Temporal changes in aqueous humor inflammatory cytokines of the contralateral eye after unilateral phacotrabeculectomy in rabbits Authors: Wangdu Luo, Min Tang, Yujie Rao, Lingling Li, Xiangji Li, Lin Xie Published: 2026 ABSTRACT AbstractTo determine whether unilateral phacotrabeculectomy induces detectable, time-dependent changes in the aqueous humor inflammatory cytokine profile of the contralateral unoperated eye, and to define the temporal fluctuation window. A unilateral phacotrabeculectomy model was established in New Zealand White rabbits. The contralateral unoperated eye was studied. Aqueous humor was collected at − 14 days preoperatively as baseline and serially at 3 days, 1 week, 2 weeks, 4 weeks, 8 weeks, and 12 weeks postoperatively from the same cohort (n = 10 rabbits). Concentrations of IL-6, IL-1β, TNF-α, MCP-1, MIP-1β, IP-10, IL-10, TGF-β2, and MMP-9 were quantified. Data are presented as mean ± SD. Each postoperative time point was compared with baseline using two-sided paired t-tests with Holm–Šidák correction for multiple comparisons. Multiple analytes in the contralateral aqueous humor showed a transient postoperative increase, predominantly in the early period. Proinflammatory and matrix remodeling–related markers (IL-6, TNF-α, IL-1β, and MMP-9) generally peaked at 3 days, and IL-6 remained significantly elevated through 2 weeks. The chemokines MCP-1 and MIP-1β most commonly peaked at 1 week; MCP-1 was significantly elevated primarily at 1 week, whereas MIP-1β showed significant upregulation from 3 days to 1 week. IP-10 peaked at 2 weeks and remained significantly elevated through 4 weeks. The immunoregulatory/repair-related cytokine IL-10 and the profibrotic factor TGF-β2 peaked at 2 weeks and remained elevated at selected time points through 4 weeks. Overall, most markers declined after the early postoperative rise, with some signals persisting to 4 weeks, and values approached stability/near-baseline levels by 8–12 weeks. Following unilateral phacotrabeculectomy, the contralateral unoperated eye showed modest, time-dependent fluctuations in aqueous humor cytokine levels, predominantly within 3 days to 2 weeks, with some signals extending to 4 weeks. Although the preliminary repeated-paracentesis control data suggest that serial aqueous humor sampling was unlikely to fully account for these patterns, subtle or very short-lived procedural effects cannot be completely excluded. These findings suggest that potential fellow-eye changes and their temporal window should be considered when designing and interpreting studies using early postoperative inflammation as an endpoint. Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1038/s41598-026-70988-3.Keywords: Phacotrabeculectomy, Contralateral unoperated eye, Aqueous humor, Inflammatory cytokines, ChemokinesSubject terms: Biomarkers, Diseases, Immunology, Medical research Introduction Glaucoma remains a leading cause of irreversible blindness worldwide1. Lowering intraocular pressure (IOP) is the most evidence-based strategy to slow glaucoma progression2. When medical therapy and laser interventions fail to achieve adequate control, surgical treatment becomes necessary. Trabeculectomy can provide robust IOP reduction by creating an alternative aqueous outflow pathway; however, its long-term success critically depends on postoperative wound healing. Excess perioperative inflammation followed by subconjunctival fibrosis can progressively restrict the filtration route, compromise bleb function, and ultimately lead to surgical failure, thereby undermining sustained IOP control3,4. Therefore, delineating the temporal dynamics of the anterior-segment immune–inflammatory microenvironment and identifying key time windows are essential for optimizing perioperative management and anti-scarring strategies4–7. In clinical practice, glaucoma frequently coexists with cataract, and phacotrabeculectomy is widely performed to improve visual function while controlling IOP. Previous studies have compared long-term outcomes and complication profiles between trabeculectomy alone and combined procedures, informing surgical decision-making8,9. Compared with filtration surgery alone, the combined procedure entails broader tissue manipulation and more complex perturbations of anterior-segment homeostasis, which may disrupt the blood–aqueous barrier and induce immune activation and repair processes with distinct timing and magnitude. However, prior work has largely focused on clinical endpoints (e.g., IOP reduction, bleb morphology, and complications)8,9, whereas comprehensive, multi-time-point longitudinal profiling of inflammatory mediators in aqueous humor—a biofluid that more directly reflects changes in the anterior-segment microenvironment—remains limited. This gap leaves several key questions unresolved, including when postoperative inflammatory peaks occur, how long they persist, when the response transitions from an “inflammation-dominant” to a “repair/remodeling-dominant” phase, and whether different molecular pathways act in a hierarchical, relay-like manner over time. Aqueous humor cytokine and chemokine profiles can sensitively reflect inflammatory initiation, immune-cell recruitment, and tissue repair, and may be linked to subsequent scarring propensity. Proinflammatory cytokines (e.g., IL-6, IL-1β, and TNF-α) contribute to inflammatory amplification and altered blood–aqueous barrier permeability; chemokines (e.g., MCP-1, MIP-1β, and IP-10) promote recruitment of monocytes/macrophages and other immune cells and can sustain local responses; the immunoregulatory cytokine IL-10 may restrain excessive inflammation via negative feedback; TGF-β2 serves as a central profibrotic signal driving fibroblast activation and extracellular matrix deposition; and matrix remodeling mediators such as MMP-9 regulate extracellular matrix degradation and reconstruction4–7. In recent years, molecular mechanisms and interventional strategies targeting fibrosis after glaucoma filtration surgery—including antifibrotic approaches and more refined concepts of wound modulation—have been systematically reviewed and summarized4–7,10. Moreover, clinical studies suggest that aqueous growth factor/cytokine levels (e.g., TGF-β–related signaling and chemokine axes) may be associated with filtration surgery outcomes, supporting the concept of using aqueous molecular profiles to identify risk windows and at-risk patients11–13. In other glaucoma procedures (e.g., drainage device surgery), aqueous mediators such as TGF-β2 and MCP-1 have also been reported to correlate with postoperative hypertensive phases or surgical outcomes14,15. Time-resolved monitoring of these mediators may therefore help to more clearly define key nodes and windows during the transition from inflammation to repair/remodeling after combined surgery, and provide finer-grained evidence to guide perioperative intervention strategies4–7,10–13. Another clinically relevant yet insufficiently characterized issue is whether unilateral surgery can induce detectable aqueous inflammatory fluctuations in the contralateral