Myopia has become a serious public health concern due to its significantly increasing prevalence, especially in East Asia (1–4). As a leading cause of vision loss worldwide (5), high myopia (HM) may result in retinal disorders, including lacquer crack formation, Forster-Fuchs' spots, chorioretinal atrophy, choroidal neovascularization, foveoschisis, and posterior staphyloma (6, 7). These complications are highly associated with morphological changes of retinal vessels (8, 9). Therefore, the change of retinal microvasculature in HM has been an important issue for several decades, which may provide a critical clue for understanding the pathophysiology of HM-associated diseases. It was previously reported that in the patients with HM, reduced retinal vessel density (10) and blood flow (11) were evident in the large retinal vessels visible on fundus photographs. Moreover, decreased choroidal blood flow not only has been found to be associated with increase of axial length (AL), but also may be a possible indication for progressive myopia (12). Recently, in retinal microvasculature, studies pointed out that students with HM experienced decreased deep perifoveal vessel density and radial peripapillary capillary (13, 14). Hence, investigating the longitudinal changes in macular microvasculature and their interaction in young adults with high myopia prior to retinal damage may reveal the underlying pathophysiology of the disorder and assist to implement an effective treatment or prevention.

Optical coherence tomography angiography (OCTA), a newly developed imaging modality, has assisted scholars to quantitatively and qualitatively measure retinal and choroidal microvasculature non-invasively (15). The present study aimed to assess changes of macular vessel density using OCTA in young adults with HM, and investigate the associations between the changes of vessel density and ocular parameters through a longitudinal study.

Seven hundred sixty participants were recruited at baseline visit, and 447 (58.82%) participants from the same cohort were followed up 21 months later. The loss to follow-up was chiefly due to school graduation. Moreover, 138 participants were further excluded at follow-up because of lacking OCTA examination (n = 128), poor-quality OCTA images (n = 5), and refractive surgery history (n = 5) and eventually, a total of 309 participants were involved in the final analysis. In the study, 40 participants had EM (12.95%), 194 had MIM or MOM (62.78%), and 75 had HM (24.27%). Demographic and ocular characteristics of participants at baseline are expressed in Table 1. There were no significant differences in age, BMI, SBP, DBP, heart rate, IOP, MOPP, central corneal thickness, and lens thickness among the three groups. However, HM group included fewer female participants (p < 0.05), presented significantly lower SE, worse BCVA, deeper ACD, and longer AL than EM group (all p < 0.05).

Table 2 and Figure 2 demonstrates the baseline and the longitudinal changes of VD and ocular parameters over follow-up period. There was no significant difference in the baseline VD, RT and GCC among the three groups (all p > 0.05). The macular VD significantly decreased while AL increased in the follow-up in all the three groups (all p < 0.05). Of note, the HM group showed a larger reduction in VD (−1.96 ± 2.57/mm) compared with the EM group (−0.61 ± 0.97/mm) (p = 0.0307). There were reductions in both RT and GCC thickness after 2-year follow-up in HM and MIM/MOM groups (both p < 0.01), which was not seen in emmetropic eyes (both p > 0.05).

The changes of macular VD in all sectors of ETDRS grid were presented in Supplementary Table 1 and Figure 3. The sectoral VD significantly decreased through follow-up in all the three groups (all p < 0.05), except for that in the outer nasal sector in EM group (p = 0.2521). In each group, the nasal sector minimally changed, whereas the temporal sector maximally changed in both inner ring and outer ring. In addition, the changes in the inner ring were more significant than those in the outer ring for the majority of sectors. Compared with EM group, HM group had notably higher change rate in the II sector (−3.15 ± 9.62% vs. −14.88 ± 20.11%, p = 0.0073), IT sector (−4.88 ± 8.43% vs. −17.22 ± 22.88%, p = 0.0298), and OT sector (−7.15 ± 8.40% vs. −16.20 ± 20.80%, p = 0.0308).

The results of the univariate regression analysis adjusted for age and gender are presented in Supplementary Tables 2–5. For the average macular VD, no variable was found to be correlated with the changes of VD (all p > 0.05). The associations of macular VD changes with other ocular parameters were further explored in selected sectors where great macular VD reduction were observed in HM group. The changes of VD significantly correlated with SE, baseline AL and the changes of AL in the II, IT, and OT sectors. BCVA and the changes of GCC thickness were negatively associated with the changes of VD in the II and IT sectors. Additionally, the changes of VD were positively correlated with baseline RT and the changes of RT in the II sector, whereas negatively correlated with ACD in the OT sector. Variables with a significance p-value <0.05 were included in the stepwise multivariate models except for SE that had a stronger correlation with AL. The results of multivariable regression analysis (Table 3 and Figure 4) revealed the baseline AL was independently associated with the changes of VD in II, IT, and OT sectors (all p < 0.05). According to the models, each 1 mm increase in baseline AL was associated with a 0.57/mm decrease in inner-inferior macular VD, a 0.73/mm decrease in inner-temporal macular VD, and a 0.44/mm decrease in outer-temporal macular VD, separately.

This prospective study, which enrolled three hundred and nine university students over 21 months, was the first longitudinal study evaluating macular microvasculature changes in young adults observed by OCTA. A significant decrease in macular VD over time was documented. In high myopes, macular VD presented larger reduction globally and in the II, IT and OT sectors. A longer baseline AL was associated with greater loss of VD in these sectors.

