New sensors, vascular images and AI can add information. Confirm reliable pressure, structural and visual-field findings before changing management; a small association study or a high AUC does not establish better vision.
Terminology and draft status
Selected medical terms were checked in the WHO UMD. Further terms remain working translations pending terminology and specialist review; source, translation and medical approval remain pending.
01Keep the clinical framework while evaluating new tests
The original update groups nine studies under pressure, visual fields and morphology. Start with the suspected disease, current damage and rate of change. Goldmann applanation tonometry (GAT), gonioscopy, optic-disc and retinal nerve-fibre-layer (RNFL) assessment, macular imaging and reliable perimetry remain complementary parts of the clinical evaluation. 10 11
A comparison of two instruments asks about agreement, not merely correlation. An imaging association asks whether findings vary together; it does not by itself establish what happens first or whether treatment based on that marker preserves vision. Assess selection, confounding, uncertainty and applicability before adopting a new test. 1 2 3 4 5 8 9
The DOG/BVA S2e guidance concerns adult open-angle-glaucoma risk and early detection, not every management decision in established glaucoma. It supports confirming a suspicious screening result with further assessment. The AAO benchmark likewise combines examination, fields and structural information; none of these documents validates a standalone research biomarker. 10 11 12
02Check pressure after DMEK without losing sight of the graft
Descemet membrane endothelial keratoplasty (DMEK) can be followed by pressure elevation. The retrospective study reviewed 415 patients and identified 54 affected eyes in 54 people, or 13%, excluding pre-existing glaucoma. Its definition was pressure >22 mmHg or an increase >10 mmHg from baseline; these are study criteria, not universal treatment thresholds. 1
Peripheral anterior synechiae (PAS) were associated with medication use and surgery. The reported odds ratio of 6.2, with 95% confidence interval 1.11–34.44, concerns surgery; assigning it to de-novo pressure elevation in general would change the outcome. Pseudoexfoliation, called PSX in the original, and prolonged steroid exposure also require attention. 1 12
After switching from dexamethasone to loteprednol etabonate 0.5%, 23 patients, or 42.5% of the affected group, had pressure control without pressure-lowering medication. This uncontrolled observation does not justify a routine switch for everyone. The clinician must balance pressure, inflammation and graft survival. 1
The original adds that 14.8% required surgery, gives an Ahmed implant as an example, and describes poorer visual improvement with PAS. Those details remain unverified without the full report. Preserve them as source questions rather than a guaranteed course; monitor postoperative pressure and investigate the mechanism before selecting treatment. 1
03Read agreement limits before exchanging tonometers
The iCare IC200 study compared rebound tonometry (RBT) followed by GAT in 402 right eyes. Spearman correlation was 0.885, but correlation does not establish interchangeability. The average RBT-minus-GAT difference was about 0.11 mmHg, with 95% agreement limits from −4.06 to +4.29 mmHg. 2
Greater central corneal thickness (CCT) was associated with higher RBT relative to GAT. Higher GAT pressure and cataract-surgery history were associated with lower RBT relative to GAT. Fixed measurement order and the actual pressure distribution matter; GAT is the comparator here, not a direct measurement of true pressure. 2
The original specifies >30 mmHg and highlights children and supine measurement. The inspected abstract does not establish that threshold or performance in those settings. Check the full protocol and the actual model instructions; interpret discordant results with the cornea, lens status and clinical findings rather than converting one reading by a generic correction. 2 10 11
04Distinguish repeated EYEMATE measurements from proven clinical benefit
EYEMATE-IO is a sulcus-based implanted pressure sensor. The reported analysis followed eight eyes from eight people in the prospective, open-label ARGOS-03 follow-up. Participants obtained at least four self-measurements at different times daily; this study schedule should not be described as uninterrupted 24-hour recording. 3
The original describes two years and 7282 measurements per patient. The actual abstract specifies a minimum two-year follow-up, a mean of 2.88 years and 7283 mean self-measurements. These differences should remain visible in source comparison rather than silently presenting the original shorthand as the study result. 3
Peak pressure and pressure fluctuation between consecutive optical coherence tomography (OCT) examinations were associated with concurrent RNFL thinning in mixed-effects models. The original gives 0.12 µm/year additional loss per 1 mmHg higher peak; the abstract displays a coefficient of −0.11 and an internally inconsistent confidence interval. Do not silently repair the interval or use an unverified slope as a decision threshold. 3