unoperated eye. Prior studies have reported that inflammatory mediators in the fellow-eye aqueous humor (e.g., MCP-1) can increase after first-eye cataract surgery16. Subsequent work has examined potential neuro–immune mechanisms underlying “fellow-eye pain/inflammation” and proposed the involvement of specific molecular pathways17. In certain clinical settings, first-eye surgery has also been associated with altered fellow-eye aqueous cytokine profiles within 1–2 weeks, suggesting that contralateral immune modulation may vary depending on the underlying ocular/systemic background18. Moreover, some studies suggest that aqueous inflammation may persist for an extended period following cataract surgery19, and preclinical animal studies further suggest that unilateral cataract surgery may perturb the stability of the contralateral intraocular microenvironment20. If such a “fellow-eye effect” truly exists, it would challenge the common assumption that the contralateral eye provides a stable within-subject control, and it could also inform the optimal interval for staged bilateral surgeries and the evaluation of perioperative strategies17,20. Nevertheless, systematic evidence regarding whether phacotrabeculectomy induces changes in the aqueous inflammatory profile of the contralateral unoperated eye—together with the magnitude and temporal window of such changes—remains limited and warrants longitudinal validation in a controlled animal model. Accordingly, we established a unilateral phacotrabeculectomy model in New Zealand White rabbits and focused on characterizing dynamic changes in the aqueous cytokine profile of the contralateral unoperated eye at baseline (2 weeks preoperatively) and multiple key postoperative time points (postoperative day 3 and weeks 1, 2, 4, 8, and 12). We quantified IL-6, IL-1β, TNF-α, MCP-1, MIP-1β, IP-10, IL-10, TGF-β2, and MMP-9. We hypothesized that the contralateral unoperated eye would exhibit a modest but quantifiable transient fluctuation predominantly within an early postoperative window, and that different mediators would display distinct peak timing, decay rates, and recovery trajectories, reflecting a hierarchical temporal pattern across multiple pathways. This study is expected to provide experimental evidence defining key time windows of fellow-eye immune microenvironment changes after combined surgery, and to inform more precise perioperative monitoring and staged-surgery strategies. Materials and methods Animals and ethics All animal procedures conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Institutional Animal Care and Use Committee of Chongqing Medical University (IACUC-CQMU-2025-0290). A total of 16 New Zealand White rabbits (2.0 ± 0.25 kg) were obtained from the Laboratory Animal Center of Chongqing Medical University, Chongqing, China. Ten rabbits were included in the main unilateral phacotrabeculectomy cohort, and six rabbits were included in a preliminary nonsurgical repeated-paracentesis control cohort established before the main experiment. The main surgical cohort included equal numbers of males and females. All animals were housed under specific pathogen-free conditions (22 ± 2 °C, 55% ± 10% humidity, 12-h light/dark cycle) with ad libitum access to food and water. The study is reported in accordance with the ARRIVE guidelines to enhance transparency and reproducibility. The sample size of the main surgical cohort (n = 10 rabbits) was determined based on commonly used sample sizes in prior longitudinal aqueous humor cytokine studies in rabbits and on study feasibility; no a priori power calculation was performed. The preliminary nonsurgical repeated-paracentesis control cohort included six rabbits and was established before the main experiment to evaluate the potential procedural effect of serial aqueous humor sampling. A within-animal paired, longitudinal repeated-measures design was used in the main surgical cohort: only the right eye underwent phacotrabeculectomy, whereas the left eye remained unoperated and served as the contralateral unoperated eye. To minimize procedural variation associated with repeated anterior chamber paracentesis, a fixed sampling volume (50 µL/eye) was used at all time points, and all sampling and assays were performed under standardized anesthesia and sample-handling protocols. Study design and sampling schedule To obtain a stable preoperative baseline and minimize procedure-related stress immediately before surgery, aqueous humor was collected from both eyes on day − 14 as baseline. Postoperative sampling time points were set at 3 days, 1 week, 2 weeks, 4 weeks, 8 weeks, and 12 weeks. At each time point, aqueous humor was collected concurrently from the operated right eye and the contralateral unoperated left eye. The sampling volume was fixed at 50 µL/eye per time point. Samples were immediately stored at − 80 °C until analysis. Serial sampling was performed at multiple postoperative time points in the same animals (repeated-measures design) to characterize the temporal dynamics of inflammatory mediators in the contralateral unoperated eye. Before the main experiment, a preliminary nonsurgical repeated-paracentesis control cohort was established to assess whether repeated anterior chamber paracentesis itself contributed to longitudinal changes in aqueous humor inflammatory mediators. Six New Zealand White rabbits underwent bilateral aqueous humor sampling using the same sampling volume, procedure, operator, anesthesia protocol, and time schedule as those used in the main surgical cohort. No ocular surgery was performed in this cohort. The results of this preliminary control experiment are presented in Supplementary Figure S1 and Supplementary Tables S1–S3. Phacotrabeculectomy All surgeries were performed under strict aseptic conditions by a single ophthalmic surgeon with 10 years of surgical experience. Antimetabolites such as mitomycin C or 5-fluorouracil were not used in this model because the aim of the study was to characterize the intrinsic aqueous humor inflammatory mediator response after unilateral phacotrabeculectomy without pharmacologic modulation of postoperative wound healing or inflammation. Phacoemulsification steps: A 2.8–3.0 mm clear corneal tunnel incision was created at the 11–1 o’clock position. A viscoelastic agent was injected to maintain the anterior chamber. Continuous curvilinear capsulorhexis (CCC) was performed, followed by hydrodissection and phacoemulsification/lens cortex removal. Residual cortex was removed by irrigation/aspiration (I/A). No intraocular lens was implanted. The viscoelastic was thoroughly removed and anterior chamber stability was maintained. Trabeculectomy steps: The conjunctiva was reflected at the same region and Tenon’s capsule was dissected. A square scleral flap (~ 3 × 3 mm; approximately one-third scleral thickness) was fashioned. A block of trabecular meshwork and adjacent scleral tissue was excised at the anterior edge of the flap to establish a filtration pathway; peripheral iridectomy was performed when necessary to reduce the risk of postoperative pupillary