Beyond a Dry Macula: Six OCT Photoreceptor Stages Predict Visual Recovery After PDT in Chronic CSC

Design: Two-center retrospective observational cohort from Milan and Turin, Italy. Population: 139 eyes of 126 patients with chronic central serous chorioretinopathy (CSC), active subfoveal subretinal fluid (SRF), and treatment with half-dose or half-fluence verteporfin photodynamic therapy (PDT). Mean age 50.7 ± 12.1 years; 86.5% of patients were male. Median symptom duration was 6 months (IQR 3–15). Baseline mean BCVA was 0.25 ± 0.29 logMAR.

Question: Can a six-stage foveal photoreceptor OCT classification predict anatomical SRF resolution and visual recovery following PDT? Primary anatomical endpoint: complete foveal SRF resolution within six months of first PDT; functional analyses evaluated change in logMAR BCVA and final BCVA.

Chronic CSC may resolve anatomically after PDT without commensurate improvement in central vision. A binary ‘intact versus disrupted’ ellipsoid zone (EZ) descriptor may obscure the continuum of photoreceptor injury. This study examines whether more granular grading of EZ and interdigitation zone (IZ) morphology provides prognostic information before treatment.

The worst morphology seen within the central 1 mm on six fovea-centered radial SD-OCT B-scans determined each eye’s grade:

1. Stage 1A — Normal architecture: distinct EZ and IZ with preserved reflectivity and thickness (6 eyes, 4.3%).
2. Stage 1B — Early fusion: EZ and IZ no longer separately resolved; overall reflectivity and thickness preserved (23 eyes, 16.5%).
3. Stage 2 — Reactive thickening: homogeneous photoreceptor thickening above SRF without focal disruption (63 eyes, 45.3%).
4. Stage 3 — Patchy EZ irregularity: notched, moth-eaten EZ reflectivity without true discontinuity (23 eyes, 16.5%).
5. Stage 4 — Structural discontinuity: thinning with a true EZ break extending to an intact external limiting membrane (16 eyes, 11.5%).
6. Stage 5 — Complete atrophy: absent EZ and IZ with residual ELM above the RPE (8 eyes, 5.8%).

Independent dual grading in 50 eyes showed 92% exact agreement, unweighted Cohen κ = 0.90 and quadratic-weighted κ = 0.96 (95% CI 0.91–0.99). This demonstrates promising internal reproducibility, but external validation and performance across OCT platforms remain unknown.

81/139 eyes (58%) achieved complete foveal SRF resolution within six months; median time to resolution among responders was 2 months (IQR 1–3). Stage-specific resolution proportions were 83.3% (1A), 69.6% (1B), 66.7% (2), 34.8% (3), 50.0% (4), and 25.0% (5). These small, uneven subgroups should not be treated as calibrated patient-level probabilities.

In univariable Cox regression, more advanced stage predicted lower resolution likelihood (HR 0.70, 95% CI 0.54–0.92; P = .011). After adjustment, stage was not independently associated with resolution (HR 0.86, 95% CI 0.63–1.17; P = .324). Symptom duration was the only significant independent anatomical predictor (HR 0.97 per month, 95% CI approximately 0.94–1.00; P = .023). A hazard ratio here concerns time to resolution, not a direct 3% absolute reduction in success probability per month.

Among eyes achieving SRF resolution, more advanced baseline photoreceptor damage was associated with less visual improvement after accounting for baseline BCVA. Relative to Stage 1A, adjusted change in BCVA was smaller for Stage 3 (β = −0.117 logMAR; P = .005), Stage 4 (β = −0.310; P < .001), and Stage 5 (β = −0.717; P < .001). Stages 1B and 2 did not differ significantly from 1A.

In the multivariable final-BCVA ANCOVA, baseline BCVA (β = +0.441; P < .001) and photoreceptor stage independently correlated with outcome. Compared with 1A, final logMAR was higher (worse) for Stage 3 (β = +0.124, 95% CI 0.009–0.239; P = .035), Stage 4 (β = +0.306, 95% CI 0.160–0.451; P < .001), and Stage 5 (β = +0.416, 95% CI 0.241–0.591; P < .001). Symptom duration was not significant for final BCVA in this model (P = .180).

Interpretation: Duration of the current episode and photoreceptor structural injury appear to provide complementary prognostic information: the former about anatomical resolution, the latter about functional outcome. These are associations, not proof of separate biological causal pathways.

Responders gained a mean 0.076 logMAR versus 0.024 in eyes without SRF resolution by six months (P = .007); the paper defines positive change as baseline logMAR minus follow-up logMAR. Responders also had larger mean reductions in central subfield thickness (−159 vs −30 µm) and subfoveal choroidal thickness (−45 vs −7 µm), both P < .001.

Among 58 eyes not resolved by six months, 23 later resolved (12 after additional PDT, 11 without additional PDT), while 35 never demonstrated resolution by their last available visit. The latter group had relatively short and variable observation (median 4 months), so ‘nonresponder’ does not establish permanent treatment failure.

Before PDT: document baseline BCVA, estimated duration of the active symptomatic episode, foveal SRF, and the EZ/IZ/ELM configuration on high-quality OCT. Distinguish relatively preserved early-stage morphology from true EZ discontinuity and complete photoreceptor atrophy.

When counseling: explain that fluid resolution and visual restoration are not synonymous. Eyes with Stage 3–5 changes may have a more limited visual ceiling even if the macula becomes dry. Avoid interpreting this study as evidence to withhold PDT from advanced-stage eyes, or to mandate a specific treatment timing.

During follow-up: monitor both fluid status and functional vision. The classification is a promising research tool, not yet an externally validated clinical prediction calculator.

• Retrospective, noncomparative design: all eyes underwent PDT; no untreated control group or randomized treatment comparison establishes causal effects of treatment or optimal timing.
• Selection and attrition: only 100/139 eyes (72%) had at least two visits recorded within the first six months; visit timing was clinician-directed, and analyses rely on available assessments.
• Small advanced-stage strata: Stage 1A comprised six eyes and Stage 5 eight eyes; regression estimates for extremes may be unstable.
• Incomplete PDT regimen data: half-dose versus half-fluence treatment could not be modeled as a covariate. Prior non-PDT treatments were present in a substantial subset.
• Potential outcome conditioning: visual-gain regression restricted to the 81 responders may introduce selection effects and should not be generalized automatically to all treated eyes.
• Limited generalizability: two Italian tertiary centers used a single OCT device family and consensus grading. Intergrader reliability was assessed by the same two graders on a 50-eye subset, not independent external raters.
• Reporting caution: the paper’s regression Table 2 displays some confidence-interval signs inconsistently with the narrative; the values reported in the article’s explanatory text are used here. The authors’ claim that Stage 3–5 damage predicts less improvement is supported by the direction of the reported adjusted coefficients.
• Endpoint interpretation: the study excluded eyes with complete foveal RPE atrophy and censored secondary MNV; findings may not extend to these clinically important subgroups.

This study offers a practical conceptual advance: baseline photoreceptor morphology predicts the quality of visual recovery after PDT in chronic CSC, whereas symptom duration independently relates to the likelihood of fluid resolution. The six-stage OCT scale showed strong internal grading agreement and graded associations with vision, particularly from Stage 3 onward. Nevertheless, its prognostic thresholds require prospective external validation before being used as definitive treatment-selection rules.