Serum microRNA-34a as a circulating biomarker for age-related macular degeneration and short-term anti-VEGF response in neovascular AMD Authors: Meng Kong, Rong Jin, Tongjie Cheng, Li Sun, Nianting Tong Published: 2026 ABSTRACT ObjectiveTo evaluate whether serum microRNA-34a (miR-34a) is associated with age-related macular degeneration (AMD) and whether baseline serum miR-34a has the potential to predict short-term response to anti-vascular endothelial growth factor (anti-VEGF) therapy in patients with neovascular AMD (nAMD).MethodsIn a cross-sectional cohort recruited at Qingdao Municipal Hospital between January and December 2024, serum miR-34a was quantified in 160 AMD patients and 122 age- and sex-matched healthy controls. AMD was staged according to Age-related Eye Disease Study (AREDS) criteria. Patients with nAMD received three consecutive monthly intravitreal anti-VEGF injections during the loading phase and were classified as good or poor responders according to prespecified anatomical and functional criteria based on optical coherence tomography (OCT) and best-corrected visual acuity (BCVA) after the loading phase. Peripheral blood samples were collected, and serum miR-34a expression was quantified using standard qPCR protocols.ResultsSerum miR-34a was significantly higher in patients with AMD than in controls (p < 0.001). Receiver operating characteristic (ROC) analysis yielded an area under the curve (AUC) of 0.740 (95% CI 0.681–0.799; p < 0.0001) for discriminating AMD from controls. Across AMD subgroups, serum miR-34a differed significantly overall; patients with geographic atrophy (GA) exhibited higher miR-34a levels than those with nAMD (p < 0.05). Among patients with nAMD, baseline serum miR-34a was remarkably higher in good responders than in poor responders after the loading phase (p < 0.001). ROC analysis showed an AUC of 0.772 (95% CI 0.687–0.857; p < 0.0001) for predicting treatment response, and multivariable logistic regression identified higher serum miR-34a as independently associated with lower odds of poor response (OR 0.143, 95% CI 0.042–0.489; p = 0.002).ConclusionElevated serum miR-34a is associated with AMD and may serve as a minimally invasive biomarker for distinguishing AMD from controls. In nAMD, baseline serum miR-34a also showed potential for predicting short-term response to loading-phase anti-VEGF therapy. These findings warrant validation in larger, longitudinal, and preferably multicenter cohorts. Introduction Age-related macular degeneration (AMD) is a leading cause of irreversible central vision loss in older adults worldwide [1, 2]. Clinically, AMD is characterized by progressive dysfunction and degeneration of the retinal pigment epithelium (RPE), photoreceptors, and adjacent outer retina in the macular region. Advanced AMD is broadly classified into geographic atrophy (GA) and neovascular AMD (nAMD), marked by choroidal neovascularization and exudative lesions [3]. Although intravitreal anti–vascular endothelial growth factor (anti-VEGF) therapy has substantially improved the management of nAMD, therapeutic response varies considerably among patients, and a meaningful proportion of eyes show suboptimal anatomical and/or functional improvement after the loading phase [4]. Therefore, clinically accessible biomarkers that may assist in AMD detection and help stratify treatment response in nAMD remain of considerable interest. AMD is a multifactorial age-related retinal disorder involving aging, oxidative stress, mitochondrial dysfunction, chronic inflammation, and dysregulation of complement activation [5–8]. Among these mechanisms, oxidative stress–induced RPE injury is considered a central pathogenic event, as it promotes cellular senescence, mitochondrial damage, and impaired retinal homeostasis. Increasing evidence indicates that microRNAs (miRNAs) participate in these stress-related processes and may represent both mechanistic mediators and clinically informative circulating molecules in AMD [9–12]. MicroRNA-34a (miR-34a) is a well-recognized regulator of aging, oxidative stress, and apoptosis. In our previous studies, we showed that miR-34a is involved in oxidative stress–related premature senescence in RPE cells and that suppression of miR-34a alleviates oxidative injury [13]. We further demonstrated that miR-34a directly targets SIRT1 and activates the p66Shc pathway, thereby reducing the tolerance of senescent RPE cells to oxidative