Diagnostic potential of OCTA for early detection of diabetic retinopathy: A literature review

Abstract

Background

Early detection of diabetic retinopathy (DR) is crucial for preventing blindness. Optical coherence tomography angiography (OCTA) offers a non-invasive method for visualizing retinal microvasculature and may facilitate earlier detection of DR. This systematic review evaluates the current evidence regarding OCTA-derived biomarkers for the identification of early DR.

Methods

A systematic review of the literature was performed. Due to substantial heterogeneity in study designs, OCTA protocols, and outcome measures, a meta-analysis was not feasible.

Results

Twenty-one studies were included. OCTA scan sizes ranged from 3 × 3 mm to 6 × 6 mm and evaluated both the superficial and deep capillary plexuses. Vessel density (VD) was the most consistently reported biomarker, with reduced VD observed in early DR in 13 of 15 studies. Enlargement of the foveal avascular zone (FAZ) was reported in 9 of 17 studies. Microaneurysms were more frequently identified in diabetic eyes in 3 of 5 studies, while increased vessel tortuosity was reported in 2 of 5 studies.

Discussion

Current evidence suggests that OCTA can detect retinal microvascular abnormalities associated with early DR before advanced clinical manifestations become evident. Reduced VD emerged as the most consistent indicator of early disease, while an increased microaneurysm burden may serve as a complementary marker. In contrast, the diagnostic value of FAZ alterations remains uncertain.

Conclusion

By identifying the OCTA parameters most strongly associated with early DR, this review suggests the potential role of OCTA in earlier diagnosis, risk stratification, and monitoring of diabetic patients.

Keywords

Diabetic retinopathy OCTA vessel density foveal avascular zone

Introduction

Diabetic retinopathy (DR) is currently the leading cause of blindness worldwide.¹ DR places a heavy burden on the medical system of both developed and developing countries,^2–4 with some studies predicting a worldwide prevalence of 7.7% by 2030, affecting nearly 450 million adults. DR is a progressive disease, which can be categorized in three clinical stages: diabetes without clinically apparent retinopathy, non-proliferative diabetic retinopathy (NPDR), and proliferative diabetic retinopathy (PDR). The risk of severe vision loss increases with disease severity at the time of diagnosis.⁵ Conversely, when DR is detected at an early stage, treatment is highly effective and disease progression can often be prevented or delayed.⁶ The importance of the early diagnosis of DR is also evident by the META-EYE study by Yau et al.,⁷ which reported that approximately 35% of patients with diabetes had DR, while nearly 10% had vision-threatening disease. Ophthalmoscopy, with identification of microaneurysms has traditionally been regarded as the standard method for the diagnosis of early DR. Another imaging modality used for the detection of DR is fluorescein angiography, which although highly sensitive, its invasive and time-consuming nature limits its routine clinical use.⁸ On the contrary, optical coherence tomography (OCT), which provides high-resolution imaging of the choroid, retina and its layers as well as the vitreous gel, has won ground against the other diagnostic tools and has been established in the diagnosis of DR.⁹ Optical coherence tomography angiography (OCTA) is a non-invasive imaging modality that uses motion contrast to generate high-resolution volumetric blood-flow data, thereby producing detailed angiographic images without the need for dye injection.¹⁰ OCTA commonly employs scan sizes of 3 × 3 mm and 6 × 6 mm, and its software allows separate evaluation of the superficial capillary plexus (SCP) and the deep capillary plexus (DCP). OCTA provides both quantitative parameters, including vessel density (VD), foveal avascular zone (FAZ), perfusion density (PD), adjusted flow index (AFI), and percentage of non-perfusion area (PAN), and qualitative features, such as microaneurysms, capillary non-perfusion, and vascular tortuosity.¹¹ As a rapid, non-invasive imaging technique that provides high-resolution visualization of the retinal microvasculature together with numerous quantitative biomarkers, OCTA has considerable potential for detecting early retinal microvascular alterations and facilitating the prompt diagnosis of DR.^12,13 Despite the growing literature on OCTA biomarkers, the evidence remains inconclusive. Studies evaluating retinal microvascular changes in diabetic patients without clinically apparent retinopathy have not always produced consistent results. Differences in study design, patient characteristics, image acquisition protocols, and OCTA devices can all influence OCTA measurements and findings. In addition, the retinal vascular layer under investigation may represent an important source of variability, as demonstrated by Pereira et al.¹⁴ Consequently, heterogeneity in imaging protocols and outcome measures continues to limit direct comparisons across studies and complicates interpretation of the available evidence. Another potential application of OCTA is the remote diagnosis and monitoring of DR through teleophthalmology and virtual clinic models. Furthermore, OCTA-derived imaging data may facilitate the development and implementation of artificial intelligence (AI)-based diagnostic systems. Although OCTA has demonstrated considerable potential for the early detection and monitoring of diabetic retinopathy, the current body of literature remains insufficient to support its widespread adoption as an established diagnostic modality. Therefore, OCTA should currently be regarded as a promising rather than fully validated clinical tool.

The aim of this systematic review is to evaluate the current evidence regarding the role of OCTA in the diagnosis of early diabetic retinopathy. Specifically, we sought to assess whether OCTA-derived biomarkers can identify retinal microvascular changes before the onset of clinically detectable DR, thereby facilitating earlier diagnosis and potentially improving patient management and visual outcomes.

Methods

The Preferred Reporting Items for Systematic Reviews and Meta Analyses (PRISMA) checklist was followed for this review (Supplementary Figures 1 and 2).¹⁵

Data sources and search strategy

We included studies evaluating the use of OCTA for the early diagnosis of diabetic retinopathy. The following inclusion criteria were applied: 1) studies including at least two of the following populations: healthy controls, diabetic patients without retinopathy (NDR) or patients with diabetic retinopathy regardless of clinical stage; 2) studies using OCTA and its principal quantitative and qualitative parameters as the primary diagnostic tool; 3) studies including data from at least 50 eyes; and 4) studies published in the English language. The exclusion criteria were: 1) studies involving pediatric patients; 2) case reports or case series with a small number of examined eyes; 3) systematic reviews or meta-analyses; 4) studies evaluating eyes with diabetic macular edema; and 5) studies in which OCTA was used solely as a complementary tool for the clinical staging of diabetic retinopathy rather than for its early diagnosis.

