This cross-sectional study included patients diagnosed with MFCPU/PIC attending follow-up visits at the Retina and Uveitis Service of Luigi Sacco Hospital in Milan from January 2024 to February 2026. The study adhered to the principles of the Declaration of Helsinki and was approved by the institutional review board (IRB). All procedures were conducted in accordance with Italian bioethical guidelines, and written informed consent was obtained from all patients before recruitment.

The study included 2 groups of patients: (1) individuals with MFCPU/PIC, and (2) a control group (non-MFCPU/PIC) including eyes with atrophic chorioretinal lesions secondary to inflammatory or infectious mimickers representing diagnostic alternatives to MFCPU/PIC considered during the validation of the SUN classification. Additionally, patients with paucifocal or multifocal ocular toxoplasmosis (OT) scars and/or punctate outer retinal toxoplasmosis (PORT) were included. ^,

Inclusion criteria for the MFCPU/PIC group patients were: (1) confirmed diagnosis based on SUN 2021 criteria ^, ; (2) presence of at least one inactive “punched-out” atrophic lesion localized in the posterior pole.

Control patients had to present with atrophic lesions of non-MFCPU/PIC origin and be supported by previous documentation confirming the etiology.

Exclusion criteria for patients of both groups included: (1) poor-quality OCT scans (<25 dB) due to significant media opacities; (2) previous retinal surgery or laser procedures or any ocular (eg, age-related macular degeneration, central serous choriorioretinopahy) or systemic (eg, uncontrolled diabetes) condition that could confound lesion interpretation.

All recruited patients underwent a comprehensive ophthalmic examination, which included best-corrected visual acuity (BCVA) assessment, biomicroscopy of the anterior and posterior segments, and applanation tonometry with a Goldmann tonometer. Retinal imaging was conducted with true-color fundus photographs (Eidon Centervue, Padua, Italy), near-infrared reflectance (NIR), short-wavelength autofluorescence (SW-AF; λ = 488 nm, 30°), and a fovea-centered 512-section 30° × 25° volume scan (241 sections, 30 µm interscan spacing, ART 25) was acquired using SD-OCT (Spectralis OCT, Heidelberg Engineering). Subsequently, the same volume was reacquired using the Heidelberg Spectralis high-resolution OCT (HR-OCT).

HR-OCT is a novel prototype developed by Heidelberg Engineering, currently available exclusively for research purposes. This system achieves enhanced axial resolution, up to 3 µm, by shortening the central wavelength from 880 to 853 nm and increasing the bandwidth from 50 to 137 nm.

To ensure precise spatial registration of OCT datasets, 30 × 30° fovea-centered volumes were exported in E2E format from the SD-OCT system and imported into the HR-OCT platform, where scans were reacquired using the Set Progression and Follow-up tracking functions. To perform an en-face analysis of the lesions, a 15 × 5 mm volume scan (256 B-scans with 6-µm spacing) centered on the enrolled lesions was acquired in a subset of patients from both groups.

To be included in the final analysis, atrophic lesions in both groups were required to meet the following criteria: (1) location within the central 30° of the posterior pole; (2) availability of clinical documentation, including SD-OCT imaging, demonstrating a co-localized previously active lesion; and (3) absence of direct involvement by CNV.

Lesions that could not be fully encompassed within the scanned volume or were localized outside the central 30° were excluded.

A senior uveitis specialist (A.I.) recruited cases and controls and identified eligible lesions according to the aforementioned criteria. To confirm the presence of previously resolved inflammatory lesions, SD-OCT imaging progression and clinical records of patients were retrospectively reviewed during this phase.

MFCPU/PIC lesions were classified as active or inactive based on multimodal imaging, including CFP, SD-OCT, and FAF.

Active lesions were defined by:

• –

  Presence of yellow-whitish, oval or round lesions with ill-defined margins on CFP.

• –

  Bruch’s membrane (BrM) disruption, retinal pigment epithelium (RPE) inflection, a moderately hyperreflective nodule extending from the choroid into the retina, and localized thickening of the underlying choroid on SD-OCT.