block. The scleral flap was secured with interrupted 10 − 0 nylon sutures, and the conjunctival incision was closed with continuous or interrupted sutures to ensure watertight closure without leakage. Anesthesia and perioperative management All surgical procedures and aqueous humor sampling procedures were conducted under strict aseptic conditions using general anesthesia combined with topical anesthesia. General anesthesia was induced by intravenous sodium pentobarbital (30 mg/kg) via the marginal ear vein. Anesthetic depth was assessed by eyelash reflex, corneal reflex, and nociceptive responses; supplemental pentobarbital, approximately 10–20% of the initial dose, was administered as needed to maintain a stable plane of anesthesia. Topical anesthesia was provided with 0.5% proparacaine eye drops. For phacotrabeculectomy, 1–2 drops were instilled into the operated eye and repeated once after a 1-min interval. For aqueous humor sampling, the same topical anesthetic regimen was applied to each eye undergoing anterior chamber paracentesis, including both eyes in the main surgical cohort and both eyes in the nonsurgical repeated-paracentesis control cohort. During surgery, animals were placed on a warming pad to maintain body temperature, and the corneal surface was intermittently moistened with sterile saline or artificial tears to prevent desiccation; viscoelastic was used when necessary to maintain anterior chamber depth and protect the corneal endothelium. Postoperatively, corneal clarity, anterior chamber depth, anterior chamber inflammation, and wound leakage were monitored daily for safety and animal welfare. Postoperative analgesia: Buprenorphine (0.05–0.1 mg/kg, subcutaneous) was administered every 6–12 h for 2–3 days to reduce pain-related stress. All animals in the main surgical cohort received the same analgesic regimen to minimize systemic bias. To avoid confounding effects on aqueous cytokine profiles, topical or systemic corticosteroids and nonsteroidal anti-inflammatory drugs were not routinely used postoperatively. Antibiotic prophylaxis: Levofloxacin eye drops were used throughout as prophylaxis. Immediately after baseline aqueous humor sampling on day − 14, both eyes received one drop, followed by QID administration for 24 h. After surgery, the operated eye received levofloxacin eye drops QID on postoperative days 0–7 and TID on days 8–14, after which treatment was discontinued. After each aqueous humor sampling procedure, one drop was instilled into each sampled eye as immediate prophylaxis. Short additional courses were permitted if paracentesis-related complications occurred. Euthanasia and endpoint management: At the study endpoint or upon reaching humane endpoints, rabbits were euthanized under deep anesthesia. After loss of corneal and pedal withdrawal reflexes, an overdose of sodium pentobarbital (≥ 100 mg/kg, intravenous) was administered via the marginal ear vein. Death was confirmed by cessation of respiration and heartbeat, with secondary confirmation performed as required by institutional guidelines. Aqueous humor collection and sample handling Under general anesthesia and topical anesthesia, aqueous humor was collected by gentle anterior chamber paracentesis using a 30-gauge needle attached to a syringe. A fixed volume of 50 µL/eye per time point was withdrawn to minimize procedural variability. For the operated eye, aqueous sampling was performed through the pre-existing corneal side-port incision created during the initial phacotrabeculectomy whenever feasible, rather than creating a new corneal entry. For the contralateral unoperated eye, the first sampling was performed via a standard clear corneal paracentesis at a designated location; subsequent samplings were consistently performed through the same paracentesis tract/location to minimize additional corneal microtrauma from repeated new punctures. At each sampling time point in the main surgical cohort, aqueous humor was collected from the contralateral unoperated left eye first, followed by the operated right eye, to minimize the possibility of carryover or cross-contamination from the surgically treated eye to the contralateral eye. Separate sterile needles and syringes were used for each eye, and samples from each eye were collected into individually labeled low-protein-binding tubes. Sampling was performed within a range that did not cause marked shallowing of the anterior chamber, and care was taken to avoid contact with the iris or other anterior segment structures. Paracentesis-related complications, including corneal abrasion, hyphema, wound leakage, and marked anterior chamber shallowing, were recorded. Samples with visible blood contamination were excluded for that eye at that time point and documented; excluded values were treated as missing and were not imputed. Statistical analyses were performed using available observations for each analyte (see Statistical analysis). Aqueous humor samples were collected into low-protein-binding tubes, placed immediately on ice, and transferred to − 80 °C within 30 min until assay. After each paracentesis, one drop of levofloxacin was instilled into each sampled eye as immediate prophylaxis. To reduce observer bias, samples were anonymized using coded identifiers prior to assay and statistical analysis; laboratory personnel and statisticians were masked to eye laterality and time points until completion of analyses. Paracentesis at each time point was performed at the same sampling location by the same operator under the same anesthesia protocol, and sampling was conducted as consistently as possible within a fixed time-of-day window to minimize procedural variability. The potential effect of repeated paracentesis was assessed in the preliminary nonsurgical control cohort described above, which was established before the main experiment and underwent the identical serial sampling protocol without phacotrabeculectomy. Cytokine quantification by ELISA Aqueous humor cytokines were quantified using commercial ELISA kits (Jiangsu Enzyme Immunoassay Industry Co., Ltd., Jiangsu, China) according to the manufacturer’s instructions. For TGF-β2, samples were activated prior to measurement (e.g., acid activation followed by neutralization) as specified by the kit protocol. The analyte panel included IL-6, IL-1β, TNF-α, MCP-1, MIP-1β, IP-10, IL-10, TGF-β2, and MMP-9. Standard curves and quality controls were included on each plate. Samples were assayed in duplicate and averaged for statistical analyses. Values below the limit of detection (LOD) were assigned as LOD/2 for analysis. To minimize batch and inter-plate effects, samples were assayed using the same kit lot whenever possible; if multiple lots were required, samples from different time points were randomly distributed across plates and inter-plate controls were included to assess consistency. Samples were anonymized before plating, and assay personnel were masked to eye laterality and time points. Statistical analysis All statistical analyses were performed using GraphPad Prism (v9.5.1). Data are presented as mean ± SD. Because