(IGR: 5-1) | International Glaucoma Review #

Selections from the Gullstrand Meeting

Subjective judgment of serial visual fields is NOT reproducible
Two general approaches for judging progressionr Series analyses
- series displays provided by perimeters
  - linear regression analysis
  - of global indices such as MD or PSD
  - point-by-point, such as Progressor
- Single field visit analyzed relative to two reliable baseline visual fields
  - clinical trials scoring methods
  - glaucoma change probability
Currently, no single algorithm for identifying and quantifying progression in serial visual fields has gained wide acceptance
Confirmation reduces number of progressed eyes, e.g., EMGT progression: 49%; after first confirmation: 14%; after second confirmation: 8%
New approaches to assess progression: more sensitive test procedures, improved perimetric programs, modeling progression, Machine Learning Classifiers.
Machine Learning Classifiers:
- are computer-based systems that learn from the data and develop their own classification algorithms
- can be supervised or unsupervised learning
- can learn complex patterns and trends in the data
- can adapt to create decision surfaces without the constraints imposed by statistical classifiers
- are unbiased and may identify meaningful patterns that experts may not see
Why is it so important to identify progression earlier?
- treatment works; visual function is the ultimate test for success
- prediction of progression not yet possible
- shorter RCTs

in large studies, the prevalence of open-angle glaucoma in persons 40 years and older is:
- -1%-3% in white populations
- ->3%-9% in black populations
the incidence of glaucoma per year varies between 0.1% and 0.6% in large studies of persons 40 years and older
the 4 years incidence in the Barbados study was 0.7% at IOP <= 17 mmHg and 18.3% at IOP > 25 mmHg (25 times higher relative risk)
high IOP was a major risk factor but (Barbados study):
1/4 OAG developed at IOP <= 13-19 mmHg
1/2 OAG developed at IOP <= 21 mmHg
2/3 OAG developed at IOP <= 25 mmHg
other risk factors: race, older age, higher IOP, lower perfusion pressure (blood pressure minus the IOP), family history, myopia; hypertension and diabetes are associated with higher IOP but not with OAG. future issues:
- increase of prevalence (aging, global demographic changes e.g. Asia, Africa)
- strategies to address modifiable and non-modifiable risk factors
- glaucoma screening and natural history
future research:
- standardized criteria for definitions
- improved diagnostic methods
- incidence and risk factors by race
- gene-environment interactions
- screening methods and effectiveness

In evidence-based studies 21% of visual fields were not eligible (mean of four OHT studies, including OHTS), 23% had no disc and retinal nerve fiber layer photos (mean of seven studies)
Profound experience is required for judging disc photographs, which may not be available in general ophthalmological practice
For confirmation of abnormality or progression, at least three visual fields (four RCTs) are needed with good reliability, same program, same instrument.
In OHTS, disc evaluation doubled the number of progressive cases; in EMGT, disc evaluation contributed little to overall progression. Why? (see Heijl in this issue of IGR)

Absolute risk reduction (ARR) in new progressive cases per year:
treated untreated ARR OHTS 1% 2% 1% EMGT 8% 10% 2%

The complete text of these three lectures will be available on a CD-rom to be issued by the Gullstrand organization.