The systematic literature search was initially performed in January 2023 and updated in December 2025 using PubMed, PubMed Central (PMC), and the Cochrane Clinical Trials Database. All eligible studies published up to December 2025 were included, with no lower publication-date restriction. The following search terms were used “OCTA” AND “Diabetic Retinopathy”, as well as “optical coherence tomography angiography” AND “Diabetic Retinopathy”. After retrieval of the search results, duplicate articles were removed, whereas in case of metachronous publications from the same institution, only the latest article or the article with the largest number of patients was included. Studies were subsequently screened by title, abstract, and full text. The authors declare no conflicts of interest and received no external funding. This review was not prospectively registered.

Study selection and data extraction

The search results were independently screened by two reviewers (LAP and AGP). Any disagreements were resolved through discussion with a third senior reviewer (MGP). The quality of the included studies was assessed using the Newcastle-Ottawa Scale (NOS) for non-randomized studies.¹⁶ No adaptations to the original NOS were made, allowing standardized quality assessment across all included studies. Using this scale, studies were awarded stars based on selection of the study groups (maximum four stars), comparability of the study groups (maximum two stars), and ascertainment of the exposure of outcome of interest (maximum three stars). Considering selection, it assesses how well the study groups were selected and whether they are representative of the target population. Comparability evaluates adjustment for important confounding factors. Finally, exposure of outcome of interest assesses the validity and reliability of outcome measurement in cohort studies. Studies receiving more than six stars were considered high quality. Studies receiving between two and six stars were considered of moderate quality, whereas studies receiving fewer than two stars were classified as low quality and were excluded.¹⁶ Despite differences in methodological quality, all included studies were analyzed and interpreted using the same approach.

The OCTA findings extracted from the included studies were categorized into qualitative and quantitative parameters. Qualitative parameters included the presence of microaneurysms, capillary non-perfusion, and vascular tortuosity. Quantitative data were extracted for vessel density (VD), perfusion index (PI), foveal avascular zone (FAZ), and capillary perfusion density (CPD).

The feasibility of a quantitative synthesis was assessed, however it was ultimately abandoned due to substantial clinical and methodological heterogeneity among the included studies. Specifically, studies differed considerably with respect to patient characteristics, definitions of study groups, OCTA devices, imaging protocols, retinal regions and vascular layers analyzed, as well as the quantitative parameters reported. Given these limitations, the reviewers agreed that a meaningful and statistically reliable meta-analysis was not feasible. Nevertheless, where appropriate, pooled means were calculated as unweighted averages of the reported study means. These values are presented solely as descriptive summaries and should not be interpreted as the results of a formal meta-analysis. Consequently, this review focuses on the qualitative synthesis of the available evidence regarding the potential role of OCTA in the early diagnosis of diabetic retinopathy.

For the purposes of this review, early diabetic retinopathy was defined as diabetic eyes without clinically detectable retinopathy (NDR) and eyes with non-proliferative diabetic retinopathy (NPDR), thereby excluding proliferative diabetic retinopathy (PDR).

Results

Search results

Figure 1 illustrates the flow diagram of the included studies. During the initial search, a total of 13658 articles were identified (157 from the Cochrane library, 2345 from PubMed and 11156 from PMC). After the removal of duplicate records and title screening, 357 articles remained for further evaluation. Following abstract screening, 264 records were excluded. Of the remaining 93 full-text articles assessed for eligibility, 72 were excluded because they did not meet the inclusion criteria described in the Methods section. Ultimately, 21 studies met the inclusion criteria and were included in the qualitative synthesis.^17–37

Study quality

Table 1 summarizes and presents the qualitative assessment of the 21 included studies. Of the included studies ten^17–26 were conducted in a retrospective design, whereas the remaining eleven^27–37 were prospective studies. The study by Veiby et al.²⁷ received seven stars (three for selection, one for comparability, and three for outcome) and was classified as a high-quality study. The remaining studies were classified as being of moderate quality, with the majority receiving six stars.

Table 1.

Qualitative assessment of the included articles.

Newcastle-Ottawa scale (NOS)
Study, date	Selection	Comparability	Outcome	Summary
Takase et al, 2015	3	1	2	6
Carnevali et al, 2017	3	1	2	6
Goudot et al, 2017	2	1	2	5
Simonett et al 2017	3	1	2	6
Nesper et al 2017	3	1	2	6
Alibhai et al 2017	3	1	2	6
Cao et al, 2018	3	1	2	6
Onishi et al, 2018	3	1	2	6
Alam et al 2018	3	1	2	6
Li et al 2019	3	1	2	6
Kim et al, 2019	3	1	2	6
Yang et al, 2019	3	1	2	6
Sacconi et al 2019	3	1	1	5
Rosen et al 2019	3	1	2	6
Temel et al 2020	3	1	2	6
Veiby et al 2020	3	1	3	7
Buyuktepe et al, 2020	3	1	2	6
Ragkousis et al, 2020	3	1	2	6
Karst et al, 2020	3	1	2	6
Ghassemi et al, 2021	3	1	2	6
Lajmi et al 2021	3	1	1	5