aqueous humor was repeatedly collected from the same animals across multiple time points, the dataset had a longitudinal repeated-measures structure. Temporal changes in each cytokine in the contralateral unoperated eye were assessed using Prism’s multiple paired t-tests: each postoperative time point (3 days, 1, 2, 4, 8, and 12 weeks) was compared with the preoperative baseline (day − 14) using two-sided paired t-tests. Normality of paired differences (postoperative value − baseline value) was evaluated using the Shapiro–Wilk test; if normality was not met, the Wilcoxon signed-rank test was used as a nonparametric alternative. To control for family-wise type I error due to multiple time-point comparisons within each analyte, Holm–Šidák correction was applied across the six “vs baseline” comparisons. A corrected P value < 0.05 was considered statistically significant. For a summary visualization of overall fluctuations, mean time-course curves were plotted for each analyte and the area under the curve (AUC) from day − 14 to week 12 was calculated using the trapezoidal rule. AUC was used for descriptive visualization of overall fluctuation burden and was not treated as a primary inferential endpoint. For the preliminary nonsurgical repeated-paracentesis control cohort, each follow-up time point was compared with the day − 14 baseline separately for each eye, and paired interocular comparisons were performed at each time point. Holm–Šidák correction was applied within each comparison family. To provide a concise summary, the smallest Holm–Šidák-adjusted P value within each comparison family was reported for each mediator in Supplementary Table S1. Results Study design, sampling timeline, and surgical workflow A unilateral phacotrabeculectomy model was established in New Zealand White rabbits to evaluate time-dependent fluctuations in the aqueous humor inflammatory cytokine profile of the contralateral unoperated eye. As shown in Fig. 1A, the right eye underwent surgery, whereas the left eye served as the contralateral unoperated control eye. Aqueous humor was sampled at day − 14 before surgery and at 3 days, 1 week, 2 weeks, 4 weeks, 8 weeks, and 12 weeks after surgery in the same main surgical cohort (n = 10 rabbits; repeated-measures observations at each time point), allowing characterization of the temporal trajectory from the acute phase to the mid-to-late postoperative period. All 10 contralateral unoperated eyes and all 10 operated eyes contributed evaluable aqueous humor samples at every scheduled time point for all nine measured mediators. Thus, the number of missing or excluded values was zero for both the contralateral unoperated and operated eyes at each time point. No sample from either eye group was excluded because of visible blood contamination, sampling failure, paracentesis-associated complications, or other predefined quality-control criteria. Fig. 1B presents representative intraoperative images highlighting key steps of phacoemulsification and trabeculectomy. Study design, aqueous humor sampling timeline, and representative intraoperative images for the unilateral phacotrabeculectomy rabbit model. (A) Schematic overview of the unilateral surgical model and aqueous humor sampling schedule in New Zealand White rabbits. The right eye was the operated eye and underwent phacotrabeculectomy, whereas the left eye served as the contralateral unoperated eye. The yellow vertical line indicates the time of phacotrabeculectomy. Aqueous humor was collected at −14 days (baseline) and at 3 days, 1 week, 2 weeks, 4 weeks, 8 weeks, and 12 weeks postoperatively. Serial sampling was performed in the same main surgical cohort (n = 10 rabbits; repeated-measures observations at each time point), with a fixed collection volume of 50 μL per eye at each time point. (B) Representative intraoperative photographs illustrating key steps of phacoemulsification (left) and trabeculectomy (right). Repeated aqueous humor sampling alone did not reproduce the analyte-specific temporal profiles observed in the main cohort Before the main experiment, a preliminary nonsurgical repeated-paracentesis control cohort of six rabbits underwent bilateral repeated aqueous humor sampling using the same sampling volume, procedure, operator, anesthesia protocol, and complete time schedule as those used in the main surgical cohort. All six rabbits contributed evaluable aqueous humor samples from both eyes at every scheduled time point; no control-cohort sample was excluded because of visible blood contamination, sampling failure, paracentesis-associated complications, or other predefined quality-control criteria. No corneal abrasion, hyphema, wound leakage, marked anterior chamber shallowing, or other paracentesis-associated complication was observed in the nonsurgical control cohort. Thus, the control cohort incorporated both the initial day − 14 paracentesis and the progressively wider intervals between subsequent sampling time points. Under this identical schedule, the first post-baseline control sample was obtained 17 days after the initial day − 14 paracentesis, corresponding to postoperative day 3 in the main surgical cohort. Although no sample was collected exactly 14 days after the initial paracentesis, no statistically significant increase in any measured mediator was detected at this nearest available post-baseline assessment after Holm–Šidák correction. Across the complete control time course, no follow-up time point differed significantly from the day − 14 baseline in either eye after Holm–Šidák correction. Specifically, none of the measured proinflammatory cytokines, including IL-6, IL-1β, and TNF-α, showed a significant increase from baseline at any scheduled follow-up time point. Likewise, no significant left–right interocular difference was detected for these or the other measured mediators at any time point after multiplicity correction. Importantly, the control cohort did not reproduce the distinct analyte-specific temporal patterns observed in the main surgical cohort, such as the early peaks of IL-6, IL-1β, TNF-α, and MMP-9, the week-1 peaks of MCP-1 and MIP-1β, or the later peaks of IP-10, IL-10, and TGF-β2. The complete control time-course profiles are presented in Supplementary Figure S1, the corresponding multiplicity-adjusted comparisons are summarized in Supplementary Table S1, and the complete numerical concentrations for the left and right eyes at each scheduled time point are provided in Supplementary Tables S2 and S3, respectively. These control data suggest that repeated paracentesis or the serial sampling schedule alone was unlikely to fully account for the structured postoperative mediator trajectories observed in the main cohort; however, subtle or very short-lived procedure-related effects cannot be completely excluded. Transient elevations in proinflammatory and chemotactic mediators in the contralateral eye are concentrated in the early postoperative period Proinflammatory cytokines and chemokines in the aqueous humor of the contralateral unoperated eye showed a transient, detectable postoperative