A meta-analysis of clinical outcomes in diabetic macular edema Authors: Christina N. Bobrek, Ishani Kapoor, Swara M. Sarvepalli, Elise Timtim, Majda Hadziahmetovic Published: 2026 ABSTRACT ObjectiveTo compare the efficacy of intravitreal anti-vascular endothelial growth factor (anti-VEGF), steroid, and laser on best-corrected visual acuity (BCVA), central retinal thickness (CRT), and treatment frequency in patients with diabetic macular edema.MethodsPubMed, EMBASE, and Web of Science were systematically searched from inception to August 2024. Mean change from baseline in BCVA (mcBCVA), mean change from baseline in CRT (mcCRT), and treatment frequency were collected using PRISMA guidelines. Statistical analysis was done on 20 studies using Review Manager 5.4.1.ResultsmcBCVA (mean difference (MD) = 0.14 logMAR, 95% confidence interval (CI): 0.05 to 0.23, P = 0.002) and mcCRT (MD = 38.73 μm, 95% CI: 13.39 to 64.08, P = 0.003) were significantly improved in the anti-VEGF group compared to laser. mcCRT (MD = −61.17 μm, 95% CI: −106.07 to −16.27, P = 0.008) was significantly greater in the steroid group compared to anti-VEGF. Mean number of treatments per year was significantly lower in the steroid treatment group compared to anti-VEGF (MD = −4.29, 95% CI: −4.5 to −4.09, P < 0.00001). In the anti-VEGF subgroup, mcCRT was significantly greater in the aflibercept group (MD = −37.52 μm, 95% CI: −62.97 to −12.07, P = 0.004) when compared to bevacizumab.ConclusionsThese data suggest anti-VEGF treatment is more effective than laser at improving BCVA and CRT, whereas steroid treatment is more effective than anti-VEGF at improving CRT outcomes. Steroid therapy, while not currently allotted a first-line therapy in the United States, may have a more significant role in reducing treatment burden and improving macular morphology. Introduction Diabetic macular edema (DME) is one of the leading causes of vision loss in patients with diabetes and diabetic retinopathy worldwide.1 It is caused by fluid accumulation secondary to disruption of the blood-retinal barrier by advanced glycated end products from diabetes.2 When left untreated, loss of two or more lines in visual acuity (VA) occurs in over half of patients with DME.3 A study from 2012 showed the global prevalence of DME was estimated to be 6.8% in patients diagnosed with diabetes, and the global prevalence of diabetes is predicted to increase to 10.9% of the population (700 million) by 2045.4,5 Due to the current and expected significant increase in patients being diagnosed with diabetes, it is crucial to assess and treat DME proactively to prevent vision loss. The three main DME treatment classes include intravitreal (IV) anti-vascular endothelial growth factor (anti-VEGF), steroid treatment, and (older) focal laser photocoagulation, with anti-VEGF therapy as the current gold standard.6 Although it is a very efficient treatment, not all patients achieve desirable clinical outcomes on anti-VEGF therapy and may benefit from switching to steroids.7 A 40% prevalence of refractory DME after two years of monthly IV ranibizumab was reported in a post-hoc analysis of the DRCR.net Protocol I.8 In a retrospective, interventional, comparative study, vision improvement was observed in 46%, 50%, and 33% of patients with refractory DME who switched to dexamethasone after three, four to six, or more than six anti-VEGF injections, respectively.9 Another consideration when treating patients with anti-VEGF is that it requires frequent treatment (8–9 injections the first year, 5–6 injections the second year) compared to steroids (3–4 injections per year) and lasers, which can be a barrier to patient care.10 With this in mind, it is important to consider not only the patients' outcomes but also their needs and capabilities when selecting a treatment plan. DME can be difficult to treat due to the variable onset, severity, and recurrence.11 There is a lack of direct head-to-head comparison of the efficacy of the three main treatment options (anti-VEGF, steroids, and focal laser photocoagulation) in treating patients with DME. This meta-analysis aims to bridge this gap by evaluating the impact of these treatments on best-corrected visual acuity (BCVA), central retinal thickness (CRT), and treatment burden. Additionally, we performed a subgroup analysis to compare the effectiveness of individual anti-VEGF agents.12, 13, 14 This