Study analysis

Table 2 summarizes the study populations and subgroups included in each study. The study populations were categorized into five groups: healthy controls (control), diabetic patients without diabetic retinopathy (NDR), diabetic patients with non-proliferative diabetic retinopathy (NPDR), a combined diabetic group (DG) comprising both NDR and NPDR patients, and patients with proliferative diabetic retinopathy (PDR). The DG group was formed by combining the NDR and NPDR groups, therefore including patients without clinically detectable retinopathy as well as those with early-stage diabetic retinopathy. This approach allowed patients with early diabetic retinal disease to be distinguished from those with proliferative diabetic retinopathy (PDR), facilitating comparisons between the early and advanced stages of the disease. Although participants with PDR were included, the primary analyses and interpretation of this review focused on the early stages of diabetic retinopathy (NDR and NPDR), with the PDR group included only to provide additional context regarding disease progression. Two studies^17,18 included only a control group and a DG group, without differentiating between NDR and NPDR, comprising a total of 208 eyes (97 control and 111 DG). Six studies^{19,20,28–31} included all five groups (control, DG, NDR, NPDR and PDR), including a total of 1154 eyes (288 control, 647 DG, 273 NDR, 374 NPDR and 219 PDR). Finally, the remaining thirteen studies^{21–27,32–37} included all the aforementioned groups except the diabetic group with proliferative retinopathy, representing a total of 1986 eyes (629 control, 1357 DG, 942 NDR and 415 NPDR). Overall, a total 3348 eyes were included. Of these, 1,014 were from healthy subjects (control), 2115 were from the combined diabetic group with no or early-stage diabetic retinopathy (DG), 1630 were from diabetic patients without retinopathy (NDR), 789 were from patients with non-proliferative retinopathy (NPDR) and 219 eyes were from patients with proliferative diabetic retinopathy (PDR). Eighteen studies^{18–21,23–25,27–37} included at least one eye per participant, with variable use of one or both eyes across participants. One study²⁶ analyzed both eyes of all enrolled patients, while in two studies,^17,22 it was not clearly defined.

Table 2.

Populations and subgroups of the included studies.

Study (First author, year)	N (participants unless indicated with^a)	Control (n, %)	DG^b (NDR + NPDR) (n, %)	NDR (n, %)	NPDR (n, %)	PDR (n, %)
Takase et al, 2015	63	19 (30.1%)	44 (69.9%)	24 (38.1%)	20 (31.8%)
Carnevali et al, 2017	50^a	25 (50%)	25(50%)	25(50%)
Goudot et al, 2017	74^a	40 (54%)	34 (46%)	34(46%)
Simonett et al 2017	51^a	23 (45.1%)	28 (54.9%)	9 (17.6%)	19 (37.3%)
Nesper et al 2017	181^a	44 (24.3%)	97 (53.6%)	45 (24.9%)	52 (28.7%)	40 (22.1%)
Alibhai et al 2017	79	40 (50.6%)	39 (49.4%)	39 (49.4%)
Cao et al, 2018	138^a	67 (48.5%)	71 (51.5%)
Onishi et al, 2018	180^a	44 (24.5%)	96 (53.3%)	44 (24.4%)	52 (28.9%)	40 (22.2%)
Alam et al 2018	160^a	40 (25%)	120 (75%)		120 (75%)
Li et al 2019	261^a	87 (33.3%)	146 (55.9%)	58 (22.2%)	88 (33.7%)	28 (10.8%)
Kim et al, 2019	70^a	30 (42.8%)	40 (58.2%)
Yang et al, 2019	372	80 (21.5%)	292 (78.5%)	292 (78.5%)
Sacconi et al 2019	66^a	32 (48.5%)	34 (51.5%)	34 (51.5%)
Rosen et al 2019	152	40 (26.3%)	74 (48.7%)	36 (23.7%)	38 (25%)	38 (25%)
Temel et al 2020	149^a	51 (34.2%)	98 (65.8%)	59 (39.6%)	39 (26.2%)
Veiby et al 2020	483^a	168 (34.8%)	315 (65.2%)	239 (49.5%)	76 (15.7%)
Buyuktepe et al, 2020	116	20 (17.2%)	96 (82.8%)	44 (38%)	52 (44.8%)
Ragkousis et al, 2020	247^a	55 (22.3%)	192 (77.7%)	103 (41.7%)	89 (36%)
Karst et al, 2020	192^a	33 (17.2%)	129 (67.2%)	51 (26.6%)	78 (40.6%)	30 (15.6%)
Ghassemi et al, 2021	188^a	40 (21.2%)	105 (55.9%)	39 (20.8%)	66 (35.1%)	43 (22.9%)
Lajmi et al 2021	76^a	36 (47.3%)	40 (52.7%)	40 (52.7%)

DG—Diabetic group including both NDR and NPDR | NDR—Diabetic group without manifested diabetic retinopathy | NPDR—Diabetic group with manifested early stage retinopathy | PDR—Diabetic group with proliferative retinopathy.

Number of eyes.

DG is a combined diabetic group including NDR and NPDR. Values for DG overlap with the NDR and NPDR subgroups and should not be interpreted as additional participants.

