Mesopic Microperimetry Captures Extensive Macular Atrophy with Pseudodrusen-Like Appearance Progression Beyond Visual Acuity: A Longitudinal Analysis

Abstract

Purpose

To quantify longitudinal changes in mesopic and scotopic retinal sensitivity (RS) and evaluate their relationship with retinal pigment epithelium (RPE) atrophy progression and foveal involvement in extensive macular atrophy with pseudodrusen-like appearance (EMAP).

Design

Retrospective longitudinal observational study.

Participants

One hundred twenty-eight eyes from 66 patients with clinically confirmed EMAP followed at a tertiary inherited retinal diseases clinic.

Methods

Eyes with ≥1 visit including mesopic and red scotopic microperimetry using a validated 20-stimulus grid and multimodal imaging (blue autofluorescence and spectral-domain OCT) were included. Associations between square root-transformed RPE atrophy size and functional metrics (best-corrected visual acuity [BCVA], mesopic RS [mRS], and scotopic RS [sRS]) were analyzed using mixed-effects models adjusted for age and intereye correlation. Longitudinal rates of functional and structural change were estimated in eyes with ≥2 visits. Cox proportional hazards regression evaluated predictors of foveal involvement.

Main Outcome Measures

Structure–function relationships, yearly rates of functional decline, and predictors of foveal involvement.

Results

Among 128 eyes (median age, 59.5 years; 69.7% female), larger RPE atrophy size was significantly associated with worse BCVA, mRS, and sRS (all P < 0.001), with the strongest association observed for paramacular mRS (R²m = 0.42). Forty-five eyes contributed to longitudinal analysis (median follow-up, 20.5 months). Retinal pigment epithelium atrophy enlarged linearly at 3.7 mm²/year, while mRS declined linearly at –1.3 dB/year. Best-corrected visual acuity and sRS demonstrated logarithmic decline patterns. Among eyes without baseline foveal involvement, faster macular mRS decline was the only significant predictor of subsequent foveal involvement (hazard ratio, 0.43 [95% confidence interval, 0.20–0.96]; P = 0.04), corresponding to a 133% increased risk per 1-dB decrease.

Conclusions

In EMAP, mRS demonstrated stronger structure–function associations with RPE atrophy and a more consistent longitudinal decline than BCVA and sRS. Decline in macular mRS independently predicted subsequent foveal involvement, supporting microperimetry as a sensitive functional endpoint for patient stratification and interventional trials.

Financial Disclosure(s)

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Extensive macular atrophy with pseudodrusen-like appearance (EMAP) is a rapidly progressive retinal degeneration that typically affects middle-aged individuals and leads to legal blindness within 5 years from diagnosis.1, 2, 3, 4 Although its pathogenesis remains partially understood,^5,6 the recent phenotypic reclassification by Antropoli et al has delineated its distinctive multimodal imaging profile.³ Macular changes range from a thick separation between the retinal pigment epithelium (RPE) and Bruch membrane to a vertically oriented pattern of multilobular RPE atrophy, considered a hallmark of the disease.⁷ Additional diagnostic requirements include the ubiquitous presence of pseudodrusen-like deposits from the posterior pole to the far periphery and the exclusion of genetic or acquired phenocopies such as Sorsby fundus dystrophy, late-onset retinal degeneration, pseudoxanthoma elasticum, and vitamin A deficiency retinopathy.³ Although peripheral paving-stone degeneration is frequently encountered,⁸ it is now regarded as complementary rather than essential. Together, these features underscore the panretinal nature of EMAP, in which best-corrected visual acuity (BCVA) declines sharply once foveal RPE atrophy, macular neovascularization (MNV),⁹ non-neovascular fibrosis,⁷ or inflammatory epiphenomena develop.^10,11

