Optical coherence tomography angiography for the assessment of retinal microvasculature characteristics in preterm-born children: A systematic review and meta-analysis

Abstract

This study aimed to evaluate the usefulness of optical coherence tomography angiography (OCTA) in assessing retinal microvascular structural changes in preterm-born children and compare them with those in term-born children. The Web of Science Library, Cochrane Library, PubMed, Embase, CNKI, Wanfang, VIP, and Sino Med databases were searched systematically to extract studies published till April 25, 2023. Two independent reviewers searched all the literature and completed the data extraction and quality assessment. Mean differences (MDs) with 95% confidence intervals (CIs) were used to assess the continuous estimates. STATA software (v15.1; StataCorp, College Station, TX) was used to analyze the data. Twelve published studies were eligible for inclusion in this study. The meta-analysis revealed that the foveal avascular zone (FAZ) area of preterm-born children was remarkably smaller than that of term-born children, with the laser photocoagulation (LP)-ROP group showing the most pronounced reduction. The foveal superficial capillary plexus vessel density (SCP-VD) and deep capillary plexus vessel density (DCP-VD) were remarkably higher in the preterm-born group than in the control group, with variations in subgroups (LP-ROP, anti-VEGF-ROP, SR-ROP, and Pre-T-ROP). The parafoveal SCP-VD was remarkably lower in preterm-born children compared to that of the controls, while no significant difference was identified in the parafoveal DCP-VD. Preterm-born children had a smaller FAZ area, higher foveal SCP-VD and DCP-VD, and lower parafoveal SCP-VD compared to their term-born counterparts. The parafoveal DCP-VD did not differ substantially between preterm- and term-born children. OCTA is an effective modality for assessing alterations in the retinal microvasculature in preterm children.

Children with a history of preterm birth are at an increased risk of developing retinopathy of prematurity (ROP). ROP occurs in preterm children before complete vascularization of the retina, which may affect the central retina and impair visual acuity.[1,2] The development of retinal blood vessels is initiated from the optic disc and extends to the central retina at a gestational age (GA) of 16 weeks. During vascular development, retinal vessels grow in a radial direction, sparing the central fovea, to form the foveal avascular zone (FAZ).[3] Data suggest that the period between 24 and 27 weeks of gestation is vital for foveal development as it encompasses vascular plexus development and foveal pit formation.[4] Therefore, preterm children are more likely to experience a delay in macular development.[5] Conventional treatment modalities for ROP include cryotherapy and laser photocoagulation (LP) for the avascular retina, which attempt to prevent disease progression. The effect of vascular endothelial growth factor (VEGF) has also been identified. Subsequent research has reported that intravitreal anti-VEGF injection could have better therapeutic value for severe ROP, leading to prompt regression.[6] Some recent studies have shown that preterm-born children who do not develop ROP or those with a history of spontaneously regressed or treated ROP do not have a completely normal retinal microstructure. Updating imaging technology is conducive to identifying subtle anatomical distinctions that may not be detectable clinically.

Optical coherence tomography angiography (OCTA) is a relatively novel and non-invasive imaging technique that facilitates better visualization and quantification of the retinal vasculature.[6] The advantages of OCTA over conventional fluorescein angiography (FA) include efficient acquisition and non-invasiveness. Importantly, OCTA can quantitatively assess the FAZ area, superficial capillary plexus vessel density (SCP-VD), and deep capillary plexus vessel density (DCP-VD). Several studies have examined and compared the variations in the retinal microvasculature in preterm-born children with those of their counterparts born at a normal GA with a normal birthweight (BW) by using OCTA, albeit with inconsistent results.[7,8,9] Therefore, a comprehensive meta-analysis of studies is required to overcome this shortcoming.

No meta-analysis has systematically investigated the structural changes in the retinal microvasculature by using OCTA in preterm-born children. Hence, the present meta-analysis assessed retinal microvascular characteristics in preterm-born children who received treatment with either LP (LP-ROP) or anti-VEGF agents (anti-VEGF-ROP), children with spontaneous regression of ROP (SR-ROP), and children who did not develop ROP (Pre-T-ROP) compared to term-born children. The results demonstrated that OCTA can be used to detect the characteristics of structural changes in the retinal microvasculature in preterm children.