upregulation (Fig. 2), with changes predominantly occurring early after surgery and subsequently declining. Statistical comparisons with baseline (day − 14) are summarized in Table 1. As noted in Sect. "Study design, sampling timeline, and surgical workflow", no contralateral-eye values were missing or excluded from the baseline-referenced analyses presented in Figs. 2 and 3; Table 1, and no paracentesis-associated complication was observed in the contralateral unoperated eyes during the serial sampling period. Therefore, subgroup comparison based on paracentesis-associated complications was not applicable. Among proinflammatory cytokines, IL-6, IL-1β, and TNF-α all peaked in the early postoperative phase. IL-6 most commonly peaked at 3 days and remained significantly elevated through 2 weeks; in contrast, increases in IL-1β and TNF-α were more short-lived, with significant differences largely confined to 3 days (Table 1). Time-course changes in proinflammatory cytokines and chemokines in the aqueous humor of the operated and contralateral unoperated eyes. Trends in aqueous humor concentrations of proinflammatory cytokines and chemokines in the contralateral unoperated left eye and the operated right eye at −14 days preoperatively (baseline) and at 3 days, 1 week, 2 weeks, 4 weeks, 8 weeks, and 12 weeks postoperatively. Panels show (A) IL-6, (B) IL-1β, (C) TNF-α, (D) MCP-1, (E) MIP-1β, and (F) IP-10. Data are presented as mean ± SD (n = 10 rabbits; repeated-measures observations across time within the same animals). Blue asterisks indicate time points at which the contralateral unoperated eye showed a significant increase from its own day −14 baseline after Holm–Šidák correction (*P < 0.05). Detailed statistical comparisons are summarized in Table 1. Table 1Peak time points and significant elevation windows relative to the preoperative baseline in aqueous humor inflammatory mediators from the contralateral unoperated eye (mean ± SD).AnalytePeak time pointPeak (mean ± SD)Significantly elevated time points (vs. day − 14; Holm–Šidák–adjusted P < 0.05)IL-6 (pg/ml)3d219.72 ± 29.24 3 d, 1w, 2wIL-1β (pg/ml)3d103.17 ± 15.953dTNF-α (pg/ml)3d1240.47 ± 199.973dMCP-1 (pg/ml)1w56.85 ± 8.601wMIP-1β (pg/ml)1w7.91 ± 0.66 3 d, 1wIP-10 (pg/ml)2w307.12 ± 30.211w, 2w, 4wIL-10 (pg/ml)2w63.52 ± 7.421w, 2w, 4wTGF-β2 (pg/ml)2w6086.95 ± 451.252w, 4wMMP-9 (pg/ml)3d13416.19 ± 1678.133d Peak time points and significant elevation windows relative to the preoperative baseline in aqueous humor inflammatory mediators from the contralateral unoperated eye (mean ± SD). Time-course changes in immunoregulatory, profibrotic, and matrix-remodeling mediators in the aqueous humor of the operated and contralateral unoperated eyes. Aqueous humor concentrations were measured at day −14 preoperatively (baseline) and at 3 days, 1 week, 2 weeks, 4 weeks, 8 weeks, and 12 weeks postoperatively. Panels (A–C) show comparative time-course curves for the contralateral unoperated left eye and the operated right eye: (A) IL-10, (B) TGF-β2, and (C) MMP-9. Panels (D–F) show the corresponding contralateral-eye profiles on expanded y-axis scales: (D) IL-10, (E) TGF-β2, and (F) MMP-9. Data are presented as mean ± SD (n = 10 rabbits; repeated-measures observations across time within the same animals). Blue asterisks indicate time points at which the contralateral unoperated eye showed a significant increase from its own day −14 baseline after Holm–Šidák correction (*P < 0.05). Detailed statistical comparisons are summarized in Table 1. Among chemokines, MCP-1 peaked at 1 week, and its significant elevation was observed primarily at 1 week; MIP-1β showed a smaller magnitude of change but remained significantly upregulated early (predominantly from 3 days to 1 week). By comparison, IP-10 exhibited a later peak at 2 weeks, with significant elevation persisting through 4 weeks (Table 1). Collectively, Fig. 2; Table 1 indicate that proinflammatory and chemotactic mediators in the contralateral eye fluctuate mainly in the early postoperative window (predominantly 3 days to 2 weeks), with some signals extending to 4 weeks, followed by a gradual return toward baseline and stabilization at later time points. Stage-dependent fluctuations in immunoregulatory, profibrotic, and matrix remodeling mediators In contrast to the early peaks observed for proinflammatory cytokines and chemokines, immunoregulatory and tissue repair–related mediators exhibited more clearly stage-dependent patterns (Fig. 3). Statistical comparisons versus baseline (day − 14) are summarized in Table 1. IL-10 increased after surgery and most commonly peaked at 2 weeks; significant elevations were mainly observed at 1 and 2 weeks and extended to 4 weeks (Table 1). The profibrotic mediator TGF-β2 also typically peaked at 2 weeks, with significant increases primarily at 2 and 4 weeks (Table 1), indicating a relatively mid-phase response. In contrast, the matrix remodeling marker MMP-9 showed an earlier peak pattern, most commonly peaking at 3 days, with significant elevation largely confined to 3 days (Table 1). Taken together, Fig. 3; Table 1 indicate a stage-dependent postoperative profile in the contralateral unoperated eye: early changes (3 days) were characterized by more prominent fluctuations in matrix remodeling–related signals, whereas mid-phase responses (1–4 weeks) were dominated by immunoregulatory and profibrotic signals, followed by gradual decline and stabilization at later time points. For a descriptive overview of the cumulative fluctuation burden across the follow-up period, AUCs of the mean time-course curves for all nine mediators are summarized in Fig. 4. Summary of overall inflammatory burden based on the area under the mean time-course curve (AUC).To provide a compressed overview of the longitudinal fluctuations in nine aqueous humor inflammation-related mediators over the follow-up period, mean time-course curves were constructed using the group mean concentration at each time point for the operated eye after unilateral phacotrabeculectomy and the contralateral unoperated eye. The area under each mean curve (AUC) from day−14 to week 12 was calculated using the trapezoidal rule and is expressed as concentration × time (days). This figure is intended for descriptive visualization of cumulative fluctuation burden. Formal statistical inference is reported for the time-course analyses and baseline-referenced comparisons in Table 1 (see Statistical Analysis). This table summarizes the temporal characteristics of nine inflammation-related mediators in the aqueous humor of the contralateral unoperated (left) eye across the follow-up period, including the peak postoperative time point, the corresponding peak value (mean ± SD), and postoperative time points that remained significantly elevated versus the preoperative baseline (day − 14) after multiplicity correction. For statistical inference, paired comparisons (two-sided paired t-tests) were performed between each postoperative time point and baseline within the same cohort (n = 10 rabbits), and Holm–Šidák correction was applied across