Table 3 summarizes the quantitative OCTA parameters. Eleven studies performed 3 × 3 mm scans, five performed 6 × 6 mm, four performed both 3 × 3 mm and 6 × 6 mm scans and one study did not clearly report the scan size used. With respect to the depth of the performed scans, twenty studies performed scans of the SCP plane, with eighteen of them performing scans also of the DCP plane. One study did not clearly specify the retinal layer in which the scans were acquired. Among the quantitative OCTA parameters, we focused on the most frequently reported biomarkers, namely vessel density (VD), perfusion index (PI), foveal avascular zone (FAZ), and capillary perfusion density (CPD). Fifteen studies reported findings on VD. Thirteen^{17–22,25–27,29,31,35,37} of them reported lower VD in the DR group compared with control group, whereas two^23,33 of them reported no significant difference. Furthermore, four studies reported progressively lower VD with increasing severity of diabetic retinopathy, whereas one study did not confirm this association. The descriptive mean parafoveal VD in the DCP was 54.70 ± 2.62% in healthy controls, 52.66 ± 3.25% in the NDR group, and 49.74 ± 3.94% in the NPDR group. In the PDR group, the mean VD was 50.94%, however, a standard deviation (SD) could not be calculated because SD data were incompletely reported across the included studies. In the SCP, the descriptive mean parafoveal VD was 47.53 ± 2.52% in healthy controls, 45.88 ± 3.31% in the NDR group, 41.68 ± 4.26% in the NPDR group and 42.28% in the PDR group. Similarly, an SD could not be calculated for the PDR group because of incomplete reporting across studies. PI was reported only in one¹⁷ study and was lower in the DR group than in healthy controls. FAZ was evaluated in seventeen of the included studies. Nine studies^{17,20,21,24–26,30,34,36} reported a larger FAZ area in the DR group than in healthy controls, whereas eight studies^{18,22,23,27,32,33,35,37} found no significant difference between the groups. Furthermore, four studies^20,21,26,30 reported a progressive increase in FAZ area with increasing severity of diabetic retinopathy. Across studies reporting FAZ measurements in the DCP, the descriptive mean FAZ area was 0.451 ± 0.157 mm² in healthy controls, 0.499 ± 0.165 mm² in the NDR group, and 0.457 ± 0.111 mm² in the NPDR group. In the SCP, the descriptive mean FAZ area was 0.288 ± 0.040 mm² in healthy controls, 0.317 ± 0.053 mm² in the NDR group, and 0.348 ± 0.056 mm² in the NPDR group. CPD was reported in two^28,36 studies. One study²⁸ reported reduction in CPD with progression of the disease, whereas the other³⁶ reported lower CPD in the DR group than in healthy controls. Adjusted flow index (AFI) was reported in two^19,20 studies and was higher in the DR group in one¹⁹ of them, while it was found to be lower in the DR group in the other one.²⁰ The percentage of non-perfusion area (PAN) was reported in two^19,20 studies and was consistently higher in the DR group. In addition, two other studies^25,32 reported larger retinal non-perfusion areas, although these were not expressed as percentages. Both studies demonstrated larger non-perfusion areas in the DR group than in healthy controls. Finally, one study³² reported a higher FAZ remodeling index in the DR group. Regarding imaging devices, 16^{18–22,24–26,29–36} of the 21 studies (76%) utilised the RTVue XR Avanti SD-OCTA system (Optovue Inc., Fremont, CA, USA) in combination with the AngioVue software, with one study additionally employing the AngioAnalytics software module. Two studies^17,37 (9.5%) used the CIRRUS HD-OCT Model 5000 OCTA system (Carl Zeiss Meditec, Inc., Dublin, CA, USA) together with the AngioPlex software, whereas two other studies^23,28 (9.5%) utilised the Plex Elite 9000 system (Carl Zeiss Meditec, Dublin, CA, USA) and its integrated software platform. Finally, one study²⁷ (5%) employed the RS-3000 Advance OCTA device (NIDEK Co., Ltd., Gamagori, Japan) in combination with the AngioScan software.

Table 3.

Quantitative measurements in OCTA.

Study date	Scan size	SCP	DCP	VD	PI	FAZ	CPD	Imaging device	Comments
Takase et al, 2015	N/A	+	+	N/A	N/A	Larger^a	N/A	RTVue XR Avanti SD-OCT
Carnevali et al, 2017	3 × 3 mm	+	+	Lower ^a,b	N/A	No difference	N/A	CIRRUS HD-OCT model 5000
Goudot et al, 2017	3 × 3 mm	+	+	No Difference	N/A	No difference	N/A	RTVue XR Avanti SD-OCT
Simonett et al 2017	3 × 3 mm	+	+	Lower^a,b	N/A	No difference	N/A	RTVue XR Avanti SD-OCT
Nesper et al 2017	3 × 3 mm	+	+	Lower^a,c	N/A	Larger^a,c	N/A	RTVue XR Avanti SD-OCT	PAN^& was increased^a,c, AFI was decreased^a,c
Alibhai et al 2017	3 × 3 mm	+	+	N/A	N/A	Larger^a	N/A	RTVue XR Avanti SD-OCT
Cao et al, 2018	6 × 6mm	+	+	Lower^a,b	N/A	No difference	N/A	RTVue XR Avanti SD-OCT
Onishi et al, 2018	3 × 3 mm	+	+	Lower^a,b	N/A	N/A	N/A	RTVue XR Avanti SD-OCT	AFI and PAN were increased^a
Alam et al 2018	6 × 6 mm	+	+	Lower^a,c	N/A	Larger^a,c	N/A	RTVue XR Avanti SD-OCT
Li et al 2019	6 × 6 mm	+	+	Lower^a,b,c	N/A	N/A	N/A	RTVue XR Avanti SD-OCT
Kim et al, 2019	3 × 3mm	+	-	Lower ^a,b	Lower ^a	Larger^a	N/A	CIRRUS HD-OCT model 5000
Yang et al, 2019	3 × 3 mm 6 × 6 mm	+	+	Lower ^a,b	N/A	No difference	N/A	RTVue XR Avanti SD-OCT
Sacconi et al 2019	3 × 3 mm 6 × 6 mm	+	+	No difference	N/A	No difference	N/A	Plex Elite 9000	Perfusion density was lower ^a
Rosen et al 2019	3 × 3 mm	N/A	N/A	N/A	N/A	Larger^a,c	N/A	RTVue XR Avanti SD-OCT
Temel et al 2020	6 × 6 mm	+	+	Lower^a,b	N/A	Larger^a	N/A	RTVue XR Avanti SD-OCT	Non Perfusion area was more evident ^a
Veiby et al 2020	3 × 3 mm	+	+	Lower^a,c	N/A	No difference	N/A	RS-3000 Advance
Buyuktepe et al, 2020	6 × 6 mm	+	+	N/A	N/A	Larger^a	Lower ^a	RTVue XR Avanti SD-OCT
Ragkousis et al, 2020	3 × 3 mm, 6 × 6 mm	+	+	Lower ^a,b	N/A	Larger^c	N/A	RTVue XR Avanti SD-OCT
Karst et al, 2020	3 × 3mm	+	+	N/A	N/A	N/A	Lower^b,c	Plex Elite 9000
Ghassemi et al, 2021	3 × 3 mm	+	+	Lower ^a,b,d	N/A	N/A	N/A	RTVue XR Avanti SD-OCT
Lajmi et al 2021	3 × 3 mm, 6 × 6 mm	+	-	N/A	N/A	No difference	N/A	RTVue XR Avanti SD-OCT	Non-perfusion zones and FAZ remodelling were more evident ^a

In the DR group in comparison with the control group.

Considering the parafoveal area.

By progression of the retinopathy.

Considering the perifoveal area.

