A new multicenter diagnostic study has developed and validated standardized OCT interpretation rules that maintained greater than 90% sensitivity and specificity across multiple countries and OCT platforms.
The findings suggest these decision-support tools could help clinicians distinguish true glaucomatous optic neuropathy from normal myopic anatomy with greater confidence.
Let’s first get some background on this
Myopia can make glaucoma surprisingly difficult to diagnose.
Case in point: Structural changes caused by elongated eyes often resemble glaucomatous damage on optical coherence tomography (OCT), meaning healthy patients may be incorrectly flagged as having disease.
- Features such as optic disc tilt, retinal nerve fiber layer (RNFL) bundle shifts and peripapillary atrophy can make healthy myopic eyes appear abnormal on OCT, contributing to unnecessary referrals and treatment
Even further: Myopic eyes may also have thinner baseline RNFL measurements and temporally displaced nerve fiber bundles that do not align well with standard reference patterns.
Go on …
Many commercial OCT normative databases include relatively few moderate or highly myopic eyes.
In fact, previous research evaluating a myopia-specific normative database found that using reference data derived from myopic eyes improved specificity without reducing sensitivity, underscoring how conventional databases can generate false-positive abnormalities in this population.
And with this in mind … the investigators behind this current study noted that prior reports found false-positive results in as many as 43% of healthy myopic eyes.
Tell me more about the new study—what was investigated?
Researchers evaluated whether a set of prespecified OCT interpretation rules could more accurately distinguish glaucomatous optic neuropathy (GON) from normal structural changes caused by nonpathologic myopia.
- Unlike approaches that rely primarily on device-generated color coding or normative databases, the investigators developed expert-derived rules focused on recognizable structural patterns in peripapillary RNFL and macular ganglion cell-inner plexiform layer measurements.
The goal: not to replace a comprehensive glaucoma assessment, but rather to create practical decision-support criteria that could be applied across different populations and OCT devices.
And how was it designed?
The multicenter diagnostic study consisted of two sequential phases:
First: 31 glaucoma and ocular imaging experts participated in a modified Delphi process to establish standardized OCT diagnostic rules before the validation analysis began.
Those prespecified rules were then tested in a cross-sectional diagnostic validation involving participants recruited from five institutions across:
- China
- Hong Kong
- Taiwan
- United States
- India
Investigators then evaluated five OCT-based rules against a clinical reference diagnosis established by masked experts.
Tell me more about the participants.
The study included: 943 adults (1,525 eyes; mean age: 45.9; 4.6% female) with moderate or high nonpathologic myopia.
Among all eligible eyes:
- 814 had confirmed glaucomatous optic neuropathy
- 711 served as myopic controls without glaucoma
A couple of notes:
- Moderate or high myopia was defined as a spherical equivalent (SE) of −3.00 diopters (D) or worse or an axial length of at least 24.5 mm.
- Eyes with pathologic myopia—including posterior staphyloma or myopic maculopathy of category 2 or higher—were excluded
What OCT rules were evaluated?
Among the five predefined diagnostic rules:
Rule A: the highest-performing qualitative rule looked for a focal dip or broad depression in the temporal-superior-inferior-nasal-temporal (TSNIT) RNFL curve.
- The researchers defined a dip as a steep, localized downward deflection within the superior or inferior peak zone. A depression represented broader flattening or loss of the curve's expected double-hump pattern.
And the other four?
The remaining quantitative rules evaluated:
- Rule B: inferior peripapillary RNFL thinning
- Rule C: inferotemporal macular ganglion cell-inner plexiform layer thinning
- Rule D: either Rule B or Rule C
- Rule E: both Rule B and Rule C
By comparing these rules, the investigators could assess whether a visual pattern-based approach performed better than isolated thickness measurements or combinations of quantitative abnormalities.
So what did they find?
The TSNIT morphology rule produced the strongest diagnostic performance.
In the internal validation cohort, it achieved:
- 96% sensitivity
- 95% specificity
In the multiethnic external validation cohort, it maintained:
- 93% sensitivity
- 93% specificity
The combined quantitative rule identifying either inferior RNFL thinning or inferotemporal mGC-IPL thinning also performed well in the external cohort, achieving:
- 90% sensitivity
- 93% specificity
In practical terms, the TSNIT rule correctly identified most eyes with glaucomatous optic neuropathy while also correctly classifying most myopic eyes without glaucoma.
Did the rules work in more difficult cases?
Yes, although performance varied somewhat with disease severity.
The rules remained accurate in highly myopic eyes and in eyes classified as having mild glaucoma, where physiological myopic changes may be especially difficult to separate from true disease.
For highly myopic eyes in the external cohort, the TSNIT rule maintained sensitivity of approximately 96% and specificity of approximately 93%.
What else?
The investigators also examined the relatively small number of misclassified eyes.
False-positive TSNIT findings were often related to physiological splitting of the RNFL bundles, which could create a dip resembling focal glaucomatous loss. False negatives tended to occur in eyes with mild or subtle structural damage.
An accompanying published comment on the study noted that all confirmed glaucoma cases had reproducible visual field (VF) abnormalities.
- Therefore, the reported accuracy reflects established perimetric glaucoma, while performance before detectable VF loss remains uncertain.
Why does this matter clinically?
OCT software generally compares a patient's measurements with an internal normative database and highlights values falling outside the expected range.
In myopic eyes, however, structural differences may cause healthy measurements to appear abnormal even when glaucoma is absent.
Give me some context on this.
Previous studies have shown that incorporating myopic or long-axial-length normative databases can reduce false-positive glaucoma classifications.
However: Specialized databases are not available on every device—and no single normative adjustment eliminates all of the diagnostic challenges created by optic disc tilt, axial elongation or altered RNFL distribution.
Rather than relying only on thickness values or color-coded warnings, the TSNIT rule evaluates the shape and architecture of the RNFL profile.
This could give clinicians another way to judge whether an abnormal-looking report reflects genuine glaucomatous loss or a physiological variation associated with myopia.
I’m sensing a but …
Still, the authors positioned the rules as decision-support tools rather than stand-alone diagnostic tests.
They reasoned: OCT findings should continue to be interpreted alongside optic nerve examination, intraocular pressure (IOP), VF testing, clinical history and evidence of progression.
Any limitations?
Yes—several. Among them:
The study excluded patients with pathologic myopia, so the findings should not be generalized to eyes with posterior staphyloma, advanced myopic maculopathy or other severe structural complications.
Additionally, all eyes classified as having GON had reproducible VF defects.
- What means: The study did not establish whether the rules perform equally well in preperimetric glaucoma, when structural damage may be present before standard VF loss becomes detectable.
The cross-sectional design also evaluated diagnostic accuracy at one point in time.
- Prospective longitudinal studies will be needed to determine whether these rules can predict structural progression or future VF conversion.
So what's next?
Although these standardized OCT rules performed well across multiple countries and imaging platforms, additional validation will be needed before they become part of routine clinical practice.
Moving forward: Future research should determine how well the rules perform in preperimetric glaucoma, where structural damage may precede detectable VF loss, as well as in patients with pathologic myopia, who were excluded from the current study.
Investigators will also need to evaluate whether these criteria can be incorporated into commercial OCT software as automated decision-support tools to improve consistency in everyday clinical practice.