the six “vs baseline” comparisons per analyte (3 days, 1, 2, 4, 8, and 12 weeks). Time points listed in the final column denote those with Holm–Šidák–adjusted P < 0.05. Summary of peak timing and baseline-referenced significant windows supports a transient fluctuation window To facilitate an integrated interpretation of temporal changes in the aqueous humor mediator profile of the contralateral unoperated eye, we summarized the peak postoperative time point for each analyte and the postoperative time points that remained significantly elevated relative to baseline (day − 14) after multiple-comparison correction (Table 1). Significance was assessed using the longitudinal repeated-measures data from the same cohort, with each postoperative time point compared against the day − 14 baseline and multiplicity controlled using the Holm–Šidák procedure (adjusted P < 0.05; see Statistical Analysis). This summary indicates that peak responses occurred predominantly in the early postoperative period. Proinflammatory and matrix-remodeling–related mediators (IL-6, IL-1β, TNF-α, and MMP-9) most commonly peaked at day 3, whereas chemokine-related mediators (MCP-1 and MIP-1β) typically peaked at week 1. In contrast, IP-10, IL-10, and TGF-β2 peaked later, most often at week 2, and for some analytes, baseline-referenced elevation extended to week 4 (Table 1). Discussion In this New Zealand White rabbit model of unilateral phacotrabeculectomy, we systematically characterized time-dependent changes in the aqueous humor inflammatory mediator profile of the contralateral unoperated eye. The main findings were as follows: although the contralateral eye did not undergo any surgical manipulation, multiple inflammatory/chemotactic and repair-related mediators showed a transient, modest but detectable postoperative increase. These changes were concentrated within 3 days to 2 weeks after surgery, with significant elevations for some analytes extending to 4 weeks, followed by a gradual decline and stabilization/return toward near-baseline levels thereafter. In addition, the mediators exhibited clear temporal stratification: proinflammatory and matrix remodeling–related markers most commonly peaked early (3 days), chemokines tended to peak at 1 or 2 weeks, and immunoregulatory and profibrotic signals (IL-10 and TGF-β2) were more prominent in the mid phase (2–4 weeks). Collectively, these findings suggest that unilateral surgery may trigger a degree of bilateral immune coupling, with important implications for experimental designs and interpretation when the contralateral eye is used as a within-subject control16–20. Evidence for a “transient fluctuation window” in the contralateral eye: linking trend plots to significant windows The contralateral aqueous mediator fluctuations were mutually supported by both the descriptive trends (Figs. 2 and 3) and the statistical summaries (Table 1). Among proinflammatory cytokines, IL-6 increased markedly at 3 days and remained significantly elevated through 2 weeks, suggesting a relatively sustained acute proinflammatory window in the contralateral eye. In contrast, TNF-α and IL-1β showed more short-lived responses, with significant differences largely confined to 3 days, consistent with a rapid initiation and resolution of acute stress/injury signaling. Among chemokines, MCP-1 and MIP-1β increased early and most commonly peaked at 1 week; MCP-1 showed significant elevation primarily at 1 week, whereas MIP-1β remained significantly upregulated early (predominantly 3 days to 1 week). Notably, IP-10 peaked later at 2 weeks and remained significantly elevated through 4 weeks, indicating a longer tail for selected chemotactic signals. For immunoregulatory and tissue repair–related mediators, IL-10 peaked at 2 weeks and remained significantly elevated through 4 weeks, suggesting engagement of negative-feedback regulation following the acute proinflammatory fluctuation. TGF-β2 likewise peaked at 2 weeks and remained significantly elevated at 2–4 weeks, consistent with a more prominent mid-phase repair/profibrotic response. In contrast, the matrix remodeling marker MMP-9 peaked at 3 days, with significance largely confined to the early phase, in keeping with an “inflammation–remodeling initiation” profile. Overall, viewed through the “peak and significant-window” framework, these findings are consistent with a transient fluctuation window in the contralateral unoperated eye after unilateral surgery, centered approximately between day 3 and week 2, with selected signals persisting to week 4. The structured timing across multiple analytes supports the possibility that the fellow-eye changes reflect a coordinated biological response rather than random variation, although the modest magnitude of these changes and the remaining procedural limitations warrant cautious interpretation16–20. Potential biological underpinnings: systemic inflammatory dissemination and neuro–immune coupling Although the magnitude of mediator upregulation in the contralateral aqueous humor was modest overall, the changes followed a distinct temporal pattern: the early phase was dominated by proinflammatory (IL-6, TNF-α, and IL-1β) and remodeling (MMP-9) signals, whereas the mid phase increasingly reflected chemotactic and regulatory/repair signals (MCP-1, IP-10, IL-10, and TGF-β2). This “inflammation first, regulation next, then repair” pattern is consistent with general wound-repair biology and aligns with established mechanistic frameworks for wound modulation and fibrosis after glaucoma filtration surgery4–7,10. Previous studies have also suggested that contralateral-eye inflammatory responses may occur after unilateral ocular surgery. Clinical reports described altered aqueous inflammatory mediator profiles in the fellow eye after first-eye cataract surgery, including increased MCP-1 and TGF-β2 under specific conditions16,18, and more recent work further identified cytokine and metabolomic changes in the contralateral eye after unilateral cataract surgery20. In parallel, fellow-eye pain/inflammatory responses during sequential cataract surgery have been linked to sympathetic and cytokine-mediated neuro–immune mechanisms17. In addition, postoperative aqueous inflammatory activity may persist longer than often assumed19. Together, these findings support the concept that the fellow eye may undergo biologically meaningful postoperative changes even without direct surgical manipulation. Several mechanisms may plausibly contribute to this phenomenon. One possibility is that local inflammatory mediator release after unilateral surgery partially disseminates systemically, leading to a mild and transient increase in inflammatory mediators in the contralateral eye. Another possibility is neurogenic modulation, whereby ocular injury and inflammation activate corneal trigeminal pathways and ciliary/autonomic circuits, thereby altering immune homeostasis bilaterally. In addition, neuro–immune interactions between ocular immunity and sensory/pain pathways contribute to regulation of the ocular surface and anterior-segment microenvironment21,22, and