Table 4 summarizes the qualitative findings of the included studies. Within this section of the review, we evaluated the reported presence of microaneurysms, capillary non-perfusion and vessel tortuosity. Five studies^{18,32,33,35,37} assessed the presence of microaneurysms. Three^18,32,35 of them reported an increased prevalence of microaneurysms in the DR population, whereas two^33,37 of them found no significant differences between the groups. Given the heterogeneity in reporting methods and the lack of standardized quantitative measures (e.g., predefined thresholds or microaneurysm counts), quantitative effect estimates and pooled analyses could not be performed for microaneurysms. Three^18,33,37 articles reported findings on capillary non-perfusion, with only one¹⁸ of them demonstrating an increased prevalence in the DR group. Vessel tortuosity was evaluated in five^{26,32,33,35,37} studies and was reported to be increased in the DR group in two^26,35 of them. Two^33,37 studies presented no significant difference in vessel tortuosity between the groups, whereas one³² of them reported increased tortuosity in only one eye among the 292 included in the study. One³⁵ study also reported increased FAZ erosion and non-perfusion area in the DR group.

Table 4.

Qualitative measurements in OCTA.

Study date	Scan size	SCP	DCP	Microaneurysms incidence	Capillary nonperfusion	Tortuosity	Comments
Takase et al, 2015	N/A	+	+	N/A	N/A	N/A
Carnevali et al, 2017	3 × 3 mm	+	+	No difference	No difference	No difference
Goudot et al, 2017	3 × 3 mm	+	+	No difference	No difference	No difference
Simonett et al 2017	3 × 3 mm	+	+	N/A	N/A	N/A
Nesper et al 2017	3 × 3 mm	+	+	N/A	N/A	N/A
Alibhai et al 2017	3 × 3 mm	+	+	N/A	N/A	N/A
Cao et al, 2018	6 × 6mm	+	+	Higher^a	Higher^a	N/A
Onishi et al, 2018	3 × 3	+	+	N/A	N/A	N/A
Alam et al 2018	6 × 6 mm	+	+	N/A	N/A	Higher^a,c
Li et al 2019	6 × 6 mm	+	+	N/A	N/A	N/A
Kim et al, 2019	3 × 3mm	+	-	N/A	N/A	N/A
Yang et al, 2019	3 × 3 mm 6 × 6 mm	+	+	Higher^a	N/A	Higher^a,b	FAZ erosion and Non perfusion area was more evident ^a
Sacconi et al 2019	3 × 3 mm 6 × 6 mm	+	+	N/A	N/A	N/A
Rosen et al 2019	3 × 3 mm	N/A	N/A	N/A	N/A	N/A
Temel et al 2020	6 × 6 mm	+	+	N/A	N/A	N/A
Veiby et al 2020	3 × 3 mm	+	+	N/A	N/A	N/A
Buyuktepe et al, 2020	6 × 6 mm	+	+	N/A	N/A	N/A
Ragkousis et al, 2020	3 × 3 mm, 6 × 6 mm	+	+	N/A	N/A	N/A
Karst et al, 2020	3 × 3mm	+	+	N/A	N/A	N/A
Ghassemi et al, 2021	3 × 3 mm	+	+	N/A	N/A	N/A
Lajmi et al 2021	3 × 3 mm 6 × 6 mm	+	-	Higher^a	N/A	No difference^d

DCP—Deep capillary plexus | SCP—Superficial capillary plexus | FAZ—foveal avascular zone | N/A—not available.

In the DR group in comparison with the control group.

was present only in 1 case out of 292.

Higher by progression of the retinopathy.

Increased vesssel tortuosity was pressent only in 1 eye out of 292 examined eyes.

Finally, three studies assessed the diagnostic performance of OCTA-derived biomarkers for distinguishing diabetic retinopathy stages from healthy controls. Alam et al.²⁶ reported high diagnostic accuracy for VD and FAZ parameters, with sensitivities of 93.62% and 91.64%, specificities of 91.69% and 90.91%, and AUC values of 0.909 and 0.878, respectively. Similarly, Ragkousis et al.²¹ demonstrated that FD-300, defined as the vessel density within a 300-μm annulus surrounding the FAZ, achieved a sensitivity of 77.2% and a specificity of 70.9% at an optimal cut-off value of 49.95%, with an AUROC of 0.833. Notably, FD-300 exhibited the strongest inverse correlation with diabetic retinopathy severity. In contrast, Rosen et al.³⁰ reported more modest diagnostic performance for VD measured within a 200-μm annulus, with an AUROC of 0.713 (95% CI: 0.598–0.827), a sensitivity of 66.7% at 95% specificity, and a specificity of 73.5% at 95% sensitivity.

Discussion

Diabetic retinopathy remains the leading cause of preventable blindness worldwide.¹ It represents the earliest and most frequently diagnosed treatable microvascular complication of diabetes.³⁸ Because treatment is most effective when initiated early, timely diagnosis is essential.⁶ Wykoff et al.⁵ demonstrated in their study that the more advanced clinical stages of DR, and particularly the proliferative disease, at the time of initial diagnosis were strongly associated with severe vision loss. OCTA is an imaging modality, that provides detailed information on retinal microvasculature but has not yet been fully exploited for the early detection of DR. The current literature contains numerous studies proposing a plethora of OCTA-derived biomarkers and quantitative metrics, without being able to provide a robust consensus about their clinical utility for the early diagnosis and staging of DR. Therefore we conducted this systematic review. During the review, substantial heterogeneity was observed across the included studies, both in the characteristics of the study populations and in the OCTA-derived biomarkers and variables evaluated.

Among the 21 included studies, six (28.5%) included participants across all clinical stages of DR, whereas the remaining fifteen studies (71.5%) excluded proliferative disease . The OCTA scan size also varied considerably across studies. In 52.3% of them, OCTA was performed using a 3 × 3 mm scan size, 23.8% of them used a 6 × 6 mm scan, an in 19% both scan sizes were used. In the remaining study it was not clear which scan size was used.

Regarding the depth of examination, SCP was the plane of examination in twenty out of 21 studies (95%), whereas DCP was the plane in eighteen studies (85.7%). Furthermore, in one study, the retinal layer examined by OCTA was not specified. With respect to quantitative OCTA metrics, VD and FAZ were the most frequently evaluated parameters with representation in 71.4% and and 80.9% of the included studies, respectively. The rest of quantitative measurements and indexes, including PI, CPD, AFI, PAN and non-perfusion area (not as percentage), were used only sporadically and were reported in fewer than 10% of the included studies.

