In this study, a total of 85 eyes of 44 patients with ESKD were enrolled, the mean HD history for these patients was 59.95 months. Based on previous medical records, the diagnosis of DME in 32 patients (57 eyes, 67.06%) prior to HD was confirmed. Of the 44 patients (85 eyes) who received long-term HD treatment for ESKD, only 11 (17 eyes, 20%) still had DME, and 7 (8.24%) of these 17 eyes had ERM. Eighty-five (100%) eyes had abnormal circulation in the retina, such as capillary nonperfusion, microaneurysm and IRMA, which indicated that the condition of DR in these patients was still serious, although 12 of them had received intravitreal anti-VEGF drug injection, and 2 had received intravitreal dexamethasone implant injection. However, the patients enrolled in this study did not receive any form of anti-VEGF or glucocorticoid treatment within the past 6 months, and the lower incidence of DME in this study may be related to long-term HD. Hwang et al. [16] evaluated the effect of HD on CRT and SFCT in a study including 15 DME patients with DN and reported that the average CRT decreased from 317.92 ± 91.41 μm to 287.77 ± 57.55 μm after the first HD and that the SFCT decreased from 313.31 ± 85.89 μm to 288.81 ± 92.02 μm after the first HD session.

The onset and progression of DME are not only related to ocular factors such as VEGF overexpression caused by retinal ischemia, retinal capillary hyperpermeability, and decreased retinal pigment epithelial (RPE) pump function, but also to the hypoproteinaemia caused by DN [17,18,19]. Since both DR and DN are chronic vascular diseases, the severity of DR correlates with the condition of DN, and DR can be used as a marker to predict the severity of DN [20, 21]. Intensive glycemic control is not only essential for preventing the occurrence and progression of DR and DME but also important for reducing the risk of DN progression [21].

For patients with ESKD, HD remains the most common treatment modality for improving quality of life and prolonging survival, even if severe DR remains [6]. In this study, the 44 patients no longer received ophthalmic treatment for DR. In department of HD, patients get HD three times a week in sessions of 3 to 5 h each. Therefore, year-round HD is the major part of their lives.

Epidemiological research shown that the incidence of DME in diabetes patients is approximately 5.5% [22], which reaches 21.1% in DR patients [23], as high as 66.2% in patients with PDR [24]. Since the condition of DN is positively correlated with the severity of DME and DR [25], theoretically, the incidence of DME is greater in patients with DN and PDR.

In this study, we found that the incidence of DME in patients undergoing HD was only 20%. After single HD session, the patients’ average plasma osmolality decreased by 13.17 ± 16.60 mOsm/kg (P < 0.001), and the average body weight decreased by 2.59 ± 0.80 kg (P < 0.001). Both the volume and composition of extracellular fluid decreased dramatically, leading to a decrease in body weight and plasma osmolality after single HD session [26], which may be related to the low incidence of DME in ESKD patients undergoing HD.

Among the 44 patients (85 eyes) enrolled in this study, 17 (20%) had DME, and all had CME. Among them, ERM was present in 7 eyes with DME (8.24%). The lack of regression of these CMEs after long-term HD may be related to the presence of advanced glycation end products (AGEs). Methylglyoxal (MGO) derived from AGEs may induce alterations in the characteristics of cystoid lesion components in diabetic CME due to posttranscriptional modification, leading to resistance of fibrin in CME to plasmin [27, 28].

ERM is a pathology caused by fibrocellular proliferation on the ILM followed by intravitreal injections for DME treatment or secondary to a wide variety of diseases, such as DN and cataract surgery [29, 30]. ERM formation causes anteroposterior and tangential forces in the retina, resulting in further damage to the blood retina barrier (BRB), the release of inflammatory factors and the deterioration of DME [29]. Clinical studies have revealed that vitrectomy has a beneficial effect on the management of DME with ERM [31].

Fang et al. [32] conducted a two-year study on 108 patients with DME who received intravitreal injections of anti-VEGF drugs to estimate changes in the glomerular filtration rate (eGFR), confirming that although DME can improve with anti-VEGF intervention, the eFGR continues to deteriorate.

Therefore, ophthalmologists should pay close attention to the renal function of patients while using anti-VEGF drugs or corticosteroid intravitreal injections to treat DME in clinical practice. If renal dysfunction meets the HD treatment standard, patients should be promptly referred to nephrologists. For DME patients who are unresponsive or have a poor response to anti-VEGF or glucocorticoid therapy, it is necessary to evaluate renal function as early as possible. Improving renal function may be more conducive to improving DME in patients with ESKD.