the concept of an “inflammatory reflex” mediated by the autonomic nervous system provides a broader framework for linking local stress to systemic or bilateral immune modulation23. Although the present study was not designed to dissect the relative contribution of each pathway, the hierarchical timing pattern observed here—an acute peak at 3 days, a chemotactic/regulatory peak at 1–2 weeks, and persistence of selected signals to 4 weeks followed by decline—is broadly consistent with the literature above16–20. Compared with these earlier studies, our work extends the existing evidence by providing a controlled longitudinal characterization of nine aqueous humor inflammatory/remodeling-related mediators in the contralateral unoperated eye after unilateral phacotrabeculectomy. Methodological and design implications: boundaries of using the contralateral eye as a control and selection of key sampling time points The contralateral eye is often considered an ideal within-subject control; however, our findings indicate that it may not represent a fully stable background in the early postoperative period. For studies in which postoperative inflammation is a primary endpoint, relying exclusively on the contralateral eye as a control within the 3-day to 2-week window may lead to misclassification of coupled contralateral fluctuations as background variation, potentially underestimating true operated-eye differences or treatment effects. Accordingly, when feasible, study designs should consider: (1) inclusion of an external nonoperated-animal control group to distinguish a fellow-eye effect from a local operated-eye effect; (2) explicit reporting of whether contralateral fluctuations occur rather than assuming a constant baseline; and (3) prioritizing high-information sampling time points (3 days, 1 week, and 2 weeks), with 4 weeks retained to capture tail effects/return, and 8–12 weeks used to confirm longer-term stability16–20. Moreover, although the longitudinal serial-sampling design enabled characterization of within-animal temporal dynamics, repeated anterior chamber paracentesis and aqueous humor removal may themselves transiently perturb the anterior-segment microenvironment. Previous studies have shown that aqueocentesis can disrupt the blood–aqueous barrier and induce acute inflammatory responses24,25. To directly evaluate this potential confounder, we established a preliminary nonsurgical repeated-paracentesis control cohort before the main experiment. Six rabbits underwent bilateral aqueous humor sampling using the same sampling volume, procedure, operator, anesthesia protocol, and complete sampling schedule as those used in the main surgical cohort, including the initial day − 14 collection and the progressively wider intervals between subsequent samplings. No measured mediator showed a statistically significant change from the day − 14 baseline in either eye, and no significant interocular difference was detected at any time point after Holm–Šidák correction (Supplementary Figure S1 and Supplementary Table S1). In particular, IL-6, IL-1β, and TNF-α showed no significant increase at any scheduled follow-up time point. Moreover, the nonsurgical control cohort did not reproduce the analyte-specific sequence observed after unilateral surgery, including early proinflammatory and matrix-remodeling peaks, intermediate chemokine peaks, and later immunoregulatory/profibrotic peaks. Because the control animals were exposed to the same serial sampling intervals, the gradual widening of the sampling intervals alone is unlikely to explain the temporal decline observed in the main cohort. These findings suggest that repeated paracentesis was unlikely to be the primary driver of the coordinated, mediator-specific postoperative trajectories observed in the main cohort. However, the control cohort did not include a sample collected exactly 14 days after the initial day − 14 paracentesis. Under the identical study schedule, the nearest post-baseline assessment was performed 17 days after the initial procedure, corresponding to postoperative day 3 in the main cohort. At this time point, no measured mediator showed a statistically significant increase from baseline after Holm–Šidák correction. Nevertheless, although the day − 14 sample represented the initial aqueous humor concentration before that sample was withdrawn, no separate sample was collected within hours after the procedure. Therefore, an immediate or very short-lived response to the initial or subsequent paracenteses cannot be directly assessed or completely excluded. Complete control concentration–time profiles, statistical comparisons, and numerical concentration data are provided in Supplementary Figure S1 and Supplementary Tables S1–S3. Potential clinical implications: staged surgery and the concept of a postoperative “window period” in the immune microenvironment Although this is an animal study, the identified contralateral fluctuation window may have implications for clinical scenarios involving staged bilateral surgery. Following unilateral surgery, the fellow eye may experience a mild alteration in its immune microenvironment; performing second-eye surgery or introducing additional stimuli within a short interval could theoretically compound inflammatory risk or complicate perioperative assessment. Based on the present observation that most mediators fluctuate early after surgery, with some signals persisting to 4 weeks before declining, the early 3-day to 2-week period and the potential tail phase extending to 4 weeks may represent potentially relevant time windows for future studies examining staged bilateral surgery. However, the present animal data do not establish an optimal interval for second-eye surgery16–20. Importantly, these inferences require further clinical validation, and the magnitude and duration of fellow-eye responses likely vary with surgical modality, baseline inflammatory status, and perioperative medications. Limitations This study has several limitations. First, although the present analyte panel included representative proinflammatory, chemotactic, immunoregulatory, profibrotic, and matrix-remodeling mediators, it was not exhaustive; therefore, other inflammatory or repair-related mediators may also contribute to the contralateral-eye response after unilateral phacotrabeculectomy. Second, we focused on aqueous humor mediator profiles and did not concurrently assess serum or peripheral immune indices or immune-cell signatures in contralateral ocular tissues; therefore, the relative contributions of systemic dissemination and neurogenic mechanisms cannot be directly determined. Third, repeated anterior chamber paracentesis and aqueous humor removal may potentially perturb the anterior-segment microenvironment, as previous studies have shown that aqueocentesis can transiently disrupt the blood–aqueous barrier and induce inflammatory responses24,25. Although the preliminary nonsurgical repeated-paracentesis control cohort showed no statistically significant longitudinal changes from baseline in either eye and no significant