Methods

This meta-analysis was performed in accordance with the Preferred Reporting Items in Systematic Reviews and Meta-analysis (PRISMA) guidelines and registered with the PROSPERO database (ID: CRD42021276504).[10] Original studies were identified via a systematic literature search of the Web of Science Library, Cochrane Library, PubMed, Embase, CNKI, Wanfang, VIP, and Sino Med databases, with a date restriction of April 25, 2023. Google Scholar and Baidu Scholar were also used to search for studies that were missing from these databases. Two independent researchers (QZ and BD) individually assessed the eligibility, extracted data, and determined the quality of the extracted studies. All discrepancies were resolved by discussion with a third researcher (FW). Moreover, the references listed in other related articles were scanned to identify eligible studies.

Search strategy

The search strategy included the following search terms: “premature birth,” “prematurity retinopathy,” “prematurity retinopathies,” “retinopathy of prematurity,” “optical coherence tomography-based microangiography,” “optical coherence tomography-based angiography,” “OCT angiography,” “optical coherence tomographic angiography,” “optical coherence tomography angiography,” “Angio-OCT,” and “OCTA.” The search strategy used for the PubMed database is presented in the Supplementary Material for representative purposes.

Inclusion criteria

The inclusion criteria for this meta-analysis were as follows: (1) studies that focused on preterm- and term-born children who could cooperate with the performance of the OCTA examination; (2) studies that provided OCTA data on the FAZ, SCP-VD, and DCP-VD; (3) OCTA data were provided as the mean ± standard deviation; (4) studies without overlapping participants; (5) original studies and those presented as full-text articles; and (6) studies published in English and Chinese.

Exclusion criteria

The exclusion criteria for this meta-analysis were as follows: (1) duplicate studies; (2) reviews, case reports, letters, animal studies, or conference articles with no data to extract; (3) studies with insufficient data; and (4) patients with ROP treated with vitrectomy.

Data extraction

Two authors (QZ and BD) independently assessed study eligibility by checking the abstract and full text and addressed disagreements through discussions with the third author (FW). All information was recorded in a predesigned table. The following information was extracted from each included study: first author and publication year, location, study design, number of eyes and participants, BW and GA, level of evidence, sex, age, OCTA device, macular scan size, and outcomes.

Quality assessment

The quality of the studies was assessed using the Newcastle–Ottawa Scale, which is a 9-point system that covers participant selection (0–4 points), comparability (0–2 points), and exposure (0–3 points).[11] Scores equaling 0–3, 4–6, and 7–9 indicated low, moderate, and high quality, respectively.

Statistical analysis

This meta-analysis was conducted using the STATA software (version 15.1; StataCorp, College Station, TX). The mean differences (MDs) and 95% confidence intervals (CIs) were calculated for the continuous estimates (FAZ area, SCP-VD, and DCP-VD). Scanning for en face angiographic visualization involved foveal and parafoveal scan patterns (1-mm and 3-mm circles of the Early Treatment Diabetic Retinopathy Study grid). Cochrane’s Q test and I² statistics were used to identify heterogeneity among the studies. A random-effect model was employed for significant heterogeneity (I² > 50% or P < 0.1), while a fixed-effect model was employed otherwise. Funnel plots were used to detect potential publication bias. A sensitivity analysis was conducted to test the robustness of the results. Subgroup analysis was conducted to explore the probable heterogeneity according to the different types of treatment received by preterm-born children (LP-ROP, anti-VEGF-ROP, SR-ROP, and Pre-T-ROP), and different OCTA devices used for examination (Zeiss, Jena, Germany; or Optovue, Fremont, CA). Statistical significance was set at P < 0.05.

Results

Search results and characteristics of eligible studies

Overall, 159 records were identified using the first search strategy. After eliminating 54 duplicates, 105 studies were assessed by their titles and abstracts. Finally, 12 studies, which included 574 children and 862 eyes, were selected for our meta-analysis according to the eligibility criteria [Fig. 1].

Figure 1.

Flowchart of the study selection process (till April 25, 2023)

Six cross-sectional and six retrospective studies were assessed. The mean age at testing was 7.49 ± 2.55 years in the preterm-born group and 8.60 ± 2.85 years in the term-born group. The mean BW and GA were 1203.80 ± 461.49 g and 29.14 ± 2.88 weeks in the preterm-born group and 3152.04 ± 392.89 g and 39.32 ± 0.99 weeks in the term-born group, respectively. The average BW and GA were higher in the Pre-T-ROP group (1800.53 ± 623.09 g and 32.37 ± 3.12 weeks, respectively) than in the LP-ROP (1014.76 ± 251.00 g and 27.88 ± 2.18 weeks, respectively), SR-ROP (1173.2 ± 394.75 g and 28.76 ± 2.44 weeks, respectively), and anti-VEGF-ROP groups (1019.32 ± 300.00 g and 27.40 ± 1.97 weeks, respectively). All included studies were published between 2017 and 2022; their salient characteristics are presented in Table 1. All involved studies were determined to be high-quality studies. The details of the quality evaluation of the involved studies are provided in the Supplementary Material.