To clarify the impact of single HD session on retina and choroid, we used SS-OCTA to analyze the CRT, CRV, SFCT and SFCV in patients before and after single HD session. We found that there was no significant change in the average CRT or CRV before and after single HD session, while the average SFCT decreased significantly from 243.11 ± 77.15 μm to 219.20 ± 72.84 μm, with an average reduction of 23.91 ± 14.15 μm (P < 0.001), and the average SFCV decreased significantly from 0.15 ± 0.10 mm3 to 0.13 ± 0.90 mm3, with an average reduction of 0.28 ± 0.23 mm3 (P < 0.001), after single HD session. Before the next single HD session, the average SFCT recovered to 245.41 ± 76.23 μm, and the SFCV recovered to 0.16 ± 0.10 mm3. There was no statistically significant difference in the CRT (P = 0.206), CRV (P = 0.154), SFCT (P = 0.108) or SFCV (P = 0.174) before the two single HD sessions.

This finding may be related to the fact that the choroid is mainly composed of blood vessels, and HD leads to the elimination of water and small molecule metabolic waste, such as BUN and creatinine, in the patient’s body. The blood volume in the systemic circulation, including the choroidal vascular system, decreases in a short period of time [33].

In this study, we also observed that within 30 min after single HD session, the average weight of patients decreased by 2.59 ± 0.80 kg (P < 0.001), and the average IOP decreased by 0.07 ± 0.63 mmHg (P = 0.282). However, the renal function of patients with ESKD tends to fail, and the water and metabolic waste discharged previously can gradually accumulate again after HD. Therefore, the SFCT and SFCV of patients in this study increased again before the next single HD session.

However, the retinal thickness did not change due to single HD session, which may be related to the fact that the patients in this study had already undergone long-term HD. Subsequently, long-term HD led to the stabilization of the macular anatomical structure of the retina. Moreover, retinal blood vessels run between the nerve fiber layer and the inner nuclear layer, while the layers outside the inner nuclear layer in the retina are avascular [34]. Even if single HD session leads to a decrease in blood volume throughout the body, including in the retinal vascular plexus, its impact on retinal thickness is difficult to detect.

Although all the patients in this study have undergone long-term HD, whose renal function is relatively stable, and DME has significantly improved, it cannot be denied that all patients still had severe DR, including the presence of nonperfused capillaries in the retina, foveal avascular zone (FAZ) enlargement, and IRMA in 85 (100%) eyes. RNV was still observed in 42 (49.41%) eyes. The average FAZ area was 0.51 ± 0.11mm2 in our study. Shahlaee et al. [35] found that the FAZ area of the superficial layer in normal adults was 0.27 ± 0.101 mm2, as measured by OCTA. FAZ surrounded by capillary network, a case control study confirmed that patients with diabetes had a larger FAZ, and patients with more severely damaged retinas had a much larger FAZ [36].

Fundus fluorescence angiography (FFA) could not be performed in ESKD patients in this study because sodium fluorescein may lead to contrast-induced nephropathy (CIN) and appears to be a possible risk factor for the progression of ESRD [37]. Stino et al. [38] demonstrated that wide-field OCTA imaging is highly reliable in the detection of PDR, and has the potential to replace FFA as a single diagnostic tool.

The basis for distinguishing IRMA and RNV was to observe the location of blood flow signals on cross-sectional OCTA images. The blood flow signal of the IRMA is located below the ILM, and if the blood flow signal protrudes from the ILM and enters the vitreous cavity, it is considered a RNV [39, 40].

These RNVs still pose a risk of causing vitreous hemorrhage and traction retinal detachment [41]. However, the impact of ESKD on quality of life and survival time may make these patients more inclined to receive frequent HD rather than ophthalmic treatment. Therefore, ESKD patients with DR still need to be examined and/or treated by an ophthalmologist in a timely manner in order to permanently preserve their limited vision.

Other studies have also confirmed that long-term hemodialysis can effectively improve CRT and visual acuity in patients with DME [5, 6]. However, our study focused on the retinal status of DR patients on long-term HD (mean 59.95 months), and analyzed the retinal and choroidal thickness before and after single HD session. Our study may provide an available reference for the follow-up and treatment of DR patients undergoing HD.

The main limitation of this study was its small sample size. Further large-scale prospective studies are needed to further clarify the impact of long-term HD on the retina, choroid and optic nerve.