Abstract
Purpose::
When expressed in comparable units (linear/linear or log/log), perimetric sensitivity can be linearly related to measures of ganglion cell number (Garway-Heath et al., 2002, IOVS 43:2213; Harwerth et al., 2004, IOVS 45:3152). These studies also found that individual patients can show substantial discordance between functional and structural measures. We assessed the effects of long-term fluctuations on this discordance.
Methods::
One eye was tested in each of twenty patients with glaucoma (mean MD = -7 +/- 6 dB, mean PSD = 6 +/- 4 dB), in two sets of visits at approximately 6-month intervals (mean =7.3 +/- 1.8 months). Neuroretinal rim area was assessed with the HRT-II. Conventional automated perimetry (CAP) was performed with 24-2 Sita Standard size III on a Humphrey Field Analyzer. Contrast sensitivity perimetry (CSP) was performed using sinusoidal Gabor patches (0.4 c/deg) on a custom testing station. Perimetric sensitivities were averaged in linear units for locations corresponding to two HRT sectors: inferior temporal (IT) and superior temporal (ST). Then perimetric averages and sector rim areas were converted to difference from mean normal in dB units (1 dB = 0.1 log unit), using lab norms and a lower limit of -10 dB to compensate for differences in dynamic range for the three instruments. A sector was considered discordant for a patient at a given visit when the difference between perimetric and rim defects was at least 3 dB.
Results::
On average, perimetric defects were slightly deeper than rim defects (1.4 +/- 2.5 dB, p < 0.02). For CAP, 10 patients showed discordance for IT, and 6 of these also showed discordance for ST. For CSP, 8 patients showed discordance for IT, and 2 of these also showed discordance for ST. In many cases discordance was only present on one of the two visits (50% of cases for CAP, 30% for CSP). For both CAP and CSP, long-term fluctuation accounted for one-third of the difference in depth of defect for perimetry versus rim area. Similar results were obtained when the analysis was conducted in linear units (percent of mean normal), with discordance defined as a difference of at least 30% between perimetric and rim defects.
Conclusions::
Our results are in agreement with prior studies which found that perimetric and imaging measures are on average linearly related when expressed in comparable units. Furthermore, we find that long-term fluctuation has a substantial contribution to the amount of discordance between structural and functional measures of glaucomatous damage.
Keywords: perimetry • optic nerve • imaging/image analysis: clinical