  

The drawing tools of the computer software PowerPoint (Microsoft Corp., Redmond, WA) then were used to convert all the fundus drawings into a database of two-dimensional retinal charts of right eyes. Based on these drawings, a custom-made MIPSpro Fortran 77 program (SGI, Sunnyvale, CA), using the computer software toolkit Netpbm (available online at http://netpbm.sourceforge.net), was used to determine automatically the geometric center of each NVE. All the central points were subsequently plotted into a single retinal chart, using the computer software Excel (Microsoft Corp.). The custom-made program also was used to merge, filter, and convert the collection of digital fundus drawings into a contour map of the fundus, displaying the number of overlapping NVE with different color codes. The same procedures were used for all the optic disc drawings, leading to a single right eye optic disc chart showing the geometric center of each NVD and a contour map of the optic disc displaying the total number of overlapping NVD with different color codes. Separate contour maps were made for specific subgroups of patients. Although the maximum number of overlapping lesions differed between these groups, each contour map was labeled with the same color scale. The dark blue color indicated areas without any neovascular lesions, and the dark red color revealed the area with the maximum number of overlapping NVE or NVD. 

Statistical Methods

The χ² goodness-of-fit test was used to analyze the distribution of the NVE and NVD under the null hypothesis that they are distributed uniformly in the retina and on the optic disc, respectively. The statistical analyses of the meridional location of all the lesions were performed under the assumption that each chart quadrant includes an equal area of the ocular fundus or optic disc. For the comparison between two groups of patients with different characteristics (binary variables), the meridional and concentric distribution of the neovascular lesions were evaluated. In eyes with multiple lesions, the mean lesion center was calculated computationally as a weighted average of the size and location of all the NVE or NVD in each eye, and used for the statistical analysis of the meridional lesion distribution. Fisher's exact test was performed for categorical variables and the Mann-Whitney U test for continuous variables. Multiple linear regression was used to determine the extent to which the NVE parameters were associated with age or diabetes type. Preliminary analysis of the data revealed a low correlation between fellow eyes. For this reason, and to avoid the bias that would occur by including all single eyes and randomly selecting only one eye of each patient with two eligible eyes, we decided to include all single and paired eyes that met our inclusion criteria. The data were analyzed using Graph Pad Prism software, version 4.0c (Graph Pad Software, Inc., San Diego, CA) and IBM SPSS Statistics, version 20 (SPSS, Inc., Chicago, IL). For all tests, two-tailed P values < 0.05 were considered statistically significant. 

Results

Patients

A total of 326 patients was found by the search strategy for appropriate diagnosis and procedure codes. The following search in the fundus image database identified 174 eyes (90 right and 84 left eyes) of 106 patients (36 women and 70 men) with PDR that met our inclusion criteria. A total of 38 (36%) patients had monocular and 68 (64%) had binocular involvement. In 26 (25%) patients, fluorescein angiography had been performed to confirm the presence of neovascularization. A total of 52 (49%) patients had type 1 diabetes, 53 (50%) had type 2 diabetes (34 on insulin and 19 on oral hypoglycemics), and 1 (1%) had diabetes secondary to pancreatitis. The median age at the time the PDR was first diagnosed was 48 years (mean 48.7, range 21–79 years). The median age at diabetes debut was 31 years (mean 31.5, range 0–68 years), and the median duration of diabetes was 17 years (mean 17.8, range 0–43 years). Among 97 (92%) patients who had undergone previous ocular examinations, the median time since the last examination to the appearance of PDR was 12 months (mean 19.2, range 2–132 months). The demographics and patient characteristics according to type 1 and type 2 diabetes are shown in Table 1. 

Table 1. 

 

Demographics and Characteristics of the Patients according to Type 1 and Type 2 Diabetes

	Type 1 (n = 52)	Type 2 (n = 53)	P
Eyes, n, right/left	42/41	47/42	0.89*
Sex, n, female/male	17/35	19/34	0.84*
Age, y, median (range)	33.5 (21–61)	64 (42–79)	<0.0001†
Age at diabetes debut, y, median (range)	12 (0–47)	50 (22–68)	<0.0001†
Duration of diabetes, y, median (range)	21 (10–43)	14 (0–39)	<0.0001†
Time since last examination, mo, median (range)	12 (2–96)	12 (3–132)	0.70†

 The patient with diabetes secondary to pancreatitis is excluded.

