Purpose
To describe the visual and anatomical changes after restarting anti-vascular endothelial growth factor (VEGF) therapy in patients with diabetic macular edema who returned with stabilized vision after being lost to follow-up (LTFU), and to explore the distinct predictors of clinical benefits upon retreatment.
Design
Retrospective clinical cohort study.
Methods
In this study, 976 patients who returned after being LTFU for ≥6 months were included. All participants had center-involved diabetic macular edema and presented with visual acuity (VA) <20/25 and central foveal thickness (CFT) ≥250 μm. VA and CFT were assessed before being LTFU, at return, and after retreatment. Patients were stratified into two groups: (1) Patients who returned with stabilized VA (the first VA upon returning ≥ the last recorded VA prior to being LTFU), and (2) patients who returned with vision decline (the first VA upon returning < the last recorded VA prior to being LTFU).
Results
After a median LTFU period of 259 days, 447 patients (45.8%) had stabilized VA, and 529 patients (54.2%) had vision decline. With similar follow-up periods (361 vs 356 days) after retreatment, visual gain was obtained in 32.7% of patients with stabilized VA, significantly lower than in those with vision decline (62.0%, P <.001). Consistently, CFT reduction by ≥10% was achieved in 56.4% of patients with stabilized vision and in 63.2% of those with decreased vision ( P =.038) over similar follow-up periods (377 vs 361 days). Multivariate logistic regression analysis indicated that the odds of visual (odds ratio = 0.29, 95% CI:0.22-0.38, P <.001) and anatomical (odds ratio = 0.78, 95% CI: 0.59-1.02, P =.073) improvements were lower in patients who returned with stabilized VA, but increased with more anti-VEGF injections administered during retreatment (both P <.001). Notably, patients presenting with stabilized VA of 20/50 or better did not obtain vision gain after retreatment despite anatomical improvement.
Conclusions
Patients who returned with stabilized vision exhibited suboptimal visual and anatomical responses to anti-VEGF retreatment. Those with VA of 20/50 or better did not improve sufficiently in vision despite anatomical improvements. The decision to readminister anti-VEGF therapy in patients returning with stabilized moderate-to-good vision should be weighed carefully, and alternative or combination therapies may be warranted.
INTRODUCTION
I ntravitreal injection of anti-vascular endothelial growth factor (VEGF) is the first-line treatment for diabetic macular edema (DME). Regular follow-up is essential to maintain visual acuity (VA) and stabilize retinal structure, as the drug is rapidly cleared from the vitreous. However, in real-world practice, patients with diabetes have limited access to eye care services and are therefore prone to being lost to follow-up (LTFU). , The rates of being LTFU are alarming, reaching 62.0% at 6 months and 54.2% at 12 months.
Previous literature reports have showed that most patients exhibited marked disease progression after being LTFU, with substantial visual decline and worsening macular edema. ,, However, some have described otherwise, suggesting that inconsistent follow-up only caused modest vision loss at return, without persistent visual impairments after resuming anti-VEGF therapy. , A recent analysis of our group disclosed that, after being LTFU for 6 months or more, about half of the patients presented with vision deterioration upon returning, while the remainder maintained stable or improved vision.
Anti-VEGF therapy is reinitiated in most patients who returned with center-involved DME. However, to our knowledge, the efficacy of retreatment in terms of visual improvement and reduction of central foveal thickness (CFT) has been rarely investigated. Another issue of particular interest is the therapeutic response in patients with stabilized VA despite treatment interruption due to being LTFU, as they may be less dependent on anti-VEGF therapy.
In this multicenter study based on a national registry, we aimed to describe the visual and anatomical changes after restarting anti-VEGF therapy in patients who returned with stabilized vision after treatment discontinuation, and to identify predictors of clinical benefit upon retreatment in this subgroup.
