That matters because a model centered on intraocular pressure alone cannot describe all the biological processes visible in the eye.
The evidence is promising, but its scope is specific. These studies measure proteins in aqueous humor, not a complete retinal or optic-nerve proteome, and they do not yet establish a screening test for people without a clinical diagnosis. The useful question is therefore not whether a single protein “detects glaucoma,” but how molecular patterns change across disease stages, and what those patterns can—and cannot—tell us.
Molecular signatures in early glaucomatous aqueous humor
Aqueous humor is a practical research specimen: it is accessible during eye procedures and contains proteins that can reflect activity within ocular tissues. It is not, however, a direct molecular census of the retina or optic nerve. A protein detected in the fluid may be produced locally, released from tissue, transported into the compartment, or altered by several interacting processes. Interpreting its abundance requires that distinction.
Comparative proteomic work has reported differentially expressed proteins in aqueous humor from people with primary open-angle glaucoma (POAG) and age-matched controls. In one study using hyper reaction monitoring mass spectrometry, 87 proteins differed between groups: 34 were significantly upregulated and 53 downregulated. That is a broad signal, not a ready-made clinical panel. The count describes differences observed in a particular study design and cohort; it does not establish that every protein will reproduce across populations, platforms or sample-handling conditions.
Targeted affinity proteomics adds another view. An Olink-based analysis comparing glaucoma patients with cataract controls identified 84 differentially expressed metabolic proteins. The control group matters: cataract controls are not equivalent to a general healthy population, and the composition of the comparison group influences which differences emerge. A signature that separates two study cohorts is not automatically specific to glaucoma in every clinical setting.
The signals are also not limited to one biological category. Reported changes touch metabolic pathways, oxidative-stress responses, inflammatory processes and extracellular-matrix remodeling. This is consistent with glaucoma molecular pathology research that treats the disease as a set of interacting tissue responses rather than a single pressure-mediated event. It does not prove that each pathway initiates disease, or that a fluid measurement identifies the original site of injury.
Aqueous humor proteomics can map a molecular response in the eye. It cannot, by itself, locate every source of that response.
Metabolic dysregulation and the ALDH1A1 signal
Metabolic findings are among the more prominent themes in the available data. In targeted proteomic profiling, ALDH1A1 was highlighted as an elevated protein associated with retinal ganglion cell apoptosis. The result positions ALDH1A1 as a candidate for further mechanistic work; it does not establish the protein as a validated diagnostic biomarker or show that its elevation is causal in human glaucoma.
That distinction is operationally important. A disease-associated protein may be a driver, a compensatory response, a marker of stressed tissue, or some combination of these. Abundance alone does not resolve direction of causality. Nor does a change in aqueous humor necessarily mean that retinal protein expression has changed in the same direction or by the same magnitude. Fluid and tissue measurements answer related but different questions.
Other reported proteins point toward mitochondrial and insulin-signaling processes. In a comparison of early and advanced glaucoma, ELISA validation confirmed expression differences in a set of 14 proteins that included IDH3A, SUCLG2, mTORC1 and mTORC2. The study’s interpretation implicated mitochondrial dysfunction and insulin signaling alongside neuroinflammation. These findings widen the mechanistic map, but they should not be read as a settled pathway sequence. The available evidence does not establish which molecular changes come first in presymptomatic disease.
A practical reading of these metabolic signals is therefore restrained:
- ALDH1A1 is a candidate linked in targeted profiling to retinal ganglion cell apoptosis, not a stand-alone clinical test.
- IDH3A and SUCLG2 connect the reported differences to mitochondrial biology.
- mTORC1 and mTORC2 appear among proteins whose expression differences were confirmed by ELISA in a disease-stage comparison.
- AQP4 and other altered proteins sit within a larger pattern that spans more than one pathway.
The value lies in the pattern and its testability. Follow-up work can ask whether the same proteins recur in independent cohorts, whether they track with defined clinical features, and whether tissue-based measurements support the interpretation made from aqueous humor.
Neuroinflammation and extracellular-matrix remodeling
The inflammatory component is not a side note in the reported proteomic profiles. In the early-to-advanced comparison, ELISA validation confirmed differences among proteins including IL2, IL4 and TNFα, alongside AQP4, TGFB2 and others. The broader interpretation implicated neuroinflammation. This supports a model in which immune signaling participates in the ocular molecular environment, but it does not show that inflammation alone explains glaucoma progression.
Extracellular-matrix remodeling is another relevant dimension. Changes in matrix-associated proteins may reflect altered tissue structure, repair responses or ongoing stress. In glaucoma, those possibilities matter because the disease affects interconnected ocular structures, while aqueous humor samples only one fluid compartment. A matrix-related signal in the fluid cannot be mapped directly onto a specific tissue location without complementary data.
The same caution applies to oxidative-stress responses. Proteomic studies report alterations in pathways associated with oxidative stress, but pathway-level interpretation is not equivalent to measuring a single causal event. A protein may rise as part of a protective response, fall because a process is impaired, or change as a consequence of tissue damage. A list of differentially expressed proteins does not resolve those alternatives by itself.
These distinctions are especially important when discussing proteomic biomarkers for optic neuropathy. A biomarker can be useful for classification without explaining mechanism; a mechanistically interesting protein may have little value for diagnosis. The field needs both forms of evidence, but they should not be conflated. For now, aqueous humor profiling is better understood as a way to characterize molecular states and generate hypotheses than as a clinically established means of identifying early glaucoma in asymptomatic people.