*   P value calculated using Fisher's exact test.

†   P value calculated using the Mann-Whitney U test.

NVE Characteristics

NVE was observed in 165 (95%) of the 174 eyes with PDR. The total number of NVE was 391, and the distribution of their central points is illustrated in Figure 1A. There was a statistically significant nasotemporal asymmetry in the distribution of the NVE, as 141 (36%) lesions were located in the temporal and 250 (64%) in the nasal hemisphere (P < 0.0001). The distribution of NVE in the superior and inferior hemisphere was 210 (54%) and 181 (46%) lesions, respectively (P = 0.14). A subanalysis of the nasotemporal distribution among three vertical zones of the ocular fundus revealed that 141 (36%) of the NVE lesions were located temporal to the macula, 77 (20%) within the zone bordered by the macula and optic disc, and 173 (44%) nasal to the optic disc. The median value of the largest distances of NVE from the fovea and optic disc was 7 mm (mean 6.8, range 1.8–14 mm) and 5.3 mm (mean 5.5, range 1.8–12.3 mm), respectively. No lesions were found in the central macular area. The median number of NVE in each eye was 2 (mean 2.4, range 1–9). A single NVE was seen in 64 (39%) eyes, whereas 101 (61%) eyes had multiple lesions. The median value of the largest NVE diameters was 1.8 mm (mean 1.8, range 0.4–4.4 mm). Retinal hemorrhages, hard exudates, and fibrous tissue were observed adjacent to 42 (11%), 14 (4%), and 28 (7%) of the NVE lesions, respectively. The distribution, number, and diameter of the NVE according to various patient characteristics are presented in Table 2. 

Figure 1. 

 

( A, B) Retinal charts with a macular center surrounded by circles representing the equator, ora serrata, and limbus. (A) The plotted central points of 391 NVE reveal the lack of neovascular lesions in the macular region, and the high NVE density nasal to the optic disc and along the vascular arcades. (B) The contour map of the merged fundus drawings illustrates the distribution and extent of 391 NVE. The colors indicate the number of overlapping NVE as given by the color scale to the right, where the top (dark red) indicates the maximum number (17) and the bottom (dark blue) no lesions.

Table 2. 

 

NVE Distribution in Various Fundus Hemispheres, Their Largest Distance from the Fovea and Optic Disc, Number and Largest Diameter, according to Binary Patient Characteristics

Binary Variables	Eyes	Hemisphere (n)						Distance (Mean ± SD, mm)				NVE (Mean ± SD, n and mm)
	n	T	N	P *	U	L	P *	Fovea	P†	Optic Disc	P†	n	P†	Diameter	P†
Female	53	19	34	0.28	27	26	0.87	6.7 ± 2.9	0.63	5.9 ± 2.5	0.39	2.4 ± 1.8	0.79	1.8 ± 1.1	0.72
Male	112	30	82		55	57		6.9 ± 2.4		5.4 ± 2.1		2.4 ± 1.5		1.8 ± 1.2
Age <50 y	94	26	68	0.61	55	39	0.01	7.3 ± 2.5	0.01	6.1 ± 2.5	0.0006	2.8 ± 1.7	<0.0001	1.9 ± 1.1	0.01
Age ≥50 y	71	23	48		27	44		6.1 ± 2.4		4.8 ± 1.6		1.9 ± 1.2		1.7 ± 1.2
Type 1 diabetes	79	20	59	0.23	46	33	0.04	7.6 ± 2.5	0.0002	6.3 ± 2.5	<0.0001	2.8 ± 1.6	0.0003	2.0 ± 1.2	0.002
Type 2 diabetes	84	29	55		35	49		6.0 ± 2.3		4.8 ± 1.7		2.0 ± 1.5		1.6 ± 1.1
Diabetes duration ≤17 y	86	21	65	0.13	40	46	0.44	6.4 ± 2.2	0.05	5.2 ± 2.0	0.03	2.4 ± 1.6	0.63	1.9 ± 1.2	0.70
Diabetes duration >17 y	79	28	51		42	37		7.3 ± 2.7		6.0 ± 2.4		2.3 ± 1.6		1.8 ± 1.1
Last examination ≤12 mo	84	26	58	1.00	42	42	0.87	6.8 ± 2.5	0.65	5.3 ± 2.1	0.05	2.0 ± 1.3	0.005	1.7 ± 1.2	0.02
Last examination >12 mo	66	21	45		32	34		7.0 ± 2.7		6.0 ± 2.3		2.7 ± 1.7		2.0 ± 1.2