stress [14]. These findings support a mechanistic link between miR-34a and key pathogenic events relevant to AMD. Because circulating miRNAs are relatively stable in peripheral blood and have shown biomarker potential in ocular and neurodegenerative disorders, serum miR-34a may provide a minimally invasive indicator of AMD-associated molecular alterations [15, 16]. In addition, given the heterogeneity of anti-VEGF responsiveness in nAMD, it is clinically relevant to explore whether baseline circulating miR-34a is associated not only with AMD status but also with short-term treatment response after the loading phase. Therefore, the present study aimed to evaluate serum miR-34a expression in patients with AMD and healthy controls, and to investigate its potential association with short-term anti-VEGF response in nAMD. By linking prior mechanistic evidence with human serum data, this study sought to assess the translational relevance of miR-34a as a circulating biomarker in AMD. Materials and methods Ethical approval This study was conducted in accordance with the tenets of the Declaration of Helsinki and was approved by the Medical Ethics Committee of Qingdao Municipal Hospital. Written informed consent was obtained from all participants prior to enrollment. Study design and participants A cross-sectional, observational study was performed at the Department of Ophthalmology, Qingdao Municipal Hospital, between January 2024 and December 2024. A total of 160 patients diagnosed with AMD and 122 age- and sex-matched healthy controls were consecutively recruited. AMD diagnosis and staging were determined based on the Age-Related Eye Disease Study (AREDS) classification system [17]. According to fundus examination and optical coherence tomography (OCT) findings, patients were categorized into early AMD (n = 12), intermediate AMD (n = 16), geographic atrophy (GA) (n = 6), and neovascular AMD (nAMD) (n = 126) subgroups. Treatment response was assessed 1 month after completion of the initial three-monthly loading injections, in accordance with the primary response time point proposed by Amoaku et al. [18]. Anti-VEGF agents used during the loading phase included ranibizumab, aflibercept, and conbercept. The overall agent distribution was ranibizumab (n = 31, 24.6%), aflibercept (n = 52, 41.3%), and conbercept (n = 43, 34.1%). Treatment assignment was not randomized; the choice of agent was determined by routine clinical practice, including physician judgment, drug availability, reimbursement or affordability considerations, and patient preference. Patients were classified using a literature-informed composite assessment based on both anatomical and functional outcomes. Good responders were defined as eyes showing resolution or marked reduction of exudative activity on OCT, including intraretinal fluid (IRF), subretinal fluid (SRF), and retinal thickening, and/or improvement in best-corrected visual acuity (BCVA) of more than 5 ETDRS letters, taking baseline visual acuity into account. Poor responders were defined as eyes showing less than 25% reduction in central retinal thickness from baseline, with persistent or new IRF/SRF and minimal functional improvement (0–4 ETDRS letters), or worsening features consistent with non-response. These criteria were adapted from the consensus definitions proposed for anti-VEGF response in neovascular AMD. For the purposes of the present analysis, eyes meeting criteria for poor response or non-response were grouped together as poor responders. Therefore, among the nAMD group in the present study, 82 patients were classified as good responders and 44 as poor responders to anti-VEGF therapy. Exclusion criteria included: (i) history of malignant tumors or autoimmune diseases; (ii) chronic neurodegenerative disorders such as Alzheimer’s or Parkinson’s disease; (iii) other severe ocular conditions (e.g., uveitis, endophthalmitis, retinal detachment); and (iv) systemic diseases affecting circulation or oxidative status (e.g., uncontrolled diabetes mellitus, severe heart failure). All participants underwent comprehensive ophthalmologic examination, including slit-lamp biomicroscopy, fundus photography, fluorescein angiography (when applicable), and OCT imaging. Demographic and clinical data (age, sex, disease duration, systemic history, and current medication) were recorded. Serum sample collection and processing Peripheral venous blood (4–5 mL) was collected from each