With respect to the qualitative parameters, the variables evaluated included microaneurysms, capillary non-perfusion, vessel tortuosity, FAZ erosion and non-perfusion area. The representation of these qualitative parameters across the included studies was substantially lower than that of the quantitative OCTA metrics. Specifically, microaneurysms were reported in 23.8% of the studies, Capillary non-perfusion in 14.2% and vessel tortuosity in 23.8%. The remaining qualitative variables were each reported in only one study. These findings underscore the considerable heterogeneity in the current literature with respect to patient selection, retinal layer segmentation, scan size, and the definition and reporting of OCTA-derived measurements.

Based on the aforementioned findings, only the indexes, measurements and parameters reported in more than 20% of the included studies were selected for further evaluation. Consequently, four parameters were included in the final analysis: FAZ, vessel density (VD), microaneurysms, and vessel tortuosity. VD was reported to be lower in diabetic eyes without diabetic retinopathy or with early stages of diabetic retinopathy in comparison with non-diabetic eyes in thirteen (86.6%) of the fifteen studies, that evaluated this parameter. In the remaining two studies no difference between the two groups was reported. Additionally, seven (87.5%) of the eight studies assessing the association between VD and disease progression reported a significant inverse relationship, with lower VD associated with more advanced stages of DR. Only one study found no such association. Regarding FAZ alterations, nine (52.9%) of the seventeen studies reported a significant enlargement of FAZ in diabetic eyes without retinopathy or with early stage retinopathy in comparison with the non diabetic eyes, whereas the remaining eight studies found no significant differences between the groups. Among the nine studies that investigated the relationship between FAZ and disease progression, four (44.4%) reported a positive correlation between increasing FAZ size and DR severity, whereas five (55.6%) found no such association. An increased prevalence of microaneurysms in diabetic eyes was reported in three (60%) of the five studies that evaluated this parameter. Likewise, increased vessel tortuosity was reported in two (40%) of the five studies, whereas the remaining studies found no significant differences between groups.

Our findings suggest that VD is the most reliable and clinically relevant OCTA-derived parameter for the early detection of diabetic retinopathy. A progressive decrease in VD, particularly within the superficial and deep capillary plexuses, has consistently been observed in patients with early stages of DR, even before the onset of clinically apparent retinal changes.^39,40 This reduction in VD likely reflects early microvascular dysfunction and capillary non-perfusion, which precedes structural damage and vision-threatening complications.⁴¹ Therefore, quantitative assessment of VD using OCTA represents a non-invasive and sensitive biomarker of early retinal microvascular impairment in patients with diabetes.

In addition to VD, the increased prevalence of microaneurysms may serve as a supportive diagnostic indicator in the early stages of DR. Although traditionally identified through fluorescein angiography, the enhanced resolution of modern OCTA devices allows more precise visualization and quantification of these lesions without the need for dye injection.⁴² However, although microaneurysms may support the diagnosis of early DR, their number and distribution are highly variable and do not consistently correlate with disease severity or progression.^43,44

Conversely, changes in the foveal avascular zone remain a controversial and less reliable parameter for the early detection of DR. Although some studies have reported an enlargement or irregularity of the FAZ in diabetic patients without clinically evident retinopathy,^45–47 others have failed to demonstrate significant differences in FAZ morphology between diabetic and non-diabetic individuals.^23,32,48 The inconsistent findings regarding FAZ measurements may be explained by several biological and technical factors. FAZ size varies considerably among individuals and can be influenced by age, axial length, refractive error and normal anatomical variations. From a technical perspective, differences in scan protocols, segmentation methods, image quality, and measurement techniques may contribute to variability across studies. Furthermore, OCTA measurements are subject to technical limitations, including projection and motion artifacts, segmentation errors, reduced signal strength, and media opacities. Finally, variability in OCTA devices and software algorithms may further limit the comparability of results across studies. Given this lack of consensus in the literature, we recommend that FAZ metrics be interpreted with caution and should not be used as standalone biomarkers for the early diagnosis of DR.

Regarding diagnostic performance, although the available studies suggest promising potential of OCTA-derived biomarkers for differentiating diabetic retinopathy stages, these findings should be interpreted with caution. The limited number of diagnostic studies, together with substantial heterogeneity in study populations, OCTA devices, imaging protocols, and biomarker definitions, restricts direct comparison between studies and limits the generalizability of the reported diagnostic estimates. Further standardized validation studies are required to confirm the clinical utility of these biomarkers. From a clinical perspective, the findings of this review suggest that VD is the OCTA-derived parameter most consistently altered in the early stages of diabetes and therefore appears to be the most promising biomarker for the early detection of DR. In contrast, FAZ metrics and qualitative markers, such as microaneurysms, may provide complementary information but currently lack sufficient evidence to support their use as standalone diagnostic biomarkers.

This review also highlights the considerable heterogeneity in OCTA devices, imaging protocols, segmentation strategies, retinal plexuses evaluated, quantitative metrics, and outcome reporting, thereby limiting direct comparability across studies and precluding definitive conclusions regarding the most reliable OCTA-derived biomarkers for the early diagnosis of DR.

Conclusion

OCTA shows promise as a tool for early diabetic retinopathy detection; however, current evidence is insufficient for routine clinical implementation. VD appears to be the most promising biomarker, while microaneurysms and FAZ alterations may provide complementary information but require further validation. Larger, standardized studies including healthy eyes, diabetic eyes without retinopathy, and early DR stages are needed to establish the clinical role of OCTA, with particular focus on VD, microaneurysms, and FAZ assessment using both SCP and DCP measurements and standardized 3 × 3 mm and 6 × 6 mm scan protocols.

