Scientists have found that a dye-free imaging technique applied to donated human retinas can reliably distinguish the protein deposits associated with amyotrophic lateral sclerosis from those linked to Alzheimer’s disease. The findings, presented through conference abstracts published in the Alzheimer’s and Dementia journal supplement, suggest that polarized light interacts differently with TDP-43 aggregates found in ALS and frontotemporal lobar degeneration than with amyloid-beta clumps tied to Alzheimer’s. If the method can eventually be adapted for living patients, it could address a persistent clinical challenge: the two diseases sometimes mimic each other in early stages, and no simple, noninvasive test currently separates them at the protein level.
How Polarized Light Reads the Eye
The core technique works by shining polarized light through flat-mounted post-mortem retinas and measuring how each type of protein deposit alters the light’s properties. Because amyloid-beta and TDP-43 have different molecular structures, they bend and scatter polarized light in distinct ways. Researchers used a label-free imaging approach, meaning the retinal tissue did not need to be stained or chemically treated before scanning. That detail matters for any future clinical translation: removing dyes from the process cuts preparation time and avoids chemical artifacts that could muddy results.
When the team fed the polarized-light interaction features into machine learning algorithms, the classifiers accurately sorted deposits into their correct disease categories. The interactions with polarized light differed significantly between retinal amyloid-beta deposits and TDP-43 deposits, according to the abstract. This is not a single-variable test; the algorithms draw on multiple optical parameters at once, which helps reduce the chance of a false match and suggests that the underlying physics of each protein aggregate leaves a recognizable optical fingerprint.
Biological Basis in the Retina
The reason this approach is plausible at all rests on a growing body of evidence that the retina mirrors brain pathology in neurodegenerative disease. For Alzheimer’s, researchers demonstrated amyloid-beta immunoreactivity in flat-mounted post-mortem retinas of confirmed Alzheimer’s cases, alongside changes in melanopsin retinal ganglion cells. Separate work has shown that retinal tau pathology correlates with brain tangle scores, Braak stage, and cognitive test performance in patients with mild cognitive impairment due to Alzheimer’s and in those with Alzheimer’s dementia. These correlations strengthen the case that the eye is not just passively accumulating protein but reflecting the same disease process unfolding in the brain.
On the ALS side, post-mortem studies have confirmed that phosphorylated TDP-43 inclusions appear in the retinas of people who had frontotemporal lobar degeneration with TDP-43 pathology and ALS. A large neuropathology review published in Acta Neuropathologica documented these inclusions while also flagging important caveats about using retinal proteinopathy as a standalone biomarker. Not every case shows the same density or distribution of deposits, and the relationship between retinal and cortical TDP-43 burden is not yet fully mapped, making it risky to assume the eye always reflects the brain one-to-one.
The broader concept that the retina can serve as a “window to the brain” has been explored across several neurodegenerative conditions. For example, analyses of retinal nerve fiber layer thickness and ganglion cell integrity have been used to probe neurodegeneration in disorders such as multiple sclerosis and Parkinson’s disease, and a comprehensive overview in ophthalmic neuroimaging underscored how structural and functional retinal changes often parallel central nervous system damage. Within this framework, the new polarized-light work extends the idea from structural measures to direct optical signatures of misfolded proteins.
A Decade of Groundwork
The polarized-light method did not appear overnight. Earlier conference work, presented through the Optica community and related imaging meetings, documented polarimetry on thioflavin-positive amyloid deposits in human post-mortem Alzheimer’s retinas years before the current ALS comparison was attempted. That initial focus on Alzheimer’s amyloid detection established the optical signatures the team later used as a reference point when they turned to TDP-43. The progression from single-disease detection to cross-disease differentiation is what makes the latest work notable: it shifts the question from “can we see disease in the eye?” to “can we tell which disease we are seeing?”
In parallel, neuropathologists have been refining the understanding of TDP-43 itself. A landmark analysis in ALS pathology outlined how phosphorylated TDP-43 accumulates in specific neuronal populations and propagates across brain regions, providing a template for where clinicians and imaging scientists might expect to find related pathology in connected visual pathways. As retinal studies began to report pTDP-43 inclusions in ALS and frontotemporal lobar degeneration, those findings were interpreted against this backdrop of staged brain involvement.
A recent review in Frontiers in Neuroscience mapped the full range of studies that have and have not found retinal pTDP-43 pathology in ALS and frontotemporal lobar degeneration cases. The review cataloged differences by genetic subtype, retinal layer involvement, and comparison to brain pathology controls. That variability is a real obstacle: if TDP-43 deposits show up inconsistently across genetic backgrounds or disease stages, a polarized-light scan could miss cases or produce ambiguous readings. It also underlines why large, well-characterized tissue banks will be crucial for training and validating any machine learning models built on these optical signatures.
What Still Separates Lab From Clinic
The most important caveat is that all of the evidence so far comes from post-mortem tissue. No study has yet demonstrated that polarized-light retinal imaging can detect or differentiate these protein deposits in a living patient’s eye. Post-mortem retinas can be flat-mounted and imaged under controlled laboratory conditions that do not translate directly to an in-office eye exam. Living tissue introduces motion, blood flow, and optical noise from the lens and vitreous humor, all of which would need to be accounted for before the technique could work in a clinical scanner.
There is also no published head-to-head comparison between this polarized-light method and established diagnostic tools such as PET imaging for amyloid or cerebrospinal fluid assays for TDP-43. Those existing tests have their own limitations, including high cost and invasiveness, but they have years of validation data behind them. Any retinal-based alternative would need to demonstrate comparable sensitivity and specificity in prospective trials before clinicians could rely on it for treatment decisions or trial enrollment.
Longitudinal data is another gap. Researchers have not yet tracked how retinal protein deposits change over time within the same individual, either in the brain or in the eye. Without that information, it is hard to know whether polarized-light signatures emerge early enough to serve as a screening tool, or whether they mainly reflect late-stage disease. It is also unclear whether the optical properties of these aggregates shift as they mature, which could complicate machine learning models trained on a narrow disease window.
Technical hurdles loom as well. Translating flat-mount polarimetry into a patient-friendly scanner would likely require integrating polarization-sensitive optics into existing retinal imaging platforms such as optical coherence tomography or scanning laser ophthalmoscopy. Engineers would need to design systems that can capture subtle polarization changes at safe light levels and in a matter of seconds, while software compensates for eye movements and variable media clarity. Regulatory pathways would then demand rigorous multicenter trials, standardized acquisition protocols, and clear quality-control metrics.
Despite these challenges, the conceptual payoff is substantial. A noninvasive eye-based test that could distinguish ALS-related TDP-43 pathology from Alzheimer’s amyloid in the clinic would give neurologists and ophthalmologists a new tool for resolving ambiguous cognitive or motor presentations. It could help stratify patients for targeted therapies, monitor disease progression without repeated lumbar punctures, and potentially flag mixed pathologies that are easy to miss with symptom-based diagnosis alone.
For now, the work remains a proof of principle anchored in donated tissue. But by showing that polarized light can read out distinct optical fingerprints from different misfolded proteins in the human retina, the studies add weight to the idea that the eye can serve as a practical, accessible extension of brain pathology. The next steps (adapting the physics to living eyes, validating against gold-standard biomarkers, and testing performance in real-world clinical cohorts) will determine whether this promising laboratory technique can make the leap into everyday practice.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