METHODS
study design
This is a retrospective cohort study based on a national registry. This network was established by the National Clinical Research Center for Eye Diseases at Shanghai General Hospital in April 2020 to promote the standardized management of vitreoretinal diseases in China. By the end of 2024, this ongoing hospital-based consortium (Official website: http://pcqs.brightnesscenter.com/ ) had expanded to cover all 31 provinces of mainland China. Detailed longitudinal treatment data for each participant were recorded in the registry. An independent expert panel of clinicians and academics was set to ensure data quality. This multicenter study was approved through a centralized review process of Institutional Review Board (IRB) (identifier, SHGH-2024-232, Supplemental File 1) and adhered to the tenets of the Declaration of Helsinki. Electronic written informed consent was obtained from all participants.
study population
Patients with newly diagnosed DME between April 2020 and July 2023 were searched in the database. After intravitreal anti-VEGF injection with either 0.5 mg Ranibizumab, 2 mg Conbercept or 2 mg Aflibercept, patients were followed up every 1 to 3 months at the physician’s discretion. Being LTFU was defined as the absence of a documented physician encounter within 6 months after any intravitreal injection. Patients were stratified into two groups: (1) Patients who returned with stabilized VA (the first VA upon returning was equal to or better than the last recorded VA prior to being LTFU), and (2) patients who returned with vision decline (the first VA upon returning was worse than the last recorded VA prior to being LTFU).
Exclusion criteria were as follows: (1) a history of other retinal diseases causing macular edema, such as retinal vein occlusion (RVO), or neovascular age-related macular degeneration; (2) resolution of DME at return, defined as CFT < 250 μm on optical coherence tomography; (3) VA of 20/25 or better at return ; (4) follow-up of less than 6 months after restarting anti-VEGF therapy; (5) prior treatment at a different practice not included in this network; or (6) incomplete data collection for major parameters, eg, VA and CFT before and after retreatments. In total, 976 and 905 patients were included in the final analyses of visual and anatomical outcomes, respectively ( Figure 1 ).
Patient flow diagram. The boxes on the left side are the remaining patients, and the boxes on the right side are the patients excluded. CFT = central foveal thickness; CI-DME = center-involved diabetic macular edema; LTFU = lost to follow-up; OCT = optical coherence tomography; VA = visual acuity; VEGF = vascular endothelial growth factor.
data collection
Baseline characteristics, including age, sex (biological and physiological characteristics that define humans as male or female), and treatment region (East China or Midwest China), were recorded for analysis. Clinical data, consisting of eye involvement (unilateral or bilateral), VA, CFT, the number of injections before and after being LTFU, the anti-VEGF agents used, and history of retinal laser photocoagulation, were documented. For patients with bilateral disease, the eye with worse VA was selected for analysis.
VA was measured as best-corrected Snellen VA by optometrists without cycloplegia. CFT, from the inner retinal surface to the inner border of the retinal pigment epithelium, was measured with the caliper tool (Heidelberg Engineering).
The primary outcome was vision gain after resuming anti-VEGF therapy, defined as an improvement of ≥0.1 the logMAR as compared with VA at return. The secondary outcome was anatomical improvement after retreatment, defined as a ≥10% reduction in CFT compared with CFT at return.
Statistical analysis
Snellen VA was converted into logMAR VA. Counting fingers, hand motion, and light perception were assigned logMAR units of 2.1, 2.4, and 2.7, respectively. All statistical analyses were performed using SPSS software (version 29, SPSS Inc,). Frequency (percentage), mean ± SD, and median (IQR) were reported for the description of categorical variables and continuous variables with normal and skewed distribution, respectively. Comparisons of means, medians and proportions were made using the Student’s t test, the nonparametric Mann–Whitney U test and the chi-square test, respectively. One-sided Wilcoxon signed-rank test was used to assess the visual and anatomical improvement after retreatment. Multivariable logistic regression analysis was performed to investigate the impact of visual changes during treatment interruption on retreatment efficacy. Odds ratios (ORs) with 95% CIs (CIs) were calculated. A P value less than.05 was considered statistically significant.