layered approach not only highlights persistent challenges in DME management but also supports a shift toward more personalized, patient-centered treatment strategies. Methods Eligibility criteria and search methods A systematic literature search was conducted on PubMed, Embase, and Web of Science using keywords "diabetic macular edema" AND "visual acuity" AND each of the following therapies: triamcinolone, dexamethasone, fluocinolone acetonide, focal laser, ranibizumab, bevacizumab, aflibercept, faricimab, or brolucizumab from inception to August 2024. Study selection The search results were downloaded and uploaded to Covidence to exclude duplicates. Two authors independently completed the search from the remaining studies to retrieve titles and abstracts for initial screening for inclusion based on relevance. Any disagreements were referred to a third researcher for discussion. Selected articles were then downloaded for full-text review to determine eligibility for inclusion. Inclusion criteria for meta-analysis were as follows: (1) randomized control trials with results available, (2) done in individuals with a DME diagnosis, (3) studied one of the three main types of DME therapy (IV anti-VEGF, IV steroids, and laser), (4) compared at least two different treatment types, (5) included follow-up period between three to 24 months, and (6) provided information on BCVA, decrease in CRT, and/or treatment frequency. Studies were excluded if they were (1) meta-analyses, reviews, case reports, retrospective studies (2) involved cell or animal models, (3) not available in English, (4) included a switch protocol, (5) included sub-tenon injections, (6) did not report the outcome of interest, and/or (8) did not have an active comparator. This protocol was registered on the PROSPERO database (CRD42024578008). Twenty studies met the inclusion/exclusion criteria. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were followed for individual data extraction by multiple researchers (Fig. 1).15Fig. 1PRISMA flow diagram. PRISMA flow diagram. Data collection and risk of bias assessment For each study, data about change in BCVA, change in CRT (defined as CRT, central macular thickness, and central subfield thickness), and mean number of treatments were collected, if reported. Data was then classified based on the treatment comparison group. Comparison groups included laser vs anti-VEGF, steroid vs anti-VEGF, steroid vs laser, and the anti-VEGF subgroup. Two independent researchers completed the search and a Risk of Bias (ROB) Assessment through the Critical Appraisal Skills Programme Checklist and then referred to a third researcher for review and discussion. The study was evaluated by the Duke University Health System Institutional Review Board of Duke University and deemed not to require ethics approval. Data and ROB forms are available upon request. Data synthesis and analysis Data for meta-analysis was pooled in Review Manager 5.4.1 using the fixed-effects model, unless heterogeneity was >60%, in which case the random-effects model was used. Heterogeneity was evaluated by calculating the I2 statistics and visual inspection of funnel plots for each comparison. Mean difference (MD) and 95% confidence intervals (CIs) were calculated for each comparison, and P-values <0.05 were considered statistically significant. Statistical measures were calculated for each comparison of laser vs anti-VEGF, steroid vs anti-VEGF, steroid vs laser, and the anti-VEGF subgroup for three outcomes of interest – difference in the change in BCVA, change in CRT, and treatment frequency. When the Early Treatment Diabetic Retinopathy Study (ETDRS) letter score was used to measure BCVA, the score was converted to approximate logarithm of the minimum angle of resolution (logMAR) using the method proposed by Beck et al., which was used in quantitative analysis.16 Results We screened 12302 studies using our inclusion and exclusion criteria, of which 8168 were duplicates and 4034 were excluded as irrelevant based on title and abstract. The remaining 100 studies were screened through full-text analysis. 