participant using serum separation tubes after overnight fasting. Within 4 hours of collection, samples were centrifuged at 3000 rpm (≈1000 × g) for 15 minutes at 4 °C to separate the serum. The supernatant was carefully aspirated, aliquoted into RNase-free microcentrifuge tubes, and stored at − 80 °C until RNA extraction. Hemolyzed samples were excluded from analysis to ensure RNA integrity. Serum RNA isolation Total RNA, including small RNAs, was extracted from 200 µL of serum using the miRcute Serum/Plasma miRNA Isolation Kit (spin column type) (Tiangen Biotech, Beijing, China), following the manufacturer’s protocol. RNA concentration and purity were determined using a NanoDrop™ 8000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Samples with A260/A280 ratios between 1.8 and 2.1 were used for subsequent analysis. Quantitative real-time reverse transcription PCR (qRT-PCR) Reverse transcription was performed using the miRcute Plus miRNA First-Strand cDNA Kit (Tiangen Biotech, China). Quantitative PCR was then carried out with the miRcute Plus miRNA qPCR Kit (SYBR Green) (Tiangen Biotech, China) on a QuantStudio™ 3 Real-Time PCR System (Thermo Fisher Scientific, USA). Specific primers used were as follows: miR-34a forward: 5′-CGCGTGGCAGTGTCTTAGCT-3′; U6 snRNA forward: 5′-CTCGCTTCGGCAGCACA-3′. The universal reverse primer was supplied in the qPCR kit. Each reaction was performed in triplicate under the following cycling conditions: 95 °C for 15 min, followed by 40 cycles of 94 °C for 20 s and 60 °C for 34 s. U6 small nuclear RNA served as the internal reference due to its stable Ct distribution across study cohorts in our dataset. Relative miRNA expression was calculated using the 2^−ΔΔCt method. Statistical analysis All statistical analyses were performed using SPSS version 27.0 (IBM Corp., Armonk, NY, USA) and GraphPad Prism 10.0 (GraphPad Software, San Diego, CA, USA). Continuous variables were expressed as mean ± standard deviation (SD) for normally distributed data or median (interquartile range) for non-normally distributed data. Differences between two groups were assessed using the Student’s t-test or Mann–Whitney U test, while comparisons among three or more groups were analyzed by one-way analysis of variance (ANOVA) or Kruskal–Wallis H test, as appropriate. When the Kruskal–Wallis test was significant, Dunn–Bonferroni post hoc pairwise comparisons were performed. Trend across ordered AMD stages was additionally explored using the Jonckheere–Terpstra test. Categorical variables were analyzed using the chi-square test or Fisher’s exact test. Correlation between continuous variables was assessed using Spearman’s rank correlation coefficient. To identify independent predictors of poor anti-VEGF treatment response in nAMD, multivariate logistic regression analysis was performed after adjustment for relevant covariates, including age, sex, baseline BCVA, and disease duration. Diagnostic performance of serum miR-34a was evaluated using receiver operating characteristic (ROC) curve analysis, and the area under the curve (AUC) with 95% confidence interval (CI) was calculated. All statistical tests were two-tailed, and p < 0.05 was considered statistically significant. Results Baseline characteristics A total of 282 participants were included in this study, comprising 160 patients with age-related macular degeneration (AMD) and 122 age- and sex-matched healthy controls. As summarized in Table 1, no significant differences were observed between the two groups regarding age, gender, hypertension, coronary heart disease, diabetes, smoking, or drinking history (all p > 0.05).Table 1Baseline characteristics of the recruited participants in the studyVariableAMD groupControl groupp valueParticipant (n)160122NAAge (years)73.34 ± 8.1673.41 ± 8.120.946Gender (M/F)94/6666/560.435Hypertension(n)78600.943Coronary heart disease(n)50360.753Diabetes(n)36300.681Smoking(n)970.968Drinking(n)23140.475 Table 1Baseline characteristics of the recruited participants in the studyVariableAMD groupControl groupp valueParticipant (n)160122NAAge (years)73.34 ± 8.1673.41 ± 8.120.946Gender (M/F)94/6666/560.435Hypertension(n)78600.943Coronary heart disease(n)50360.753Diabetes(n)36300.681Smoking(n)970.968Drinking(n)23140.475 Baseline characteristics of the recruited participants in the study Correlation of serum miR-34a with age Spearman’s correlation analysis showed that serum miR-34a levels were not significantly