Table 1.

General characteristics of the included studies

First author, year		Location		Design		Number of eyes/patients		BW (g)/GA (wk)		Classification of ROP		Gender (M/F)		Age (years)		OCTA device and macular scan size (mm)		Outcomes
Vinekar A. 2021[12]		Indian		Cross-sectional study		LP-ROP: 21/21 SR-ROP: 21/21 Pre-T: 3 9/39 Control: 8/8		NA NA NA 2837.00±70.00/38.00±1.10		NA NA NA		NA NA NA 2/6		NA NA NA 7.80±1.50		Optovue 3×3		FAZ aera, SCP-VD, DCP-VD
Mataftsi A. 2021[13]		Greece		Cross-sectional study		SR-ROP: 26/13 Pre-T: 32/16 Control: 34/17		1129.00.00±338/28.84±2.62 1734.00±586.00/31.83±2.26 3367.00±303.00/38.47±0.67		zone II/III, stage 1/2/3		6/7 2/14 11/6		7.29±0.73 7.87±0.50 7.35±0.86		Optovue 6×6		SCP-VD, DCP-VD
Liang Z Q. 2021[2]		China		Retrospective study		LP-ROP: 29/15 Control: 26/15		1318.46±273.10/30.57±2.08 NA		zone I/II, stage 3		8/7 8/7		7.73±1.83 8.70±1.77		Optovue 3×3		FAZ area
Vural, A. 2020[9]		Turkey		Cross-sectional study		Anti-VEGF-ROP: 18/18 LP-ROP: 19/19 SR-ROP: 18/18 Control: 16/16		1122.50±324.92/28.44±1.72 1079.47±274.25/28.21±2.22 1443.89±585.68/29.56±2.61 3502.50±584.12/39.69±0.60		zone I/II, stage 2/3 zone II, stage 2/3		8/10 5/14 11/7 6/10		5.06±1.11 5.68±1.15 5.50±1.24 5.94±1.12		Optovue 6×6		FAZ area, SCP-VD, DCP-VD
Tiryaki Demir S. 2020[14]		Turkey		Retrospective study		SR-ROP: 40/20 Control: 40/20		1040.00±253.00/28.70±1.60 3197.00±333.00/39.70±0.70		zone I/II/III, stage 1/2		13/7 9/11		7.00±0.80 7.00±0.80		Optovue 3×3		FAZ area, SCP-VD, DCP-VD
Takagi M. 2019[15]		Japan		Retrospective study		LP-ROP: 38/21 Control: 66/36		954.00/27.80 NA/≥37.00		Zone II/III, stage 3		NA 20/16		8.80±2.60 10.5±3.20		Optovue 3×3		FAZ area
Wu Z Q. 2019[16]		China		Cross-sectional study		SR-ROP: 12/6 Control: 16/8		1461.70±395.90/31.20±2.20 3222.50±237.10/39.60±0.70		zone I, stage 1/2; zone II, stage 3		3/3 5/3		7.30±2.80 8.30±1.60		Zeiss 3×3		FAZ area, SCP-VD
Miki A. 2019[17]		Japan		NA		LP-ROP: 14/14 Pre-T: 17/17 Control: 41/41		870.29±213.61/26.86±2.31 1925.76±688.40/33.41±4.20 NA		NA		8/6 6/11 12/29		10.00±1.57 10.00±2.87 9.41±2.40		Zeiss 3×3		FAZ area
Chen Y C. 2019[18]		Taiwan, China		Retrospective study		anti-VEGF-ROP: 42/23 control: 51/27		916.13±239.09/26.96±1.92 NA		Zone I/II, stage3		11/12 12/25		6.56±0.95 6.70±1.38		Optovue 3×3		FAZ area, SCP-VD, DCP-VD
Balasubramanian S. 2019[7]		USA		Cross-sectional study		LP-ROP: 12/7 Pre-T: 18/9 Control: 30/24		NA NA NA		NA		NA NA NA		7.83±2.89 8.11±3.95 8.93±3.20		Optovue 3×3		FAZ area, SCP-VD, DCP-VD
Leng Y X. 2018[8]		China		Cross-sectional study		LP-ROP: 15/14 Control: 40/20		880.50±158.90/26.40±1.30 2878.80±322.70/38.80±1.20		zone II/III, stage 3		8/6 11/9		11.50±3.70 11.10±3.80		Optovue 3×3		FAZ area, SCP-VD
Falavarjani KG. 2017[19]		USA		Cross-sectional study		LP-ROP: 10/6 SR-ROP: 18/9 Control: 15/11		871.10±172.60/25.80±1.20 1070.00±290.10/28.00±2.40 2992.60±217.80/39.00±1.00		NA		4/2 6/3 7/4		7.10±2.60 5.60±1.50 7.60±2.20		Optovue 3×3		FAZ area, SCP-VD