 For eyes with multiple NVE, the mean hemispheric location is presented. L, lower; N, nasal; T, temporal; U, upper.

*   P values calculated using Fisher's exact test.

†   P values calculated using the Mann-Whitney U test.

The topography of the neovascular lesions, visualized by the merged retinal charts, corresponded clearly with the aforementioned numerical distributions. In general, the NVE lesions were located throughout the midperiphery of the posterior pole with the majority concentrated inferiorly and nasally to the optic disc, and along the superior vascular arcades (Fig. 1B). According to Table 2, statistically significant differences were found between patients with types 1 and 2 diabetes, as the NVE in type 1 diabetes were located farther from the fovea and optic disc, were more numerous, and larger than in eyes of patients with type 2 diabetes (Figs. 2A, 2B). Similar significant differences were observed in patients aged <50 years compared to those aged ≥50 years at the time the PDR was first diagnosed. Multiple linear regression analysis revealed that age group was associated significantly with the number of NVE (P = 0.004), while diabetes type was associated with the NVE distance from the fovea (P = 0.04). No significant associations were found for the NVE diameter or distance from the disc. The number and diameter of the NVE lesions were significantly higher in those patients for whom the time interval from the last examination before the appearance of PDR exceeded 12 months. The studied NVE parameters did not differ significantly between female and male patients, or between right and left eyes. 

Figure 2. 

 

(A, B) Contour maps showing the distribution and extent of the NVE according to diabetes type. The same color scale as in Figure 1B is used for both images. The maximum number of overlapping NVE (indicated by the dark red color) is given by the following numbers in parentheses. (A) Type 1 diabetes (13). (B) Type 2 diabetes (11).

NVD Characteristics

NVD was observed in 64 (37%) of the 174 eyes with PDR. In 9 (14%) of these eyes, there was no concurrent NVE. The total number of NVD was 73, and the distribution of their central points is illustrated in Figure 3A. There was a statistically significant nasotemporal asymmetry in the distribution of the NVD, as 46 (63%) were located in the temporal and 27 (37%) in the nasal sectors of the optic disc (P = 0.03). The distribution of the NVD in the superior and inferior sectors was 42 (58%) and 31 (42%) lesions, respectively (P = 0.20). The NVD lesions were located within the optic cup in 2 (3%) eyes, within the optic disc margin in 27 (42%) eyes, and showed peripapillary extension in 35 (55%) eyes. The median number of NVD in each eye was 1 (mean 1.1, range 1–3), and the median value of the largest NVD diameters was 1.3 mm (mean 1.9, range 0.4–8.8 mm). Hemorrhages, hard exudates, and fibrous tissue were seen adjacent to 10 (14%), 1 (1%), and none of the NVD lesions, respectively. The distribution, number, and diameter of the NVD according to various patient characteristics are presented in Table 3. 

Figure 3. 

 

(A, B) Optic disc charts with circles representing the neuroretinal rim and an “x” indicating the location of the fovea. (A) The plotted central points of 73 NVD reveal the high density of neovascular lesions in the upper temporal sector of the optic disc rim. (B) The contour map of the merged optic disc drawings illustrates the distribution and extent of 73 NVD. The colors indicate the number of overlapping NVD as given by the color scale to the right, where the top (dark red) indicates the maximum number (34) and the bottom (dark blue) no lesions.