Currently, no therapies exist to halt or slow RPE atrophy progression in EMAP. The emergence of complement inhibitors for geographic atrophy (GA) has renewed interest in extending these strategies to allied atrophic disorders,¹² particularly EMAP, where complement dysregulation is well documented.^5,6 However, BCVA—despite its widespread use as the primary endpoint in retinal clinical trials—poorly captures the functional burden of macular diseases.13, 14, 15 Notably, recent pivotal trials evaluating pegcetacoplan and avacincaptad pegol showed limited functional gains when relying on BCVA as the main outcome, emphasizing the need for more sensitive and spatially resolved functional metrics.^16,17 In this context, microperimetry offers such granularity.¹⁸ Our group recently demonstrated that mesopic and red scotopic retinal sensitivity (mRS and sRS) correlate strongly with structural biomarkers in EMAP, including RPE atrophy extent, foveal involvement, and choriocapillaris loss.¹⁹ This is clinically meaningful because BCVA often remains relatively preserved until late-stage disease—specifically stage 3 described by Romano et al—when the fovea becomes affected.^7,20 Despite these insights, no longitudinal data on the natural history of mRS and sRS decline are currently available.

To support the development of microperimetry as a functional endpoint for future EMAP trials, we aimed to quantify yearly mRS and sRS loss, compare these trajectories with BCVA decline, and evaluate whether functional impairment predicts progression toward foveal involvement in a large EMAP cohort followed at an Italian tertiary retina center.

Methods

Design and Population

This retrospective, longitudinal observational study included patients diagnosed with EMAP between May 2018 and July 2024 and followed at the Inherited Retinal Diseases Clinic of Ospedale Luigi Sacco (University of Milan). The study adhered to the tenets of the Declaration of Helsinki and national bioethical regulations. Written informed consent was obtained from all participants. The study was conducted under institutional ethics committee approval (Ethics Committee of Azienda Sanitaria dell’Alto Adige) associated with the EMAP genetic research protocol (Protocol No. 86/2024).

Inclusion criteria were as follows: (1) a clinical diagnosis of EMAP based on the criteria proposed by Antropoli et al,³ supported by multimodal imaging and negative genetic testing for Sorsby fundus dystrophy (OMIM #136900), late-onset retinal degeneration (OMIM #605670), and pseudoxanthoma elasticum (OMIM #264800); (2) baseline BCVA >30 ETDRS letters (∼20/240 Snellen); (3) absence of MNV throughout follow-up; (4) availability of ≥1 visit including mesopic and red scotopic microperimetry using a previously validated 20-stimulus grid (MAIA-S, CenterVue/iCare)¹⁹ along with multimodal retinal imaging including blue autofluorescence (BAF) and spectral-domain OCT (SD-OCT; Spectralis HRA + OCT2, Heidelberg Engineering GmbH). Exclusion criteria were as follows: (1) other ocular diseases (e.g., glaucoma, retinal vascular disorders, keratoconus): or uncontrolled systemic conditions (e.g., severe hypertension, neurological disorders) that could confound the retinal phenotype or functional results; (2) refractive errors with spherical equivalent > |6| diopters; (3) suboptimal imaging quality due to media opacities, including visually significant cataract or corneal opacities; and (4) unstable fixation on microperimetry, defined as >30% fixation losses. The baseline BCVA threshold was selected to reduce floor effects in BCVA and retinal sensitivity (RS) measurements, preserve sufficient dynamic range for longitudinal analyses, and improve the reliability of microperimetry in eyes with advanced disease.