RESULTS
descriptive statistics
Among the 976 patients who returned after being LTFU, 447 patients (45.8%) had stabilized VA, and 529 patients (54.2%) experienced VA decline after a median LTFU time of 259 days (IQR, 213-344 days; range, 182-942 days). The demographic and clinical characteristics stratified by visual status at return are summarized in Table 1 . Notably, patients with vision decline had larger proportion of being bilaterally involved (60.1% vs 51.0%, P =.004). Before being LTFU, the two groups received a similar number of injections (4.0 ± 3.1 injections vs 3.9 ± 3.0 injections, P =.557). Eyes returning with stabilized VA had worse VA (logMAR VA: 0.7 [0.4-1.0] vs 0.4 [0.2-0.6], P <.001), and more severe foveal thickening (321 [261-449] μm vs 305 [255-400] μm, P =.017) at the last visit before being LTFU. The median duration of being LTFU was 250 (209-314) days in the stabilized VA group and 272 (217-371) days in the vision decline group ( P <.001). At return, patients with stabilized VA exhibited better VA (logMAR VA: 0.5 [0.3-0.8] vs 0.8 [0.5-1.2], P <.001) and less foveal thickening (355 [289-465] μm vs 386 [300-543] μm, P =.001) than those with vision decline. After resuming anti-VEGF therapy, eyes with stabilized VA received a mean of 3.7 ± 2.2 injections, compared with 3.9 ± 2.2 injections in those with vision decline ( P =.179). At the final visit, patients with stabilized vision had better VA than those with vision decline; however, the difference did not reach statistical significance (logMAR VA: 0.6 [0.4-1.0] vs 0.7 [0.4-1.0], P =.050).
TABLE 1
Clinical Characteristics of Patients With Diabetic Macular Edema by the Visual Status After Being Lost to Follow-Up.
| Variables | Total | With Stabilized VA | With Decreased VA | P |
|---|---|---|---|---|
| Age at registration (y) ( n = 976) | 60 (53-67) | 61 (53-67) | 59 (53-66) | .253 |
| Sex ( n = 976) | .431 | |||
| Male | 491 (50.3) | 231 (51.7) | 260 (49.1) | |
| Female | 485 (49.7) | 216 (48.3) | 269 (50.9) | |
| Region ( n = 976) | .798 | |||
| East China | 428 (43.9) | 198 (44.3) | 230 (43.5) | |
| Midwest China | 548 (56.1) | 249 (55.7) | 299 (56.5) | |
| Eye involvement ( n = 976) | .004 | |||
| Unilateral | 430 (44.1) | 219 (49.0) | 211 (39.9) | |
| Bilateral | 546 (55.9) | 228 (51.0) | 318 (60.1) | |
| LogMAR VA at last visit before being LTFU ( n = 976) | 0.5 (0.3-0.8) | 0.7 (0.4-1.0) | 0.4 (0.2-0.6) | <.001 |
| Last VA before being LTFU ( n = 976) | <.001 | |||
| 20/40 or better | 296 (30.3) | 62 (13.9) | 234 (44.2) | |
| 20/50-20/200 | 551 (56.5) | 287 (64.2) | 264 (49.9) | |
| Worse than 20/200 | 129 (13.2) | 98 (21.9) | 31 (5.9) | |
| CFT at the last visit before being LTFU (μm) ( n = 912) | 310 (257-415) | 321 (261-449) | 305 (255-400) | .017 |
| No. of anti-VEGF injections before being LTFU ( n = 908) | 3.9 ± 3.1 | 4.0 ± 3.1 | 3.9 ± 3.0 | .557 |
| Last anti-VEGF agent used before being LTFU ( n = 924) | .250 | |||
| Ranibizumab | 430 (46.5) | 210 (49.4) | 220 (44.1) | |
| Conbercept | 299 (32.4) | 128 (30.1) | 171 (34.3) | |
| Aflibercept | 195 (21.1) | 87 (20.5) | 108 (21.6) | |