20 studies matched our inclusion criteria for meta-analysis, representing 4314 eyes. PRISMA guidelines were followed for the selection process (Fig. 1).15 The studies were published between 2007 and 2022 with an average follow-up duration of 14 months (range: 3–24 months) (Table 1). Four studies included treatment-naïve participants.17, 18, 19, 20Table 1Baseline characteristics of the included studies.Table 1Study (year)DesignPeriod (months)Number of ParticipantsTreatment GroupsLaser vs anti-VEGFBaker 201921RCT24446Focal/grid, AFLBrown 201522RCT24308Focal laser, AFLChen 202023RCT12251Focal laser, AFLComyn 201424RCT1233Focal laser, RBZFaghihi 200819,aRCT489Focal laser, BVZIshibashi 201525RCT12164Focal laser, RBZLi 201926RCT12384Focal laser, RBZMitchell 201127RCT12227Focal laser, RBZRajendram 201228RCT2465Focal laser, BVZSoheilian 200717,aRCT370Focal laser, BVZSoheilian 201220,aRCT24100Focal laser, BVZYaseri 201418,aRCT subset2419Focal laser, BVZSteroid vs anti-VEGFGillies 201429RCT1288DEX implant, BVZMeyer 202230RCT1258DEX implant, BVZShah 201631RCT750DEX implant, BVZSteroid vs LaserDRCRN 200832RCT24584Triamcinolone, Focal laserLam 200733RCT675Triamcinolone, Focal laseranti-VEGF SubgroupBaghi 201614RCT381AFL, BVZVader 202012RCT6166BVZ, RBZWells 201613RCT24660AFL, BVZ, RBZRCT = randomized control trial, AFL = aflibercept, BVZ = bevacizumab, RBZ = ranibizumab, DEX = dexamethasone.aIndicates patients had treatment-naïve DME. Table 1Baseline characteristics of the included studies.Table 1Study (year)DesignPeriod (months)Number of ParticipantsTreatment GroupsLaser vs anti-VEGFBaker 201921RCT24446Focal/grid, AFLBrown 201522RCT24308Focal laser, AFLChen 202023RCT12251Focal laser, AFLComyn 201424RCT1233Focal laser, RBZFaghihi 200819,aRCT489Focal laser, BVZIshibashi 201525RCT12164Focal laser, RBZLi 201926RCT12384Focal laser, RBZMitchell 201127RCT12227Focal laser, RBZRajendram 201228RCT2465Focal laser, BVZSoheilian 200717,aRCT370Focal laser, BVZSoheilian 201220,aRCT24100Focal laser, BVZYaseri 201418,aRCT subset2419Focal laser, BVZSteroid vs anti-VEGFGillies 201429RCT1288DEX implant, BVZMeyer 202230RCT1258DEX implant, BVZShah 201631RCT750DEX implant, BVZSteroid vs LaserDRCRN 200832RCT24584Triamcinolone, Focal laserLam 200733RCT675Triamcinolone, Focal laseranti-VEGF SubgroupBaghi 201614RCT381AFL, BVZVader 202012RCT6166BVZ, RBZWells 201613RCT24660AFL, BVZ, RBZRCT = randomized control trial, AFL = aflibercept, BVZ = bevacizumab, RBZ = ranibizumab, DEX = dexamethasone.aIndicates patients had treatment-naïve DME. Baseline characteristics of the included studies. RCT = randomized control trial, AFL = aflibercept, BVZ = bevacizumab, RBZ = ranibizumab, DEX = dexamethasone. Indicates patients had treatment-naïve DME. Ten studies did not report hemoglobin A1c (HbA1c) (Table 2).17, 18, 19, 20, 21, 27, 28, 31, 32, 33 Four studies did not report age.18,32,21,24 Two studies did not report sex.18,28 Nine studies did not report race.17, 18, 19, 20, 28,33,27,29,30 Ten studies did not report duration of diabetes.18,19,28,21, 31, 33,24,26,25Table 2Demographics and clinical characteristics of the comparison groups.12, 13, 14,17, 18, 19, 20, 28, 32, 21, 33, 31, 27, 24, 29, 30, 26, 25, 22, 23Table 2Laseranti-VEGFSteroidanti-VEGFSteroidLaseranti-VEGF SubgroupAge, years61.35 (9.08)60.68 (9.30)61.91 (9.51)61.70 (8.87)67.2 (9.8)66.2 (8.2)61.48 (10.51)SexFemale (%)44.0546.0242.4740.0059.4652.6346.79Male (%)55.9553.9857.5360.0040.5447.3753.21RaceCausasian (%)43.6332.9288.8995.6584.3385.2677.50Black or African American (%)18.2937.6311.114.3511.9810.8814.58Asian (%)34.0125.980.000.002.312.461.11American Indian or Alaskan Native (%)0.140.100.000.000.920.700.14Native Hawaiian or Pacific Islander (%)0.410.000.000.000.460.350.56Other (%)3.523.370.000.000.000.356.11HbA1c, %7.52 (1.35)7.62 (1.33)n/an/an/an/an/aDuration of Diabetes, years13.95 (8.77)14.14 (8.82)14.78 (9.68)14.98 (9.69)n/an/a16.44 (11.32) Table 2Demographics and clinical characteristics of the comparison groups.12, 13, 14,17, 18, 19, 20, 28, 32, 21, 33, 31, 27, 24, 29, 30, 26, 25, 22, 23Table 2Laseranti-VEGFSteroidanti-VEGFSteroidLaseranti-VEGF SubgroupAge, years61.35 (9.08)60.68 (9.30)61.91 (9.51)61.70 (8.87)67.2 (9.8)66.2 (8.2)61.48 (10.51)SexFemale (%)44.0546.0242.4740.0059.4652.6346.79Male (%)55.9553.9857.5360.0040.5447.3753.21RaceCausasian (%)43.6332.9288.8995.6584.3385.2677.50Black or African American (%)18.2937.6311.114.3511.9810.8814.58Asian (%)34.0125.980.000.002.312.461.11American Indian or Alaskan Native (%)0.140.100.000.000.920.700.14Native Hawaiian or Pacific Islander (%)0.410.000.000.000.460.350.56Other (%)3.523.370.000.000.000.356.11HbA1c, %7.52 (1.35)7.62 (1.33)n/an/an/an/an/aDuration of