Response of OCT Biomarkers Based on New Diabetic Macular Edema Classi cation TCED-HFV to Anti-VEGF Therapy and Its Prognostic Role in Vision Bene ts

Meng Yan (  yankeyanmeng@126.com ) The A liated Hospital of Qingdao University https://orcid.org/0000-0002-0890-6515 Wenying Wang The A liated Hospital of Qingdao University Zhaoxia Zheng The A liated Hospital of Qingdao University Lu Li The A liated Hospital of Qingdao University Duo Zhang The A liated Hospital of Qingdao University Lina Zhang The A liated Hospital of Qingdao University https://orcid.org/0000-0002-3327-1160


Introduction
Diabetic macular edema (DME) is the most common cause of vision impairment in patients with diabetic retinopathy (DR) and affects approximately 7% of all patients with diabetes [1,2].Approximately 50% of patients with DME lose two or more lines of visual acuity within 2 years if left untreated [3].
At present, uorescein angiography (FFA) and optical coherence tomography (OCT) represent the techniques of choice for evaluating diabetic maculopathy, providing several quantitative and qualitative data points concerning DME [4]. Despite the fact that Early Treatment Diabetic Retinopathy Study (ETDRS) has traditionally been regarded the gold standard for classi cation and laser therapy planning in DR for many years, OCT is the most used modality today for anti-VEGF therapy management, evaluation and monitoring of individual treatment responses. New classi cation systems for DME were proposed in the past years, attempting to classify DME according to its extent (focal vs diffuse) [5], location (center-vs non-center-involving)[6], or nature (vasogenic vs non-vasogenic) [7,8]. However, a morphological grading classi cation of DME, including all its relevant features visible on OCT is missing.
Patients with other retinal diseases, vitreous hemorrhage, previous vitrectomy, ocular in ammation, macular ischemia on FFA, myopia ≥ 6D, media opacities, uncontrolled glaucoma, ocular trauma surgery within the last 6 months and those whose OCT images were of poor quality were excluded from the study. The study was in accordance with the Tenets of Helsinki Declaration and was approved by the Institutional Review Board of the participating hospital. Written informed consents were obtained by all patients to use their data.
The following data were recorded: baseline demographics; previous treatments; ocular examination, FA, and OCT ndings of the patients; the numbers of intravitreal injections; additional laser treatments; need for cataract surgery; and development of vitreous hemorrhage (VH). A venous blood sample was drawn from an antecubital vein in the morning after at least 8 hours of fasting. Determination of albumin, creatinine, blood ureanitrogen, total cholesteral, glycosylated albumin and concentrations of HbA1c were acquired. All measurements were performed at the department of diagnostic testing, the A liated Hospital of Qingdao University using commercially. Patients underwent a comprehensive ophthalmological examination in baseline and follow-up visits, which included slit lamp biomicroscopy, funduscopy, tonometry (intraocular pressure-IOP), FFA, and the spectral domain Cirrus HD-OCT 5000 (Carl Zeiss Meditec, Dublin, CA, USA). Retinal analysis of the patients before the operation and at the earliest 1 year after the operation was performed using 6 × 6 mm images taken with the macular cube 512 × 128 protocol on the Cirrus HD-OCT device in our clinic. The grade was judged by scanning through a poorly formed fovea, or by the best visibility of the retinal structure in the case of middle layer opaci cation. A minimum signal strength ≥7 was required for the study.
The OCT images which were taken at baseline and follow-up visits were analyzed to determine the following: morphological characterization of DME (1) CMT calculated automatically by the instrument; (2) the size of IRC; (3) the visibility of ELM and/or EZ at the fovea; (4) the presence of DRIL; (5) the presence of SF; (6) the presence and the number of HF; and (7) the vitreoretinal relationship. SF was categorized as temporary if resolution was observed after intravitreal anti-VEGF injections; if SF existed during the whole treatment, usually de ned as persistent SF. A manual count of the total number of HF, de ned as small (<30 mm), punctiform discrete white lesions with re ectivity similar to the nerve ber layer, absence of back-shadowing, and location in both inner and outer retina, was performed between the two vertical lines and calculated in the area of 3000 µm centered on the fovea [9].
