Aqueous Humor Cytokine Profiles in Congenital Cataracts With Posterior Polar Abnormality and Associations With Postoperative Complications

Abstract

Purpose

To characterize the aqueous humor (AH) cytokine profiles across clinical subtypes of congenital cataracts with posterior polar abnormality (PPA) and to assess their associations with postoperative complications.

Methods

This prospective study included 96 eyes with congenital cataracts divided into a PPA group (n = 78) and non-PPA controls (n = 18). The PPA group was stratified into four clinical subtypes: congenital cataract with persistent fetal vasculature (PFV, n = 27), posterior polar cataract (PPC; n = 16), posterior lenticonus (PL; n = 17), and posterior capsule defect (PCD; n = 18). Baseline demographics and biometrics were collected. Thirteen cytokines in AH were measured using Luminex xMAP technology. Postoperative complications, including visual axis opacification (VAO) and glaucoma-related adverse events, were evaluated with a minimum follow-up of 12 months.

Results

The PPA group showed differences in several cytokines compared with non-PPA controls, including higher vascular endothelial growth factor A (VEGF-A), fibroblast growth factor 2 (FGF2), and neurotrophin-4 (NT-4) and lower fibroblast growth factor 1 (FGF1) and platelet-derived growth factor-AA (PDGF-AA) (all q < 0.05, false discovery rate adjusted). Cytokine patterns varied across PPA subtypes, with substantial overlap. PFV showed higher levels of several markers, whereas PPC and PL showed lower levels. In exploratory Firth logistic regression, higher NT-4 (odds ratio [OR] = 1.16; 95% confidence interval [CI], 1.02–1.36; P = 0.028) and PDGF-AA (OR = 1.06; 95% CI, 1.01–1.12; P = 0.027) were associated with postoperative VAO.

Conclusions

PPA showed AH cytokine differences compared with non-PPA controls. Across PPA subtypes, cytokine distributions showed substantial overlap despite exploratory differences in several markers. Higher NT-4 and PDGF-AA were associated with postoperative VAO in exploratory analyses and warrant further validation.

Congenital cataract is a primary contributor to childhood visual impairment worldwide, with an estimated 200,000 children affected by cataract-related blindness annually.^1,2 This condition imposes a heavy burden, particularly in developing countries, where it contributes significantly to human morbidity, economic loss, and social challenges.³ Although surgical techniques have continuously advanced, the ultimate visual outcomes remain largely dependent on the presence of concurrent ocular structural abnormalities, particularly those affecting the posterior pole of the lens.^4–6

Congenital cataract with posterior polar abnormality (PPA) not only significantly increases the complexity and risks of surgery but is also closely associated with the occurrence of severe postoperative complications such as visual axis opacification (VAO) and glaucoma-related adverse events, posing a persistent threat to the long-term visual function of affected children.^7–9 Clinically, congenital cataract with PPA exhibits diverse manifestations, including persistent fetal vasculature (PFV), posterior polar cataract (PPC), posterior lenticonus (PL), and posterior capsule defect (PCD). PFV is widely recognized as a failed regression of the hyaloid vascular system.¹⁰ PPCs are opacities of the subcapsular cortex in the polar regions of the lens,¹¹ and PL manifests a localized protrusion of the posterior lens capsule.¹² PCD is diagnosed as a congenital posterior capsule discontinuity, usually presenting intraoperatively as a well-circumscribed defect with thickened margins and whitish granules.¹³ Previous studies have often discussed these subtypes as separate disease entities; however, a large-scale clinical study proposed that the failure of regression of the hyaloid vascular system may be the initial event that leads to all these abnormalities.¹⁴ Whether such clinical relationships are accompanied by differences in aqueous humor (AH) cytokine profiles remains unclear. Specifically, it is unknown whether congenital cataracts with PPA differ from those with normal posterior poles and whether distinct cytokine patterns are present across PPA subtypes.

To investigate the cytokine profiles associated with PPA, the AH offers a unique window into the eye's microenvironment. As it directly interfaces with intraocular tissues, its cytokine profile serves as a sensitive, real-time indicator of inflammation, tissue remodeling, and other key biological functions.¹⁵ Altered AH cytokine expression is well established as a contributor to the pathogenesis and progression of ocular diseases. Furthermore, Luminex multiplex technology enables multiplex cytokine quantification from ≤50 µL of fluid, which has been widely used in serial monitoring and therapeutic targeting recently.¹⁶ Recent evidence suggests that congenital cataracts associated with PCD may compromise the intraocular barrier, as inflammatory mediators such as monocyte chemoattractant protein-1 (MCP-1) and transforming growth factor-β2 (TGF-β2) are elevated in the AH, reflecting a localized inflammatory response.¹⁷ Given that abnormal posterior capsule development is a key feature of PPA, we hypothesized that PPA may be associated with alterations in the AH cytokine profile beyond the observed structural abnormalities. Furthermore, PPA is a known risk factor for postoperative complications such as VAO and glaucoma^8,9,14; however, whether preoperative cytokine profiles in the AH are associated with these outcomes remains unexplored.