Limitations

This systematic review has several limitations. First, diabetes type, disease duration and metabolic control were not considered for evaluation in the analysis. While all 3 are important factors in the development and progression of diabetic retinopathy, examining their association with OCTA findings was beyond the scope of this study. In addition, these variables were not consistently reported across the included studies, making direct comparisons difficult. Further studies are needed to clarify how disease type and duration as well as glycaemic control influence OCTA-derived biomarkers in the early stages of diabetic retinal disease. Another limitation of this review is the inconsistent reporting of diabetic retinopathy stages across included studies. In several cases, it was not clearly specified whether diabetic eyes were classified as NDR or NPDR and in some studies the specific NPDR stage (mild, moderate, or severe) was not reported. In addition, some studies included only one of these subgroups. To allow a review of the available evidence, NDR and NPDR were therefore grouped together as early-stage diabetic retinal disease, which may introduce a degree of clinical heterogeneity within this category. Additionally, although study quality was assessed using the Newcastle–Ottawa Scale, quality ratings were not used to weight the evidence, and the findings should therefore be interpreted with appropriate caution. Finally, diagnostic accuracy metrics were inconsistently reported across studies, with only three studies providing sensitivity, specificity, and AUC values. Consequently, conclusions regarding the diagnostic potential of OCTA-derived biomarkers should be interpreted with caution, as the available evidence remains limited and heterogeneous.

Supplemental Material

sj-zip-1-ejo-10.1177_11206721261476180 - Supplemental material for Diagnostic potential of OCTA for early detection of diabetic retinopathy: A literature review

Supplemental material, sj-zip-1-ejo-10.1177_11206721261476180 for Diagnostic potential of OCTA for early detection of diabetic retinopathy: A literature review by Loukia A. Politi, Apostolos G. Pitoulias, Matthaios G. Pitoulias and Fotis Topouzis in European Journal of Ophthalmology

Footnotes

ORCID iD

Loukia A. Politi

Ethical approval and informed consents statement

As this article is a literature review, ethical approval and informed consent was not necessary.

Author contribution

Concept and design: LAP, FT. Analysis and interpretation: LAP, AGP, MGP, FT. Data collection: LAP, AGP, MGP, FT. Writing the article: LAP, AGP, MGP, FT. Critical revision of the article: LAP, AGP, MGP, FT. Final approval of the article: LAP, AGP, MGP, FT. Obtained funding: none. Overall responsibility: LAP, FT.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

Data are available by request from the corresponding author.

Supplemental material

Supplemental material for this article is available online.

References

Sabanayagam

Yip

Ting

, et al. Ten emerging trends in the epidemiology of diabetic retinopathy. Ophthalmic Epidemiol 2016; 23: 209–222.

Mendrinos

Stangos

Pournaras

. Diabetic retinopathy. BMJ Clin Evid 2007; 2007: 0702.

Azagew

Yohanes

Beko

, et al. Determinants of diabetic retinopathy in Ethiopia: a systematic review and meta-analysis. PLoS One 2023; 18: e0286627.

Brar

Sahoo

Behera

, et al. Prevalence of diabetic retinopathy in urban and rural India: a systematic review and meta-analysis. Indian J Ophthalmol 2022; 70: 1945–1955.

Wykoff

Khurana

Nguyen

, et al. Risk of blindness among patients with diabetes and newly diagnosed diabetic retinopathy. Diabetes Care 2021; 44: 748–756.

Gupta

Chhikara

. Diabetic retinopathy: present and past. Procedia Comput Sci 2018; 132: 1432–1440.

Yau

Rogers

Kawasaki

, et al. Global prevalence and Major risk factors of diabetic retinopathy. Diabetes Care 2012; 35: 556–564.

Kwiterovich

Maguire

Murphy

, et al. Frequency of adverse systemic reactions after fluorescein angiography. Ophthalmology 1991; 98: 1139–1142.

Corcóstegui

Durán

González-Albarrán

, et al. Update on diagnosis and treatment of diabetic retinopathy: a consensus guideline of the working group of ocular health (spanish society of diabetes and spanish vitreous and retina society). J Ophthalmol 2017; 2017: 8234186.

10.

de Carlo

Romano

Waheed

, et al. A review of optical coherence tomography angiography (OCTA). Int J Retina Vitreous 2015; 1: 5.

11.

de Barros Garcia

Isaac

Avila

. Diabetic retinopathy and OCT angiography: clinical findings and future perspectives. Int J Retina Vitreous 2017; 3: 14.

12.

Ishibazawa

Nagaoka

Takahashi

, et al. Optical coherence tomography angiography in diabetic retinopathy: a prospective pilot study. Am J Ophthalmol 2015; 160: 35–44.e1.

13.

Sambhav

Abu-Amero

Chalam

. Deep capillary macular perfusion indices obtained with OCT angiography correlate with degree of nonproliferative diabetic retinopathy. Eur J Ophthalmol 2017; 27(6): 716–729. doi: https://doi.org/10.5301/ejo.5000948

14.

Pereira

Faria

Domingues

, et al.

Foveal avascular zone area measurement in diabetic patients: superficial, deep or combined retinal vascular complex?

Microvasc Res 2025; 157: 104743.

15.

Moher

Liberati

Tetzlaff

, et al. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med 2009; 6: e1000097.

16.

Wells

Shea

O’Connell

, et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. Ottawa Hospital Research Institute, 2000.

17.

Kim

, et al. Progressive retinal neurodegeneration and microvascular change in diabetic retinopathy: longitudinal study using OCT angiography. Acta Diabetol 2019; 56: 1275–1282.

18.

Cao

Yang

Huang

, et al. Optical coherence tomography angiography discerns preclinical diabetic retinopathy in eyes of patients with type 2 diabetes without clinical diabetic retinopathy. Acta Diabetol 2018; 55: 469–477.

19.

Onishi

Nesper

Roberts

, et al. Importance of considering the middle capillary Plexus on OCT angiography in diabetic retinopathy. Invest Ophthalmol Vis Sci 2018; 59: 2167–2176.

20.

Nesper

Roberts

Onishi

, et al. Quantifying microvascular abnormalities with increasing severity of diabetic retinopathy using optical coherence tomography angiography. Invest Ophthalmol Vis Sci 2017; 58: BIO307–BIO315.

21.