Comparing early and advanced disease
Disease-stage comparisons add a useful axis to case-control studies. One quantitative proteomic study compared 20 people with early glaucoma and 20 with advanced glaucoma. Early disease was defined by a Humphrey visual field 24-2 mean deviation between −6 and −12 dB; advanced disease was defined by a mean deviation below −20 dB. Across the comparison, 53 proteins changed with severity: 21 were downregulated and 32 upregulated.
The distribution is important. Progression was not represented as a uniform increase in all measured proteins. Some signals rose and others fell. That pattern is compatible with shifts in cellular stress, metabolism, inflammation and tissue remodeling, but it does not reveal a simple molecular clock. Nor can a cross-sectional comparison establish the temporal order in which those proteins changed in each individual.
| Evidence frame | Main result | What it supports | What it does not establish |
|---|---|---|---|
| POAG versus age-matched controls, HRM-MS | 87 differentially expressed proteins: 34 upregulated and 53 downregulated | Broad aqueous humor proteomic differences associated with POAG in the studied cohorts | A universal glaucoma signature or an individual diagnostic test |
| Glaucoma versus cataract controls, Olink targeted profiling | 84 differentially expressed metabolic proteins; ALDH1A1 was highlighted | A metabolic signal worth testing in additional cohorts | That ALDH1A1 alone causes disease or is clinically adopted for screening |
| Early versus advanced glaucoma | 53 proteins differed with severity: 21 downregulated and 32 upregulated | Proteomic composition varies across defined clinical stages | The presymptomatic sequence or the trajectory within an individual |
| ELISA validation in a stage comparison | Differences confirmed for 14 proteins, including AQP4, IDH3A, IL2, TGFB2 and TNFα | Selected discoveries can be tested with an orthogonal assay | That all candidates will generalize across platforms and populations |
The stage data give molecular profiling a more informative structure than a simple case-control contrast. But the clinical categories are still based on visual-field status. They do not answer whether the same protein changes arise before measurable field loss, or whether a particular pattern distinguishes primary open-angle glaucoma from normal-tension glaucoma at an early stage. Those remain open questions.
Stage-associated protein changes are evidence of molecular variation across groups, not a timeline of disease in one person.
The measurement pipeline: from sample to interpretable signal
Proteomic results depend on more than the assay platform. The sample is part of the measurement system. Collection context, processing, storage, protein degradation and cohort definition can all shape the observed signal. In ocular research, aqueous humor and donor eye tissue also have different procurement pathways and different analytical limits. A fluid sample may be available in a clinical workflow; postmortem tissue profiling must contend with procurement latency and the condition of the specimen at collection.
That matters for donor eye proteomic profiling and for attempts to connect fluid biomarkers to retinal or optic-nerve biology. Aqueous humor studies can identify candidate protein changes, but they do not substitute for spatially resolved tissue analysis. Conversely, donor tissue can provide anatomical context that fluid lacks, while introducing its own constraints around postmortem interval, tissue integrity and donor metadata. The two evidence streams are complementary, not interchangeable.
Assay choice also affects what becomes visible. Mass spectrometry can survey many proteins and support quantitative comparisons; targeted affinity platforms focus on selected analytes and may offer a different balance of sensitivity and coverage. ELISA validation provides an independent check for chosen candidates, but validating a subset does not confirm the whole discovery profile. Differences across platforms may reflect assay design and analyte coverage as well as biology.
A robust research pipeline therefore needs to preserve the links between specimen and interpretation:
1. Define the clinical comparison. POAG versus age-matched controls, glaucoma versus cataract controls, and early versus advanced disease are distinct questions. Their results should not be merged as if the cohorts were interchangeable.
2. Record the phenotype precisely. Visual-field criteria, diagnostic category and disease stage determine what a protein difference means in context.
3. Keep the specimen type explicit. Aqueous humor measurements should not be described as direct retinal protein expression unless paired tissue evidence supports that claim.
4. Separate discovery from validation. A broad proteomic screen generates candidates; targeted follow-up tests whether selected differences persist.
5. Treat biomarker performance as a separate endpoint. Differential expression is not equivalent to validated sensitivity, specificity or clinical utility.
For retinal protein expression in glaucoma, this separation is central. Aqueous humor can carry signals associated with ocular disease, but it cannot provide a complete map of retinal cell states. Single-cell RNA sequencing, tissue proteomics and other molecular approaches can address different layers of the problem, though each brings its own sampling and interpretation constraints. A protein-level result in fluid should not be silently translated into a cell-specific retinal claim.
What the evidence supports—and where it stops
The current findings support a measured conclusion: ocular proteome changes in early glaucoma include signals associated with metabolism, oxidative stress, neuroinflammation and matrix remodeling. Multiple approaches have reported substantial numbers of differentially expressed proteins, and selected proteins have been validated with ELISA. The convergence is enough to justify deeper study of molecular states beyond an intraocular-pressure-only model.
It is not enough to claim that one protein can screen for early glaucoma, that the observed proteins define a universal signature, or that the temporal sequence of presymptomatic changes is known. The available evidence also does not establish whether an aqueous humor profile can reliably distinguish early primary open-angle glaucoma from normal-tension glaucoma without clinical visual-field information.
The next analytical step is not a larger list of candidates for its own sake. It is replication across well-described cohorts, explicit separation of discovery and validation, and coordination between fluid proteomics and donor-tissue molecular profiling. That is how a candidate signal becomes interpretable across sample types and research settings.
For now, aqueous humor proteomics offers a useful map of biological activity associated with glaucoma, not a stand-alone diagnostic instrument. Its strongest contribution is to broaden the research model: pressure remains relevant, but metabolic, inflammatory and tissue-remodeling signals also belong in the account.