Table 3. 

 

NVD Distribution in Various Meridional Sectors of the Optic Disc, Their Largest Extent from the Optic Disc Center, Number and Largest Diameter, according to Binary Patient Characteristics

Binary Variables	Eyes	Meridional Sector (n)						Extent (n)				NVD (Mean ± SD, n and mm)
	n	T	N	P *	U	L	P *	Intrapapillar		Peripapillar	P *	n	P†	Diameter	P†
								Cup	Disc
Female	20	14	6	0.58	12	8	1.00	1	10	9	0.42	1.2 ± 0.4	0.71	1.8 ± 1.6	0.62
Male	44	27	17		25	19		1	17	26		1.1 ± 0.4		1.9 ± 1.9
Age <50 y	40	25	15	0.79	23	17	1.00	0	16	24	0.31	1.2 ± 0.4	0.58	2.2 ± 2.0	0.04
Age ≥50 y	24	16	8		14	10		2	11	11		1.1 ± 0.4		1.2 ± 1.1
Type 1 diabetes	37	23	14	0.80	21	16	1.00	0	16	21	0.80	1.1 ± 0.4	0.70	1.9 ± 1.6	0.30
Type 2 diabetes	27	18	9		16	11		2	11	14		1.1 ± 0.4		1.8 ± 2.1

 For eyes with multiple NVD, the mean meridional location and the largest extent are presented.

*   P values calculated using the Fisher's exact test.

†   P values calculated using the Mann-Whitney U test.

The merged optic disc charts confirmed that the NVD were asymmetrically distributed, as the majority of the lesions were located on the neuroretinal rim in the upper temporal sector of the disc (Fig. 3B). Except for a slightly larger NVD diameter in patients aged <50 years, there were no statistically significant differences in the NVD parameters among different age groups or types of diabetes (Figs. 4A, 4B). Neither were there any significant differences between sexes, right and left eyes, nor between patients with different diabetes duration or time interval since the last examination before the diagnosis of PDR. 

Figure 4. 

 

(A, B) Contour maps showing the distribution and extent of the NVD according to diabetes type. The same color scale as in Figure 3B is used for both images. The maximum number of overlapping NVD (indicated by the dark red color) is given by the following numbers in parentheses. (A) Type 1 diabetes (23). (B) Type 2 diabetes (15).

Discussion

The study group represented a population of diabetic patients seen under routine conditions at a tertiary care ophthalmology department. The included patients were distributed equally between those with types 1 and 2 diabetes, and as anticipated, the age at diagnosis and duration of the disease differed significantly between these two groups. The male predominance was somewhat unexpected and its origin is unclear. Although many investigators have reported a higher risk of diabetic retinopathy among men, ^23–25 the sex distribution in our material could have been influenced by a higher proportion (60%) of male patients participating in the department's fundus photography screening program. Generally, we found no significant differences in the number, size or topography of the neovascular lesions between male and female subjects. 

Earlier studies have shown that the majority of NVE lesions are located in the superotemporal quadrant of the ocular fundus. ^7,8 In our study, the numerical distribution and the merged fundus drawings showed that neovascular lesions have a predilection for the nasal hemisphere. A significant number of the NVE were located inferonasally to the optic disc and along the superior vascular arcades. The discrepancy between our results and previous studies may be due partly to the use of different retinal charts. To perform statistics on the number of NVE in equal areas of the fundus, we used retinal charts with a macular center instead of an optic disc center. However, further analysis of the NVE distribution revealed that a substantial proportion (44%) of the lesions were confined to the hemisphere nasal to the optic disc, which is of clinical relevance for early detection and diagnosis of PDR. Other potential reasons for the discrepancies between the present and previous studies could be related to the different study populations and the time periods in which the studies were conducted. 