NA: not available; FAZ: foveal avascular zone; SCP-VD: superficial capillary plexus-vessel density; DCP-VD: deep capillary plexus-vessel density; LP: laser photocoagulation; SR: spontaneously regressed; Pre-T: preterm; anti-VEGF: anti-vascular endothelial growth factor; M/F: male/female; BW: birth weight; GA: gestation age

Analysis of the FAZ area in preterm- and term-born children

The FAZ area was analyzed in 11 studies that included 408 eyes of preterm-born children and 486 term-born control eyes. The FAZ area was significantly smaller in preterm-born children (MD: −0.17, 95% CI: −0.19 to − 0.15, P = 0.00; Fig. 2). Moderate heterogeneity was identified among these 11 studies (I² = 69.9%, P = 0.00). Sensitivity analysis revealed that the results were robust, and the funnel plot did not reveal any evidence of publication bias [Supplementary Material].

Figure 2.

Forest plot for the analysis of the foveal avascular zone area in the eyes of preterm- and term-born children

Subgroup analyses were conducted to explore the probable heterogeneity with respect to the different types of treatment received by preterm-born children and different OCTA devices used for examination. Meta-regression analysis revealed that subgroup analysis according to the different types of treatment for preterm-born children could explain the source of heterogeneity. Subgroup analysis according to the different subgroups of preterm-born children indicated that the FAZ area of the LP group was remarkably smaller than that of the healthy control group (MD: −0.19, 95% CI: −0.23 to − 0.16, P = 0.00; Table 2) with significant heterogeneity (I² = 72.9%, P = 0.00; Table 2). Patients in the SR-ROP, anti-VEGF-ROP, and Pre-T-ROP groups also had a smaller FAZ area than the healthy controls (MD: −0.16, 95% CI: −0.20 to − 0.12, P = 0.00; MD: −0.16, 95% CI: −0.21 to − 0.10, P = 0.00; MD: −0.12, 95% CI: −0.17 to − 0.08, P = 0.00, respectively; Table 2), with no significant heterogeneity (I² = 46.5%, P = 0.11; I² = 52.7%, P = 0.15; I² = 23.6%, P = 0.27, respectively; Table 2). Subgroup analysis by OCTA device revealed that two studies employed a Zeiss device (MD: −0.16, 95% CI: −0.25 to − 0.06, P = 0.00; I² = 72.5%, P = 0.03), and nine used an Optovue device (MD: −0.17, 95% CI: −0.20 to − 0.15, P = 0.00; I² = 71.5%, P = 0.00).

Table 2.

Main outcomes of subgroup analysis according to the different treatment methods of ROP

Outcome Variables		Parameters		Total		LP-ROP		SR-ROP		anti-VEGF ROP		Pre-T-ROP
FAZ area		No. of eyes MD		894 −0.17		410 −0.19		204 −0.16		127 −0.16		153 −0.12
		95% CI		−0.19, −0.15		−0.23, −0.16		−0.20, −0.12		−0.21, −0.10		−0.17, −0.08
		P		0.000		0.000		0.000		0.000		0.000
		I2 test (%)		69.9		72.9		46.5		52.7		23.6
SCP-VD (fovea)		No. of eyes MD		720 6.16		196 7.68		236 5.90		127 7.20		161 3.03
		95% CI		4.69, 7.63		4.96, 10.4		3.92, 7.88		5.23, 9.16		−0.14, 6.21
		P		0.000		0.000		0.000		0.000		0.061
		I2 test (%)		53.7		53.9		18.2		0.0		53.5
SCP-VD (parafovea)		No. of eyes MD		532 −1.71		106 −3.40		231 −0.95		34 −4.27		161 −0.99
		95% CI		−2.64, −0.77		−4.86, −1.94		−1.76, −0.14		−7.19, −1.35		−3.26, 1.28
		P		0.000		0.000		0.022		0.004		0.392
		I2 test (%)		60.4		0.0		0.0		0.0		79.9
DCP-VD (fovea)		No. of eyes MD		597 6.08		106 6.51		203 6.55		127 7.52		161 4.40
		95% CI		4.99, 7.16		3.81, 9.21		4.51, 8.58		5.28, 9.76		2.49, 6.31
		P		0.000		0.000		0.000		0.000		0.000
		I2 test (%)		0.0		0.0		0.0		0.0		10.4
DCP-VD (parafovea)		No. of eyes MD		401 0.01		71 −0.45		169 0.27		NA NA		161 −0.13
		95% CI		−0.75, 0.76		−2.28, 1.38		−0.82, 1.35		NA		−1.43, 1.16
		P		0.986		0.629		0.631		NA		0.839
		I2 test (%)		0.0		0.0		42.2		NA		0.0

