Chinese Herbal Medicine Combined with Intravitreal Antivascular Growth Factor Agents in Treatment of Macular Edema Secondary to Retinal Vein Occlusion: A Systematic Review and Meta-Analysis

Background Retinal vein occlusion (RVO) is the second most common retinal vascular disease in the world after diabetic retinopathy. Moreover, macular edema (ME) is the main cause of visual impairment in RVO patients. Intravitreal injection of antivascular endothelial growth factor (VEGF) agents is recommended for RVO-ME. However, repeated injections severely limit their efficacy. Chinese herbal medicine (CHM) is widely used in RVO-ME as adjuvant therapy in China. Objective The study aims to evaluate the efficacy and safety of anti-VEGF combined with CHM for RVO-ME and to provide reliable evidence for clinical application. Methods Seven databases were searched without language or publication status restrictions. Randomized controlled trials (RCTs) comparing anti-VEGF combined with CHM (anti-VEGF + CHM) versus anti-VEGF in participants with RVO-ME were included in this study. The “risk of bias assessment tool” of the Cochrane Handbook was applied to assess the quality of included trials, and RevMan 5.3 software was used for data analysis. Results A total of 10 relevant trials with 743 patients were identified. The results showed that BCVA of the anti-VEGF + CHM group significantly improved at 3 months (P < 0.00001), 6 months (P=0.008), and 12 months (P=0.01), and CMT significantly reduced at 1 month (P=0.02), 2 months (P=0.0009), 3 months (P < 0.05), 6 months (P < 0.0001), and 12 months (P < 0.00001) compared with the anti-VEGF group alone. At the same time, the anti-VEGF + CHM group has a better performance in reducing the number of injections (P < 0.05) and improving the total effective rate (P < 0.0001). However, regarding adverse events, there was no statistical difference between the two groups (P=0.09). Conclusions Our results provide promising evidence that anti-VEGF therapy combined with CHM may be more beneficial to patients than anti-VEGF therapy alone. However, because of the low quality and small sample size of the included studies, more rigorous and larger-scale trials were necessary to validate our results. Registration Number. CRD42021270262.


Introduction
Retinal vein occlusion (RVO) is an obstruction of the normal venous system of the retina and the second most common retinal vascular disease after diabetic retinopathy, which seriously afects the visual function and quality of life of patients [1,2]. Te overall prevalence of RVO is 0.72% and it is estimated that RVO will afect 21 million people in Asia in 2040 [3]. Macular edema (ME) is the most serious complication of RVO, which can lead to irreversible visual impairment [4,5]. ME occurs in 5%-15% of patients with branch retinal vein occlusion (BRVO) within 1 year of onset. Within 2 years of central retinal vein occlusion (CRVO), ME occurs in approximately 30% of non-ischemic CRVO patients and 75% of ischemic CRVO patients [6,7]. Vascular endothelial growth factor (VEGF) is a key factor in the pathogenesis of RVO-ME. It increases the permeability of blood vessels by increasing the phosphorylation of tight junction proteins, leading to the destruction of the bloodretinal barrier and vascular leakage, triggering the occurrence of ME. Anti-VEGF has therefore become a major treatment option for RVO-ME with promising results [8][9][10]. However, anti-VEGF agents have a short half-life and require multiple injections, which increased the safety risks for patients, such as intraocular infammation, intraocular pressure elevation, and ocular hemorrhage [11]. At the same time, it brings a heavy psychological and economic burden to patients and may even lead to family and social problems. Terefore, the study of complementary therapy based on anti-VEGF may have great clinical and social signifcance.
CHM has the characteristics of comprehensive efect and low price. In the theory of traditional Chinese medicine (TCM), RVO belongs to the category of "sudden blindness." RVO-ME is a collateral disease in TCM. Te basic pathogenesis is vein stasis, and blood stasis turns into edema. Promoting blood circulation and removing blood stasis, promoting diuresis, and eliminating dampness are the most important therapeutic methods in clinical research [12]. TCM is widely used in complementary therapy of RVO-ME in China, but there is a lack of evaluation and evidence on the efcacy and safety of this add-on therapy.
Te purpose of this systematic review is to evaluate the efcacy and safety of anti-VEGF combined with CHM Versus anti-VEGF alone in the treatment of RVO-ME and to provide evidence for clinical practice in this specifc feld.

Methods.
Tis study followed the preferred reporting items for systematic reviews and meta-analysis (PRISMA) statements checklist [13]. Te PRISMA 2020 checklist is provided in Supplementary Material S1. Besides, our systematic review protocol was registered at the international prospective register of systematic reviews (PROSPERO) with the registration number CRD42021270262.

