June 2015
Volume 56, Issue 7
Free
ARVO Annual Meeting Abstract  |   June 2015
Trend-based Progression Analysis (TPA): A New Algorithm for Visualizing the Topology of Progressive Retinal Nerve Fiber Layer (RNFL) Thinning in Glaucoma
Author Affiliations & Notes
  • Christopher Kai-Shun Leung
    Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, Hong Kong, Hong Kong
  • Chen Lin
    Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, Hong Kong, Hong Kong
  • Marco Yu
    Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, Hong Kong, Hong Kong
  • Footnotes
    Commercial Relationships Christopher Leung, Carl Zeiss Meditec (C), Carl Zeiss Meditec (F), Carl Zeiss Meditec (R); Chen Lin, None; Marco Yu, None
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science June 2015, Vol.56, 3980. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Christopher Kai-Shun Leung, Chen Lin, Marco Yu; Trend-based Progression Analysis (TPA): A New Algorithm for Visualizing the Topology of Progressive Retinal Nerve Fiber Layer (RNFL) Thinning in Glaucoma. Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):3980.

      Download citation file:


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

      ×
  • Supplements
Abstract
 
Purpose
 

To evaluate and compare the performance of TPA with Guided Progression Analysis (GPA) (a commercially available event analysis for RNFL progression) to detect progressive RNFL thinning in glaucoma patients.

 
Methods
 

244 eyes of 140 glaucoma patients were followed 4 monthly for RNFL imaging with the Cirrus HD-OCT and visual field (VF) testing over a mean of 5.8 years (range: 4.0-6.9 years). Individual superpixel RNFL thickness data from serial RNFL thickness maps (50x50 superpixels) were exported for linear regression analysis. Each eye had a mean of 17 visits for analysis. To minimize Type 1 error for multiple testing, TPA controlled the false discovery rate (FDR) (area detected with false positives) at ≤5%. The impact of TPA and GPA on VF progression (determined by the EMGT criteria) was examined with Cox proportional hazards modeling. The survival probabilities of eyes detected with progression by TPA and GPA were compared by Kaplan-Meier estimator. The specificities of TPA and GPA were calculated from 20 normal eyes followed for 8 consecutive weeks.

 
Results
 

TPA detected significantly more eyes with progressive RNFL thinning compared with GPA (121 eyes (49.6%) and 56 eyes (23.0%), respectively). While the inferotemporal sector was the most frequent location where progression was detected for both algorithms, the area of progressive RNFL thinning detected by TPA was greater than that by GPA (p<0.001). Among the 32 eyes detected with progression by both algorithms, 78.1% (25 eyes) had progression detected by TPA earlier than or at the same time as by GPA (Fig.1). The survival probability of TPA was significantly lower than that of GPA (p<0.001) (Fig.2). No normal eyes showed RNFL progression by TPA and GPA (i.e. the specificities were 100%). Eyes detected with TPA progression had a higher risk of VF progression (Hazard ratio: 5.1, p<0.001) than with GPA (HR: 2.1, p=0.022).

 
Conclusions
 

TPA outperformed GPA in detecting more eyes with progressive RNFL thinning at similarly high specificity. With the inclusion of FDR and rates of change of RNFL thinning, TPA can offer a more informative approach to analyze progressive RNFL thinning than GPA.  

 
FIG.1 A glaucomatous eye with progressive RNFL thinning detected by TPA before GPA.<br />
 
FIG.1 A glaucomatous eye with progressive RNFL thinning detected by TPA before GPA.<br />
 
 
FIG.2 Survival probabilities of eyes detected with RNFL progression by TPA and GPA. Shaded area: 95% CI
 
FIG.2 Survival probabilities of eyes detected with RNFL progression by TPA and GPA. Shaded area: 95% CI

 
×
×

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.

×