Purchase this article with an account.
S. K. Gardiner, C. A. Johnson, N. G. Strouthidis, S. Demirel; Using Heidelberg Retinal Tomography (HRT) to Predict Future Functional Change in Early Glaucoma. Invest. Ophthalmol. Vis. Sci. 2009;50(13):4395.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Studies suggest that HRT can be used to distinguish between normal and glaucomatous eyes, and to predict whether eyes with ocular hypertension will develop glaucoma. In this study, we ask whether HRT is useful to predict the rate of future functional change, and to identify those eyes at highest risk for rapid progression.
Data were taken from an ongoing longitudinal study of subjects with ocular hypertension or early glaucoma. For inclusion, subjects were required to have undergone at least 7 annual visits, including standard automated perimetry (SAP) and HRT. For subjects with longer series, the most recent 7 visits were used. 136 subjects fit this criterion. Stereometric parameters generated using the HRT in Years 1 and 2 were used to predict each subject's subsequent functional change, defined as the rate of change in SAP mean sensitivity (from linear regression over time) over Years 3-7. The eye with most rapid functional loss for each subject was selected for analysis.
In Years 1 and 2, subjects in the worst tertile for subsequent functional change had significantly lower rim volume (p=0.016 and 0.035 respectively, using a nonparametric Mann-Whitney test), retinal nerve fiber layer (RNFL) thickness (p=0.049 and 0.019) and RNFL cross-sectional area (p=0.037 and 0.014); and lower rim area in Year 2 (p=0.039). Subjects in the worst tertile were also older, and had lower baseline SAP mean sensitivity (p<0.001 in each case). Other HRT parameters did not reach statistical significance. Logistic regression was then used to predict the probability of being in this worst tertile after adjusting for age and baseline mean sensitivity. Rim volume, RNFL thickness and RNFL area were significant predictors in Year 1 (p=0.022, 0.017 and 0.029 respectively). Due to improved predictability of subsequent change from mean sensitivity alone, in this model RNFL area was the only HRT parameter that remained a significant predictor in Year 2 (p=0.030).
Although it has been reported that rim area is well correlated with perimetric sensitivity, it may offer little independent prognostic information once sensitivity is known. RNFL parameters gave additional information about future prognosis.
This PDF is available to Subscribers Only