| Switching anti-VEGF drugs before being LTFU ( n = 948) | 112 (11.8) | 55 (12.5) | 57 (11.2) | .543 |
| Retinal photocoagulation before being LTFU ( n = 976) | 137 (14.0) | 59 (13.2) | 78 (14.7) | .489 |
| Time of being LTFU (d) ( n = 976) | 259 (213-344) | 250 (209-314) | 272 (217-371) | <.001 |
| LogMAR VA at return ( n = 976) | 0.7 (0.4-1.0) | 0.5 (0.3-0.8) | 0.8 (0.5-1.2) | <.001 |
| VA at return ( n = 976) | <.001 | |||
| 20/40 or better | 176 (18.0) | 125 (28.0) | 51 (9.6) | |
| 20/50-20/200 | 602 (61.7) | 278 (62.2) | 324 (61.2) | |
| Worse than 20/200 | 198 (20.3) | 44 (9.8) | 154 (29.1) | |
| CFT at return (μm) ( n = 976) | 368 (295-506) | 355 (289-465) | 386 (300-543) | .001 |
| No. of anti-VEGF injections after return ( n = 976) | 3.8 ± 2.2 | 3.7 ± 2.2 | 3.9 ± 2.2 | .179 |
| First anti-VEGF agent used after return ( n = 930) | .846 | |||
| Ranibizumab | 405 (43.5) | 189 (44.5) | 216 (42.8) | |
| Conbercept | 315 (33.9) | 143 (33.6) | 172 (34.1) | |
| Aflibercept | 210 (22.6) | 93 (21.9) | 117 (23.2) | |
| Last anti-VEGF agent used after return ( n = 907) | .964 | |||
| Ranibizumab | 340 (37.5) | 153 (37.6) | 187 (37.4) | |
| Conbercept | 318 (35.1) | 144 (35.4) | 174 (34.8) | |
| Aflibercept | 249 (27.5) | 110 (27.0) | 139 (27.8) | |
| Switching anti-VEGF drugs in resuming therapy ( n = 911) | 441 (48.4) | 197 (47.5) | 244 (49.2) | .604 |
| Retinal photocoagulation after return ( n = 976) | 152 (15.6) | 66 (14.8) | 86 (16.3) | .522 |
| LogMAR VA at final visit ( n = 976) | 0.6 (0.4-1.0) | 0.6 (0.4-1.0) | 0.7 (0.4-1.0) | .050 |
| VA at final visit ( n = 1475) | .059 | |||
| 20/40 or better | 206 (21.1) | 107 (23.9) | 99 (18.7) | |
| 20/50-20/200 | 578 (59.2) | 263 (58.8) | 315 (59.5) | |
| Worse than 20/200 | 192 (19.7) | 77 (17.2) | 115 (21.7) | |
| CFT at final visit (μm) ( n = 976) | 323 (260-434) | 326 (260-419) | 320 (260-450) | .955 |
Data presented as median (interquartile range) or mean ± standard deviation or n (%). East China included Shanghai, Beijing, Tianjin Province, Hebei Province, Jiangsu Province, Zhejiang Province, Fujian Province, Shandong Province, Guangdong Province, Hainan Province, Heilongjiang Province, Jilin Province, and Liaoning Province. Midwest China included Shanxi Province, Anhui Province, Jiangxi Province, Henan Province, Hubei Province, Hunan Province, Inner Mongolia Autonomous Region, Guangxi Autonomous Region, Chongqing, Sichuan Province, Guizhou Province, Yunnan Province, Tibet Autonomous Region, Shaanxi Province, Gansu Province, Qinghai Province, Ningxia Autonomous Region, and Xinjiang Autonomous Region.
primary outcome
With comparable follow-up periods (361 [257-518] days vs 356 [259-498] days, P =.590) after reinitiation of anti-VEGF therapy, vision gain was obtained in 32.7% of patients who returned with stabilized VA, which was remarkably lower than that in patients with vision decline (62.0%, P <.001, Table 2 ). This trend persisted in subgroup analyses after stratifying patients by VA at return as 20/40 or better (15.2% vs 43.1%, P <.001), 20/50-20/200 (37.8% vs 59.0%, P <.001), and worse than 20/200 (50.0% vs 74.7%, P =.002).