Diabetes, years13.95 (8.77)14.14 (8.82)14.78 (9.68)14.98 (9.69)n/an/a16.44 (11.32) Demographics and clinical characteristics of the comparison groups.12, 13, 14,17, 18, 19, 20, 28, 32, 21, 33, 31, 27, 24, 29, 30, 26, 25, 22, 23 The meta-analysis was done to determine if BCVA, CRT, and injection frequency outcomes showed greater improvement with anti-VEGF, steroid, or laser treatment (Fig. 2). Both mean change in BCVA (MD = 0.14 logMAR, 95% CI: 0.05 to 0.23, P = 0.002) and CRT (MD = 38.73 μm, 95% CI: 13.39 to 64.08, P = 0.003) were significantly improved in the anti-VEGF group compared to laser. The mean number of treatments per year was significantly lower in the laser group compared to anti-VEGF (MD = −0.59, 95% CI: −1.05 to −0.13, P = 0.01). The change in CRT (MD = −61.17 μm, 95% CI: −106.07 to −16.27, P = 0.008) was significantly greater in the steroid (dexamethasone implant only) group compared to anti-VEGF (bevacizumab only). The mean number of treatments per year was significantly lower in the steroid treatment group compared to anti-VEGF (MD = −4.29, 95% CI: −4.5 to −4.09, P < 0.00001). There was no significant difference in BCVA in the steroid group when compared to laser or anti-VEGF.Fig. 2Comparison of BCVA, CRT, and injection frequency outcomes in the three main treatment groups.(A) Comparison of BCVA and CRT outcomes in laser vs anti-VEGF treatment groups. There were 10 studies included for the BCVA outcomes comparison with 686 eyes and 926 eyes in the laser and anti-VEGF groups, respectively.17, 18, 19, 20,28,27,22,23,25,26 Studies were 24 months (4), 12 months (4), 4 months (1), and 3 months (1) in duration. 10 studies were included for the CRT outcome comparison with 684 eyes and 908 eyes in the laser and anti-VEGF groups, respectively.17, 18, 19, 20,28,21,27,24,26,22 Studies were 24 months (5), 12 months (3), 4 months (1), and 3 months (1) in duration. 3 studies were included for the mean annual treatment outcome comparison with 392 eyes and 606 eyes in the laser and anti-VEGF groups, respectively.21,26,23 Studies were 24 months (1) and 12 months (2) in duration. (B) Comparison of BCVA, CRT, and injection frequency outcomes in steroid vs anti-VEGF treatment groups. There were 3 studies included for the BCVA and CRT outcome comparisons with 97 eyes and 93 eyes in the steroid and anti-VEGF groups, respectively.31,29,30 Studies were 12 months (2) and 7 months (1) in duration. 2 studies were included for the injection frequency outcome comparison with 51 eyes and 51 eyes in the steroid and anti-VEGF groups, respectively.31,30 Studies were 12 months and 7 months in duration. (C) Comparison of BCVA outcomes in steroid vs laser treatment groups. There were 2 studies included for the BCVA outcome comparison with 243 eyes and 309 eyes in the steroid and laser groups, respectively.32,33 Studies were 24 months and 6 months in duration. Comparison of BCVA, CRT, and injection frequency outcomes in the three main treatment groups. (A) Comparison of BCVA and CRT outcomes in laser vs anti-VEGF treatment groups. There were 10 studies included for the BCVA outcomes comparison with 686 eyes and 926 eyes in the laser and anti-VEGF groups, respectively.17, 18, 19, 20,28,27,22,23,25,26 Studies were 24 months (4), 12 months (4), 4 months (1), and 3 months (1) in duration. 10 studies were included for the CRT outcome comparison with 684 eyes and 908 eyes in the laser and anti-VEGF groups, respectively.17, 18, 19, 20,28,21,27,24,26,22 Studies were 24 months (5), 12 months (3), 4 months (1), and 3 months (1) in duration. 3 studies were included for the mean annual treatment outcome comparison with 392 eyes and 606 eyes in the laser and anti-VEGF groups, respectively.21,26,23 Studies were 24 months (1) and 12 months (2) in duration. (B) Comparison of BCVA, CRT, and injection frequency outcomes in steroid vs anti-VEGF treatment groups. There were 3 studies included for the BCVA and CRT outcome comparisons with 97 eyes and 93 eyes in the steroid and anti-VEGF groups, respectively.31,29,30 Studies were 12 months (2) and 7 months (1) in duration. 