No association was found for the in-office mean GAT measurements in that small sample. A nonsignificant comparator association does not demonstrate general GAT inferiority or prove that implant-guided treatment preserves vision. Selection, treatment changes, model stability and independent validation still matter. 3
Repeated self-measurement may reveal pressure patterns missed at a visit. Any clinical use must be weighed against implantation burden, complications, availability and whether the result would change management. This small structural-outcome analysis alone is not an indication for sensor implantation or pressure-lowering surgery. 3 10 11
05Interpret vascular change with the structural and visual-field course
The choroidal microvascular dropout (MvD) study selected 103 eyes from 92 people with substantial macular retinal ganglion-cell damage at baseline. Over about three years, it compared eyes without central visual-field defects, eyes developing new defects and eyes with defects already present. It did not study an unselected healthy screening population. 4
Eyes developing new central defects showed a mean MvD angular increase of 17.8°, with enlargement toward the disc–fovea axis. This accompanied the functional change. Similar baseline RNFL and ganglion-cell–inner plexiform-layer (GCIPL) group averages do not prove that structural damage was absent or that vascular injury universally came first. 4
The investigators accounted for two-eye clustering but excluded poor-quality scans; the deep slab and projection artifacts still limit interpretation. Central defects were not evaluated with 10-2 testing. The association does not establish that escalating treatment because of MvD prevents field loss. 4
A separate retrospective study followed 187 primary open-angle glaucoma eyes for a mean of 3.67 years. Visual-field progression occurred in 29.9%, alongside an association between optic-disc vessel-density (ODVD) reduction and progression. The inspected abstract does not define that percentage as central-only progression. 5
Both vessel-density change and RNFL thinning were associated with progression in adjusted analyses. The original claims of a superior intradisc marker and “function before structure” require the full definitions and analyses. Review circumpapillary and intradisc signals, segmentation, scan comparability and the actual visual fields; do not treat an OCTA association as a validated escalation algorithm. 5 10 11
06Use myopic disc orientation to check the matching field region
The myopic open-angle-glaucoma study followed 53 eyes from 53 people over three years. Inferior disc rotation occurred in about 70%, and rotation corresponded with the visual-field defect in about 77%. Describe the actual disc orientation and compare it with the field rather than diagnosing progression from shape alone. 6
Inferior rotation with a corresponding superior defect was associated with faster progression in the superior peripheral region; superior rotation with a corresponding inferior defect showed the converse regional association. The original statement that speed was unaffected is too broad. Regional change must be distinguished from a global field index. 6
The original suggests 90° perimetry. That testing rule is not verified from the inspected abstract. Choose reliable central and peripheral testing for the clinical question, and check the full study for field strategy, confounding and multiplicity before applying a universal protocol to myopic eyes. 6 10 11
07Use AS-OCT to complement a complete angle assessment
The swept-source anterior-segment OCT (AS-OCT) study assessed 125 eyes from 68 Thai participants already suspected of narrow angles using Van Herick screening. It used CASIA SS-1000 images before contact tests, with masked image readers and gonioscopy as the reference. This selected sample differs from population screening. 7
The qualitative analysis reported 92.16% sensitivity and 73.91% specificity. Quantitative iris–trabecular contact (ITC) analysis at ≥55% reported 86.36% and 92.86%, respectively, but used only 58 quality-selected eyes. The original rounded 90% detection claim should not replace the stated definition and denominator; these are not universal cutoffs. 7
The original table is retained below with its comparator columns. Overall interobserver kappa was 0.69 for AS-OCT and 0.55 for gonioscopy; AS-OCT intraobserver values were 0.71–0.80. Gonioscopy agreement was based on only 12 eyes. More consistent reading of one selected image set does not by itself prove superior management or outcomes. 7
Lid shadowing and uncertain scleral-spur identification particularly affected superior and inferior assessment. Bilateral observations, selected usable images and an imperfect reference contribute uncertainty. The source also has unresolved abstract-versus-table totals and table inconsistencies; do not silently repair these or treat reported precision as independent external validation. 7