Search Strategy.
Te following electronic databases were searched: PubMed, Embase, Cochrane Library, China National Knowledge Infrastructure (CNKI), Wanfang Database, VIP information resource integration service platform (cqvip), and China Biology Medicine (SinoMed). Te search time was from the inception of the database to February 12, 2022, and the search method of combining subject words and free words was adopted. Te following grouped keywords were used as search strategy and modifed according to diferent databases: "retinal vein occlusion," "RVO," "BRVO," "CRVO," "macular edema," "macular edema, cystoid," "central retinal edema," "cystoid macular dystrophy," "drugs, Chinese herbal," "Chinese herbal medicine," "TCM," "alternative medicine," "vascular endothelial growth factors," "anti-VEGF," "ranibizumab," "bevacizumab," "afibercept," and "randomized controlled trial "RCT." No restriction on language was applied. We also manually retrieved relevant articles and clinical studies to obtain the literature as much as we can. No language and status restriction was set in this review. Te detailed search strategies are provided in Supplementary Material S2.

Inclusion Criteria.
Inclusion criteria were as follows: (1) subjects: patients with RVO-ME; (2) research type: randomized controlled trial (RCT), regardless of whether the literature is blind or not; (3) interventions: patients in the treatment group received anti-VEGF + CHM (including decoction, pill, capsule, Chinese medicinal preparations, and no restricted dosage form). Te control group received the same anti-VEGF agents as the treatment group, with or without a placebo; (4) languages of publication are not restricted; (5) one or some related outcomes, including bestcorrected visual acuity (BCVA) and central macular thickness (CMT), number of intravitreal injections, total efective rate, and the incidence of adverse events (AE) were reported in the studies; and (6) study had a minimum follow-up of 6 months.

Exclusion
Criteria. Exclusion criteria were as follows: (1) case reports, abstracts, reviews, and reports with incomplete data about the macular thickness or best-corrected visual acuity; (2) combined with other retinal diseases; (3) patients received other TCM therapies except for CHM, including acupuncture, moxibustion, cupping, and tuina; (3) trials without a control group; (4) the original document or complete raw data could not be obtained or the raw data could not be extracted or merged; and (5) duplicate published articles and duplicate data.

Data Extraction.
Two authors will independently extract data. Any disagreement will be resolved by discussion until consensus is reached or by consulting a third author. Te following data will be extracted: author, published year, country, age, gender and number of participants, duration of RVO, intervention measures, follow-up period, BCVA, CMT, number of intravitreal injections, total efective rate, and AE. Any diferences in the assessment process will be resolved through consultation. If such diferences are not resolved, a third researcher will join the discussion and make a fnal decision.
Te logarithm of the minimum angle of resolution (LogMAR) data on BCVA was extracted from studies. If studies reported BCVA in other units, the values were transformed to LogMAR as described [14].

Quality Evaluation.
Based on the bias risk assessment of the Cochrane collaboration tool [15], the methodological quality of the selected studies was evaluated by XS and XL. Te assessed items included: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcomes assessment (detection bias), incomplete outcomes data (attrition bias), selective reporting (reporting bias), and other biases. Tree levels were used to assess the following methodological quality: "low risk of bias (+)," "high risk of bias (−)," and "unclear risk of bias (?)." Consensus on this issue was researched with the help of a third party when necessary.

Statistical Analyses.
Review Manager software (Rev-Man, Version 5.3; Cochrane Collaboration) was utilized to conduct the data analysis of dichotomous and continuous outcome measures. Te risk ratio (RR) was adopted for dichotomous variable data as the index of efect value, and mean diference (MD) was adopted for the continuous variables. All of these were expressed as a 95% confdence interval (CI). Te heterogeneity was estimated by the chisquared test, and it was presented as signifcant when I 2 statistics > 50% or P < 0.05 [16]. Te random-efects model was applied for the meta-analysis if signifcant heterogeneity was identifed. Otherwise, the fxed-efects model was applied. Feature analyses were performed based on subgroup analyses of the study to explore the sources of heterogeneity. For the study reporting median and interquartile ranges, mean and standard deviation were estimated using the method described by McGrath et al. [17]. In addition, sensitivity analyses were performed to assess the stability of the meta-analysis results. Publication bias was assessed by funnel plot analysis and statistically investigated by Egger's test. P < 0.05 was considered statistically signifcant, except where otherwise specifed.

Study Identifcation.
According to the literature retrieval strategies and inclusion and exclusion criteria mentioned previously, 140 articles with possible relevance were retrieved from seven electronic databases. After the removal of 57 duplicates, 83 reports were retrieved. After going through the titles and abstracts, 52 articles were excluded because they failed to meet the inclusion criteria, and the remaining 31 articles were retrieved in full text for further assessment. After further reading, 21 articles were excluded for the reasons listed in Supplementary Material S3. Finally, a total of 10 eligible studies were included in this meta-analysis, with 743 patients in total. Te fow chart of the literature screening is shown in Figure 1.