Study and Imaging Protocol

Medical records of eligible EMAP patients were reviewed at baseline and subsequent follow-up visits to collect demographic and clinical data, including age, sex, smoking status, and lens status (phakic or pseudophakic). Patients underwent a standardized ophthalmic examination annually, with the exception of those whose fellow eye had active MNV and therefore required more frequent monitoring. At each visit, examinations included BCVA testing measured on a Snellen chart and converted to ETDRS letter scores, slit-lamp biomicroscopy of the anterior segment, and intraocular pressure measurement prior to pupil dilation using phenylephrine 2.5% and tropicamide 1%. Microperimetry (MAIA-S; CenterVue/iCare) was performed by 2 clinical trial-certified orthoptists (R.D. and A.B.) under both mesopic (white stimuli on a white background) and red scotopic conditions (red stimuli on a dark background, λ = 627 nm). Scotopic testing was preceded by 30 minutes of dark adaptation (DA) at approximately 0.1 lux.¹⁸ Participants without prior perimetric experience underwent a short training session to familiarize with the testing procedure. To evaluate test–retest variability, microperimetry was repeated twice at baseline within a short time interval in a subset of patients (16 eyes). The multimodal imaging protocol included (ultra-)widefield fundus photography using the true-color EIDON device (CenterVue/iCare) and the pseudocolor Optos California system (Optos plc); 30° and 55° BAF imaging; and coregistered near-infrared reflectance and SD-OCT scans (30° × 30° area, ≥61 B-scans) obtained with the Spectralis HRA + OCT2 platform (Heidelberg Engineering GmbH).

Imaging and Microperimetry Analysis

Imaging studies from each visit were independently reviewed by 2 graders with expertise in retinal imaging (A.T. and F.R.). Discrepancies in categorical variables or differences >10% for continuous measures were adjudicated by a senior grader (G.S.).

30° × 30° BAF images were analyzed to quantify macular RPE atrophy. Atrophy size was measured using the semiautomated Heidelberg RegionFinder tool (version 2.6.5.0) and defined as well-demarcated areas of definitely or moderately decreased autofluorescence that corresponded to hyperreflective regions on near-infrared reflectance and to areas of complete RPE and outer retinal atrophy on coregistered SD-OCT.^2,3,21 To ensure longitudinal consistency, RPE atrophy quantification was limited to the standardized 30° × 30° BAF acquisition field. When atrophy extended beyond image boundaries, only the measurable portion within the imaging frame was included. This multimodal approach, consistent with prior literature, ensures accurate delineation of atrophic margins while excluding other sources of hypoautofluorescence, such as macular pigment, hemorrhage, or subretinal fibrosis.²² Foveal involvement was defined as extension of RPE atrophy or non-neovascular subretinal fibrosis with overlying outer retinal loss across the foveal centerpoint on SD-OCT. Retinal pigment epithelium atrophy size (in mm²) was square root-transformed (mm) for statistical purposes.²³ Macular neovascularization was excluded throughout follow-up using multimodal imaging. Swept-source OCT angiography represented the primary screening modality, particularly to distinguish MNV from non-neovascular subretinal fibrosis, with fluorescein and indocyanine green angiography performed in equivocal cases.

Microperimetry was performed using a customized and previously validated 20-stimulus grid, centered on the fovea and employing a 4–2 staircase strategy.¹⁹ Sensitivity values ranged from 0 to 36 dB for both testing modalities. Mesopic microperimetry was always performed before red scotopic testing, which followed 30 minutes of DA and was interpreted as a mixed dark-adapted response rather than a purely rod-isolating measure. This sequence was adopted to avoid potential carry-over effects of prolonged DA on mesopic measurements. To ensure precise alignment of stimuli across longitudinal visits, the automated follow-up function of the MAIA-S microperimeter was activated prior to each subsequent examination. For both mesopic and scotopic tests, microperimetry maps were reviewed to extract the following parameters: total mRS/sRS (mean sensitivity across all 20 stimuli), macular mRS/sRS (mean sensitivity of the central 8 stimuli, 4°-radius eccentricity), and paramacular mRS/sRS (mean sensitivity of the 12 peripheral stimuli, 8°–10° from the foveal center) (Fig 1).

Figure 1.