MD: mean difference; CI: confidence interval; FAZ: foveal avascular zone; SCP-VD: superficial capillary plexus-vessel density; DCP-VD: deep capillary plexus-vessel density; LP: laser photocoagulation; SR: spontaneously regressed; Pre-T: preterm; anti-VEGF: anti-vascular endothelial growth factor; SD: standard deviation; BW: birth weight; GA: gestation age

Analysis of the foveal and parafoveal SCP-VD in preterm- and term-born children

Overall, eight studies analyzed the foveal SCP-VD by using the Optovue device in 720 eyes (359 eyes of preterm-born children and 361 eyes of term-born children). The pooled MD for the SCP-VD between the preterm-born children and controls was 6.16 (95% CI: 4.69–7.63, P = 0.00; Fig. 3), with moderate heterogeneity across studies (I² = 53.7%, P = 0.01), demonstrating that the SCP-VD of the fovea was remarkably higher in the preterm-born group than in the term-born children group. Sensitivity analysis showed that the results were robust, and the funnel plot did not reveal any evidence of publication bias [Supplementary Material]. Subgroup analysis was performed for this group. The pooled results indicated that the foveal SCP-VD of the LP-, SR-, and anti-VEGF-ROP groups was significantly higher than that of the healthy control group (MD: 7.68, 95% CI: 4.96–10.4, P = 0.00; MD: 5.90, 95% CI: 3.92–7.88, P = 0.00; MD: 7.20, 95% CI: 5.23–9.16, P = 0.00, respectively; Table 2), whereas substantial heterogeneity was identified in the studies with respect to the LP-ROP group (I² = 56.9%, P = 0.07; Table 2), and no heterogeneity was observed among the studies with respect to the SR-ROP and anti-VEGF-ROP groups (I² = 18.2%, P = 0.23; I² = 0.0%, P = 0.49, respectively; Table 2). One study[19] may have been a potential source of heterogeneity arising from the LP-ROP group. After excluding this study, the foveal SCP-VD in the LP-ROP group was 6.77 (95% CI: 4.77–8.76, I² = 29.7%, P = 0.23). There was no significant difference in the foveal SCP-VD between the Pre-T-ROP group and term-born controls (MD: 3.03, 95% CI: −0.14 to 6.21, P = 0.06; I² = 53.7%, P = 0.01).

Figure 3.

Forest plot for the analysis of the foveal superficial capillary plexus vessel density in the eyes of preterm- and term-born children

Furthermore, six studies analyzed the parafoveal SCP-VD in 587 eyes (276 eyes of preterm-born children and 256 eyes of term-born children). The MD of the parafoveal SCP-VD between the two groups was −1.71 (95% CI: −2.64 to −0.77, P = 0.00; Fig. 4), demonstrating that the parafoveal SCP-VD was remarkably lower in preterm-born children. Moderate heterogeneity was identified (I² = 60.4%, P = 0.00) in these studies. Moreover, according to subgroup analysis, the pooled MD of the parafoveal SCP-VD in the LP-ROP group was significantly lower, with the studies showing no heterogeneity (MD: −3.40, 95% CI: −4.86 to − 1.94, P = 0.00; I² = 0.0%, P = 1.0; Table 2). The pooled mean parafoveal SCP-VD was − 0.95 (95% CI: −1.76 to −0.14, P = 0.02; I² = 0.0%, P = 0.50; Table 2) in the SR-ROP group, and − 4.27 (95% CI: −7.19 to − 1.35, P = 0.00; I² = 0.0%, P=0.00; Table 2) in the anti-VEGF-ROP group, respectively. The difference between the Pre-T-ROP group and term-born children was not significant (MD: −0.99, 95% CI: −3.26 to 1.28, P = 0.39; I² = 79.9%, P = 0.01). One study may have been a potential source of heterogeneity in the Pre-T-ROP group. After excluding this study, the parafoveal SCP-VD in the Pre-T-ROP group was 0.06 (95% CI: −1.01 to 1.13, I² = 0.3%, P = 0.32).[12]

Figure 4.

