Researchers investigate diagnostic performance of imaging technologies in suspected glaucoma

Results of an observation cohort study show retinal nerve fiber layer imagingwith the scanning laser polarimeter (GDx VCC, Carl Zeiss Meditec) performed better thantopographic optic disc assessment with the confocal scanning laser ophthalmoscope (HeidelbergRetinal Tomograph 3.0, Heidelberg Engineering) for detecting early damage in eyes withsuspected glaucoma, said Felipe Medeiros, MD, PhD.

The study included 82 eyes with suspected glaucoma based on optic disc appearance, said Dr.Medeiros, Hamilton Glaucoma Center, University of California San Diego. All patients had normalvisual field results at the time they underwent structural imaging and at least 5 years ofprior follow-up. The eyes were divided into glaucoma (n = 40) and control (n = 42) groups basedon the presence or absence, respectively, of progressive glaucomatous optic neuropathy onprior, serial stereophotographs.

For the scanning laser polarimeter, the best parameter for discriminating between the glaucomaand control groups was the Nerve Fiber Indicator (NFI), which had an area under the receiveroperating curve (ROC) of 0.83. Rim volume was the best performing parameter for the confocalscanning laser ophthalmoscope, with an ROC of 0.70.

The ROC for the NFI was significantly greater than for the rim volume and also compared withthe Glaucoma Probability Score. Assuming borderline results as normal, the Moorfield RegressionAnalysis classification had a sensitivity of 48% for a specificity of 69%. For a similarspecificity of 70%, NFI had a significantly higher sensitivity of 83%, Dr. Medeirossaid.

"Most studies evaluating the accuracy of diagnostic equipment performance have used patients[with glaucoma] with confirmed visual field loss and controls who are usually healthyindividuals without any suspicious findings," Dr. Medeiros said. "While these studies areimportant for initial evaluation of any diagnostic test, in clinical practice we performdiagnostic testing in [eyes] suspected of having the disease and not in those with confirmeddisease.

"Since these study population samples may not be representative of the ones in which we apply the diagnostic test in everyday practice, estimates of diagnostic accuracy obtained from these studies may not be directly applicable in clinical practice," Dr. Medeiros added. "In fact, in our study of [suspected] glaucoma, the diagnostic accuracies of the two imaging techniques were lower than those reported for patients with established disease."