2 studies were included for the injection frequency outcome comparison with 51 eyes and 51 eyes in the steroid and anti-VEGF groups, respectively.31,30 Studies were 12 months and 7 months in duration. (C) Comparison of BCVA outcomes in steroid vs laser treatment groups. There were 2 studies included for the BCVA outcome comparison with 243 eyes and 309 eyes in the steroid and laser groups, respectively.32,33 Studies were 24 months and 6 months in duration. A subgroup analysis comparing individual anti-VEGF treatments to each other was done (Fig. 3). The change in CRT was significantly greater in the aflibercept group (MD = −37.52 μm, 95% CI :−62.97 to −12.07, P = 0.004) when compared to bevacizumab. There was no significant difference between aflibercept and bevacizumab for BCVA and injection frequency outcomes. The change in CRT was significantly greater in the ranibizumab group (MD = 45.02 μm, 95% CI: 23.14 to 66.90, P < 0.0001) when compared to bevacizumab. There was no significant difference between ranibizumab and bevacizumab for BCVA outcomes.Fig. 3Comparison of BCVA, CRT, and injection frequency outcomes of the anti-VEGF subgroup.(A) Comparison of BCVA, CRT, and injection frequency outcomes in aflibercept vs bevacizumab treatment groups. There were 2 studies included for all three outcome comparisons with 240 eyes and 218 eyes in the aflibercept vs bevacizumab subgroups, respectively.13,14 Studies were 24 months and 3 months in duration. (B) Comparison of BCVA, CRT, and injection frequency outcomes in bevacizumab vs ranibizumab treatment groups. There were 2 studies included for all three outcome comparisons with 269 eyes and 273 eyes in the bevacizumab vs ranibizumab subgroups, respectively.12,13 Studies were 24 months and 6 months in duration. Comparison of BCVA, CRT, and injection frequency outcomes of the anti-VEGF subgroup. (A) Comparison of BCVA, CRT, and injection frequency outcomes in aflibercept vs bevacizumab treatment groups. There were 2 studies included for all three outcome comparisons with 240 eyes and 218 eyes in the aflibercept vs bevacizumab subgroups, respectively.13,14 Studies were 24 months and 3 months in duration. (B) Comparison of BCVA, CRT, and injection frequency outcomes in bevacizumab vs ranibizumab treatment groups. There were 2 studies included for all three outcome comparisons with 269 eyes and 273 eyes in the bevacizumab vs ranibizumab subgroups, respectively.12,13 Studies were 24 months and 6 months in duration. Discussion While previous randomized controlled trials have established anti-VEGF as the gold standard for treating DME, our study's strength lies in utilizing meta-analysis to evaluate additional outcomes, including CRT and injection frequency. The results of this meta-analysis, which pooled 20 studies, suggest that anti-VEGF treatment is more effective than laser at improving BCVA and CRT, whereas steroid treatment is more effective than anti-VEGF at improving CRT outcomes and requires fewer yearly treatments. Due to data limitations, we were unable to assess the effect of steroid treatment on visual acuity at the three-month time point, which would likely provide more meaningful insights. Additionally, with several studies excluding baseline HbA1c levels, diabetes duration, and patient race, there is a decreased generalizability of the results. Long-term visual outcomes in the steroid groups were likely influenced by cataract development, leading to a gradual decline in vision over time. CRT outcomes did not differ between steroid and laser treatments, as focal macular edema suitable for laser therapy is typically characterized by focal microaneurysms rather than diffuse macular thickening. In the anti-VEGF subgroup analysis, aflibercept proved more effective than bevacizumab at reducing CRT; however, this outcome may be confounded by patients in these groups having different baseline visual acuity. Additionally, while aflibercept was superior in reducing CRT, there was no significant difference in BCVA improvement compared to bevacizumab, suggesting that structural changes may not directly translate to functional outcomes. Steroid treatment resulting in better CRT outcomes compared to anti-VEGF confirms the well-known significance of the inflammatory component associated with DME, and steroids' role in effectively blocking multiple inflammatory pathways.34,35 Another potential benefit of steroids is the reduced frequency of injections (on average, every three to six months, in some instances even longer, depending on the compound).36,37 While CRT outcomes were improved, BCVA outcomes were likely inhibited by cataract development. In a five-year study by Gilles et al., 71% of phakic patients treated with IV