Ragkousis

Kozobolis

Kabanarou

, et al. Vessel density around foveal avascular zone as a potential imaging biomarker for detecting preclinical diabetic retinopathy: an optical coherence tomography angiography study. Semin Ophthalmol 2020; 35: 316–323.

22.

Simonett

Scarinci

Picconi

, et al. Early microvascular retinal changes in optical coherence tomography angiography in patients with type 1 diabetes mellitus. Acta Ophthalmol 2017; 95: e751–e755.

23.

Sacconi

Casaluci

Borrelli

, et al. Multimodal imaging assessment of vascular and neurodegenerative retinal alterations in type 1 diabetic patients without fundoscopic signs of diabetic retinopathy. J Clin Med 2019; 8: 1409.

24.

Takase

Nozaki

Kato

, et al. Enlargement of foveal avascular zone in diabetic eyes evaluated by en face optical coherence tomography angiography. Retina 2015; 35: 2377–2383.

25.

Temel

Batıoğlu

. Optical coherence tomography angiography findings of diabetic patients with and without retinopathy. Eur J Ophthalmol 2021; 31: 3124–3132.

26.

Alam

Zhang

Lim

, et al. Quantitative optical coherence tomography angiography features for objective classification and staging of diabetic retinopathy. Retina 2020; 40: 322–332.

27.

Veiby

NCBB

Simeunovic

Heier

, et al. Associations between macular OCT angiography and nonproliferative diabetic retinopathy in young patients with type 1 diabetes Mellitus. J Diabetes Res 2020; 2020: 8849116.

28.

Karst

Heisler

, et al. Evaluating signs of microangiopathy secondary to diabetes in different areas of the retina with swept source OCTA. Invest Ophthalmol Vis Sci 2020; 61: 8.

29.

Ghassemi

Fadakar

Berijani

, et al. Quantitative assessment of vascular density in diabetic retinopathy subtypes with optical coherence tomography angiography. BMC Ophthalmol 2021; 21: 82.

30.

Rosen

Andrade Romo

Krawitz

, et al. Earliest evidence of preclinical diabetic retinopathy revealed using optical coherence tomography angiography perfused capillary density. Am J Ophthalmol 2019; 203: 103–115.

31.

Almansoob

Zhang

, et al. Quantitative analysis of retinal and choroid capillary ischaemia using optical coherence tomography angiography in type 2 diabetes. Acta Ophthalmol 2019; 97: 240–246.

32.

Lajmi

Hmaied

Othmen

, et al. Optical coherence tomography angiography microvascular changes in diabetics without diabetic retinopathy. Saudi J Ophthalmol 2020; 34: 156–159.

33.

Goudot

Sikorav

Semoun

, et al. Parafoveal OCT angiography features in diabetic patients without clinical diabetic retinopathy: a qualitative and quantitative analysis. J Ophthalmol 2017; 2017: 8676091.

34.

Alibhai

Moult

Shahzad

, et al. Quantifying microvascular changes using OCT angiography in diabetic eyes without clinical evidence of retinopathy. Ophthalmol Retina 2018; 2: 418–427.

35.

Yang

Wang

Yan

, et al. Microvascular retinal changes in pre-clinical diabetic retinopathy as detected by optical coherence tomographic angiography. Graefes Arch Clin Exp Ophthalmol 2020; 258: 513–520.

36.

Celik Buyuktepe

Özmert

Demirel

, et al. Role of inflammation in retinal microcirculation in diabetic eyes: correlation between aqueous flare and microvascular findings. Ophthalmologica 2020; 243: 391–398.

37.

Carnevali

Sacconi

Corbelli

, et al. Optical coherence tomography angiography analysis of retinal vascular plexuses and choriocapillaris in patients with type 1 diabetes without diabetic retinopathy. Acta Diabetol 2017; 54: 695–702.

38.

Shukla

Tripathy

. Diabetic Retinopathy 2023. In: StatPearls [Internet]. StatPearls Publishing, 2024.

39.

Zheng

, et al. Early neurovascular changes in the retina in preclinical diabetic retinopathy and its relation with blood glucose. BMC Ophthalmol 2021; 21: 220.

40.

Isaac

DLC

Pires

ACF

Neves

, et al. Effect of disease duration on foveal microvasculature assessed by OCTA in type 2 diabetes mellitus without clinical diabetic retinopathy. Int J Retina Vitreous 2025; 11: 66.

41.

Santos

Almeida

, et al. Retinal capillary nonperfusion in preclinical diabetic retinopathy. Ophthalmic Res 2023; 66: 1327–1334.

42.

Ong

CJT

Wong

MYZ

Cheong

, et al. Optical coherence tomography angiography in retinal vascular disorders. Diagnostics 2023; 13: 1620.

43.

Romero-Aroca

Garcia-Curto

Pascual-Fontanilles

, et al. Distribution of microaneurysms and hemorrhages in accordance with the grading of diabetic retinopathy in type diabetes patients. Diagnostics 2024; 14: 1547.

44.

Rasmussen

Broe

Frydkjaer-Olsen

, et al. Microaneurysm count as a predictor of long-term progression in diabetic retinopathy in young patients with type 1 diabetes: the Danish Cohort of Pediatric Diabetes 1987 (DCPD1987). Graefes Arch Clin Exp Ophthalmol 2015; 253: 199–205.

45.

Borodi

P-G

Slevin

Gavriș

, et al. Evaluating foveal avascular zone alterations in type 2 diabetes Mellitus and their association with C-reactive protein: a comparative study with healthy controls. Diabetology 2025; 6: 63. 10.3390/diabetology6070063

46.

Weihong

Xiao

, et al. A morphological study of the foveal avascular zone in patients with diabetes mellitus using optical coherence tomography angiography. Graefes Arch Clin Exp Ophthalmol 2016; 254: 873–879.

47.

Alzogool

Xiao

, et al. Optical coherence tomography angiography findings of neurovascular changes in type 2 diabetes mellitus patients without clinical diabetic retinopathy. Acta Diabetol 2018; 55: 1075–1082.

48.

Sacconi

Lamanna

Borrelli

, et al. Morphofunctional analysis of the retina in patients with type 1 diabetes without complications after 30 years of disease. Sci Rep 2020; 10: 206.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.98 MB