Representative longitudinal multimodal imaging from a patient with extensive macular atrophy with pseudodrusen-like appearance followed for ∼3 years. Top row: 30° × 30° blue autofluorescence (Spectralis HRA + OCT2, Heidelberg Engineering GmbH) images show progressive radial enlargement of hypoautofluorescent retinal pigment epithelium (RPE) atrophy over time. Middle row: Mesopic retinal sensitivity (mRS) maps acquired with a customized 20-stimulus grid (MAIA-S; CenterVue/iCare). Stimuli are subdivided into macular points (central 8 stimuli, 0°–4°) and paramacular points (12 peripheral stimuli, 8°–10°). A progressive reduction in total and paramacular mRS is evident over time, paralleling the enlargement of RPE atrophy. Bottom row: red scotopic retinal sensitivity (sRS) maps showing an early and profound decline with minimal residual dynamic range over time.

Statistical Analysis

All analyses were performed in RStudio (version 4.3.3; PBC). Two-sided tests were used with significance set at P < 0.05. Continuous variables were summarized as mean ± standard deviation or median (interquartile range) based on Shapiro–Wilk normality testing; categorical variables were reported as counts (percentages). For each metric, linear, logarithmic, and exponential relationships with time were compared using Akaike and Bayesian Information Criteria (AIC/BIC), and their comparison was used to identify the optimal functional form for structure–function associations. Intergrader agreement for atrophy measures was assessed using intraclass correlation coefficients (ICCs; 2-way random-effects model) and Cohen κ for categorical assessments (e.g., foveal involvement). For patients undergoing 2 same-day microperimetry tests, test–retest variability was assessed using ICCs and Bland–Altman analysis, reporting mean bias and 95% limits of agreement.

Associations between square root-transformed RPE atrophy size and functional outcomes were estimated using generalized linear mixed-effects models adjusted for age. All predictors were z-score standardized to allow comparison of standardized β coefficients (β∗). Patient-level nesting (with eyes nested within patients) was included as a random intercept in all structure–function models. Model fit was reported using β∗ (95% confidence interval), marginal and conditional R², and AIC/BIC. Longitudinal changes in BCVA, mRS, sRS, and RPE atrophy size were quantified in eyes with ≥2 visits using linear mixed-effects models with time (years) as a fixed effect and random intercepts and slopes at the patient level. An additional random intercept for eye-level clustering was included. Predictions at clinically relevant timepoints (baseline to year 3) were derived from fixed-effect estimates. Among eyes without foveal involvement at baseline, the risk of progression was assessed using Cox proportional hazards regression, with time-to-event defined as the interval from baseline to the visit when foveal involvement was first detected. Functional metrics were modeled as continuous predictors. Results are presented as hazard ratios with 95% confidence intervals, and Kaplan–Meier curves illustrate cumulative incidence.

Results

A total of 128 eyes from 66 patients with EMAP met the inclusion criteria. The cohort included 46 females (69.7%), with a median age of 59.5 (57.8–61.5) years and a median baseline BCVA of 70.5 (55.0–82.0) ETDRS letters (∼20/40 Snellen equivalent). Based on multimodal imaging, 38 eyes (29.7%) were classified as stage 1, 60 (46.9%) as stage 2, and 30 (23.4%) as stage 3.

The longitudinal cohort included 45 eyes (23 patients) with a median follow-up of 20.5 (12.1–31.6) months, whereas 83 eyes (43 patients) had baseline data only. The 2 groups were similar in sex distribution, baseline BCVA, foveal involvement, fibrosis, mRS, and sRS (all P > 0.05); however, eyes with follow-up were slightly younger (P = 0.01) and had larger baseline RPE atrophy (P = 0.03). Among the 45 eyes included in longitudinal analysis, 12 eyes reached ≥36 months of follow-up. Complete demographic and clinical data are summarized in Table 1. Intergrader agreement was excellent for RPE atrophy size measurements (ICC: 0.91 [0.87–0.94]) and near-perfect for foveal involvement (κ: 0.84 [0.81–0.86]). Test–retest variability analysis demonstrated excellent repeatability for both modalities: mRs had a mean bias of 0.21 (–0.02 to 0.43) dB with limits of agreement from –0.68 to 1.10 dB, whereas sRS had a mean bias of 0.12 (–0.08 to 0.32) dB with limits of agreement between –0.69 and 0.93 dB. Consistently, ICCs showed excellent test–retest reliability for both mRS (0.99 [0.98–1.00], P < 0.001) and sRS (0.99 [0.97–1.00], P < 0.001) (Fig 2).