interocular differences after Holm–Šidák correction, its relatively small sample size limits the ability to exclude subtle procedure-related effects. In addition, although the day − 14 sample provided the initial baseline concentration before that sample was withdrawn, no separate post-paracentesis sample was collected shortly thereafter, and no control sample was obtained exactly 14 days after the initial paracentesis. The nearest post-baseline control assessment was performed 17 days after the initial procedure and showed no statistically significant increase in any measured mediator after Holm–Šidák correction. Nevertheless, an immediate inflammatory response to the initial or subsequent paracenteses could not be directly assessed, and very short-lived procedure-related responses may not have been captured. Repeated sampling also carries a small risk of procedure-related complications and missing observations, although standardized sampling conditions were used to minimize these effects. Fourth, although several procedural precautions were implemented to minimize potential inter-eye sample carryover, inadvertent cross-contamination originating from the operated phacotrabeculectomy eye during aqueous humor sampling, sample transfer, handling, or downstream processing cannot be completely excluded. Specifically, aqueous humor from the contralateral unoperated eye was collected before that from the operated eye at each time point, and separate sterile needles, syringes, and individually labeled low-protein-binding tubes were used for the two eyes. These procedures were intended to minimize the possibility of cross-contamination; however, they cannot absolutely eliminate the possibility of low-level inadvertent carryover during the overall sampling and processing workflow. Given that the cytokine changes detected in the contralateral eye were generally modest, even minimal contamination could theoretically influence the measured mediator concentrations. Therefore, this potential source of bias should be considered when interpreting the observed fellow-eye cytokine fluctuations. Fifth, antimetabolites such as mitomycin C or 5-fluorouracil were not used in this rabbit phacotrabeculectomy model. This design was chosen to avoid pharmacologic modulation of postoperative wound healing and inflammatory mediator profiles, thereby allowing characterization of the intrinsic surgery-associated aqueous humor cytokine response. However, antimetabolites are commonly used in clinical trabeculectomy or phacotrabeculectomy to reduce subconjunctival fibrosis, and they may alter the magnitude, duration, or composition of postoperative inflammatory responses. Therefore, the present findings may not fully reflect cytokine dynamics under antimetabolite-assisted clinical surgical conditions. Future studies incorporating antimetabolite-treated surgical models will be needed to determine how wound-modulating agents influence contralateral-eye inflammatory mediator profiles. Sixth, although the 12-week follow-up allowed us to define the principal postoperative fluctuation window and showed that most analytes approached near-baseline levels by 8–12 weeks, this time frame may still represent a subacute stage rather than a fully resolved long-term state for all mediators. Longer-term follow-up will therefore be necessary to determine whether subtle inflammatory or remodeling-related alterations persist beyond 12 weeks and to better define the complete resolution profile of the contralateral-eye response. Finally, we did not correlate contralateral mediator changes with phenotypic endpoints such as anterior chamber inflammation grading or intraocular pressure. Integrated phenotype–molecule analyses, together with imaging and tissue-level validation, may further enhance the biological interpretability of these findings. Conclusions In a New Zealand White rabbit model of unilateral phacotrabeculectomy, we observed modest, time-dependent fluctuations in the aqueous humor inflammatory mediator profile of the contralateral unoperated eye. The primary window was concentrated within 3 days to 2 weeks after surgery, with significant elevations of some mediators persisting to 4 weeks, followed by gradual decline and stabilization/return toward near-baseline levels by 8–12 weeks. Most proinflammatory and matrix remodeling–related mediators peaked early (predominantly at 3 days), whereas chemokines and immunoregulatory/repair-related signals were more prominent at 1–2 weeks. These findings suggest that the contralateral eye may not always provide a completely stable background during the early postoperative period, and potential fellow-eye changes and their temporal window should be considered when designing and interpreting studies that use early postoperative inflammation as a primary endpoint. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Acknowledgements We thank Professor Chao Wang’s team at the College of Basic Medicine, Chongqing Medical University, for providing experimental facilities and technical support. Author contributions W.L. wrote the manuscript and performed the experiments. M.T. performed the experiments, revised the manuscript, and contributed to data interpretation. Y.R. performed the experiments and revised the manuscript. L.L. performed the statistical analysis. X.L. established the animal model. L.X. provided experimental guidance and supervised the study. All authors reviewed and approved the final manuscript. Funding This work was supported by the National Natural Science Foundation of China (No. 82571207); the Chongqing Natural Science Foundation—Innovation Development Joint Project (CSTB2024NSCQ-LZX0023); the Internal Incubation Project of the Third Affiliated Hospital of Chongqing Medical University (No. KY23047); the 2026 Doctoral Research Start-up Project (No. 26BSQD28); the Chongqing Young Talent Program (cstc2022ycjh-bgzxm0163); the Chongqing High-Level Young and Middle-Aged Medical Talent Program; the Chongqing Returned Overseas Scholars Innovation and Entrepreneurship Project (cx2023019); and the Hunan Provincial Natural Science Foundation of China (2023JJ70048). Data availability The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request. Declarations Competing interests The authors declare no competing interests. Ethics approval and consent to participate All experimental procedures complied with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The animal protocol was approved by the Institutional Animal Care and Use Committee of Chongqing Medical University (IACUC-CQMU-2025-0290) and conducted in accordance with institutional guidelines. Consent for publication Not applicable (this study did not involve human participants). Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Wangdu Luo, Min Tang and Yujie Rao contributed equally to this work. Contributor Information Xiangji Li, Email: lixiangji@hospital.cqmu.edu.cn. Lin Xie, Email: xielin@hospital.cqmu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1 Data Availability Statement The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.