triamcinolone acetonide required cataract extraction.38 Similarly, Campochiaro et al. reported that most patients treated with fluocinolone acetonide vitreous inserts developed cataracts; however, their post-surgical BCVA was comparable to that of patients who were pseudophakic at baseline.39 In the anti-VEGF subgroup analysis, when aflibercept, bevacizumab, and ranibizumab were used to treat patients with mild vision loss (20/32 to 20/40), as previously reported, there was not a significant difference in mean visual acuity after one year of treatment.12, 13, 14 However, patients with worse baseline visual acuity saw less improvement in reducing CRT with bevacizumab than aflibercept and ranibizumab.12, 13, 14 In the 12-month study by Wells et al., patients with worse baseline visual acuity saw greater improvement in BCVA with aflibercept than bevacizumab or ranibizumab.13 Another important consideration was the lack of significant difference in BCVA and treatment frequency (mean injections per year) between aflibercept and bevacizumab despite aflibercept having a greater improvement in CRT. This may be attributable to the inclusion of only two studies, which had largely different follow-up times of three months and 24 months. Despite our search including fluocinolone acetonide, faricimab, and brolucizumab, there were not enough studies to conduct an analysis, or the data were not provided in the format needed for analysis. The YOSEMITE and RHINE trials are two important randomized, double-masked, non-inferiority trials that compared 6.0 mg faricimab to 2.0 mg aflibercept over 12 months.40 In the YOSEMITE study, the difference in BCVA between faricimab and aflibercept was −0.2 ETDRS letters (97.52% CI: −2.0 to 1.6).40 The RHINE study resulted in a difference in BCVA of 1.5 ETDRS letters (97.52% CI: −0.1 to 3.2) between faricimab and aflibercept.40 These results show faricimab to be as effective as aflibercept at improving BCVA. Additionally, the PHOTON study assessed the efficacy of aflibercept 8 mg to aflibercept 2 mg.41 The results at 48 weeks showed that the mean change in BCVA for the 8 mg dose every twelve weeks (8.8 letters, SD 9.0) or every sixteen weeks (7.9 letters, SD 8.4) was non-inferior to the mean change in BCVA for the 2 mg dose every eight weeks (9.2 letters, SD 9.0, respectively).41 Given the improvement in outcomes with each treatment, it is also important to note their adverse effects. A three-year, randomized trial demonstrated that one-third of patients experienced elevated intraocular pressure (IOP) after the second and third doses of dexamethasone.42 It was found that the increases in IOP were predictable and regressed by six months with no increased incidences in the second or third years of treatment.42 Similarly, there is an increased risk of cataract formation with dexamethasone use. A randomized sham-controlled trial showed that rates of cataracts after three years of treatment in DME patients were 67.9% and 20.4% in 0.7 mg dexamethasone and sham groups, respectively.36 Interestingly, the study found that vision loss caused by cataract formation was restored after surgery, and dexamethasone showed improvements in visual acuity regardless of phakic or pseudophakic status at baseline.36 In assessing the risks of anti-VEGF treatment, a systematic review of 278 studies found that ranibizumab, bevacizumab, and pegaptanib therapies resulted in fewer than one serious adverse event per 100 injections.43 The RISE/RIDE studies showed monthly 0.3 mg ranibizumab, 0.5 mg ranibizumab, and sham to have a mortality rate of 4.4%, 6.4%, and 2.8%, respectively, after 36 months.44 Since the efficacy of the 0.3 mg and 0.5 mg doses of ranibizumab was equal, they concluded that treating patients with 0.3 mg ranibizumab in the setting of diabetic retinopathy may be beneficial in reducing the risk of adverse outcomes.44 Adverse effects of focal laser treatment include subretinal fibrosis and scarring, and increased risk of choroidal neovascularization.45 Regardless, with available effective pharmacological therapies, this treatment modality seems to be coming out of favor. When considering the future of DME studies, the use of disorganization of retinal inner layers (DRIL)as a biomarker for patient outcomes should be explored, especially in patients who fail to achieve improved BCVA. DRIL has been associated with decreased improvement in BCVA when