Abstract
Purpose: :
The fact that recent clinical trials of glaucoma have shown little concordance between progression in optic nerve structure and function suggests that the information contained in the two methods is complementary. Based on this, we hypothesize a combined measure of structure and function might perform better than either alone. We therefore designed and tested an index that combines visual field testing and optic nerve imaging using retinal nerve fiber layer anatomy.
Methods: :
We identified patients with automated perimetry and Heidelberg Retina Tomograph (HRT) imaging performed on the same day using a database of all such tests performed by the Wilmer Glaucoma Service over an 8 year period. Using billing codes produced by glaucoma specialists, we identified data for 2228 eyes of glaucoma suspects and for 1353 eyes of patients diagnosed with open angle glaucoma. We used 1000 of the glaucoma suspect eyes to define normative distributions for the total deviation at each point in the visual field and for the numeric value of the Moorfields regression function for each optic disc sector. We then calculated a new measure, the structure function index (SFI), for the remaining subjects. The SFI is calculated at each point in the visual field using the probability that the visual field total deviation is abnormal, the probability that corresponding optic nerve sectors are abnormal, and the probability that the two are linked by nerve fiber layer anatomy. The output of this calculation is the probability of an abnormal structure-function relationship for each point in the visual field. A modified version of the glaucoma hemifield test (GHT) was calculated using the SFI for each patient. This SFI hemifield test (SFI-HT) was then used to perform receiver operating characteristic (ROC) analysis. In cases where a patient had two eligible eyes, the eye with the worse mean deviation was used.
Results: :
Using the SFI-HT, the area under the ROC curve was 0.77. The optimal sensitivity and specificity using this method were 80% and 65%. For comparison, sensitivity and specificity were 58% and 84% using the Humphrey Field Analyzer GHT and were 64% and 67% using the overall HRT Moorfields classification.
Conclusions: :
Distinguishing patients with glaucoma from those who are merely suspects is a particularly difficult task in a tertiary care subspecialty clinic. We have shown that the SFI is a fair test for distinguishing glaucoma and that its ability to distinguish glaucoma from glaucoma suspect is comparable to current methods. Since it is not possible to demonstrate superiority of any new diagnostic criteria using cross-sectional data, we are proceeding with analysis of longitudinal data using the SFI.
Keywords: imaging/image analysis: clinical • visual fields • optic nerve