June 2015
Volume 56, Issue 7
Free
ARVO Annual Meeting Abstract  |   June 2015
Patterns of early glaucomatous visual field loss and their evolution over time
Author Affiliations & Notes
  • Tobias Elze
    Harvard Medical School, Schepens Eye Research Institute, Boston, MA
    Max Planck Institute for Mathematics in the Sciences, Leipzig, Germany
  • Lucy Q Shen
    Mass. Eye and Ear Infirmary, Harvard Medical School, Boston, MA
  • Janey L Wiggs
    Mass. Eye and Ear Infirmary, Harvard Medical School, Boston, MA
  • Michael V Boland
    Wilmer Eye Institute, Johns Hopkins University, Baltimore, MD
  • Sarah Wellik
    Bascom Palmer Eye Institute, Miami, FL
  • Peter J. Bex
    Northeastern University, Boston, MA
  • Louis R Pasquale
    Mass. Eye and Ear Infirmary, Harvard Medical School, Boston, MA
  • Footnotes
    Commercial Relationships Tobias Elze, Provisional Patent No. 61/869,627 (P); Lucy Shen, None; Janey Wiggs, None; Michael Boland, None; Sarah Wellik, None; Peter Bex, Provisional Patent No. 61/869,627 (P); Louis Pasquale, None
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science June 2015, Vol.56, 3178. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Tobias Elze, Lucy Q Shen, Janey L Wiggs, Michael V Boland, Sarah Wellik, Peter J. Bex, Louis R Pasquale; Patterns of early glaucomatous visual field loss and their evolution over time. Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):3178.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract
 
Purpose
 

Early detection of glaucomatous visual field (VF) loss is important for the timely initiation of therapy. We applied a mathematical method to VF data from five clinical glaucoma practices to identify patterns of initial VF loss and their evolution over time.

 
Methods
 

For pattern learning, the most recent reliable automated VFs (fixation loss≤33%, false negative/false positive rates ≤20%, SITA Standard 24-2) for each eye with mean deviation (MD) from 0.5 to -5.5 dB were partitioned into 6 categories based on MDs (Fig. 1). A statistical learning procedure called Mixture of Gaussian Clustering was applied to pattern deviation (PD) values to identify a statistically optimal set of patterns. To assess the evolution of patterns over time, each VF with multiple tests was then computationally assigned to its closest pattern and compared to the patterns of all its available subsequent fields.

 
Results
 

98,140 VFs from 98,140 eyes were used to generate patterns of early VF loss. Fig. 1 illustrates that the optimal number of patterns increases from 2 (MD=0 dB, unspecific patterns) to 6 (MD=-3 dB, mild VF loss) and then converges to 5 (larger VF loss patterns). 172,201 VFs from 56,536 eyes (mean number of follow-up visits: 3.6, mean follow-up time: 3.3 years) were analyzed for evolution over time. We identified four classes of VF evolution (nasal, superior, paracentral and inferior). Fig. 2 shows each pattern within these classes together with its most frequent evolution in lower MD categories. Superior defects from MD=-2 dB to -4 dB either merge with nasal defects or with paracentral defects. Inferior defects occur initially at MD=-3 dB. Paracentral defects at MD=-2 dB can remain paracentral or develop inferior defects. There is a main effect of stability over the four classes (ANOVA, p=0.009), with nasal and inferior defects being most likely to change.

 
Conclusions
 

Our large VF dataset allowed us to mathematically identify typical patterns for early glaucomatous VF loss. The stability estimates for the specific patterns may help identify patients at high risk for visual loss at early disease stages where treatment may be most effective.  

 
Plots of computationally optimal pattern schemes for each MD category.
 
Plots of computationally optimal pattern schemes for each MD category.
 
 
Changes of the patterns over time for progressions to categories of lower MD.
 
Changes of the patterns over time for progressions to categories of lower MD.

 
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×