TABLE 2
Visual and Anatomical Outcomes After Resuming Anti-VEGF Therapy for Patients With Diabetic Macular Edema Who Were Lost to Follow-Up.
| Variables | Total | With Stabilized VA at Return | With Decreased VA at Return | P |
|---|---|---|---|---|
| With logMAR VA improved by ≥0.1 after resuming anti-VEGF therapy a ( n = 976) | 474 (48.6) | 146 (32.7) | 328 (62.0) | <.001 b |
| Stratified by VA at return | ||||
| 20/40 or better ( n = 176) | 41 (23.3) | 19 (15.2) | 22 (43.1) | <.001 b |
| 20/50-20/200 ( n = 602) | 296 (49.2) | 105 (37.8) | 191 (59.0) | <.001 b |
| Worse than 20/200 ( n = 198) | 137 (69.2) | 22 (50.0) | 115 (74.7) | .002 b |
| With CFT reduction by ≥10% after resuming anti-VEGF therapy a ( n = 905) | 544 (60.1) | 233 (56.4) | 311 (63.2) | .038 b |
| Stratified by CFT at return | ||||
| <350 μm ( n = 391) | 177 (45.3) | 89 (45.2) | 88 (45.4) | .971 |
| ≥350 μm ( n = 514) | 367 (71.4) | 144 (66.7) | 223 (74.3) | .043 b |
Data presented as n (%).
CFT = central foveal thickness; logMAR = logarithm of the minimum angle of resolution; VA = visual acuity; VEGF = vascular endothelial growth factor.
The demographic and clinical characteristics of patients with and without vision gain after retreatment are compared in Table 3 . Notably, patients who achieved vision gain had worse VA at return (logMAR VA: 0.8 [0.5-1.2] vs 0.5 [0.3-0.9], P <.001). During retreatments, these patients received more anti-VEGF injections (4.2 ± 2.5 vs 3.3 ± 1.9, P <.001). At the final visit, patients with vision gain exhibited better VA (logMAR VA: 0.5 [0.3-0.9] vs 0.7 [0.4-1.1], P <.001) and less thickening foveal (307 [256-411] μm vs 336 [270-459] μm, P =.001) than those without vision gain.
TABLE 3
Clinical Characteristics of Patients With Diabetic Macular Edema by the Visual Changes After Resuming Anti-VEGF Therapy.