Table 1.

Demographic and Baseline Clinical Characteristics of the EMAP Cohort

EMAP Cohort
Participants, eyes (patients)	128 (66)
Age, yrs, median (IQR) [range]	59.5 (57.8-61.5) [49.53–72.45]
Sex
Male, n (%)	20 (30.3%)
Female, n (%)	46 (69.7%)
Race/ethnicity
White non-Hispanic, n (%)	66 (100%)
Other, n (%)	0 (0%)
Smoking status
Never, n (%)	57 (86.4%)
Former, n (%)	9 (13.6%)
Current, n (%)	0 (0%)
Lens status
Phakic, eyes (%)	125 (97.7%)
Pseudophakic, eyes (%)	3 (2.3%)
Baseline BCVA, ETDRS letters, median (IQR) [range]	70.5 (55.0–82.0) [30.0–85.0]
Baseline total mesopic RS, dB, median (IQR) [range]	6.7 (3.1–11.2) [0.1–20.4]
Baseline total scotopic RS, dB, median (IQR) [range]	2.4 (0.5–6.1) [–1.0 to 18.9]
Baseline RPE atrophy size, mm2, median (IQR) [range]	12.8 (5.2–23.6) [0.0–53.9]
Baseline SQRT RPE atrophy size, mm, median (IQR) [range]	3.6 (2.3–4.9) [0.0–7.24]
Participants with F/U, eyes (patients)	45 (23)
Follow-up duration, mo, median (IQR) [range]	20.5 (12.1–31.6) [8.6–70.2]

BCVA = best-corrected visual acuity; dB = decibel; EMAP = extensive macular atrophy with pseudodrusen-like appearance; F/U = follow-up; IQR = interquartile range; n = frequency; RPE = retinal pigment epithelium; RS = retinal sensitivity; SQRT = square root-transformed.

Figure 2.

Test–retest variability of mesopic and red scotopic retinal sensitivity in extensive macular atrophy with pseudodrusen-like appearance (EMAP) using the 20-stimulus microperimetry grid. A and B, Bland–Altman plots comparing test–retest differences for mesopic (A) and red scotopic (B) retinal sensitivity. The blue line indicates the mean bias; dashed red lines denote the 95% limits of agreement. Mesopic testing shows minimal bias and narrow limits, whereas scotopic testing demonstrates slightly greater (still acceptable) variability. C and D, Scatter plots comparing test 1 vs. test 2 values for mesopic and red scotopic sensitivity. Points are color-coded by EMAP stage. Mesopic sensitivity demonstrates excellent repeatability across disease stages, while red scotopic values cluster near the lower end of the sensitivity scale, reflecting early plateauing.

Structure–function Analysis

After square-root transformation and z-score standardization, larger RPE atrophy size was strongly associated with reduced BCVA (β∗ = –0.56, P < 0.001), mRS (β∗ range –0.63 to –0.42, all P < 0.001), and sRS (β∗ range –0.42 to –0.33, all P < 0.001). The strongest association was observed for paramacular mRS (β∗ = –0.63 [–0.78 to –0.48], P < 0.001). Among the functional outcomes, total and paramacular mRS showed the highest marginal and conditional R² (R²m = 0.43 and 0.42; R²c = 0.94 and 0.92, respectively), indicating that RPE atrophy size explained 43% and 42% of their variability. Models predicting paramacular mRS also demonstrated the lowest AIC (306.7) and BIC (333.4), supporting the strongest overall efficiency and predictive performance for these functional endpoints. Corresponding results are shown in Figure 3A to D.

Figure 3.