In summary, this study aimed to characterize AH cytokine profiles in congenital cataracts with PPA, to explore variation across clinical subtypes, and to assess associations between preoperative cytokine levels and postoperative complications, including VAO and glaucoma-related events. Figure 1 shows the proposed approach in a graphical abstract.

Figure 1.

Graphical abstract of the study workflow, illustrating patient enrollment, AH collection, data collection and classification, and postoperative outcome assessment. PCD, posterior capsule defect; PFV, persistent fetal vasculature; PL, posterior lenticonus; PPA, posterior polar abnormality; PPC, posterior polar cataract; ZOC, Zhongshan Ophthalmic Center.

Methods

Ethics Statement

This study adhered to the tenets of the Declaration of Helsinki and was approved by the Institutional Review Board of Zhongshan Ophthalmic Center (2022KYPG099-6). As the participants were all minors, written informed consent was obtained from a parent or legal guardian.

Participants

This prospective observational study enrolled 96 pediatric patients with congenital cataracts between October 2019 and June 2024 at Zhongshan Ophthalmic Center. All pediatric patients underwent lens aspiration combined with limited anterior vitrectomy; primary intraocular lens (IOL) implantation was performed at the surgeon's discretion, based on age, ocular biometry, and intraoperative findings. Clinical subtypes were determined based on preoperative examinations and confirmed by intraoperative video recordings, which were independently reviewed by two experienced congenital cataract surgeons (HC, WRC).

Diagnostic Criteria and Group Definitions

Congenital cataract with PPA was defined as congenital cataract accompanied by a congenital structural abnormality involving the posterior lens pole, posterior capsule, or retrolental space, classified on standardized preoperative assessment and confirmed intraoperatively.^18–21 Congenital cataract with PFV was diagnosed and classified according to the classic classification frameworks proposed by Pollard¹⁸ and Goldberg.¹⁹ Cases included in this study were predominantly anterior PFV, with minor posterior involvement permitted.²²

PPC was defined as a central posterior subcapsular plaque or disc-like opacity at the posterior pole with well-circumscribed margins.^20,23 PL was defined as a localized protrusion of the posterior lens capsule.⁸ PCD was defined as a pre-existing posterior capsule defect, confirmed intraoperatively.²¹ Non-PPA controls were defined as congenital cataract eyes without evidence of posterior polar abnormality on standardized preoperative assessment and intraoperative confirmation. The final group assignment was adjudicated intraoperatively by microscopic inspection of the posterior capsule and retrolental space. The same exclusion criteria were applied to both groups, as described above, to minimize ocular or systemic conditions that may alter aqueous cytokine profiles or compromise postoperative assessment. AH sampling and processing followed the standardized protocol described in the following AH Sample Collection section and were applied identically across groups. Perioperative management and the surgical workflow were standardized across groups, and all surgeries were performed using the same operative approach by the same experienced surgeon(WRC).

Exclusion criteria included any condition that may alter the aqueous cytokine profiles or compromise postoperative assessment, including but not limited to ocular infection, steroid-related cataract, congenital glaucoma, uveitis, isolated posterior PFV, PFV cases with clear and significant posterior segment involvement, optic-nerve or fundus abnormalities, retinopathy of prematurity, previous intraocular surgery, history of ocular trauma, recognized syndromic conditions or systemic disorders known to be associated with congenital cataracts, or follow-up shorter than 12 months.

If both eyes met the inclusion criteria, one eye was randomly chosen for the analysis. The selection was made by consulting a random number table, where odd numbers corresponded to the left eye and even numbers to the right eye, following the approach of Meng et al.²⁴ All seven bilateral patients exhibited identical subtypes in both eyes, so random selection did not distort the subtype distribution.

AH Sample Collection

All surgical interventions were conducted under general anesthesia by an experienced cataract surgeon (WRC). Before surgery, the corneal and conjunctival surfaces were thoroughly disinfected with povidone-iodine. AH (100–150 µL) was collected under sterile conditions via anterior-chamber paracentesis with a 27-gauge needle attached to a 1-mL sterile syringe before any intraocular invasive procedure. All AH samples were immediately sealed in Eppendorf tubes and stored at −80°C until cytokine analysis.

Measurement of Cytokine Levels

We selected 13 cytokines, including angiogenic factors (vascular endothelial growth factor A [VEGF-A], placental growth factor [PlGF], angiopoietin-1, fibroblast growth factor 1 [FGF1], fibroblast growth factor 2 [FGF2], and platelet-derived growth factor-AA [PDGF-AA]); neurotrophins (brain-derived neurotrophic factor [BDNF], glial cell line-derived neurotrophic factor [GDNF], neurotrophin-4 [NT-4], and β-nerve growth factor [β-NGF]); and antiangiogenic or tissue-remodeling molecules (endostatin, thrombospondin-2, and angiogenin). These cytokines were chosen to represent pathways of ocular development, angiogenesis, fibrosis, and neurotrophic signaling, all of which may be relevant to phenotypic variation in PPA and to postoperative healing responses. This panel was selected a priori as a targeted, hypothesis-driven set rather than an exhaustive inflammatory screen. Given the limited aqueous volume obtainable in pediatric eyes and the need to limit multiplicity while preserving interpretability, we focused on these 13 cytokines. Accordingly, several classical inflammatory and fibrotic mediators were not included in the prespecified assay panel, and the current study was therefore not intended to provide a comprehensive mechanistic characterization of the AH microenvironment. Angiogenin values exceeded the assay upper limit of quantification (ULOQ; 4710 pg/mL, per the manufacturer's calibration range) in all samples (96/96, 100%) and were recorded as “>ULOQ.” Therefore, angiogenin was not included in the primary inferential analyses (between-group comparisons or regression modeling) due to the absence of quantifiable variability.

Ophthalmic Examinations and Measurements

Comprehensive ophthalmic evaluations were performed preoperatively and at 1 day, 1 week, 1 month, 3 months, 6 months, and at least 12 months postoperatively. All eligible patients were sedated using chloral hydrate (0.6 mL/kg; maximum dose, 10 mL) for a comprehensive ophthalmologic examination. Under the ideal conditions of general anesthesia, slit-lamp–adapted anterior segmental photography (BX 900; Haag-Streit, Köniz, Switzerland) was conducted. Additionally, intraoperative photographs were taken during surgery to document the posterior lens morphology. The intraocular pressure (IOP) was measured by rebound tonometry (iCare PRO; iCare Finland, Vantaa Finland), which was measured six times for each eye, and then the mean values were recorded. Best-corrected visual acuity (BCVA) was assessed preoperatively with age-appropriate instruments, including Teller Acuity Cards (<3 years), LEA Symbols (3–6 years), and Early Treatment Diabetic Retinopathy Study (ETDRS) charts (>6 years).²⁵ All measurements were converted to the logarithm of the minimum angle of resolution (logMAR) for analysis. Axial length (AL) was measured preoperatively using an IOLMaster 700 (Carl Zeiss Meditec, Jena, Germany). All of the measurements were performed by the same technician.

Postoperative Complications

VAO was counted as any fibrocellular membrane, lens epithelial cell proliferation, or residual opacities along the visual axis that obscure the red reflex or retinoscopy and cover the central visual axis, requiring Nd:YAG laser or surgical intervention.²⁶ Glaucoma-related adverse events included both glaucoma and glaucoma suspect. A glaucoma suspect was defined as elevated IOP (>21 mmHg on two occasions post-steroid) or the use of IOP-lowering medication without anatomical changes. Glaucoma was diagnosed if IOP exceeded 21 mmHg accompanied by anatomical changes such as corneal enlargement assessed by Scheimpflug imaging (Pentacam HR; Oculus, Wetzlar, Germany), progressive axial myopia confirmed by repeated AL measurements using A-scan ultrasonography (Aviso; Quantel Medical, Cournon d'Auvergne, France), increased optic nerve cupping (≥0.2) on dilated fundus examination, or the need for surgical IOP control.^27,28

Statistical Analysis

All data were analyzed using the R 4.4.3 (R Foundation for Statistical Computing, Vienna, Austria). The Shapiro–Wilk test was used to assess data normality, and the Mann–Whitney U test and Kruskal–Wallis test were applied for intergroup comparisons. Associations between each cytokine and postoperative complications were evaluated using Firth logistic regression to mitigate sparse data bias in this low-event setting.²⁹ Given the limited number of postoperative events, covariate adjustment was prespecified and restricted to age, sex, and PPA status (PPA vs. non-PPA), with one cytokine entered per model. For VAO (n = 13) and glaucoma-related adverse events (n = 10), each primary Firth model included one cytokine plus three prespecified covariates (age, sex, and PPA status), and sensitivity analyses additionally adjusted for primary IOL implantation (total of five covariates). For between-group comparisons across multiple cytokines, we report both unadjusted P values and Benjamini–Hochberg false discovery rate (FDR)-adjusted q values (q < 0.05). For the prespecified Firth regression models, we report unadjusted P values (no multiple-testing correction). Principal component analysis (PCA) was performed on log-transformed cytokine concentrations.

Results

Demographic Data of the Enrolled Patients

Baseline demographic and clinical parameters are summarized in Tables 1 and 2. As shown in Table 1, the PPA group (n = 78) and the non-PPA group (n = 18) were comparable in age, sex distribution, preoperative IOP, preoperative BCVA, and AL. Similarly, when the PPA group was further categorized into four subtypes, no significant differences were found among the PFV (n = 27), PPC (n = 16), PL (n = 17), and PCD (n = 18) groups regarding these baseline characteristics (all P > 0.05) (Table 2). During a mean follow-up of 33.26 ± 12.34 months, postoperative complications were observed in the study cohort. Among the 96 eyes, 13 eyes developed VAO (13.5%), and 10 eyes experienced glaucoma-related adverse events (10.4%).

Table 1.

Baseline Demographic and Clinical Parameters of Patients With PPA and Non-PPA Groups

Characteristic	PPA Group (n = 78)	Non-PPA Group (n = 18)	P
Age (mo), mean ± SD	29.12 ± 25.72	33.00 ± 33.92	0.891*
Sex (M/F), n	39/39	9/9	1.000†
Axial length (mm), mean ± SD	20.65 ± 1.74	20.98 ± 0.92	0.265*
Preoperative IOP (mmHg), mean ± SD	14.96 ± 2.80	15.14 ± 2.50	0.844*
Preoperative BCVA (logMAR), mean ± SD	1.03 ± 0.58	0.89 ± 0.52	0.321*

^* P values were calculated using the Mann–Whitney U test for continuous variables.

^† P values were calculated using the χ² test for categorical variables.

Table 2.

Baseline Demographic and Clinical Characteristics of PPA Subtypes

Characteristic	PFV (n = 27)	PPC (n = 16)	PL (n = 17)	PCD (n = 18)	P
Age (mo), mean ± SD	27.93 ± 22.29	27.06 ± 22.25	39.76 ± 31.59	22.67 ± 26.40	0.293*
Sex (M/F), n	12/15	7/9	10/7	10/8	0.721†
Axial length (mm), mean ± SD	20.66 ± 1.63	21.17 ± 1.01	21.23 ± 1.06	20.08 ± 2.23	0.125*
Preoperative IOP (mmHg), mean ± SD	15.86 ± 2.57	15.60 ± 2.85	14.91 ± 2.84	18.42 ± 2.42	0.516*
Quantitative BCVA (logMAR)	1.04 ± 0.47	0.97 ± 0.59	1.21 ± 0.77	1.10 ± 0.49	0.674*

^* P values were calculated using the Kruskal–Wallis test for continuous variables.

^† P values were calculated using the χ² test for categorical variables.

Comparison of Cytokine Profiles Between the PPA Group and the Non-PPA Group

The levels of BDNF, GDNF, FGF1, FGF2, NT-4, PlGF, VEGF-A, angiopoietin-1, endostatin, β-NGF, PDGF-AA, and thrombospondin-2 in 96 AH samples are reported as the median (interquartile range [IQR]) in Table 3. After FDR correction, five cytokines showed significant differences between the PPA and control groups. Compared to the non-PPA group, the PPA group exhibited significantly elevated levels of FGF2 (q = 0.018), VEGF-A (q = 0.029), and NT-4 (q = 0.035) (Fig. 2). Conversely, levels of FGF1 (q = 0.029) and PDGF-AA (q = 0.035) were significantly reduced in the PPA group (Fig. 2). No intergroup differences were observed for the remaining cytokines measured. The distribution of all cytokines is presented in Supplementary Figure S1.

Table 3.

Comparison of Aqueous Humor Cytokine Levels Between PPA and Non-PPA Groups

Cytokine (pg/mL)	Non-PPA Group (n = 18) Median (IQR)	PPA Group (n = 78) Median (IQR)	P	q
BDNF	12.56 (5.49)	11.04 (5.44)	0.417	0.556
GDNF	3.77 (2.23)	2.60 (1.56)	0.106	0.212
NT-4	12.06 (4.36)	19.11 (7.21)	0.0146	0.035
β-NGF	0.86 (0.80)	0.86 (0.59)	0.661	0.661
FGF1	79.95 (40.63)	54.45 (53.09)	0.0072	0.029
FGF2	4.80 (4.52)	7.75 (6.35)	0.0015	0.018
PDGF-AA	27.52 (17.47)	21.92 (13.74)	0.0135	0.035
Thrombospondin-2	226.62 (487.84)	221.37 (472.04)	0.589	0.643
VEGF-A	89.32 (133.27)	182.79 (143.23)	0.0067	0.029
PlGF	1.19 (1.17)	1.42 (1.35)	0.342	0.513
Angiogenin	>ULOQ	>ULOQ	>ULOQ	>ULOQ
Endostatin	10615.09 (13130.43)	9283.27 (6524.86)	0.235	0.403
Angiopoietin-1	86.98 (56.36)	83.83 (51.85)	0.521	0.626

P values were obtained using the Mann–Whitney U test. The q values were Benjamini–Hochberg adjusted for multiple testing; q < 0.05 was considered statistically significant. >ULOQ indicates that angiogenin exceeded the ULOQ (4710 pg/mL) in all samples; no P or q values were calculated. Bold values indicate statistical significance at q < 0.05.

Figure 2.

Comparison of AH cytokine levels between the PPA and non-PPA groups: (A) FGF1, (B) FGF2, (C) NT-4, (D) VEGF-A, and (E) PDGF-AA. The PPA group exhibited significantly lower FGF1 and PDGF-AA, but higher FGF2, VEGF-A, and NT-4 levels compared with the non-PPA group. Each point represents an individual sample; violin plots display the data distribution with median and interquartile ranges. *P < 0.05, **P < 0.01, ***P < 0.001.

Differences in the Levels of Cytokines Among the Four Subtypes of PPA

Representative clinical photographs of the four PPA subtypes are shown in Figure 3. Kruskal–Wallis analysis revealed significant intergroup differences in nine cytokines, including BDNF, GDNF, FGF1, PlGF, angiopoietin-1, endostatin, β-NGF, PDGF-AA, and thrombospondin-2 (all q < 0.05, FDR adjusted) (Table 4). NT-4, FGF2, and VEGF-A showed no significant intergroup differences (all q > 0.05, FDR adjusted) (Table 4). Post hoc pairwise comparisons further demonstrated that the PFV group exhibited relatively higher levels of multiple angiogenic factors (PlGF, angiopoietin-1), anti-angiogenic factors (thrombospondin-2), and neurotrophic factors (BDNF, GDNF, β-NGF) compared with the PPC and PL groups, whereas PDGF-AA was relatively elevated in the PCD group (Table 4; Fig. 4). Subgroup distribution plots for these nine cytokines are provided in Supplementary Figure S2 to facilitate more transparent visualization of the data distributions across subtypes. However, cytokine distributions showed substantial overlap across subtypes.

Figure 3.

Representative clinical photographs of the four PPA subtypes. (A) PFV: Slit-lamp photograph (oblique illumination) shows a fibrovascular stalk adherent to the posterior capsule (arrow). (B) PPC. (C) PL. (D) PCD: Slit-lamp photograph (oblique illumination) shows a round-like defect seen on the posterior capsule (arrow). (E) PCD: intraoperative photograph shows a round-like defect in the posterior capsule after partial aspiration of the lens cortex and nucleus (arrow).

Table 4.

Differences in the Levels of Cytokines Among the Four Subtypes of PPA

	Subtype, Median (IQR)				q
Cytokine (pg/mL)	PFV (n = 27)	PPC (n = 16)	PL (n = 17)	PCD (n = 18)	Among Four Groups	PFV vs. PPC	PFV vs. PL	PFV vs. PCD	PPC vs. PL	PPC vs. PCD	PL vs. PCD
BDNF	12.83 (11.04–17.81)	9.42 (8.17–10.77)	11.39 (8.35–14.28)	10.32 (8.75–13.01)	0.030	0.032	0.123	0.084	0.581	0.644	0.792
GDNF	3.38 (2.60–5.30)	1.82 (1.62–2.60)	1.82 (1.03–2.60)	2.60 (1.82–4.44)	0.009	0.041	0.016	0.711	0.664	0.109	0.046
NT-4	10.61 (6.74–14.85)	8.33 (5.02–9.11)	7.56 (4.19–12.06)	10.59 (7.56–12.06)	0.222	0.343	0.323	0.735	0.760	0.445	0.492
FGF1	71.90 (55.23–103.30)	42.48 (37.09–56.00)	42.48 (30.68–61.39)	68.90 (42.41–99.14)	0.008	0.029	0.012	0.565	0.678	0.125	0.057
FGF2	10.17 (6.65–13.91)	6.32 (5.44–8.08)	6.65 (4.89–7.64)	9.07 (7.86–11.57)	0.070	0.142	0.140	0.855	0.797	0.126	0.188
PlGF	1.50 (1.42–3.37)	1.03 (0.93–1.24)	1.26 (0.95–1.42)	2.14 (1.19–2.55)	0.002	0.004	0.010	0.727	0.777	0.01	0.026
VEGF-A	163.53 (130.03–308.57)	154.69 (124.53–228.73)	175.35 (117.92–219.81)	232.81 (183.11–370.64)	0.316	0.753	0.768	0.448	0.886	0.384	0.540
Angiopoietin-1	108.90 (83.83–194.66)	71.15 (58.35–83.83)	58.35 (58.35–90.13)	83.83 (72.74–105.78)	<0.001	0.001	<0.001	0.104	0.719	0.127	0.088
Endostatin	12520.49 (8388.41–16811.49)	7603.39 (5459.32–8773.35)	7779.55 (4357.43–9750.39)	9740.55 (7399.24–12187.21)	0.007	0.012	0.008	0.209	0.903	0.203	0.186
β-NGF	0.86 (0.86–1.41)	0.86 (0.27–0.86)	0.57 (0.27–0.86)	0.86 (0.27–0.86)	0.004	0.024	0.002	0.074	0.513	0.551	0.232
PDGF-AA	23.19 (17.58–33.75)	19.06 (17.16–24.24)	18.11 (16.41–23.61)	36.80 (23.93–43.97)	0.004	0.253	0.074	0.088	0.540	0.01	0.002
Thrombospondin-2	625.1 (225.62–2352.05)	144.7 (123.31–246.80)	144.7 (127.60–204.39)	331.3 (174.56–541.46)	<0.001	0.001	<0.001	0.194	0.771	0.062	0.035

The q values were obtained using the Kruskal–Wallis test followed by Dunn's post hoc test with Benjamini–Hochberg FDR correction. Bold values indicate statistical significance at q < 0.05.