Earlier studies reported that age was determinant of macular VD measured by OCTA (20–22). However, these studies were generally conducted in healthy eyes. In our longitudinal study, macular microvasculature was significantly decreased in both emmetropes and myopes over a 21-month period. Jo et al. (20) reported that the majority of the sectoral peripapillary and macular VDs were significantly reduced with increase of age in healthy eyes but the VD of papillomacular bundle area was not age-dependent. The present study also demonstrated that the macular VD in the ON sector did not decrease in emmetropic students, while this sectoral VD reduced in myopic students, which indicating the effect of myopia on macular perfusion reflected in the area of papillomaular bundle area. It was revealed that the ischemic injury of the papillomacular bundle could predict poor vision in both central retinal artery occlusion and branch retinal artery occlusion (23, 24). Thus, the effects of VD loss in the papillomacular bundle area on visual impairment in myopes and its underlying mechanism still require further investigations.

Several studies used OCTA to investigate the changes of retinal microvascular in HM, and they have reported controversial findings (13, 25–28). Guo et al. (28) pointed out that HM causes a lower superficial peripapillary microvascular density, but no significant difference was found in parafoveal microvascular density among all groups. Wang et al. (26) also found a decreased vessel density in the peripapillary area, rather than in the parafoveal area of high myopic eyes. On the contrary, our previous study and a number of cross-sectional studies (13, 25, 27) have yielded a reduced parafoveal microvascular density in HM. It may lead to controversial results because the vessel density was quantified in different ways. In Guo's and Wang's studies, the vessel density was defined as the proportion of the total area occupied by vessels, while the vessel density was defined as the total length of perfused vasculature per unit area in this study. Moreover, the previous studies were all cross-sectional studies with different simple sizes. The present research prospectively evaluated the changes of macular VD and showed a significantly greater loss of VD in HM group than in EM group. However, no significant difference of global and sectoral VD changes was noted between MIM/MOM group and HM group, which was in line with Yang et al.'s findings (29). When it comes to sectoral changes of macular VD, there was no unified understanding. In the comparison with EM group, HM group presented larger longitudinal changes of macular VD in the II, IT, and OT sectors in this study, indicating these sectors might be critical focus areas in clinical practice.

We previously compared the peripapillary and parafoveal vessel density with AL and retina thickness, and found a significant negative correlation between AL and vessel density (13). Li et al. (27) pointed out that microvascular density was negatively correlated with AL in myopia in both superficial and deep vascular plexuses. Ucak et al. (30) reported that the macular vessel density reduced with increase of AL and decrease of GCC thicknesses in patients with HM. The current study uncovered that the longitudinal changes of macular VD were only correlated with baseline AL in the II, IT, and OT sectors, suggesting the macular vessel loss mainly results from the elongation of eyeballs. It is suggested that with the increase of AL in myopia, the eyeball stretches, causing mechanically expanding and thinning of retina, resulting in a narrowing of the vessel diameter, which leads to the decrease of vessel density (27). Meanwhile, the macular thickness showed different rates of change in different sectors according to age and refractive status (31, 32). Since the oxygen demands of those sectors might reduce resulting from the thinning of retina, the vessels supplying blood to those regions was likely to decrease accordingly. As well as the mechanical stretch, there might be some changes in myopia-related signaling pathways, resulting in subsequent changes in vessel density. However, it is not clear which of structural changes or biochemical changes came first. Despite this, the mechanisms of sectoral changes in macular vessel density still require further study.

The capillary networks in the human retina mainly distributed in the inner five layers of the retina, of which the superficial capillary network is mainly distributed in the nerve fiber layer and the ganglion cell layer, while the deep capillaries are mainly distributed in the inner plexiform layer and the inner nuclear layer (33). There has been increasing evidence that HM has longer AL, lower peripapillary VD, and thinner peripapillary retinal nerve fiber layer thickness (26, 28, 34, 35). In addition, AL was also noted to be negatively correlated with thickness of the outer nuclear layer and photoreceptor outer segment layer (36). But the RT and GCC thickness were not related with the loss of macular VD in our multivariate regression analysis. The underlying biochemical mechanism might give reason to changes of macular VD as well as the mechanical stretch. Dopamine, secreted by dopaminergic amacrine cell (DAC) in the inner layer of the retina, was indicated to be a key molecule in the retinal signaling pathway during myopia development, working with Melatonin and retinal ganglion cells (RGCs) (37–39). The structural changes in myopia may affect the DAC, RGCs, and outer photoreceptor cells, leading to subsequent changes in DA and melatonin synthesis and release, thereby influencing further development of myopia.

This study has several limitations. Firstly, the low follow-up rate made it not representative. Secondly, the examinations were only performed at two visits, so that we could not observe the dynamic changes during this period. Thirdly, the small sample size, especially for EM and HM groups, may significantly impact the reliability of our findings. Fourthly, OCTA deep slab images were not included in the analysis because several projection artifacts prevented qualified images.

In conclusion, the macular VD significantly decreased over time in all the sectors in young myopes, while the macular VD in the papillomacular bundle area was not age-dependent in emmetropes. A larger reduction of macular VD was observed in HM group compared with healthy eyes and correlated with longer baseline AL in the II, IT and OT sectors. Although the underlying mechanisms of physiological and pathological macular VD changes remain unknown, the reduction of macular VD might be considered as a preclinical characteristic in patients with HM. Therefore, further studies with longer follow-up periods are needed to investigate correlations between structure and function.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethical committee of Shanghai General Hospital, Shanghai Jiao Tong University, Shanghai, China. The patients/participants provided their written informed consent to participate in this study.

YF, JZ, JH, and XX: study concept and design. YS, QC, GH, and YY: data collection and management. YS and LY: analysis and interpretation of data. YS: writing the manuscript. JZ, JH, ZZ, and HZ: critical revision of the manuscript. XX: supervision. All authors read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.