Structure–function model performance and longitudinal estimates of atrophy and functional decline in extensive macular atrophy with pseudodrusen-like appearance. A–D, Structure–function performance metrics from generalized linear mixed-effects models evaluating the association between square root-transformed retinal pigment epithelium (RPE) atrophy size and each functional parameter. A, Standardized β coefficients show the strongest associations for paramacular and total mesopic retinal sensitivity (mRS), followed by macular mRS and best-corrected visual acuity (BCVA); scotopic RS (sRS) parameters demonstrated weaker associations. B, Marginal R² values were highest for total (R²m = 0.43) and paramacular mRS (R²m = 0.42), indicating greater explanatory power of atrophy size on mRS than on BCVA or sRS. C and D, Akaike Information Criteria and BIC comparisons identified the paramacular mRS model as the most efficient (AIC = 306.7; BIC = 333.4), followed closely by the total mRS model. (E-H) show the longitudinal mixed-effects model estimates for the rate of change over 3 years. E, Retinal pigment epithelium atrophy enlarged linearly at 3.7 (2.7–4.7) mm²/year (0.40 [0.31–0.49] mm/year). F, BCVA declined logarithmically at –7.0 (–10.7 to –2.6) ETDRS letters/year. G, Total mRS declined linearly at –1.3 (–1.7 to –0.9) dB/year, with similar linear loss across macular and paramacular regions. H, Total red sRS declined logarithmically at –0.3 (–0.4 to 0.3) dB/year, reflecting early rod dysfunction and limited dynamic range. AIC = Akaike Information Criteria; BIC = Bayesian Information Criteria.

Longitudinal Analysis

Retinal pigment epithelium atrophy enlarged linearly over time, both in raw and square root-transformed scales. The estimated rate of progression was 3.7 (2.7, 4.7) mm²/year (0.40 [0.31, 0.49] mm/year, square root-transformed), increasing from 18.3 (17.3, 19.3) mm² (4.0 [3.9, 4.1] mm) at baseline to 29.4 (28.4, 30.4) mm² (5.2 [5.1, 5.3] mm) at year 3. Best-corrected visual acuity followed a logarithmic pattern of decline, with model-based estimates decreasing from 57.3 (52.2, 62.9) letters at baseline to 37.3 (33.9, 41.0) letters at year 3, corresponding to an average modeled change of –7.0 ETDRS letters/year over this interval. Total mRS decreased linearly at –1.3 (–1.7, –0.9) dB/year, from 6.3 (6.0, 6.7) dB to 2.4 (2.0, 2.8) dB over the same period. Total sRS also followed a logarithmic trajectory, decreasing from 1.9 (1.7, 2.1) dB at baseline to 0.4 (0.3, 0.5) dB at year 3, corresponding to an average modeled change of –0.3 dB/year over the same interval. Macular and paramacular subanalyses for mRS and sRS, along with corresponding trajectories, are presented in Figure 3E to H.

Foveal Involvement Prediction

A total of 25 eyes (15 patients) without foveal involvement at baseline had serial study visits. These eyes developed foveal involvement at an incidence rate of 14.7 (6.0, 23.4) events per 100 eye years, with a median survival-free time of 3.5 years (Fig 4A). Univariable Cox proportional hazard regression identified the rate of macular mRS decline as the only significant predictor of foveal involvement (hazard ratio 0.43 [0.20–0.96], P = 0.04), indicating that each 1-dB reduction in macular mRS was associated with a 133% increased risk of subsequent foveal involvement (Fig 4B).

Figure 4.

Time to foveal involvement and functional predictors of structural progression in extensive macular atrophy with pseudodrusen-like appearance. A, Kaplan–Meier curve showing cumulative probability of retaining foveal sparing in 25 eyes without foveal involvement at baseline. The incidence rate of foveal involvement was 14.7 events per 100 eye years, with a median foveal-sparing survival of 3.5 years. Shaded areas denote 95% confidence intervals, and numbers at risk are reported above the x-axis. B, Univariable Cox regression results for functional predictors of foveal involvement. Macular mesopic retinal sensitivity (mRS) decline emerged as the only significant predictor (hazard ratio = 0.43 [0.20–0.96]; P = 0.04), indicating a 133% increased risk of foveal involvement for each 1-dB reduction in macular mRS. No significant associations were observed for BCVA, total/paramacular mRS, or any scotopic RS metric. BCVA = best-corrected visual acuity; RS = retinal sensitivity.