Our results regarding the distance of the NVE lesions from the optic disc correspond very closely to previously reported values. ^7,8 We found a mean distance between the NVE and the optic disc margin of 5.5 mm, which was significantly shorter than the length from the fovea. The merged fundus drawings further revealed that the nasal lesions lie closer to the optic disc than the temporal lesions. Diabetic neovascular lesions occur typically in the postequatorial retina. In our study, only 6 (2%) NVE were located beyond 10 mm from the optic disc, and they all were well within the photographic fields of the fundus. The majority (61%) of the eyes showed multiple NVE, suggesting that neovascular formation usually starts simultaneously at different locations in the retina. A particularly interesting finding was the significant difference between types 1 and 2 diabetic patients in the number, size, and location of the NVE. Whether this is a diabetes type–specific feature or has any other pathophysiologic relevance remains unknown. Nevertheless, patients with type 1 diabetes had more numerous, larger, and more peripherally located NVE, which may reflect the severity of the diabetic retinopathy. The similar results found in the age group below 50 years are probably related to the large age difference between the two diabetes types. The number and size of the NVE were also significantly higher in patients with a time interval longer than one year from the last ophthalmologic examination before the onset of PDR, a finding that may be an argument in favor of annual rather than more extended intervals for the routine screening of diabetic retinopathy. 

We found that only 9 (14%) of the eyes with NVD had no signs of concurrent NVE. This is in sharp contrast with the results obtained by Feman et al., ⁸ who reported that almost all eyes with neovascularization originating from the optic disc had no other proliferative changes. Although this discrepancy may have been influenced by the selection criteria of the study groups, we think that our results support the common clinical belief that the presence of NVD indicates a more advanced stage of PDR that usually develops after the onset of NVE. ^2,26,27 Similar to the retinal neovascularization, the NVD were not distributed uniformly. A significant number of the lesions were located on the neuroretinal rim in the upper temporal sector of the optic disc. As neovascularization may be induced by local tissue ischemia or the deposition of VEGF produced somewhere else in the fundus, ⁶ several factors could influence the location of the NVD. A possible explanation for the ischemia-induced neovascularization on the temporal half of the optic disc could be that this side has fewer blood vessels and a lower blood flow compared to the nasal side. ^28,29 Another explanation could be that the temporal part of the optic nerve head is more frequently situated within the so-called watershed zone, which is particularly susceptible to ischemia. ³⁰ However, as the presence of NVD strongly correlates with the extent of retinal nonperfusion, ² it is more likely that VEGF produced elsewhere in the fundus is responsible for the NVD development. Therefore, the location of the neovascular lesions may be determined by the peripapillary accumulation of VEGF or the distribution of VEGF receptors within the vasculature and glial cells of the anterior optic nerve. ³¹ Except for a slightly larger lesion diameter in the younger patients, there were no statistically significant differences in the NVD parameters between the studied groups. 

The main limitation of our study is its retrospective design. Other considerations include possible spherical distortions in the periphery when the NVE parameters were converted to the retinal charts, and the assumption that each quadrant or hemisphere comprises an equal area of the ocular fundus. The clinical estimation of the NVE size, using the optic disc diameter as a reference, may also lead to an overestimation of the most peripherally located lesions. 

In summary, we demonstrated that diabetic neovascular lesions are distributed asymmetrically throughout the retina and on the optic disc. The majority of the NVE lesions were located inferonasal to the optic disc and along the superior vascular arcades, while the NVD had a predilection for the temporal half of the optic disc. Improved knowledge of the preferential sites of new vessel formation can be helpful for the examiner, particularly in the very early stages of PDR where there may be some doubt about the nature of atypical retinal or optic disc vessels. There was a significant difference between the types of diabetes, as the NVE in type 1 diabetes were located more peripherally, and were more numerous and larger than in type 2. The number and diameter of the NVE were also significantly higher in patients with an examination interval longer than one year. Our findings may facilitate the search for early stages of diabetic neovascularization and give further insight into the process of ocular angiogenesis. As timely panretinal photocoagulation for PDR has a key role in reducing visual impairment and blindness, improved strategies for screening and early detection of the neovascular lesions are essential. 

Footnotes

 Disclosure: R.W. Jansson, None; T. Frøystein, None; J. Krohn, None