Risk of Bias of Included Trials.
Te results of the risk of bias assessment are shown in (Figures 2(a) and 2(b)). In this systematic review, all 10 trials were reported as RCTs. However, only six studies described their random method [18,19,[21][22][23]26], among which four studies mentioned the "random number table method" explicitly [18,19,22,26], one study was generated by SAS software [21], and one study was mentioned as the "complete random method" [23]. Te remaining research only mentions the word "random," without detailed information about this method. Because no sham injection and placebo were used, the blinding of participants and personnel was assessed as high risk. None of the included studies mentioned the blind method of distribution hiding and result evaluation explicitly. Due to a lack of information, other biases in all studies were also assessed as unclear.

Description of the CHM. Te ingredients of CHM in each
RCT are listed in Supplementary Material S4. A total of 9 herbal decoctions were used in 10 included studies. Te number of herbal components in the formulas varied from 9 to 19. Details of the most commonly used herbs in included studies are listed in Table 2. Te most popular form of Chinese herbal medicine is a decoction, which was used in 8/ 10 studies. All herbal medicines are taken orally. All studies reported the complete ingredients of herbal formulas, among which seven studies mentioned the full dose of related ingredients [18,[20][21][22][24][25][26]. Te meta-analysis of BCVA at 6 months showed moderate heterogeneity (P � 0.01, I 2 � 69%), so we conducted a subgroup analysis. Due to the small number of studies, we did not conduct a subgroup analysis based on the types of anti-VEGF agents. In the RVO subtype, the average BCVA improvement between the two groups was signifcant at the RVO subgroup (P � 0.0008), while it was not statistically signifcant at the BRVO subgroup (P � 0.22) and the CRVO subgroup (P � 0.85). Meanwhile, in the course of treatment, the average BCVA improvement was signifcant at ≥12 W subgroup (P � 0.004), and it was not signifcant at <12 W subgroup (P � 0.  Figure 4). Due to the high heterogeneity at 3 months, the descriptive analysis was performed. Tree studies showed that there was no statistical diference in CMT between the two groups [20,23,27]. Five studies showed that the anti-VEGF + CHM group had better CMT than the anti-VEGF group alone, and the diference was statistically signifcant (P < 0.05) [18,19,21,25,26].
Te CMT improvement between the two groups was signifcant at the CRVO subgroup (P � 0.004), bevacizumab subgroup (P � 0.003), and ≥12 W subgroup (P � 0.008) at 1 month. At the same time, the CMT improvement between the two groups was signifcant in all 3 subgroups at 6 months, except for the unclear anti-VEGF agents' subgroup (P � 0.26) (Supplementary Materials S5 and S6).

Number of Intravitreal Injections.
Six studies reported the number of anti-VEGF injections during the follow-up period, with high heterogeneity among studies (P < 0.00001, I 2 � 98%) [19][20][21][22][23]27]. We speculated that the reason for the heterogeneity might be the diference in injection schemes and reinjection standards among studies, so we made a descriptive analysis. One study showed that there was no statistical diference in the number of anti-VEGF injections between the two groups [22]. Five studies showed that the anti-VEGF + CHM group, compared with the anti-VEGF group, could reduce the number of intravitreal injections, and the diference was statistically signifcant (P < 0.05) [19-21, 23, 27].

Sensitivity
Analysis. In addition, we performed a sensitivity analysis. Generally, when one particular study was deleted, the remaining studies were reanalysis to see if omitted studies afected the overall estimate signifcantly. As shown in Supplementary Material S8, sensitivity analysis indicated that Hao's study [21] was the main source of CMT heterogeneity at 1 month. Heterogeneity was reduced after the removal of this study (P � 0.

GRADE Evidence
Profle. An overview of the level of evidence is presented in Table 3. Due to the signifcant ROB and heterogeneity in the study methods, the overall quality of evidence was assessed as very low quality and low quality, which indicated that these estimates were uncertain and that further studies may afect our confdence in estimating the efects of CHM.