Discussion

In this retrospective study, we provide the first longitudinal evaluation of mesopic and red scotopic microperimetry in EMAP and its relationship with RPE atrophy progression. Using an integrated structural–functional approach, we found that atrophy size correlated strongly with all functional measures, most prominently mRS, and that mRS exhibited a linear decline paralleling the linear expansion of atrophy. Notably, faster reductions in macular mRS predicted earlier foveal involvement. Together, these findings reinforce a central role for microperimetry—using a customized 20-stimulus grid—in monitoring EMAP progression and informing endpoint selection for clinical trials.

Visual function assessment remains a major challenge across macular degenerations.^14,24 Traditional endpoints such as BCVA lack spatial specificity and often fail to capture early extrafoveal dysfunction.^13,15 The experience from recent complement–inhibition trials in GA has highlighted the need for sensitive, anatomically aligned functional measures capable of detecting meaningful change. This is particularly relevant in EMAP, a panretinal condition in which patients may retain normal or near-normal BCVA until late stages despite profound DA difficulties and extensive extrafoveal damage.^1,19,25 Thus, BCVA has limited sensitivity as an early marker or prognostic tool in this disease. Our prior work demonstrated that mRS correlates strongly with RPE atrophy and foveal involvement, whereas sRS aligns more closely with choriocapillaris alterations,^19,26 consistent with pseudodrusen-like deposits and RPE-Bruch membrane separation that impair vitamin A cycling and selectively affect rod recovery.²⁷ The present study confirms and extends these observations in a larger cohort, adding a longitudinal dimension that is largely lacking in the microperimetry literature. After adjusting for age and patient-level nesting, atrophy size remained significantly associated with all functional metrics, with total and paramacular mRS showing the strongest structure–function relationships. Their high marginal R² values and lowest AIC/BIC support greater explanatory power and model efficiency. The paramacular region (8°–10° eccentricity, approximately 2.3–2.9 mm from the foveal center) warrants particular emphasis. Extensive macular atrophy with pseudodrusen-like appearance shows extensive extrafoveal involvement long before foveal integrity is threatened, and functional losses in this region appear to precede BCVA decline.²⁸ This aligns with the spatial progression of EMAP, in which vertically oriented atrophy expands centrifugally before reaching the fovea. Paramacular mRS, therefore, offers a topographically aligned and highly sensitive indicator of disease progression. Although our predefined 20-stimulus grid was not designed to formally classify stimuli according to their distance from the atrophic border, this finding is conceptually consistent with observations from GA studies showing that RS measurements near lesion margins are often more sensitive to disease progression than global averages.²⁹ Given that the paramacular region in EMAP frequently includes the junctional zone between preserved and atrophic retina, future studies using denser grids and border-based stimulus classification may further refine functional endpoints. By contrast, sRS showed weaker associations with RPE atrophy, reflecting the well-recognized floor effect in EMAP: predominantly rod-mediated dysfunction is already severe in stage 1 and undergoes minimal decline thereafter.^20,30 This pattern is consistent with early RPE-Bruch membrane dysfunction, impaired retinoid flux, and electrophysiologic evidence of early mixed photoreceptor impairment.^5,25,31

The longitudinal findings of this study provide important natural history metrics for functional loss in EMAP. Approximately one-third of eyes had ≥2 visits with a median follow-up >20 months, enabling estimation of annual rates of change. Total and paramacular mRS declined linearly at approximately 1.3 dB/year, paralleling the linear expansion of RPE atrophy (∼3.7 mm²/year). The rate of mRS decline appears faster than in other macular diseases, including GA (∼0.6–1.0 dB/year) and Stargardt disease (∼0.5–1.0 dB/year), further underscoring the aggressive functional trajectory of EMAP.^32,33 In contrast, BCVA and sRS were better described by logarithmic models; therefore, their annualized estimates should be interpreted as average modeled changes over the study interval rather than constant rates of decline. These distinct linear versus nonlinear dynamics have important implications for trial design: linear mRS decline supports straightforward modeling of treatment effects, while logarithmic endpoints may require larger samples and nuanced time-to-event approaches. Differentiating these trajectories may refine endpoint selection and improve the interpretability of functional change in therapeutic studies.