correlated with age in the AMD cohort (r=−0.068, p=0.395), whereas a weak positive correlation was observed in the control group (r=0.216, p=0.017) (Fig. 1A)Fig. 1Serum miR-34a in healthy controls and patients with AMD. (A) correlation between serum miR-34a levels and age in the AMD and control groups, assessed using Spearman’s rank. (B) relative serum miR-34a expression in AMD and control groups. Data are presented as median (interquartile range). (C) comparison of serum miR-34a among AMD subgroups (early AMD, intermediate AMD, GA, and nAMD). Statistical significance was assessed using the Kruskal–Wallis test followed by Dunn–Bonferroni post hoc analysis. (D) ROC curve showing the discriminatory performance of serum miR-34a for AMD versus control status (AUC = 0.740, 95% CI 0.681–0.799; p < 0.0001). The solid line represents the ROC curve and the shaded area indicates the 95% confidence band Serum miR-34a in healthy controls and patients with AMD. (A) correlation between serum miR-34a levels and age in the AMD and control groups, assessed using Spearman’s rank. (B) relative serum miR-34a expression in AMD and control groups. Data are presented as median (interquartile range). (C) comparison of serum miR-34a among AMD subgroups (early AMD, intermediate AMD, GA, and nAMD). Statistical significance was assessed using the Kruskal–Wallis test followed by Dunn–Bonferroni post hoc analysis. (D) ROC curve showing the discriminatory performance of serum miR-34a for AMD versus control status (AUC = 0.740, 95% CI 0.681–0.799; p < 0.0001). The solid line represents the ROC curve and the shaded area indicates the 95% confidence band Relative expression of serum miR-34a Serum miR-34a levels were markedly elevated in AMD patients compared with controls (median 1.458 vs. 0.116; p < 0.001) (Fig. 1B). When stratified by disease stage, the Kruskal–Wallis test showed an overall difference in serum miR-34a among early AMD, intermediate AMD, GA, and nAMD subgroups (Kruskal–Wallis H = 10.99, p = 0.012). Dunn–Bonferroni post hoc analysis indicated that serum miR-34a was higher in GA than in nAMD (adjusted p = 0.013), whereas other pairwise comparisons were not significant (all adjusted p > 0.05). Although serum miR-34a appeared to increase progressively from early to intermediate and to GA, the Jonckheere–Terpstra trend test did not reach statistical significance (p = 0.08). Given the small size of the GA subgroup, this stage-related observation should be interpreted cautiously (Fig. 1C). Diagnostic accuracy of serum miR-34a for AMD ROC curve analysis demonstrated an AUC of 0.740 (95% CI 0.681–0.799; p < 0.0001), indicating moderate discrimination of AMD from healthy controls by serum miR-34a (Fig. 1D). Baseline characteristics of patients with neovascular AMD Among 126 patients with nAMD who completed the loading phase of anti-VEGF therapy, 82 (65.1%) were classified as good responders and 44 (34.9%) as poor responders according to the predefined OCT- and BCVA-based criteria. The anti-VEGF agents administered were ranibizumab (n = 31, 24.6%), aflibercept (n = 52, 41.3%), and conbercept (n = 43, 34.1%). The distribution of anti-VEGF agents was 22/35/25 in good responders and 9/17/18 in poor responders, respectively, with no statistically significant between-group difference (chi-square test, p = 0.473). As summarized in Table 2, there were no significant differences between the two groups in baseline demographic characteristics, systemic comorbidities, or ocular imaging parameters, including the presence of polypoidal choroidal vasculopathy (PCV), intraretinal fluid (IRF), subretinal fluid (SRF), pigment epithelial detachment (PED), subretinal hyperreflective material (SHRM), and baseline BCVA (all p > 0.05).Table 2Baseline characteristics for the patients with neovascular AMDVariableGood responded groupPoor responded groupp valueAge (years)71.52 ± 7.6773.66 ± 10.030.185Gender (M/F)44/3821/230.525Hypertension(n)42220.896Coronary heart disease(n)17100.795Diabetes(n)28120.429Smoking(n)720.493Drinking(n)1360.740PCV(n)941.000IRF, mm20.00 (0.00,0.08)0.00 (0.00,0.75)0.698SRF, mm20.00 (0.00,0.13)0.02 (0.00,0.10)0.515PED, mm234.00 (0.00,89.25)58.00 (0.00,100.50)0.525SHRM, mm20.12 (0.02,0.15)0.07 (0.01,0.16)0.172BCVA, logMAR0.70 (0.40,0.94)0.59 (0.52,0.82)0.861 Table 2Baseline characteristics for the patients with neovascular AMDVariableGood responded groupPoor responded groupp valueAge (years)71.52 ± 7.6773.66 ± 