Discussion
We conducted an extensive literature search and identifed 10 studies (743 patients) for analysis. Te results showed that the overall quality of the evidence was low. Compared with anti-VEGF therapy alone, the statistical results showed that anti-VEGF agents combined with CHM had a potentially positive efect on RVO-ME in improving vision, promoting the resolution of macular edema, and reducing the number of intravitreal injections. In terms of safety, there was no signifcant diference between the anti-VEGF + CHM group and the anti-VEGF group alone, which indicated that CHM can be safely used in RVO-ME. At present, the evidence showed that CHM is efective and safe for RVO-ME, but it needs to be verifed by a more rigorous design of large-scale clinical trials.
Ischemia and hypoxia in the retina will promote the synthesis of VEGF, induce neovascularization and increase vascular permeability, which plays an important role in promoting the generation and progression of RVO and ME. Anti-VEGF is the preferred treatment option for RVO-ME at this stage [28]. However, due to the economic burden caused by repeated injections, patients with RVO-ME are more willing to use CHM alone or in combination with anti-VEGF agents. In their opinion, CHM is efective in improving systemic symptoms, improving visual function, and reducing the number of injections of anti-VEGF. In addition, CHM has been used for thousands of years, and it seems to be relatively safe. However, the efcacy and safety of CHM for RVO-ME are still inconclusive. To the best of our knowledge, this is the frst systematic review and metaanalysis of published RCTs to summarize the efcacy and safety of CHM combined with anti-VEGF in the treatment of RVO-ME and provide patients, decision-makers, and clinicians with the latest updated level of evidence.
Te eight most commonly used herbs include the following: Chinese angelica root (Radix Angelicae Sinensis), Radix Rehmanniae (Rehmannia glutinosa), Safower fower (Flos Carthami Tinctorii), Poria (Scierotium Poriae Cocos), Szechuan lovage root (Rhizoma Ligustici Chuanxiong), Peach kernel (Prunus persica), Alisma (Rhizoma Alismatis), and red Peony root (Paeonia veitchii Lynch). Tese herbs achieve the efects of nourishing and harmonizing blood, removing blood stasis. It is considered that the basic pathogenesis of RVO-ME in TCM is blood stasis in the eye context. Based on the basic pathogenesis, TCM focuses on the overall regulation and syndrome diferentiation and has the characteristics of giving consideration to both the symptoms and the root causes and fexible medication. Pharmacological studies have demonstrated the mechanisms of some drugs in treating RVO-ME. For example, Z-Ligustilide in Chinese angelica root is involved in regulating the secretion of VEGF in ischemic/ hypoxic retinopathy and protecting the function and morphology of the retina [29]. Radix Rehmanniae can dilate blood   vessels, reduce the permeability of capillary, and inhibit infammation of vascular endothelium [30]. Quercetin, luteolin, baicalein, and β-sitosterol in safower fower have such biological activities as anti-infammation, oxidation resistance, and regulation of immunity, which are the material basis for the treatment of RVO [31]. Te ethanol extract of Poria can inhibit the nuclear factor-kappa B signaling pathway induced by lipopolysaccharide in RAW 264.7 macrophages, thus reducing the secretion of infammatory mediators TNF-α and IL-1β. Poria targets the response of macrophages via the inactivation of the NF-κB signaling pathway [32]. Our fndings are consistent with previously published systematic reviews and meta-analyses. Liu et al. evaluated the efcacy of TCM combined with diferent western medicine therapies in the treatment of RVO-ME and found that the combined therapy had better efcacy than the western medicine therapy in improving the visual acuity, reducing the CMT, and improving the clinical efcacy [33]. However, most of the included studies have a short observation time. RVO-ME is prone to relapse, so long-termfollow-up and observation are needed to obtain the long-term efect of CHM on RVO-ME. Terefore, our study stipulated that the included study should have a minimum follow-up time of 6 months. At the same time, we took the number of anti-VEGF injections as a secondary outcome, because it was necessary and important for evaluating the economy and safety of CHM.
Our meta-analysis has some limitations. First, the overall quality of the included studies is not high and the methodological quality is poor, which may lead to selection bias, implementation bias, and result measurement bias. Secondly, due to the diference in prescription and herbal dosage, there is considerable clinical heterogeneity among these CHMs. Terefore, the combined analysis can only draw the trend of overall efcacy, but cannot draw positive or negative conclusions, which leads to a certain limitation of the extrapolation of the research results. At the same time, none of the included studies focused on the outer layer structure of optical coherence tomography, which is closely related to visual acuity. Future studies are suggested to take external limiting membrane and ellipsoid zone into account, which is more conducive to the interpretation of the results. In addition, since we only retrieved articles in English and Chinese, articles published in other languages, such as Korean and Japanese (CHM is commonly used in East Asian countries), may not be recognized and included in this review, which may lead to deviations in the results. It is suggested that future research should report the generation of a random distribution sequence and the concealment of a random scheme in detail, as well as the withdrawal and dropping of of participants in the study. Te sample size calculation method, as well as the composition and dose of CHM, should also be reported in detail. At the same time, the design and report of the study should follow the CONSORT statement.

Conclusion
In conclusion, the current data are promising but inconclusive. Te combination of CHM and anti-VEGF may be more benefcial to RVO-ME than anti-VEGF alone.
However, due to the limitation of sample size and methodological quality, a large sample size and multicentre welldesigned clinical study are required.