Statistical analyses were performed using the SPSS 26.0 for Windows package. The numerical data are expressed as mean and standard deviation and the categorical variables as absolute frequency and percentage. For the analysis of best central visual acuity (BCVA), CMT, and the amount of HF, examined variables did not present a normal distribution as veri ed by the Shapiro-Wilk test and therefore, the Wilcoxon test has been used. We performed statistical comparisons between groups using the Chi Square test for categorical variables, Mann-Whitney test, ANOVA analysis, Kruskal-Wallis H test or paired Student's t test for numerical variables. The comparison of measurements between baseline and follow-up visits were performed with Cochran's tests. A P value lower than 0.05 was considered statistically signi cant.

Results
A total of 166 eyes (122 patients) were included in the study and were followed for 1 year. Descriptive characteristics of the enrolled patients were shown in Table 1. Overall, the average age was 60.72±9.97 years and 48(39.34%) of the patients were male. The mean durations of diabetes was 12.61±7.39 years. All patients received a mean of 6.24±2.52 injections in a year. Before accepting anti-VEGF treatment, 94(56.63%) and 38(22.89%) eyes had undergone PRP and cataract surgery, respectively. SF was detected in 86(51.81%) eyes, DRIL in 30(18.07%) eyes. More than 40% of the eyes had vitreomacular interface abnormality (VMIA), and many are ungraded, which is likely to  Severe DME (%) 12(7. No signi cant differences were detected between BCVA in DME eyes with or without PRP and between phakic or pseudophakic eyes, as well as in men or women and right or left eyes (P values were 0.115, 0.747, 0.624 and 0.782, respectively). The number of HF was 57.00±7.78 in the early DME, 79.00±26.33 in the advanced DME and 119.00±38.55 in the severe stage (P=0.026). The average BCVA has been demonstrated a signi cant improvement at 3M and 6M, while the average CMT and HF decreased noticeably. An increased number of HF and thickness of CMT were generally accompanied by the recurrence of DME (Table 2).  respectively, P=0.269). A persistence SF was a negative factor, meaning that the mean visual gain was less compared with SF-absent group (0.05±0.11 log MAR versus 0.26±0.23 log MAR, respectively, P=0.045).
In addition, a meaningless conclusion in the relationship between SF and serological indexes was drawn in our cohort, which includes albumin, creatinine, blood ureanitrogen, total cholesteral, glycosylated albumin and concentrations of HbA1c (Figure 1).

Discussion/conclusion
As the wide application of OCT in the diagnosis and monitoring of retinal diseases, we have discovered new pathologies related to retinal diseases, and researchers have begun to investigate the relationship between these pathologies and visual outcomes. Similar progress has been made in patients with DME because OCT has been able to identify various pathologies in these patients, such as CMT, HF, SF, EZ/ELM irregularity and DRIL, and these ndings have begun to clarify the pathogenesis and prognosis of the disease. A large number of studies have been conducted on DME patients receiving anti-VEGF injections, and it has been con rmed that the microstructural changes seen at the baseline by OCT scans can predict the treatment response [14,10,15]. The TCED-HFV classi cation, as a non-invasive diagnostic method, had shown its potential for early diagnosis of DME and could quantitatively response the therapeutic effect to anti-VEGF treatment.
Our ndings suggested that the anti-VEGF treatment strategy signi cantly improved BCVA and reduced CMT. These observation indexes were associated with better results and higher patient's satisfactions rate after 3-and 6-months injections when compared with 1 year. Lam and Lai [16] reported that changes of BCVA and central foveal thickness in 48 eyes with DME treated with IVB during 6-month follow-up. The results showed signi cant improvements between baseline and 6-month mean BCVAs, as well as the reductions between baseline and 6month mean central foveal thickness. They demonstrated the best treatment effect at 3 months, followed by 6 months, which had better results compared with the baseline. Similar conclusion can be drawn by Kumar and Sinha [17]. IVB resulted in a signi cant decrease in macular thickness and a signi cant improvement in visual acuity after 3 months, but this effect was weakened at the end of 6 months, although it was still statistically signi cant. The current ndings compare favorably with those reported, and con rm their ndings through longer follow-up and more patients.