10.030.185Gender (M/F)44/3821/230.525Hypertension(n)42220.896Coronary heart disease(n)17100.795Diabetes(n)28120.429Smoking(n)720.493Drinking(n)1360.740PCV(n)941.000IRF, mm20.00 (0.00,0.08)0.00 (0.00,0.75)0.698SRF, mm20.00 (0.00,0.13)0.02 (0.00,0.10)0.515PED, mm234.00 (0.00,89.25)58.00 (0.00,100.50)0.525SHRM, mm20.12 (0.02,0.15)0.07 (0.01,0.16)0.172BCVA, logMAR0.70 (0.40,0.94)0.59 (0.52,0.82)0.861 Baseline characteristics for the patients with neovascular AMD Relative expression of serum miR-34a and prediction in nAMD Baseline serum miR-34a levels were significantly higher in the good response group (median 1.938) than in the poor response group (median 0.353; Mann–Whitney U = 826; p < 0.001) (Fig. 2A). In multivariable logistic regression, higher serum miR-34a was independently associated with lower odds of poor response (OR 0.143; 95% CI 0.042–0.489; p = 0.002), while IRF, SRF, PED, SHRM, and BCVA were not significant predictors (Table 3). ROC analysis for baseline serum miR-34a yielded an AUC of 0.772 (95% CI 0.687–0.857; p < 0.0001) for discriminating good from poor responders after the loading phase (Fig. 2B).Fig. 2Baseline serum miR-34a and short-term anti-VEGF response in nAMD. (A) relative serum miR-34a expression in the good-response and poor-response groups after the loading phase of anti-VEGF therapy. Data are shown as median (interquartile range). (B) ROC curve for baseline serum miR-34a in discriminating good from poor responders after the loading phase (AUC = 0.772, 95% CI 0.687–0.857; p < 0.0001). The solid line represents the ROC curve and the shaded area indicates the 95% confidence bandTable 3Multivariable logistic regression for poor anti-VEGF response (dependent variable: 1 = poor, 0 = good)βStd.errorWaldp valueOR95%CIIRF, mm2−0.4700.4301.1940.2740.6250.269–1.452SRF, mm20.7260.4672.4150.1202.0660.827–5.159SHRM, mm2−0.1270.2820.2020.6530.8810.506–1.532PED, mm20.1160.2430.2280.6331.1230.697–1.809BCVA, logMAR0.0820.2610.1000.7521.0860.651–1.812miR-34a−1.9420.6269.6200.0020.1430.042–0.489 Baseline serum miR-34a and short-term anti-VEGF response in nAMD. (A) relative serum miR-34a expression in the good-response and poor-response groups after the loading phase of anti-VEGF therapy. Data are shown as median (interquartile range). (B) ROC curve for baseline serum miR-34a in discriminating good from poor responders after the loading phase (AUC = 0.772, 95% CI 0.687–0.857; p < 0.0001). The solid line represents the ROC curve and the shaded area indicates the 95% confidence band Table 3Multivariable logistic regression for poor anti-VEGF response (dependent variable: 1 = poor, 0 = good)βStd.errorWaldp valueOR95%CIIRF, mm2−0.4700.4301.1940.2740.6250.269–1.452SRF, mm20.7260.4672.4150.1202.0660.827–5.159SHRM, mm2−0.1270.2820.2020.6530.8810.506–1.532PED, mm20.1160.2430.2280.6331.1230.697–1.809BCVA, logMAR0.0820.2610.1000.7521.0860.651–1.812miR-34a−1.9420.6269.6200.0020.1430.042–0.489 Multivariable logistic regression for poor anti-VEGF response (dependent variable: 1 = poor, 0 = good) Discussion AMD is a complex age-related retinal disorder driven by oxidative stress, mitochondrial dysfunction, inflammation, and senescence-related signaling, all of which contribute to marked biological and clinical heterogeneity [6, 19]. In nAMD, although intravitreal anti-VEGF therapy has substantially improved visual outcomes, a considerable proportion of patients show suboptimal anatomical and/or functional improvement after the loading phase [20]. Therefore, minimally invasive biomarkers that may assist in identifying AMD and in stratifying short-term treatment responsiveness remain of substantial clinical interest. Among microRNAs implicated in aging-related retinal injury, miR-34a has emerged as a plausible candidate because of its established role in cellular senescence, oxidative stress, and apoptosis. In our previous studies, miR-34a promoted oxidative stress-induced premature senescence in retinal pigment epithelium (RPE) cells and aggravated mitochondrial injury by targeting SIRT1 and activating the p66Shc pathway [13, 14]. These findings suggest that miR-34a is mechanistically linked to pathogenic processes relevant to AMD and provide a biological rationale for evaluating circulating miR-34a in a clinical setting [6, 13, 14, 19, 20]. In the present study, serum miR-34a levels were significantly elevated in patients with AMD compared with controls, supporting an association between circulating