Forest Mapper — Retinal Biomarkers in Schizophrenia Spectrum Disorders

Design: Systematic review + meta-analysis (44 studies pooled)
Design classification. PRISMA2020 systematic review and meta-analysis (registered PROSPERO CRD4202235795) synthesizing 65 studies (44 in meta-analysis: 37 OCT, 4 OCTA, 7 fERG) — Module F applies. Downgrade note: the source figures (Fig. 2A etc.) list per-study SMDs, but the individual study rows extracted from the PDF's embedded forest-plot image were too garbled (misaligned columns from OCR) to reliably re-cite one study per node without risking fabricated numbers. Per the module's own constraint ("never fabricate... display only what's stated"), this mapper instead treats each retinal/ERG parameter as the unit, using the paper's own clearly-stated pooled SMD, CI, and k (number of studies/eyes) for each (Komatsu et al., Mol Psychiatry 2024;29:464-482).
pRNFL average thickness
SMD -0.30 [-0.46,-0.14], k=26 (thinner in SSD)
Macular average thickness
SMD -0.52 [-0.74,-0.29], k=11 (thinner in SSD)
Macular GCL-IPL
SMD -0.41 [-0.57,-0.25], k=9 (thinner in SSD)
Optic cup volume
SMD +0.26 [0.05,0.46], k=4 (enlarged in SSD)
fERG a-wave (photopic)
SMD -1.03 [-1.49,-0.57], k=6 (reduced in SSD)
fERG b-wave (photopic)
SMD -0.28 [-0.44,-0.12], k=7 (reduced in SSD)
fERG a-wave (scotopic)
SMD -0.80 [-1.34,-0.27], k=4 (reduced in SSD)
fERG b-wave (scotopic)
SMD -0.73 [-1.07,-0.39], k=4 (reduced in SSD)
a-wave latency (photopic)
SMD -0.25 [-0.48,-0.03], k=6 (shorter in SSD — ns after adjustment)
Hub / textbook-path summary

The dominant cross-parameter signal: SSD participants show consistent structural thinning (pRNFL, macula, GCL-IPL) and consistent functional attenuation (all four fERG amplitude measures), forming a coherent "textbook path" of both trait- and state-level retinal change. This pattern survived sensitivity analyses excluding diabetes/hypertension for all parameters except pRNFL inferior/nasal thickness (Results, Komatsu et al. 2024).

Outlier / divergence summary
NodeWhy notable
Optic cup volume (OCV)The only parameter that goes the opposite direction (enlarged, not thinned) — an outlier relative to the otherwise uniform thinning pattern, and its significance disappeared after adjusting for publication bias (Table S5).
photopic a-wave latencyStatistically significant pooled estimate that became non-significant after removing one influential study (Fridel et al. 2022) and after publication-bias correction — a fragile, low-certainty node (GRADE "low").
pRNFL vs macular thickness and illness durationDisease duration correlates negatively with pRNFL thickness but positively with macular thickness — a genuine directional divergence between two structurally adjacent retinal regions, discussed at length as evidence of two different underlying processes (neurodegeneration vs macular edema) rather than a single linear trajectory.
References

Komatsu H, Onoguchi G, Silverstein SM, Jerotic S, Sakuma A, Kanahara N, Kakuto Y, Ono T, Yabana T, Nakazawa T, Tomita H. Retina as a potential biomarker in schizophrenia spectrum disorders: a systematic review and meta-analysis of optical coherence tomography and electroretinography. Mol Psychiatry. 2024;29:464-482. https://doi.org/10.1038/s41380-023-02340-4 (PROSPERO CRD4202235795)