• –

  Partial or complete hyper-autofluorescent halos with a central hypo-autofluorescent core on FAF. ^,,,,

Inactive lesions were characterized by:

• –

  Well-defined, round, punched-out atrophic areas, often with pigmented margins on CFP.

• –

  Chorioretinal atrophy involving disruption of the ellipsoid zone (EZ), RPE, and BrM on SD-OCT.

• –

  Homogeneous hypo-autofluorescence on FAF. ^,,,,

All eligible lesions were assigned a unique alphanumeric identifier and randomly sampled using a computer-based randomization software to include a maximum of 3 lesions per eye in the final dataset, thereby minimizing potential intra-eye clustering.

A second uveitis specialist (F.Z.) analyzed the enrolled lesions on SD-OCT, measuring the size of RPE atrophy (RPE-AS) and BrM defects (BrM-DS) using the built-in caliper tool of the Heidelberg Spectralis OCT, positioning it at the location of the maximum horizontal diameter. The tool was also used to measure choroidal thickness at the center of the lesion (sub-lesional choroidal thickness, SLCT) and at 2 sites 200 µm from the lateral margins (perilesional choroidal thickness, PLCT1 and PLCT2). To determine the maximum choroidal thickness at each site, the caliper was positioned between the inner scleral boundary and the BrM/RPE complex, or, in cases of choroidal atrophy, at the innermost location where a choroidal vascular pattern was detectable.

To minimize the influence of age, sex, axial length, eccentricity, and diurnal variations on choroidal thickness, a choroidal thickness coefficient (CTC), representing the ratio of SLCT to the average of PLCT1 and PLCT2, was calculated for each lesion ( Figure 1 ). For accuracy, co-registered SD-OCT and NIR images were reviewed simultaneously during all phases of the process.

Image analysis of an inactive MFCPU/PIC lesion. NIR images and structural OCT were simultaneously reviewed to ensure accuracy in identifying the site of the maximum lesion diameter (A). The in-built caliper tool provided by Heidelberg Spectralis was used to measure the maximum size of BrM defects (B, red caliper) and RPE atrophy (B, yellow caliper). Choroidal thickness was measured at 3 locations: 2 sites 200 µm from the lesion margins, named PLCT1 and PLCT2 (B, yellow calipers), and at the lesion center (C, red caliper, SLCT). The ratio of the SLCT to the average of PLCT1 and PLCT2 was then calculated to determine the CTC. Two blinded graders independently analyzed the lesions to identify structural findings within each retinal layer. In (D), white arrows highlight RNFL, GCL, and IPL subsidence. The orange dashed line indicates INL herniation through a disrupted BrM (D, gray dashed line). A reduction in choroidal thickness is observed beneath the lesion (D, white dashed lines), where an FCE is visible (D, red arrow).

To enhance the visualization of concentric co-localization of RPE atrophy and BrM defects, manual verification of the HR-OCT volume segmentation was performed. Subsequently, images were processed using the Heidelberg Spectralis software for 3D visualization and en-face imaging, isolating the RPE and BrM layers ( Figure 2 ).

En-face HR-OCT analysis of RPE atrophy and BrM defects in MFCPU/PIC and controls. Panels A1 and B1 display color fundus photographs (Eidon Centervue, Padua, Italy) of multifocal atrophic lesions involving the posterior pole in 2 patients diagnosed with MFCPU/PIC and toxoplasmosis, respectively. High-resolution OCT (HR-OCT) volumes were acquired with a dense scan pattern and 6 µm spacing, encompassing all lesion margins (A2, B2). Notably, MFCPU/PIC lesions exhibit Bruch’s membrane (BrM) interruption (green arrowheads), which is absent in toxoplasmosis lesions (red arrowhead). To enhance visualization of the concentric localization of retinal pigment epithelium (RPE) atrophy and BrM damage, en-face HR-OCT images were segmented to isolate the RPE and BrM (A3-A4 and B3-B4). Interestingly, in MFCPU/PIC lesions, both the disrupted RPE and BrM appear hyporeflective, with the BrM defect being smaller and concentrically localized within the RPE atrophy (A5, A6). In contrast, in toxoplasmosis lesions, the atrophic RPE appears hyporeflective (B5), while the intact BrM is hyperreflective (B6).