Forest plot for the analysis of the parafoveal superficial capillary plexus vessel density in the eyes of preterm- and term-born children

Analysis of the foveal and parafoveal DCP-VD in preterm- and term-born children

A total of six studies that included 597 eyes (306 eyes of preterm-born children and 291 eyes of term-born children) analyzed the foveal DCP-VD by using the Optovue device. The MD in the foveal DCP-VD between preterm- and term-born children was 6.08 (95% CI: 4.99–7.16, P = 0.00; Fig. 5), indicating that the foveal DCP-VD was significantly higher in preterm-born children; no heterogeneity was identified among the studies (I² = 0.0%, P = 0.63; Fig. 5). The subgroup results also demonstrated that parafoveal DCP-VD was remarkably higher in the LP-, SR-, anti-VEGF-, and Pre-T-ROP groups than that in the control group (MD: 6.51, 95% CI: 3.81–9.21, P = 0.00; MD: 6.55, 95% CI: 4.51–8.58, P = 0.00; MD: 7.52, 95% CI: 5.28–9.76, P = 0.00; MD: 4.40, 95% CI: 2.49–6.31, P = 0.00, respectively; Table 2), with no heterogeneity identified among the studies (I² = 0.0%, P = 0.68; I² = 0.0%, P = 0.80; I² = 0.0%, P = 0.96; I² = 10.4%, P = 0.33, respectively; Table 2).

Figure 5.

Forest plot for the analysis of the foveal deep capillary plexus vessel density in the eyes of preterm- and term-born children

Four studies reported the parafoveal DCP-VD (examined using the Optovue device) in a total of 401 eyes (209 eyes of preterm-born children and 192 eyes of term-born children). The MD in the parafoveal DCP-VD between the two groups was 0.01 (95% CI: −0.75 to 0.76, P = 0.99; Fig. 6), indicating that parafoveal DCP-VD was higher in preterm-born children; however, this difference was not significant, and no heterogeneity was found in the outcomes of the studies (I² = 0.0%, P = 0.61; Fig. 6). The results of subgroup analysis showed a lower parafoveal DCP-VD in the LP-ROP and Pre-T-ROP groups and higher parafoveal DCP-VD in the SR-ROP group (MD: −0.45, 95% CI: −2.28 to 1.38, P = 0.63; MD: −0.13, 95% CI: −1.43 to 1.16, P = 0.84; MD: 0.27, 95% CI: −0.82 to 1.35, P = 0.63, respectively; Table 2); however, the difference did not attain statistical significance. No heterogeneity was identified among these studies (I² = 0.0%, P = 0.55; I² = 0.0%, P = 0.59; I² = 42.2%, P = 0.18, respectively; Table 2).

Figure 6.

Forest plot for the analysis of the parafoveal deep capillary plexus vessel density in the eyes of preterm- and term-born children

Discussion

Summary of the main results

To the best of our knowledge, the present study is the first meta-analysis of studies to assess and compare variations in the retinal microvasculature, as revealed by OCTA, in preterm- and term-born children. Twelve valid studies that incorporated 862 eyes were selected to assess the differences in the retinal microvasculature between preterm- and term-born children. The present meta-analysis pooled several parameters of the retinal microvasculature, including the FAZ, foveal and parafoveal SCP-VD, and DCP-VD. Our data showed that preterm-born children exhibited a smaller FAZ area, significantly increased foveal SCP-VD and DCP-VD, and decreased parafoveal SCP-VD compared to term-born children.

FAZ

Previous studies found a smaller or absent FAZ in the eyes of preterm-born children by using FA.[4,20] Kothari et al.[21] demonstrated that arm-mounted OCTA is a feasible tool for screening ROP in extremely low-birth-weight neonates, enabling the observation of the FAZ area, retinopathy staging, and treatment assessment. OCTA, as compared to conventional FA, offers superior visualization of ocular foveal morphology and vasculature. This advanced imaging technique yields high-quality images and precise data for evaluating the FAZ and VD. The study employed two OCTA devices, Zeiss and Optovue, for FAZ detection. It is worth noting that variations in results can occur due to the use of different devices, software, scanning speeds, and areas of focus. The MDs in the FAZ area between preterm- and term-born children, as detected by the Zeiss and Optovue devices, were found to be − 0.16 and − 0.17, respectively. Notably, meta-regression analysis failed to elucidate the source of heterogeneity, suggesting that factors such as VEGF levels and VEGF-lowering treatments may influence the FAZ area.