In our study, a signi cantly decreased HF could be found under adequate anti-VEGF treatment. This result is consistent with the results observed in previous studies that hyperre ective foci or spots decreased after anti-VEGF treatment[18, 14,10]. The exact formation mechanism of HF remains still unclear despite several theories have attempt to explain the special marker. Bolz et al. [19] reported that HF may represent tiny intraretinal protein and/or lipid deposits after inner blood-retinal destruction, playing as precursors of hard exudates.
Additionally, some researchers suggested that HF might correspond to microglial cells, debris of photoreceptors, and RPE hyperplasia [20,21].
All these studies stated that HF might be associated with retinal in ammatory response. With the aggravation of retinal in ammation, microglia cells are activated, increasing the number, changing morphology and gathering together, which act as HF on OCT [22][23][24]. Anti-VEGF therapy may inhibit the activation of VEGFR1-dependent microglia and restore structural damage [25]. The results suggested that a decreased number of HF may be connected with good response to anti-VEGF treatment for patients with DME. Additionally, the number of HF was associated with different stages of DME and especially high in severe DME. Thus, there may be a strong link between HF and exudative retinal disease and HF could be used as a predictive marker of visual function.
Previous studies have shown that the prevalence of SF in patients with DME ranges from 11.4-52.4% [26,27]. Our nding shows that SF existed approximately 50.00% of the patients at baseline, which was consistent with the results reported in the previous literature. The high incidence rate of SF occurred in early and advanced DME, and the lowest in severe stage. With the advancement of OCT technology, higher quality images and neglected pathology can be easily and clearly obtained, which is also the reason for the increasing prevalence of SF. A great deal of literatures has reported positive, negative or no effects of SF on vision[28-31]. Zur et al. [30] suggested that better BCVA was related to the existence of baseline SF. On the contrary, Sophie et al. [28] proposed the presence of SF caused a poor visual acuity which improved signi cantly through aggressive and sustained suppression of VEGF. In addition, a post-hoc analysis of evaluating the effect of SF on treatment outcomes in patients with DME in the VIVID and VISTA studies indicated baseline SF status play an meaningless impact on treatment outcomes with IAI [31]. Similar ndings were reported from Vujosevic et al. [29]. It is suggested that the effect of SF on the visual acuity is controversial for patients with DME. The degree of vision growth was greater in the absent SF group than that in the temporary SF and persistent SF groups, although baseline BCVA was mildly poorer compared with the other two groups. No signi cantly difference in BCVA was found in patients with persistent SF during treatment and follow-up, which means that the persistence SF may be a negative factor for visual outcome. Visual acuity after anti-VEGF therapy was greatly improved in both temporary SF and absent SF groups. In addition, SF reappeared in only 22.22% of patients in the present SF group, indicating that anti-VEGF therapy played an important role in the regression of SF.
Meanwhile, researchers explored the in uence of systemic condition on SF, one of which is that poor renal function. De Benedetto et al. [32] and Melissa et al. [33] reported the cases of the hypoalbuminemia leading to subretinal uid. Tsai et al. [34] reported that the presence of subretinal uid before treatment was associated with lower EGFR stages and lower albumin levels. The reduction of serum albumin may decrease the intravascular osmotic pressure and increase the hydrostatic pressure, resulting in the retention of uid in the subretinal space. All these ndings suggested serum albumin is a sensitive marker of SF in chronic kidney disease. Conversely, we came to a meaningless conclusion in the relationship between SF and kidney markers. The discrepancy might be partly explained by the different population choices, different follow-up periods and different treatments. Interestingly, the persistent SF was observed in 14 patients with chronic kidney disease.
As a result, larger sample sizes, more detailed imaging and longer follow-up periods are needed to con rm the clinical impact of SF.
The pathological mechanism underlying the formation of SF is not fully known, but one of the possible mechanisms is the EZ/ELM condition [29]. The apoptosis and proliferation of vascular endothelial cells may lead to vasodilation and increased vascular permeability [35].