miR-34a and AMD status. The observed diagnostic AUC of 0.740 indicates moderate discriminatory performance, suggesting that serum miR-34a may have potential as a minimally invasive biomarker for distinguishing AMD from unaffected individuals [21, 22]. Notably, we also observed differences among AMD subgroups, with higher serum miR-34a in GA than in nAMD. However, this finding should be interpreted cautiously, because the study was cross-sectional and the GA subgroup was very small. Prior work supports an anti-angiogenic role of miR-34a via repression of Notch1/HIF-1α-VEGF signaling, which could help explain the relatively lower levels in neovascular phenotypes [23, 24]. A clinically relevant aspect of this study was the evaluation of short-term anti-VEGF response in nAMD. We found that baseline serum miR-34a levels were significantly higher in good responders than in poor responders after the loading phase, and that serum miR-34a remained independently associated with lower odds of poor response in multivariable analysis. These findings suggest that circulating miR-34a may provide useful information beyond baseline clinical examination alone. Mechanistically, this association may reflect differences in the underlying oxidative stress- and senescence-related molecular milieu, or possibly variation in angiogenesis-related signaling linked to miR-34a activity; however, these interpretations remain hypothetical and require further validation in prospective translational studies [23, 24]. These hypotheses merit prospective validation integrating intraocular and circulating readouts [24]. Classical OCT biomarkers remain important in the assessment of nAMD, but they do not fully capture the molecular heterogeneity that may underlie treatment variability. In our cohort, baseline IRF, SRF, SHRM, and PED were not independently associated with poor response, which further supports the potential complementary value of circulating molecular biomarkers. Recent clinical datasets indicate that persistent IRF is generally linked to poorer vision, whereas SRF can associate with equal or better outcomes under treat-and-extend paradigms—yet these imaging markers show modest certainty and are context-dependent [25]. Thus, rather than replacing structural imaging, serum miR-34a may eventually contribute to a multimodal approach to response stratification in nAMD. From a translational perspective, our findings extend prior mechanistic observations on the miR-34a–SIRT1/p66Shc axis to a clinically measurable circulating biomarker. Circulating miRNAs are detectible and relatively stable in serum and therefore are attractive candidates for minimally invasive biomarker development, provided that pre-analytical and analytical conditions are carefully standardized [21, 26–28]. Together, these data support serum miR-34a may represent a biologically informed biomarker linking AMD-associated stress pathways with clinically accessible readouts. Several limitations should be acknowledged. First, this was a single-center, cross-sectional study, which may limit causal inference and preclude assessment of longitudinal disease progression. Second, the subgroup distribution was markedly imbalanced, particularly for GA, thereby limiting the robustness of subtype comparisons. Third, treatment response was assessed after the initial loading phase, and longer-term outcomes were not evaluated. Fourth, patients received different anti-VEGF agents, including ranibizumab, aflibercept, and conbercept, and treatment assignment was not randomized but followed routine clinical practice; therefore, treatment heterogeneity and potential selection bias may have introduced residual confounding. Finally, although serum miR-34a was associated with AMD and short-term anti-VEGF response, the present study does not establish whether circulating miR-34a directly reflects intraocular molecular activity. Future multicenter studies with larger and more balanced cohorts, standardized treatment protocols, and longitudinal follow-up are needed to validate these findings. In summary, serum miR-34a is elevated in patients with AMD and is independently associated with a favorable anti-VEGF response in nAMD. Together with prior mechanistic evidence implicating miR-34a in oxidative stress-related RPE senescence and injury, these findings support the potential translational relevance of serum miR-34a as a minimally invasive circulating biomarker in AMD.