Use AS-OCT for a defined supplementary question and confirm the clinical angle mechanism. EGS6 states that anterior-segment imaging cannot replace gonioscopy; indentation distinguishes appositional from synechial closure. PAS, pigment and secondary causes may be missed by imaging alone, and device-dependent measurements do not automatically determine treatment. 7 10 11
| Metric | Qualitative AS-OCT | Quantitative AS-OCT, ITC ≥55% | Gonioscopy reference |
|---|---|---|---|
| Sensitivity | 92.16% | 86.36% | — |
| Specificity | 73.91% | 92.86% | — |
| Observer agreement | Interobserver kappa 0.69; intraobserver 0.71–0.80 | — | Interobserver kappa 0.55; selected subset |
08Keep iris vessel density at the level of an exploratory biomarker
The iris pilot compared 31 glaucoma eyes with 30 normal eyes using anterior-segment OCT angiography (AS-OCTA). It included both open-angle and angle-closure glaucoma. Subtype denominators differ between the abstract and a later table; the pooled comparison should not be silently relabelled as a validated result for one subtype. 8
Iris vessel density was lower across the examined regions. The nasal 6–12 mm region had an area under the receiver operating characteristic curve (AUROC) of 0.87, followed by a temporal region. This describes discrimination in a small selected sample, not 87% clinical accuracy, an individual risk estimate or proof of earlier useful diagnosis. 8
The original reports a negative cup-to-disc-ratio association and no association with pressure or RNFL thickness; those specific claims remain unverified in the inspected material. Image quality, angle anatomy, ethnicity, treatment and multiple region comparisons affect applicability. Same-operator repeatability and post-hoc power do not remove selection bias or establish prognosis. 8
Quantitative iris blood-flow imaging is different from unsupported iridology. Do not use the original aside to imply validation of iris-based alternative diagnosis. This cross-sectional work cannot establish that following iris density improves vision or supplies a routine treatment threshold. 8
09Read the AI test design before accepting a high AUC
The original “Girira, Atlanta” example corresponds to Michaël J. A. Girard and colleagues. Their retrospective cross-sectional study converted segmented OCT tissue boundaries into three-dimensional point clouds. An encoder compressed the shape, a decoder reconstructed it and a classifier separated healthy, highly myopic, glaucomatous and highly myopic–glaucomatous eyes. 9
The study recruited 754 subjects and analysed 685 usable scans. It excluded pathological myopia and inadequate imaging; all scans used one Spectralis device in a selected Singapore-Chinese cohort. These exclusions matter when considering the difficult myopic eyes encountered in routine practice. 9
The 70%/15%/15% training, validation and test split separated subjects. Five-fold evaluation reported micro-average AUC 0.92 ± 0.03, pooling classification decisions across classes. This is internal discrimination, not 92% accuracy or external clinical validation. The reported spread is not automatically a confidence interval for future patients. 9
RNFL morphology and optic-cup shape, including excavation depth, contributed to class separation. Synthetic changes in a latent representation help describe structural phenotypes; they are not observed disease progression. Segmentation quality, disease severity, selection and generalisation across devices still require attention. 9
The original claim of superiority over standard single-parameter methods requires the actual comparator and relevant endpoints. No better visual outcome or solved myopic diagnostic problem was established here. The source discloses Girard’s co-founder role in Abyss Processing; the supplement and cited underlying reports have not been independently appraised. 9
10Let reliable clinical change guide the next step
If pressure, morphology and visual fields disagree, review acquisition quality, cornea, segmentation, reliability and previous surgery, then compare a reliable time course. Repeated measurements can add context, but an isolated vascular image or AI label should not automatically trigger escalation. 2 3 4 5 10 11
Retain the three complementary pillars: pressure, functional perimetry and structural assessment. Sensors and vascular or anterior-segment imaging can answer selected questions; distinguish their research promise from demonstrated patient benefit. Follow-up and treatment remain individual decisions based on risk, confirmed damage and progression. 10 11 12
The original Christian Mardin lecture attribution, recording and slides are not independently verified. This research discussion complements the existing diagnostic fundamentals and retains every original study topic. Open numerical, source and translation questions remain explicit; urgent clinical assessment must not wait for a research test. 10 11
What are the main learning points of this update?