Vascular endothelial growth factor may also in uence the rupture and vasodilation of blood-retinal barrier (BRB) [36]. In addition, BRB destruction will also increase the osmotic pressure of intraocular uid, which forces proteins moving in all directions [37]. However, it is di cult for proteins and cells to pass through intact EZ/ELM [37,29]. Oncotic pressure will be created by this accumulation of protein, bind water and thus create a condition of retinal edema. Accumulated proteins, lipids and uid may more easily enter the subretinal space as the EZ/ELM is interrupted [29], resulting in the accumulation of subretinal uid. In our study, SF appeared more likely in eyes with disrupted or absent EZ/ELM. In 8 eyes of the disrupted EZ/ELM group, EZ interruption and SF appeared at the same time. With the recovery of EZ, SF disappeared, which also con rmed our point of view.
Our data showed that patients with severe DME had signi cantly worse vision than those with early and advanced DME. The EZ/ELM condition was the most decisive indictor to distinguish advanced DME and severe DME. The status of EZ/ELM was referred to photoreceptor integrity and is considered to predict the visual prognosis. A study by uji et al. [21] reported the intact EZ/ELM was associated to better visual outcomes. Similar nding was found by Muftuoglu et al.[38]. The present study showed poorer visual acuity outcomes in the eyes with absent EZ/ELM, while the best visual gain was obtained in eyes with intact EZ. Therefore, anti-VEGF therapy could achieve better therapeutic effect in early DME stage, as EZ/ELM was complete.
The incidence of VMIA in our study, including incomplete posterior vitreous detachment (IVD), PVD, VMT and ERM, stays the similar range to the previously reportion. The prevalence of ERM has been shown to vary from 6.6-43% [39][40][41][42]and is 32.53% in our study. The relationship between anti-VEGF therapy and ERM formation was not fully known. Some researchers indicated intravitreal injection of anti-VEGF agent might suppress ERM formation [43]. On the contrary, Anti-VEGF injections have been reported to promote tissue brosis and ERM formation [44][45][46]. The present date con rmed this, where the increasing prevalence of ERM is correlated with frequent intravitreal anti-VEGF injections. Das et al. [9] suggested the presence of ERM has a negative in uence on the anatomical structure of corresponding region and was associated with poorer nal visual acuity. This means that the effect of anti-VEGF treatment was decreased after ERM formation and explains why the best therapeutic effect of anti-VEGF injections came in 3-and 6-months. This was consistent with previous reports that ERM might serve as a physical barrier and reduce drug permeability [47].
Another indicator emphasized in patients with DME was DRIL, which was de ned as the unidenti able boundaries separating the inner layers of the retina (the ganglion cell layer (GCL)-inner plexiform layer (IPL) complex, inner nuclear layer (INL), and outer plexiform layer (OPL), predicted worse visual outcomes in the DME patients [27,48,49]. Previous studies hypothesized that the formation of DRIL was connected with the disruption and loss of Müller cells, bipolar cells, horizontal cells, and amacrine cells within the retina. DRIL occurred when the bipolar axon damaged as the elasticity of cells disappeared because of severe edema [50][51][52]. Additionally, PDR may form a brous vascular membrane and produce traction on the macula. Therefore, a variety of mechanisms related to vascular abnormalities and mechanical stress may be involved in the formation of DRIL [49]. The present results, showing the statistically difference between the eyes with and without DRIL, supported the ndings of previous literatures, and DRIL may act as a negative indictor forecasting treatment outcome in treatment-naïve diabetic eyes.
Our retrospective study was limited by the relatively small number of graded patients and its retrospective character. The impact of Periodization was usually ignored while the great majority of present studies focused on the relationship between OCT ndings and nal visual acuity in patients with DME. Consequently, the controversial impact on therapeutic effect of indictors, such as HF and SF were reported repeatedly by researchers. Further prospective studies with more patients and an appropriate control group are needed to determine the longterm effect of OCT biomarkers on visual prognosis of patients in the different stages based on the grading protocol TCED-HFV. Further studies are warranted to evaluate TCED-HFV's potential in a clinical and a study setting, and to correlate all graded morphological characteristics in different stages to retinal function and disease severity.

Declarations
Con ict of Interest: The authors have no con icts of interest to declare.

Acknowledgement
The author would like to thank Xiaobin Zhou for helping to interpret the statistical analyzes.
Ethical approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed consent: Informed consent was obtained from all individual participants included in the study.
Compliance with Ethical Standards: