The study protocol was approved by the Institutional Review Board of Kyushu University Hospital (Fukuoka, Japan), and patient’s informed consent was obtained in accordance with the Declaration of Helsinki. Patients with ocular conditions other than RP, such as glaucoma, age-related macular degeneration (AMD), or uveitis, and those under anti-inflammatory treatment were excluded. We enrolled 37 eyes of 37 patients with typical RP (29 patients without CME and eight with CME) who underwent cataract surgery at Kyushu University Hospital during the period 2019–2023.

Aqueous humor samples were collected from each patient during the cataract surgery, and paired peripheral blood samples were obtained the same day. In cases in which both of the patient’s eyes were eligible, only the right eye and its corresponding serum sample were included for analysis.

The diagnosis of typical RP was established based on the patient’s clinical history, including night blindness and progressive peripheral visual field constriction or ring scotoma, along with hallmark findings, e.g., attenuated retinal vessels, bone spicule-like pigmentation in the mid-to-peripheral retina, and markedly diminished electroretinography (ERG) responses. In all eight of the RP patients with CME, the CME was clinically evident on a fundus examination and was confirmed by optical coherence tomography (OCT) prior to cataract extraction. Genetic inheritance patterns were inferred from identified mutations.

The patients’ baseline data were retrieved from their electronic medical records, including: age, sex, weight, body mass index (BMI), medication history, macular pathology, systemic comorbidities (i.e., hypertension, hyperlipidemia, diabetes mellitus, fatty liver disease, and autoimmune disorders), and lifestyle factors (tobacco smoking and alcohol consumption habits).

Each patient’s best-corrected visual acuity (BCVA) was assessed using either a Landolt C decimal chart (CV-6000, Tomey, Nagano, Japan, or AVC-36, Kowa Pharmaceuticals, Tokyo, Japan) at a testing distance of 5 m or with single optotype cards (HP-1258, Handaya, Tokyo, Japan) when necessary. Acuity values were converted to logarithm of the minimum angle of resolution (logMAR) units for statistical analyses. The smallest Landolt ring correctly identified by a patient in ≥60% (in three or more out of five) of presentations was used as the patient’s BCVA value.

For all 37 patients, the following measurements were obtained as described (10, 15): the result of an automated static perimetry test by a Humphrey field analyzer (HFA) (Humphrey Instruments, San Leandro, CA, USA), the averaged retinal sensitivity at 4 or 12 central points, and the baseline spectral-domain (SD) OCT values (Cirrus, Carl-Zeiss Meditec, Dublin, CA, USA). All of the patients also underwent a comprehensive fundus evaluation with a multimodal retinal imaging system.

Aqueous humor and serum samples were collected as described (10), and a Q-Plex™ Human Cytokine multiplex immunoassay (Quansys Biosciences, West Logan, UT, USA) was used to determine the concentrations of 15 cytokines [interleukin 1 alpha (IL-1α), IL-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12p70, IL-13, IL-15, IL-17, and IL-23; interferon gamma (IFN-γ), tumor necrosis factor alpha (TNF-α), and TNF-β] and nine chemokines [eotaxin, growth-related oncogene alpha (GROα), I-309, IL-8, interferon gamma-inducible protein 10 (IP-10), monocyte chemotactic protein 1 (MCP-1), MCP-2, RANTES (regulated activation normal T-cell expressed and secreted), and thymus and activated-regulated chemokine (TARC)]. Detailed assay protocols followed the manufacturer’s instructions and have been described (10).

Continuous variables such as age, BMI, BCVA, central foveal thickness (CFT), ellipsoid zone length (EZL), mean deviation (MD), and the four-point and 12-point central retinal sensitivities are expressed as the mean ± standard deviation (SD) and were compared using the Mann–Whitney test. Categorical variables including sex, smoking and alcohol consumption habits, systemic and ocular complications, and inheritance patterns are summarized as counts (percentages) and were analyzed using Fisher’s exact test. The cytokine, chemokine, and systemic blood parameters are summarized using the median and interquartile range (IQR) due to their non-normal distribution. Group comparisons between the RP patients with and without CME were performed using the Mann–Whitney test for the aqueous and serum molecule concentrations and the peripheral immune parameters, including the lymphocyte percentage (%LYMPH), the monocyte percentage (%MONO), the white blood cell count (WBC), the C-reactive protein (CRP), and the lymphocyte-to-neutrophil ratio (LNR). A two-tailed p-value <0.05 was considered significant.

To explore the integrated structure of the inflammatory and clinical variables in the 37 RP patients with/without CME, we performed a multiple-factor analysis (MFA) using 40 continuous variables, including the cytokine/chemokine levels, the CFT, and the peripheral immune parameters. The inflammatory variables were grouped according to previously defined clusters based on the results of a hierarchical clustering analysis (10). These included six cytokine/chemokine groups: group A (aqueous RANTES, TARC, IP-10, MCP-1, and serum RANTES), group B (aqueous IL-6, serum MCP-1, and TARC), group C (aqueous IL-23, eotaxin, GROα, IL-8, I-309, serum IL-23, and IFN-γ), group D (serum eotaxin, GROα, I-309, IP-10, and MCP-2), group E (serum IL-10, IL-17, IL-2, IL-8, IL-4, IL-15, IL-6, IL-12, IL-1α, IL-13, and IL-5), and group F (aqueous MCP-2 and serum TNFα).

In addition, two new parameters were incorporated into the MFA: CFT and Peripheral_immune (five variables: %LYMPH, %MONO, WBC, CRP, and LNR). To distinguish between patients with and without CME, the MFA was conducted using the FactoMineR software package ver. 2.10 (Agrocampus Ouest, Rennes, France), and visualization was performed using the factoextra package ver. 1.0.7 (Kassambara, Marseille, France) in R.

Table 1 summarizes the patients’ baseline clinical characteristics. The mean ages were 71 years (range, 48–81 years) in the eight RP patients with CME and 62 years (range, 41–81 years) in the 29 patients without CME. There was a statistically significant difference between groups (p = 0.032). No significant differences were observed between the groups in terms of sex distribution, visual acuity, BMI, tobacco or alcohol habits, CFT, EZL, MD, central four-point or 12-point retinal sensitivity, systemic diseases, or causative gene distributions.

Compared to the RP eyes without CME, those with CME displayed higher aqueous concentrations of IL-23, I-309, and GROα (IL-23: p = 0.002, I-309: p = 0.039, GROα: p = 0.042), as shown in Figure 1 and Table 2. Among them, IL-23 showed the greatest fold increase, which is highlighted in the volcano plot in Figure 2. In contrast, no serum molecules were elevated in patients with CME. Furthermore, a multivariable logistic regression model including age and aqueous IL-23 confirmed that aqueous IL-23 was significantly associated with CME status (p = 0.02), whereas age showed no significant association (p = 0.06) (Supplementary Table S1).

We next analyzed the associations between peripheral leukocytes and RP with CME (RP-CME). The peripheral blood analysis revealed a statistically higher %LYMPH in the RP-CME patients compared to patients without CME (Figure 3, Table 3).

Building on our research group’s previous identification of systemic–ocular inflammatory networks in RP, we performed an MFA using the same biologically defined inflammatory clusters to explore their relationships across the 37 RP patients with and without CME (10). The MFA in the present study was conducted to integrate the systemic and ocular inflammatory molecules, the CFT, and the peripheral immune parameters, aiming to determine how patterns of inflammatory molecules relate to RP-CME.

The analysis results revealed that the first two dimensions of the MFA accounted for 37.4% of the total variance, with dimension 1 (Dim1) and dimension 2 (Dim2) explaining 22.1% and 15.3% of the variability, respectively (Figure 4A). As shown in Figure 4B, the plot of the quantitative variables demonstrated that the following were closely aligned with CFT: aqueous IL-23, IL-8, eotaxin, GROα, and I-309, as well as serum IL-23 and IFN-γ from group C; aqueous IL-6, serum MCP-1, and TARC from group B; serum eotaxin, GROα, I-309, and IP-10 from group D; and the peripheral immune markers %LYMPH and LNR. These results suggest potential associations between these immune profiles and the central macular structure. In addition, these vectors were oriented in the same direction as that in the RP-CME group, indicating that these inflammatory signatures may be related to the development of CME.

In contrast, the variables from group A (aqueous RANTES, TARC, IP-10, MCP-1, and serum RANTES) and group F (aqueous MCP-2 and serum TNFα) were oriented orthogonally to the CME axis, indicating minimal associations with the patients’ CME status.