Finally, 2 masked readers (A.T. and D.F.) independently evaluated the SD-OCT and HR-OCT B-scans of the randomly selected and coded lesions, focusing on grading each retinal layer and BrM for subsidence, herniation, or disruption/atrophy, and for evaluating the presence of loss of vascular structures in the choroid. Changes in retinal layers were categorized as “disruption” when a layer was interrupted, or its normal structure could no longer be identified, “subsidence” when a layer sagged without disruption, and “herniation” when a layer completely displaced into the choroidal space through a BrM defect. The choroid was classified as “atrophic” when abnormal thinning was observed at the lesion site, accompanied by partial or complete loss of vascular structures. If the defect was confined to the inner or intermediate layers of the choroid, with partial preservation of the outer vessels, the term “focal choroidal excavation” (FCE) was applied ( Figure 1 ). ^,

Discrepancies between readers were resolved by a third senior grader (M.O.).

All registered values were recorded in digital charts for data visualization and subsequent statistical analysis.

The normality of continuous variables was assessed using the Shapiro-Wilk test. Data were reported as mean (SD) or median (interquartile range [IQR]), as appropriate, and as counts and percentages for categorical variables. All analyses were performed at the lesion level, accounting for intrapatient clustering using cluster-robust SEs. Linear regression models were used to assess the association between RPE-AS and BrM-HS and to analyze CTC, while logistic regression models (adjusted for age when appropriate) were used to compare structural alterations between groups. Results were reported as odds ratios (ORs) with 95% confidence intervals (CIs), and adjusted predictive margins were calculated for Bruch’s membrane involvement. Model assumptions were evaluated using the Shapiro-Wilk test (normality), Breusch-Pagan test (heteroscedasticity), variance inflation factor (multicollinearity), and residual-vs-fitted plots (linearity). Agreement between SD-OCT and HR-OCT was assessed using McNemar’s test and Cohen’s kappa coefficient. All analyses were performed using Stata/MP version 18.5, with statistical significance set at α < 0.05.

A total of 59 patients were included. Overall, 1302 atrophic lesions were identified, and 106 lesions were selected for analysis after randomization. The final groups included 27 patients (27 females; 57 lesions) with MFCPU/PIC and 32 patients (16 females; 49 lesions) with non-MFCPU/PIC scars. Median age was 52 years (IQR 47-63) in the MFCPU/PIC group and 51 years (IQR 34-67) in the control group ( P =.71).

Non-MFCPU/PIC lesions were distributed across the following diagnoses: acute posterior multifocal placoid pigment epitheliopathy (APMPPE) (n = 10; 20.4%), birdshot chorioretinopathy (BSCR) (n = 2; 4.1%), serpiginous choroiditis (SC) (n = 7; 14.3%), serpiginous-like tubercular choroiditis (n = 8; 16.3%), sarcoidosis (n = 8; 16.3%), syphilis (n = 5; 10.2%), and ocular toxoplasmosis (n = 9; 18.4%).

MFCPU/PIC lesions did not demonstrate significant differences in the overall involvement of the retinal nerve fiber layer (RNFL; χ² = 2.48, P =.115), ganglion cell layer (GCL; χ² = 0.43, P =.512), or inner plexiform layer (IPL; χ² = 0.37, P =.544) compared with non-MFCPU/PIC lesions.

In MFCPU/PIC lesions, INL herniation through Bruch’s membrane defects was observed in 33 lesions (57.9%), compared with only 2 lesions (4.3%) in the non-MFCPU/PIC group ( P <.001).

The outer plexiform layer (OPL) was disrupted in 98.2% (56/57) of MFCPU/PIC lesions compared with 93.9% (46/49) of non-MFCPU/PIC lesions ( P =.239). In contrast, outer nuclear layer (ONL) disruption was observed in 100% (57/57) of MFCPU/PIC lesions and in 91.8% (45/49) of non-MFCPU/PIC lesions (χ² = 4.84, P =.028). External limiting membrane (ELM) disruption occurred in 100% (57/57) of MFCPU/PIC lesions and in 83.7% (41/49) of non-MFCPU/PIC lesions (χ² = 10.07, P =.002). The photoreceptor layer (PRL) was interrupted in all analyzed lesions.