| Variables | Total | With Vision Gain b | Without Vision Gain b | P |
|---|---|---|---|---|
| Age at registration (IQR) (y) ( n = 976) | 60 (53-67) | 60 (54-67) | 60 (53-67) | .745 |
| Sex ( n = 976) | .226 | |||
| Male | 491 (50.3) | 229 (48.3) | 262 (52.2) | |
| Female | 485 (49.7) | 245 (51.7) | 240 (47.8) | |
| Region ( n = 976) | .685 | |||
| East China | 428 (43.9) | 211 (44.5) | 217 (43.2) | |
| Midwest China | 548 (56.1) | 263 (55.5) | 285 (56.8) | |
| Eye involvement ( n = 976) | .780 | |||
| Unilateral | 430 (44.1) | 211 (44.5) | 219 (43.6) | |
| Bilateral | 546 (55.9) | 263 (55.5) | 283 (56.4) | |
| LogMAR VA at last visit before being LTFU ( n = 976) | 0.5 (0.3-0.8) | 0.5 (0.3-0.8) | 0.5 (0.3-0.8) | .740 |
| Last VA before being LTFU ( n = 976) | .905 | |||
| 20/40 or better | 296 (30.3) | 141 (29.7) | 155 (30.9) | |
| 20/50-20/200 | 551 (56.5) | 271 (57.2) | 280 (55.8) | |
| Worse than 20/200 | 129 (13.2) | 62 (13.1) | 67 (13.3) | |
| Last CFT before being LTFU (μm) ( n = 912) | 310 (257-415) | 314 (260-416) | 308 (256-413) | .461 |
| No. of anti-VEGF injections before being LTFU ( n = 908) | 3.9 ± 3.1 | 4.0 ± 3.4 | 3.9 ± 2.6 | .438 |
| Last anti-VEGF agent used before being LTFU ( n = 924) | .119 | |||
| Ranibizumab | 430 (46.5) | 191 (43.0) | 239 (49.8) | |
| Conbercept | 299 (32.4) | 153 (34.5) | 146 (30.4) | |
| Aflibercept | 195 (21.1) | 100 (22.5) | 95 (19.8) | |
| Switching anti-VEGF drugs before being LTFU ( n = 948) | 112 (11.8) | 57 (12.5) | 55 (11.2) | .513 |
| Retinal photocoagulation before being LTFU ( n = 976) | 137 (14.0) | 64 (13.5) | 73 (14.5) | .640 |
| Time of being LTFU (d) ( n = 976) | 259 (213-344) | 265 (215-363) | 255 (211-333) | .138 |
| LogMAR VA at return ( n = 976) | 0.7 (0.4-1.0) | 0.8 (0.5-1.2) | 0.5 (0.3-0.9) | <.001 a |
| VA at return ( n = 976) | <.001 a | |||
| 20/40 or better | 176 (18.0) | 41 (8.6) | 135 (26.9) | |
| 20/50-20/200 | 602 (61.7) | 296 (62.4) | 306 (61.0) | |
| Worse than 20/200 | 198 (20.3) | 137 (28.9) | 61 (12.2) | |
| CFT at return (μm) ( n = 976) | 368 (295-506) | 378 (300-523) | 361 (291-483) | .110 |
| No. of anti-VEGF injections after return ( n = 976) | 3.8 ± 2.2 | 4.2 ± 2.5 | 3.3 ± 1.9 | <.001 a |
| First anti-VEGF agent used after return ( n = 930) | .192 | |||
| Ranibizumab | 405 (43.5) | 203 (44.7) | 202 (42.4) | |
| Conbercept | 315 (33.9) | 160 (35.2) | 155 (32.6) | |
| Aflibercept | 210 (22.6) | 91 (20.0) | 119 (25.0) | |
| Last anti-VEGF agent used after return ( n = 907) | .505 | |||
| Ranibizumab | 340 (37.5) | 158 (36.0) | 182 (38.9) | |
| Conbercept | 318 (35.1) | 162 (36.9) | 156 (33.3) | |
| Aflibercept | 249 (27.5) | 119 (27.1) | 130 (27.8) | |
| Switching anti-VEGF drugs in resuming therapy ( n = 911) | 441 (48.4) | 226 (51.5) | 215 (45.6) | .074 |
| Retinal photocoagulation after return ( n = 976) | 152 (15.6) | 80 (16.9) | 72 (14.3) | .275 |
| LogMAR VA at final visit ( n = 976) | 0.6 (0.4-1.0) | 0.5 (0.3-0.9) | 0.7 (0.4-1.1) | <.001 a |
| VA at final visit ( n = 976) | <.001 a | |||
| 20/40 or better | 206 (21.1) | 124 (26.2) | 82 (16.3) | |
| 20/50-20/200 | 578 (59.2) | 285 (60.1) | 293 (58.4) | |
| Worse than 20/200 | 192 (19.7) | 65 (13.7) | 127 (25.3) | |
| CFT at final visit (μm) ( n = 976) | 323 (260-434) | 307 (256-411) | 336 (270-459) | .001 a |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