Figure 4.

Comparison of aqueous cytokine levels among the four PPA subtypes: PFV, PCD, PL, and PPC. The cytokines measured included (A) BDNF, (B) GDNF, (C) FGF1, (D) FGF2, (E) NT-4, (F) PlGF, (G) VEGF-A, (H) angiopoietin-1, (I) endostatin, (J) β-NGF, (K) PDGF-AA, and (L) thrombospondin-2. Each point represents an individual sample. Boxes indicate the IQR with medians; whiskers represent 1.5 × IQR. Letters above boxes denote pairwise group differences based on Dunn's post hoc test with Benjamini–Hochberg correction; groups sharing the same letter are not significantly different. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Expression Patterns of AH Cytokines by PCA in Four Subtypes of PPA

PCA was applied to assess the overall distribution of cytokine profiles among the four PPA subtypes. The first principal component (PC1) accounted for 54.1% of the total variance, and the second principal component (PC2) accounted for 9.4%, with a cumulative contribution of 63.5% (Fig. 5A). In the score plot, subgroup distributions showed substantial overlap, although PFV tended to be displaced toward higher PC1 values; 95% confidence ellipses are provided to aid visualization of the subgroup distributions (Fig. 5A). Violin plots of PC1 scores showed an overall difference across the four subtypes (P < 0.001, Kruskal–Wallis), with PFV generally higher and PPC/PL generally lower (Fig. 5B). The loading plot demonstrated that FGF1, GDNF, and endostatin had the highest loadings on PC1, consistent with the elevated PC1 scores observed in PFV (Fig. 5C). Taken together, these results suggest exploratory heterogeneity across clinically defined PPA subtypes, with substantial overlap among groups.

Figure 5.

Expression patterns of AH cytokines by principal component analysis in four PPA subtypes. (A) Score plot of the first two principal components (PC1, 54.1%; PC2, 9.4%; cumulative, 63.5%). Each point represents one sample, colored by subgroup; ellipses denote 95% CIs. Variation was observed mainly along PC1, with PFV tending toward higher PC1 values, although subgroup distributions overlapped substantially. (B) Violin plots of PC1 scores illustrate the distribution within each group; boxplots inside indicate medians and IQRs. PC1 scores differed overall among the four subtypes (P < 0.001, Kruskal–Wallis). PFV generally showed higher PC1 scores than the other groups, although the distributions overlapped substantially. (C) PC1 loading plot shows the contribution of individual cytokines; FGF1, GDNF, and endostatin had the largest positive loadings, driving the high PC1 scores in PFV.

Logistic Regression Analysis of AH Cytokines and Postoperative Complications

In exploratory Firth logistic regression analyses adjusted for age, sex, and PPA status, higher preoperative NT-4 (odds ratio [OR] = 1.16; 95% confidence interval [CI], 1.02–1.36; P = 0.028) and PDGF-AA (OR = 1.06; 95% CI, 1.01–1.12; P = 0.027) levels were associated with postoperative VAO in this cohort (Table 5). No cytokine was significantly associated with glaucoma-related adverse events (Table 6). In a sensitivity analysis further adjusting for primary IOL implantation (30 eyes, 31.25 %), the associations of NT-4 and PDGF-AA with VAO showed similar effect estimates (NT-4: OR = 1.15; 95% CI, 1.01–1.31; P = 0.032; PDGF-AA: OR = 1.06; 95% CI, 1.01–1.12; P = 0.029), and primary IOL implantation (yes/no) was not independently associated with VAO (P = 0.96).

Table 5.

Adjusted Associations of Cytokines With Visual Axis Opacification Based on Firth Logistic Regression

Cytokine	OR	95% CI	P
BDNF	1.02	0.91–1.15	0.670
GDNF	1.03	0.72–1.45	0.850
FGF1	1.01	0.99–1.03	0.329
FGF2	1.05	0.97–1.16	0.239
NT-4	1.16	1.02–1.36	0.028
PlGF	1.02	0.73–1.34	0.894
VEGF-A	1.00	1.00–1.00	0.156
Angiopoietin-1	1.00	0.99–1.00	0.604
Endostatin	1.00	1.00–1.00	0.857
β-NGF	0.88	0.26–2.74	0.828
PDGF-AA	1.06	1.01–1.12	0.027
Thrombospondin-2	1.00	1.00–1.00	0.321

Odds ratios and 95% confidence intervals were estimated using Firth's logistic regression, adjusting for the potential confounders of age, sex, and PPA status (PPA vs. non-PPA). Events (VAO): n = 13. Bold values indicate statistical significance at P < 0.05.

Table 6.