present at baseline.46,47 Furthermore, changes in DRIL have been shown to predict future changes in BCVA.47 In the study by Chakraborty et al., the percentage of eyes with DRIL in the dexamethasone group decreased from 36% to 25%, whereas no change was observed in the aflibercept group over 12 months.42 Similarly, a 12-month multicenter, retrospective study observed a significant improvement in DRIL when treated with dexamethasone implants; however, baseline VA did not differ between eyes with or without DRIL.48 Despite DRIL being an excellent marker for DME studies, our meta-analysis could not use DRIL as a biomarker for analysis due to its absence in the included studies. Two other important biomarkers in DME treatment include the external limiting membrane (ELM) and ellipsoid zone (EZ). Treatment of DME with steroid or anti-VEGF has been associated with reduced disruption of the ELM and EZ markers, accompanied by improvements in BCVA.49,50 A retrospective study including 100 eyes showed improvement in ELM integrity (1575.9 ± 285.9 μm to 1711.7 ± 244.0 μm) and EZ integrity (1531.5 ± 317.1 to 1694.3 ± 252.8 μm) from baseline to three months of dexamethasone implant treatment.49 Changes in visual acuity were found to be significantly correlated with ELM and EZ integrity.49 Similarly, results from a prospective study including 44 patients who received three monthly doses of 1.25 mg bevacizumab showed a decrease from 81.8% to 9.1% and 79.5% to 11.4% in global ELM and EZ disruption, respectively, with significant visual acuity improvements.50 While these studies have limited population sizes and follow-up times, VEGF has been linked to increased retinal thickness and disruption of ELM and EZ, making these markers promising for future studies.50,51 Sensitivity analysis Sensitivity analysis showed similar results when comparing fixed-effects and random-effects model outcomes (Supplementary Fig. 1). This was assessed for BCVA and injection frequency in the steroid vs anti-VEGF group. Limitations Although our study provides insight into treating DME in a clinical setting, certain limitations should be considered. First, due to the extensive volume of data, we were not able to assess the change in BCVA between each individual anti-VEGF and each individual steroid, as this would demand resources beyond the scope of this study. For the same reasons, we were not able to analyze studies that included combination groups (e.g., anti-VEGF vs anti-VEGF with laser). Second, the included studies varied in their end points and treatment dosages. This may affect the visual outcomes analysis of patients due to anti-VEGF producing the most significant change in BCVA within the first three months, followed by a gradual improvement over time.31,29,42 Third, due to the high risk of cataract development associated with steroid injections, evaluating BCVA over an extended period (e.g., beyond three months) may not provide a fair assessment of treatment efficacy, particularly in a phakic patient population.42 However, a five-year follow-up study reported that the maximum change in BCVA and CRT outcomes occurred at two years with anti-VEGF, and our included studies are within a follow-up range of three to 24 months.52 Another limitation in comparing anti-VEGF and steroid treatments is that patients receiving steroid injections are often those previously treated, with chronic or refractory DME, which may introduce confounding variables related to disease duration, prior treatments, and baseline severity.53 Similarly, this study included a wide timeframe from 2007 to 2022, and as a result, treatment standards and approaches to patient management may have varied. Fourth, we were not able to compare visual outcomes between anti-VEGF and steroid treatments in pseudophakic eyes due to insufficient data. In the study by Gillies et al., BCVA outcomes showed no significant difference between the bevacizumab and dexamethasone implant groups after 12 months in pseudophakic eyes.29 A twofold increase in cataract density was observed more frequently in the dexamethasone group (13%, 6/46) compared to bevacizumab (4.8%, 2/42).29 Additionally, cataract surgery was required in three patients from the dexamethasone group and one patient from the bevacizumab group.29 Fifth, while switch protocols were excluded, rescue treatment was not. Overall, seven of the included papers offered rescue