Diagnostic Accuracy of the Moorfields Regression Analysis Using the Heidelberg Retina Tomograph in Glaucoma Patients with Visual Field Defects

Abstract

Purpose

This study was designed to determine the specificity and sensitivity of the Heidelberg Retina Tomograph I (HRT) using the Moorfields Regression Analysis (MRA) in differentiating healthy from glaucomatous eyes.

Methods

In this cross-sectional study, 74 eyes of 37 healthy subjects and 87 eyes of 47 glaucoma patients were examined with Octopus standard automated perimetry and HRT. Only one eye per patient was used for statistical analysis. According to visual field index mean defect (MD) glaucoma patients were divided into three groups with early (MD <6 dB), moderate (6 dB < MD <12 dB), and advanced visual field loss (MD >12 dB). The sensitivity and specificity of optic nerve head examinations using the MRA of HRT were evaluated by two criteria (criteria 1, as diseased if classified by MRA as outside normal limits; criteria 2, as diseased if classified by MRA as borderline). The correlations between the topographic parameters and visual field index MD were measured by correlation coefficient and presented by scatter plot.

Results

The specificity and sensitivity of HRT-MRA examination were, respectively, 100% and 68.1% when borderline cases were considered normal (criteria 1) and 97.3% and 85.1% when borderline cases were considered glaucomatous (criteria 2). The sensitivity of the MRA in eyes with different stages of visual field loss was 59.1% for early, 54.5% for moderate, and 92.8% for advanced visual field loss with criteria 1; the figures were, respectively, 81.8%, 72.7%, and 100% with criteria 2. The statistically significant correlations with moderate strength of association (r=0.40–0.59) were found for rim area, rim volume, cup to disc area ratio, mean retinal nerve fiber layer (RNFL) thickness, and RNFL cross-sectional area.

Conclusions

The MRA showed an excellent specificity and good sensitivity using criteria 2 including as glaucomatous optic discs those classified by MRA as borderline. Although correlations of moderate strength were found between rim area, rim volume, mean RNFL thickness, RNFL cross-sectional area, and visual field index MD, great interindividual variation limits the prediction of one parameter from the other. Therefore, in clinical practice both structural and functional examinations should be performed in order to characterize glaucomatous damage. (Eur J Ophthalmol 2007; 17: 216–22)