June 2015
Volume 56, Issue 7
Free
ARVO Annual Meeting Abstract  |   June 2015
Disease-modeling-based prediction of recurrence patterns in anti-VEGF therapy from OCT analyses
Author Affiliations & Notes
  • Ursula Schmidt-Erfurth
    Christian Doppler Lab for Ophthalmic Image Analysis, Vienna Reading Center, Dept of Ophthalmology, Medical Univ of Vienna, Vienna, Austria
  • Wolf-Dieter Vogl
    Christian Doppler Lab for Ophthalmic Image Analysis, Vienna Reading Center, Dept of Ophthalmology, Medical Univ of Vienna, Vienna, Austria
  • Dominika Podkowinski
    Christian Doppler Lab for Ophthalmic Image Analysis, Vienna Reading Center, Dept of Ophthalmology, Medical Univ of Vienna, Vienna, Austria
  • Ana-Maria Glodan
    Christian Doppler Lab for Ophthalmic Image Analysis, Vienna Reading Center, Dept of Ophthalmology, Medical Univ of Vienna, Vienna, Austria
  • Bianca S. Gerendas
    Christian Doppler Lab for Ophthalmic Image Analysis, Vienna Reading Center, Dept of Ophthalmology, Medical Univ of Vienna, Vienna, Austria
  • Alessio Montuoro
    Christian Doppler Lab for Ophthalmic Image Analysis, Vienna Reading Center, Dept of Ophthalmology, Medical Univ of Vienna, Vienna, Austria
  • Jing Wu
    Christian Doppler Lab for Ophthalmic Image Analysis, Vienna Reading Center, Dept of Ophthalmology, Medical Univ of Vienna, Vienna, Austria
  • Christian Simader
    Christian Doppler Lab for Ophthalmic Image Analysis, Vienna Reading Center, Dept of Ophthalmology, Medical Univ of Vienna, Vienna, Austria
  • Georg Langs
    Christian Doppler Lab for Ophthalmic Image Analysis, Vienna Reading Center, Dept of Ophthalmology, Medical Univ of Vienna, Vienna, Austria
  • Sebastian M Waldstein
    Christian Doppler Lab for Ophthalmic Image Analysis, Vienna Reading Center, Dept of Ophthalmology, Medical Univ of Vienna, Vienna, Austria
  • Footnotes
    Commercial Relationships Ursula Schmidt-Erfurth, Alcon (F), Bayer (F), Boehringer Ingelheim (F), Novartis (F); Wolf-Dieter Vogl, None; Dominika Podkowinski, None; Ana-Maria Glodan, None; Bianca Gerendas, None; Alessio Montuoro, None; Jing Wu, None; Christian Simader, None; Georg Langs, None; Sebastian Waldstein, None
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science June 2015, Vol.56, 2771. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to Subscribers Only
      Sign In or Create an Account ×
    • Get Citation

      Ursula Schmidt-Erfurth, Wolf-Dieter Vogl, Dominika Podkowinski, Ana-Maria Glodan, Bianca S. Gerendas, Alessio Montuoro, Jing Wu, Christian Simader, Georg Langs, Sebastian M Waldstein; Disease-modeling-based prediction of recurrence patterns in anti-VEGF therapy from OCT analyses. Invest. Ophthalmol. Vis. Sci. 2015;56(7 ):2771.

      Download citation file:


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

      ×
  • Supplements
Abstract

Purpose: The prediction of therapeutic outcomes is a key requirement for efficient disease management. Spectral-domain optical coherence tomography (SD-OCT) is capable to identify the pathomorphological fingerprint of retinal disease and is therefore most appropriate to predict individual response patterns.The aim of this study was to predict the recurrence of macular edema in patients receiving ranibizumab therapy for macular edema due to neovascular AMD (nAMD) and central retinal vein occlusion (CRVO).

Methods: Patients with nAMD / CRVO enrolled in phase III multicenter clinical trials with monthly SD-OCT available at the Vienna Reading Center were included. All patients received a loading dose of three ranibizumab injections followed by pro-re-nata treatment. OCT volumes were processed using an automated image analysis pipeline consisting of image denoising, motion correction, layer, vessel and foveal segmentation and registration to a common population reference frame. The individual treatment response patterns were defined as “recurrent” or “non-recurrent” based on a threshold of ≥29 µm increase in regional retinal thickness at any time point. A machine learning algorithm (sparse logistic regression with elastic-net regularization) was finally used to predict recurrent or non-recurrent patterns based on baseline and loading phase OCT morphology.

Results: Images from a total of 120 nAMD and 250 CRVO eyes were processed. In CRVO, 528 OCT volumes were analyzed. 86% of patients showed recurrent and 14% showed non-recurrent response patterns. Recurrence patterns were predictable by the machine-learning system with a sensitivity of 100% and a specificity of 88%, yielding a receiver-operated characteristics area under the curve of 0.99 for the prediction of non-recurrence. Central macular thickness at baseline was significantly greater in the recurrent group (p<0.001). Similarly for nAMD and CRVO, prediction of disease prognosis was feasible after a single injection specifying interval and location of recurrence.

Conclusions: Computational disease-modeling of longitudinal OCT data enables precise prediction of intensity, timing and location of recurrence in anti-VEGF treatment on an individual patient level. On a population base, precise and objective disease management allows effective use of resources.

×
×

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.

×