Adjusted Associations of Cytokines With Glaucoma-Related Adverse Events Based on Firth Logistic Regression

Cytokine	OR	95% CI	P
BDNF	1.03	0.90–1.16	0.641
GDNF	1.18	0.80–1.72	0.405
FGF1	0.99	0.97–1.01	0.362
FGF2	1.04	0.94–1.13	0.386
NT-4	1.01	0.87–1.16	0.930
PlGF	0.95	0.62–1.26	0.741
VEGF-A	1.00	0.99–1.00	0.550
Angiopoietin-1	1.00	0.99–1.00	0.848
Endostatin	1.00	1.00–1.00	0.825
β-NGF	0.58	0.14–2.12	0.416
PDGF-AA	1.04	0.99–1.09	0.163
Thrombospondin-2	1.00	1.00–1.00	0.970

Odds ratios and 95% confidence intervals were estimated using Firth logistic regression, adjusted for age, sex, and PPA status (PPA vs. non-PPA). Events (glaucoma-related adverse events): n = 10.

Discussion

To our knowledge, this is the first study to characterize AH cytokine profiles in patients with PPA. Our findings suggest that the intraocular microenvironment in PPA may involve alterations extending beyond a purely inflammatory profile. The observed cytokine differences across clinically defined PPA subtypes suggest exploratory heterogeneity in this cohort; however, substantial overlap among subtypes warrants cautious interpretation.

These findings may be viewed in relation to two broad molecular patterns. The first is the significant dysregulation of cytokines governing angiogenesis and vascular remodeling, including VEGF-A and members of the fibroblast growth factor (FGF) family. VEGF-A is a central regulator of ocular angiogenesis and fetal vascular regression, and its overexpression has been linked to persistent hyaloid vasculature.^30–34 Consistent with this, the elevated VEGF-A levels observed in the AH of children with PPA may be relevant to processes related to angiogenesis and fetal vascular regression. Similar VEGF-A elevations have been reported in ocular neovascular diseases, including wet age-related macular degeneration, where VEGF-A overexpression drives pathological angiogenesis and fibrosis.^35,36

The complexity of this microenvironment is further exemplified by the contradictory regulation observed within the FGF family, which is known for its highly context-dependent roles in eye development, angiogenesis, and inflammation.³⁷ We observed a significant increase in FGF-2 levels in the PPA group, which may be relevant to inflammatory, injury-related, and fibrotic processes. FGF-2 can be induced by inflammatory stimuli and cooperate with TGF-β to promote fibrotic processes such as epithelial–mesenchymal transition (EMT), which is relevant to posterior capsule abnormalities.³⁸ Similar elevations have been reported in other ocular conditions with tissue remodeling and stress.^39,40 Collectively, these findings are compatible with involvement of injury-repair and tissue-remodeling–related processes.

In contrast, FGF-1 levels were markedly decreased, consistent with reports linking reduced FGF-1 to early vascular dysfunction.⁴¹ Liu et al.⁴² also observed lower AH FGF-1 in diabetic retinopathy, associated with impaired retinal blood flow. Together, these findings are consistent with reduced signaling related to vascular stabilization under pro-inflammatory and fibrotic conditions.

The concurrent increase in FGF-2 and decrease in FGF-1 represent a distinctive pattern in PPA. This pattern may be compatible with relatively enhanced tissue-remodeling activity accompanied by reduced signaling related to vascular stabilization, although its biological significance remains to be clarified.

Another notable finding was lower PDGF-AA levels in PPA, although the biological significance of this observation remains unclear. PDGF-AA is essential for lens epithelial proliferation, fiber differentiation, and vascular development.^43,44 In this study, PDGF-AA levels were relatively reduced in eyes with PPA, which may be associated with altered developmental processes in these eyes. In contrast, PDGF-AA levels are reportedly elevated under ischemic and proliferative conditions, including uveitic glaucoma and proliferative diabetic retinopathy, wherein it functions as a pathological mediator.^45,46 Taken together, these observations suggest that the pattern of PDGF-AA observed in PPA may differ from that reported in other ocular conditions, but its significance remains uncertain.

In contrast, NT-4, a key factor for neuronal survival, was increased in the PPA group. NT-4 is well known as a neuronal survival promoter.⁴⁷ Current research on NT-4 in the AH is still limited. In our study, the increased NT-4 levels in the AH may indicate altered neurotrophic signaling within the intraocular microenvironment, although its biological significance remains unclear. Taken together, the observed PDGF-AA and NT-4 differences may provide preliminary clues for further investigation into developmental and neurotrophic pathways in PPA. To further characterize the AH microenvironment across different subtypes of PPA, we compared the cytokine profiles of PFV, PCD, PL, and PPC and applied PCA to explore overall variation in cytokine distributions across clinically defined subtypes.