Because loss of foveal integrity marks a rapid transition to legal blindness, predicting this event is of major clinical importance. In our cohort, among eyes without baseline foveal involvement, new foveal involvement occurred at 14.7 events per 100 eye years. Faster decline in macular mRS (0°–4°, corresponding to ∼0–1.2 mm from the fovea) emerged as the only significant predictor of foveal involvement, with each 1-dB decrease conferring a 133% increased risk. This suggests that subtle cone dysfunction near the fovea may precede structural breakdown visible on OCT. The observation parallels findings in AMD, where reduced contrast sensitivity and scotopic dysfunction predict progression from intermediate AMD to advanced atrophy.^15,34 Our study extends this concept by incorporating temporal change: dynamic mRS decline, rather than baseline values alone, stratifies risk of imminent stage 3 conversion. This metric may therefore help identify patients requiring closer monitoring, counseling, or prioritization for early intervention once therapies become available. Although foveal involvement is a major determinant of BCVA, it represents a binary marker of disease severity, whereas atrophy size provides a continuous measure of disease burden and progression. These markers should therefore be considered complementary rather than interchangeable. Total and paramacular mRS provide longitudinal functional measures aligned with structural disease growth, while macular mRS may offer valuable prognostic information regarding subsequent foveal involvement. Beyond its research applications, microperimetry may be integrated into routine EMAP evaluation, particularly when BCVA remains deceptively preserved despite extensive structural compromise. Lastly, although microperimetry should not be considered a direct surrogate of patient-reported quality of life, studies in other macular diseases have shown stronger associations between RS and vision-related quality-of-life measures than with BCVA. Future studies incorporating patient-reported outcomes are warranted to establish the significance of mesopic microperimetry at the patient level.

Several limitations warrant mention. First, the retrospective design may have introduced potential selection bias, although all EMAP patients underwent a standardized imaging and microperimetry protocol. Only 45 eyes were included in longitudinal analyses; although most metrics were comparable between eyes with and without follow-up, the longitudinal cohort had younger age and larger baseline atrophy. Additionally, only 12 eyes reached ≥36 months; thus, model-derived 3-year trajectories should be interpreted cautiously and validated in larger cohorts. Importantly, we did not incorporate other psychophysical or electrophysiologic tests, which could complement microperimetry but are difficult to implement routinely. The sample size for predicting foveal involvement was modest, and larger studies will be needed to confirm the prognostic utility of macular mRS. Fixation instability in advanced disease may influence microperimetry reliability, though our exclusion criteria likely mitigated this concern. Finally, our 20-stimulus grid was designed for EMAP to limit test duration while preserving longitudinal comparability with the original protocol, but it does not include a central foveal stimulus and does not assess the recently described temporally spared region. Targeted testing of this preserved area and denser point-wise registration to atrophy borders may further refine structure–function correlations.

In conclusion, this study provides longitudinal microperimetric benchmarks for EMAP and shows that mRS is a sensitive marker of functional loss associated with RPE atrophy progression. Decline in macular mRS may also help identify eyes at higher risk of foveal involvement. These findings underscore the limitations of BCVA as the sole functional endpoint and support the incorporation of microperimetry into future prospective studies and interventional trials. By defining natural history trajectories and potential predictors of central vision loss, this work provides a foundation for evaluating emerging therapies aimed at slowing atrophy progression and preserving meaningful visual function.

Manuscript no. XOPS-D–26–00221.