VEGF is critical for the growth of retinal vessels in the fovea, whereas angiostatin inhibits vascular ingrowth.[13] The VEGF–angiostatin balance is vital for maintaining the FAZ and determining the foveal and parafoveal VD. Thus, elevation in the VEGF level during the period of FAZ formation in preterm-born children may disrupt this balance, resulting in a smaller FAZ and greater foveal VD.[17] Because the ingrowth of the radial vasculature into the foveal area occurs only during the early stage of ocular development, newborns with a smaller GA may be affected more due to the elevation in VEGF.[13,17] In our meta-analysis, the mean GA was 29.14 ± 2.88 weeks in the preterm-born group and 39.32 ± 0.99 weeks in the term-born group, and the FAZ was remarkably smaller among the eyes of preterm-born children in contrast to term-born children. Among the subgroup analyses, the mean GA in the Pre-T-ROP group was higher (32.37 ± 3.12 weeks) than that in the LP-, SR-, and anti-VEGF-ROP groups (27.88 ± 2.18, 28.76 ± 2.44, and 27.40 ± 1.97 weeks, respectively), and the reduction in FAZ in the Pre-T-ROP group was lower than that in the other three groups.

In addition, the treatment modality (anti-VEGF or LP) may also have exerted different effects on FAZ formation. In the subgroup analysis, we observed that the degree of FAZ area reduction was more pronounced in the LP-ROP group than in the anti-VEGF-ROP group, which may be attributed to the various mechanisms of action of the VEGF-lowering treatments. Intravitreal anti-VEGF drugs are likely to promptly downregulate VEGF levels via intravitreal VEGF binding. In contrast, LP reduces intravitreal VEGF through ablation of the avascular retina, with the possibility of slower and less complete results during the immediate postoperative period. Therefore, a higher VEGF level in the LP-treated eyes is likely to be present during the vital foveal vascular development period, possibly resulting in a smaller FAZ and higher foveal VD. As such, a smaller FAZ was correlated with the reduction in angiogenesis caused by VEGF blockade by LP and anti-VEGF treatment. By using an arm-mounted OCTA device in neonates who had ROP treatment, including primary laser photocoagulation or anti-VEGF injection, Kothari N.[21] demonstrated that eyes treated with anti-VEGF had more mature FAZ and foveal development. This finding leads to the question of whether laser can cause arrest of foveal maturation. Other factors, such as different zones and stages of ROP, may also contribute to the size of the FAZ. In the LP-ROP group, most patients had stage 3, zone I/II disease, whereas most patients in the SR group had stage 1/2, zone I/II/III disease, which means that patients in the LP-ROP group were likely to have more severe disease and higher levels of VEGF, which ultimately leads to the ingrowth of the retinal vasculature and a smaller FAZ.

VD

In this meta-analysis, the SCP-VD and DCP-VD were significantly higher at the fovea, and the SCP-VD was lower at the parafovea in preterm children compared to term-born children. This study also found a slight and insignificant increase in the parafoveal DCP-VD in preterm-born children compared to term-born controls. The results indicated that the SCP and DCP at 1 and 3 mm may demonstrate variable features as the vascular insults originating from preterm birth, treatment for ROP, or both could impact vascular parameters. As previously discussed, a younger GA and higher VEGF levels in preterm-born children may be responsible for a smaller FAZ and higher foveal SCP-VD and DCP-VD.[22] The results of this study imply that lower VEGF levels during vascular development in the parafoveal region due to the suppression after anti-VEGF treatment or LP may be responsible for the lower parafoveal SCP-VD. The differences in the parafoveal DCP-VD between the preterm- and term-born children were not significant.

This study also assessed the foveal and parafoveal SCP-VD and DCP-VD in four distinct categories (LP, anti-VEGF, spontaneously regressed, and no ROP) of preterm-born children. Subgroup analyses revealed similar foveal and parafoveal SCP-VD and DCP-VD with remarkably higher foveal VD and lower parafoveal SCP-VD in the three ROP subgroups than in term-born children, whereas the differences in the foveal SCP-VD and parafoveal DCP-VD between pre-T-ROP children and term-born children were not significant. The study by Falavarjani et al.,[19] which reported a prominently higher foveal SCP-VD in the LP-ROP group compared to term-born controls, may be a potential source of heterogeneity with respect to the foveal SCP-VD. The higher foveal SCP-VD in the LP-ROP group may be associated with the substantially lower GA. Chen et al.[18,23] showed that patients with ROP undergoing laser treatment had a remarkably smaller FAZ, higher foveal VD, and lower parafoveal VD than those undergoing anti-VEGF treatment. However, only two studies with an anti-VEGF group assessed the VD, and more data are needed to determine whether laser or anti-VEGF treatment affected the development of vessel density around the fovea.