Our comparisons revealed significant differences in multiple cytokines among the four PPA subtypes, suggesting exploratory heterogeneity in their AH microenvironment. PCA further demonstrated that PC1 accounted for 54.1% of the variance and was influenced by several cytokines, including FGF1, GDNF, endostatin, PlGF, BDNF, angiopoietin-1, and VEGF-A. In contrast, PC2 explained only 9.4% of the variance and was mainly influenced by FGF2, NT-4, and PDGF-AA, with limited biological relevance. PC1 scores differed overall across subtypes, with PFV generally higher and PL/PPC generally lower, although substantial overlap remained among groups. Notably, VEGF-A, FGF2, and NT-4 did not differ significantly across the four groups, suggesting a partial overlap in cytokine profiles across subtypes.

These findings may be discussed in the context of prior clinical observations suggesting overlapping features among PPA subtypes. Müllner-Eidenböck et al.⁴⁸ observed that subtle PFV remnants frequently underlie unilateral congenital cataracts and proposed posterior lenticonus as a mild PFV variant, suggesting possible phenotypic relatedness in some cases. Furthermore, Khokhar et al.⁴⁹ described bilateral PL associated with PFV stalks in the same patient, directly linking the two anomalies.

Collectively, these reports suggest that PPA phenotypes may share certain clinical features. In the present study, however, the findings remain exploratory and should be interpreted with caution.¹⁹ One exploratory finding of this study is that specific preoperative AH cytokine levels were associated with postoperative VAO. Higher preoperative NT-4 and PDGF-AA levels were associated with postoperative VAO in this cohort. These observations may provide preliminary molecular clues for further investigation.

The observed association between PDGF-AA and VAO may have a plausible biological basis, although it should be interpreted cautiously. VAO is not merely a result of mechanical proliferation of lens epithelial cells (LECs), but rather a wound-healing and fibrotic process driven by inflammation.⁵⁰ Cataract surgery–induced tissue injury and breakdown of the blood–aqueous barrier may create a sustained inflammatory milieu in the anterior chamber, which in turn is associated with increased levels of various growth factors, including PDGF-AA.⁵¹ Prior studies suggest that PDGF-AA may have a context-dependent role in wound-healing responses. Although moderate PDGF-AA activity may support tissue repair, sustained expression has been associated with LEC migration, proliferation, and EMT, processes that may contribute to VAO formation.⁵² Although the overall median PDGF-AA level in the PPA cohort was lower than that in the non-PPA group, patients who developed postoperative VAO showed relatively higher PDGF-AA levels within the PPA group. This pattern may be compatible with possible involvement of PDGF-AA–related processes in postoperative fibrotic responses.

Studies have shown that inflammatory and injury signals can induce the upregulation of NT-4 expression, which may support neuronal survival and tissue repair.⁵³ Nevertheless, studies addressing the impact of NT-4 on LECs remain scarce. In corneal regeneration and ocular injury models, NT-4 has been shown to promote epithelial cell migration and proliferation.⁵⁴ In the present study, higher NT-4 levels were associated with VAO in this cohort. Although the underlying mechanisms remain uncertain, this association may suggest a possible involvement of NT-4–related pathways in postoperative tissue response and remodeling.

From a translational perspective, the AH cytokine features observed across PPA subtypes and in exploratory analyses of VAO may provide preliminary signals for future mechanistic and prospective studies. If independently replicated and standardized, the observed associations of NT-4 and PDGF-AA with postoperative VAO may help inform future investigation of perioperative follow-up strategies. In addition, the observed cytokine differences across clinically defined subtypes may support further investigation into whether different morphologic presentations are associated with different postoperative monitoring needs. Furthermore, these associations highlight pathway-level candidates for future mechanistic and prospective investigations, particularly in relation to postoperative tissue response and fibrotic processes. Overall, our findings provide exploratory molecular observations in congenital cataracts with PPA and may help inform future studies of postoperative surveillance. In addition, this study did not identify any cytokine associated with glaucoma-related adverse events, which may reflect the multifactorial nature of postoperative ocular hypertension. The limited number of glaucoma cases in our cohort also reduced the statistical power to detect subtle associations.

This study has several limitations. The modest cohort size and limited number of postoperative events, including 13 VAO events and 10 glaucoma-related adverse events, constrained the precision and stability of the complication analyses, particularly for subgroup comparisons and regression modeling. Although Firth logistic regression was used to reduce small-sample bias under sparse-event conditions, these analyses should still be interpreted as exploratory and hypothesis generating rather than definitive. In addition, cytokine measurement was restricted to a prespecified 13-analyte panel and did not capture a broader range of inflammatory and fibrotic mediators. Moreover, future studies with larger event numbers, broader molecular profiling, and independent multicenter cohorts will be needed to validate these exploratory findings and to address the underlying mechanistic questions more comprehensively. Incorporating genetic testing alongside standardized phenotyping may also help clarify potential genotype–phenotype relationships.

Conclusions

Congenital cataracts with PPA showed AH cytokine differences compared with non-PPA controls, with exploratory heterogeneity across clinically defined subtypes. Higher preoperative NT-4 and PDGF-AA levels were associated with postoperative VAO in exploratory analyses and warrant further validation in larger cohorts.