treatment to patients whose visual outcomes failed to achieve adequate improvement.13,32,21,22,23,25,26 Rescue protocols included switching from laser treatment to anti-VEGF/steroid treatment or vice versa, participants leaving the study to seek alternative treatment, or the addition of steroid treatment to those receiving anti-VEGF treatment. Since this meta-analysis focused on endpoint data, only studies that retained patients placed on a rescue protocol through the entirety of the study were included in the analysis, while those that did not were excluded. The outcomes of these studies were not separated based on which patients did or did not receive the rescue treatment, thus, the further analysis comparing these two was not possible. Excluding switch protocols helped control confounding factors but limited the generalizability of the findings to real-world practice. Sixth, several potential contributors to heterogeneity were considered, including baseline BCVS. However, formal comparison of baseline BCVA differences was limited because these data were inconsistently reported. Specifically, baseline BCVA was not reported in all studies, and some studies provided only pooled baseline values rather than values stratified by comparison group. In analyses where only two studies contributed to each comparison (Fig. 2, Fig. 3), further subgroup or sensitivity analyses were not statistically meaningful. Thus, heterogeneity may reflect differences in baseline disease severity, treatment exposure, follow-up duration, imaging definitions, and study design. There were 20 RCTs assessed for the risk of bias. Random sequence generation and allocation concealment were clear for 20/20 (100%) of the RCTs. The RCTs had a low risk of bias with respect to blinding of participants (12/20, 60%), blinding of trial personnel (4/20, 20%), and blinding of outcome assessment (14/20, 70%). Conclusions What makes our study unique is its comprehensive and comparative approach. While previous studies have evaluated individual treatments for DME, few have systematically compared all three main therapeutic options, anti-VEGF agents, corticosteroids, and focal laser photocoagulation, in a single, rigorous meta-analysis. Most importantly, direct head-to-head comparisons remain limited in the literature, especially those assessing not only anatomical and functional outcomes such as CRT and BCVA, but also treatment burden, a critical factor in real-world management. By integrating data from both randomized clinical trials and real-world evidence, we bridge the gap between controlled environments and everyday clinical practice, offering insights that are more broadly applicable and reflective of patient diversity. Our meta-analysis confirms that anti-VEGF therapy yields greater improvements in BCVA and CRT outcomes compared to laser treatment. Additionally, steroid treatment shows a greater improvement in CRT outcomes with fewer injections compared to anti-VEGF therapy, while there is no difference in BCVA outcomes when comparing steroid and laser treatment. Additionally, aflibercept was shown to be more effective than bevacizumab at reducing CRT in the anti-VEGF subgroup, but this outcome may be confounded by the initial severity of DME. Further studies are needed to assess the use of steroids as first-line treatment and the use of combination treatments. These analyses, along with the use of DRIL as a biomarker for patient outcomes, will provide a comprehensive assessment of best treatment plans for patients. Study approval Not Applicable. Author contributions The authors confirm contribution to the paper as follows: CB: Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing - original draft, Writing - review and editing; IK: Formal analysis, Investigation, Methodology, Project administration, Supervision, Validation, Visualization, Writing - review and editing; SS: Formal analysis, Investigation, Methodology, Project administration, Supervision, Validation, Writing - review and editing; ET: Formal analysis, Methodology, Project administration, Supervision, Validation, Writing - review and editing; MH: Conceptualization, Data curation, Investigation, Methodology, Project administration, Supervision, Validation, Writing - review and editing. All authors reviewed the results and approved the final version of the manuscript. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.