ROP Assessment: Limitations and future prospects

This study had some limitations. First, data on ROP severity were insufficient, and subgroup analysis was not conducted with respect to the ROP zone and stage. This is of particular importance because the severity of ROP may be related to the degree of retinal microvascular alteration. Further studies should be conducted to investigate the differences between the phases of ROP. Second, this review is restricted to the data of vessel density and FAZ area reported in children who could cooperate for a sit-down OCTA scan; however, data collected with handheld devices from premature infants in the NICU or operating room would be crucial for determining the best course of treatment alongside the impact on macular development. Third, the involved studies incorporated a cross-sectional and retrospective design, which is especially subject to selection bias and information bias. Future studies with a prospective longitudinal design are necessary to assess the retinal microstructure in patients with ROP to validate these results.

Conclusion

In summary, preterm-born children exhibited structural and vascular variations in the macular area. Preterm-born children had a smaller FAZ area, higher foveal SCP-VD and DCP-VD, and lower parafoveal SCP-VD compared to term-born children. No significant difference was identified in the parafoveal DCP-VD between the two groups. OCTA is a useful non-invasive modality for assessing the retinal structure and microvascular alterations in preterm-born children.

Abbreviations

BW, birthweight; CI, confidence interval; DCP-VD, deep capillary plexus vessel density; FA, fluorescein angiography; FAZ, foveal avascular zone; GA, gestational age; LP, laser photocoagulation; MD, mean difference; OCTA, optical coherence tomography angiography; ROP, retinopathy of prematurity; SCP-VD, superficial capillary plexus vessel density; SR, spontaneous regression; VEGF, vascular endothelial growth factor

Financial support and sponsorship

This work was supported by the Sichuan Science and Technology Program (2022YFS0611).

Conflicts of interest

There are no conflicts of interest.

Supplementary Material

1. List of Captions

Captions		Contents
Table 1		Evaluation of study quality using the Newcastle-Ottawa Scale
Figure 1 (1.1MB, tif) , 3 (1.1MB, tif)		Sensitivity analyses
Figure 2 (1.1MB, tif) , 4 (1.1MB, tif)		Funnel plots
Appendix		Search terms

2. Tables

Table 1.

Evaluation of study quality using the Newcastle-Ottawa Scale

Author, year		Newcastle-Ottawa Scale
		Selection		Comparability		Exposure
Vinekar A. 2021		★★★★		★★		★★
Mataftsi A. 2021		★★★★		★★		★★
Liang Z Q. 2021		★★★★		★★		★★
Vural, A. 2020		★★★★		★★		★
Tiryaki Demir S. 2020		★★★★		★★		★
Takagi M. 2019		★★★		★★		★★
Wu Z Q. 2019		★★★		★★		★★
Miki A. 2019		★★★		★★		★★
Chen Y C. 2019		★★★★		★★		★★
Balasubramanian S. 2019		★★★★		★★		★★
Leng Y X. 2018		★★★		★★		★★
Falavarjani K G. 2017		★★★★		★★		★★

3. Figure legends

Figure 1

Sensitivity analysis for FAZ aera in preterm- and term-born children

Figure 2

Funnel plot for FAZ aera in preterm- and term-born children

Figure 3

Sensitivity analysis for fovea SCP-VD in preterm- and term-born children

Figure 4

Funnel plot for fovea SCP-VD in preterm- and term-born children

4. Search terms

Appendix search terms for PubMed

((("Premature Birth"[Mesh]) OR ("Retinopathy of Prematurity"[Mesh])) OR ((prematurity retinopathy[Title/Abstract]) OR (prematurity retinopathies[Title/Abstract]))) AND (((((((optical coherence tomography based microangiography[Title/Abstract]) OR (optical coherence tomography based angiography[Title/Abstract])) OR (oct angiography[Title/Abstract])) OR (optical coherence tomographic angiography[Title/Abstract])) OR (optical coherence tomography angiography[Title/Abstract])) OR (Angio-OCT[Title/Abstract])) OR (OCTA[Title/Abstract]))