Combine pressure, fields and anatomy. Read agreement limits, identify the actual outcome and denominator, and separate a selected association or internal AUC from a validated clinical decision. Preserve uncertainty when a full source or protocol has not been checked.
Which findings need urgent assessment?
Sudden vision loss, a new field defect, severe pain, marked redness or inflammation, pressure-related symptoms or suspected infection or neurological disease require prompt assessment. Do not wait for a sensor, scan or AI result.
References and access status
- 1Asfuroğlu et al. · DMEK pressure · epub2025/issue2026
Complete actual NCBI abstract read; retrospective cohort. Full tables,14.8%surgery/implant details and treatment selection unread.
- 2Zhang et al. · iCare IC200 versus GAT ·2025
Complete actual NCBI abstract read; fixed measurement order. Full protocol,>30mmHgsubgroup and device instructions unread.
- 3EYEMATE-IO · Pressure and RNFL thinning ·2025
Complete actual NCBI abstract read;eight participants/repeated self-measurements. Full report/protocol and inconsistent displayed CI require checking.
- 4Lee et al. · MvD and central field ·2025
Complete actual abstract and selected main methods/results/discussion/limitations read;actual images,all tables/supplements and underlying studies unread.
- 5Suh et al. · Disc vessel density ·2025
Complete actual NCBI abstract read. Full outcome definitions,confidence intervals,model scaling and supplements unread.
- 6Tu et al. · Myopic disc rotation ·2025
Complete actual NCBI abstract read;regional associations. Full analyses,confounding and90°testing claim unchecked.
- 7Yaisiri et al. · AS-OCT and gonioscopy ·2025
Actual abstract/main text and tables1–6text read;selected125eyes/68participants and58ITCeyes. Actual images and underlying reports unread;source discrepancies unresolved.
- 8Li et al. · Iris vessel density ·2025
Complete actual abstract and selected publisher methods/results/discussion text read. Full tables/actual images,subtype discrepancy and original correlations remain open.
- 9Sharma et al. ·3D optic-nerve phenotype ·2026
Complete actual abstract;author-manuscript methods2.1–2.5/results3.1/tabletext/selected discussion/funding read. Not all results,actual images,supplement or cited reports.
- 10EGS6 · Examination and angle imaging ·2025
Actual selected text II.1.2.4 p91and dynamic gonioscopy p87reread;not a new full-guideline appraisal.
- 11AAO · Glaucoma Summary Benchmarks · December2025
Actual complete nine-page summary text read. A summary,not the still-unread complete PPP;no new dose or universal schedule imported.
- 12DOG/BVA S2e045-015 · Risk and early detection ·2024
Actual selected scope/method text pp1–3and screening/confirmation text pp20–22read through the primary PDF web representation. Not the whole guideline/report;separate direct download returned HTML.
For patients
Pressure readings, field tests and eye images examine different aspects of eye health. A new device or a detailed scan can add useful information, but no single result replaces your clinician’s examination and the pattern over time.
Small research studies may find associations without proving that the new test improves sight. Your clinician considers image quality, previous surgery, medicines and your individual risk before changing treatment. Do not change prescribed drops or steroids on your own.
Seek urgent care for sudden loss of vision, a new visual-field defect, severe eye pain or marked redness. These educational notes do not replace individual assessment or a treatment plan.
Seek urgent medical assessment for sudden vision loss, a new field defect, severe pain, marked redness or inflammation, pressure-related symptoms or suspected infection or neurological disease. Do not wait for a routine research test or imaging appointment.