Patients and methods

PATIENTS

A total of 79 white patients (34 patients with normal tension glaucoma, 20 with primary open angle glaucoma, 13 with pseudoexfoliation glaucoma, and 12 with ocular hypertension), together with 21 control subjects, were included in this study, which followed the tenets of the Declaration of Helsinki. Individuals were examined between January and October 1999 at the department of ophthalmology, University of Bern, Switzerland, each having given their consent to undergo optical low coherence reflectometry.

Patients with normal tension glaucoma had untreated IOPs of less than 22 mm Hg on at least two diurnal curves, an open, normal angle, a glaucomatous optic disc, and glaucomatous visual field defects. Individuals with primary open angle glaucoma had untreated IOPs of 22 mm Hg or higher, an open, normal angle, a glaucomatous optic disc, and glaucomatous visual field defects. Subjects with pseudoexfoliation glaucoma had untreated IOPs of 22 mm Hg or higher, an open angle, typical pseudoexfoliation material within the angle and upon the lens, a glaucomatous optic disc, and glaucomatous visual field defects. Patients with ocular hypertension had untreated IOPs of 22 mm Hg or higher, an open, normal angle, normal optic discs, normal visual fields, and no family history of glaucoma.

METHODS

In all patients, the registered condition was bilateral, one eye per individual being randomly selected for investigation.

IOP was monitored according to a standard protocol using a calibrated Goldmann applanation tonometer.

Visual fields were mapped with an automated perimeter Octopus 101, using the G2 glaucoma program in conjunction with a normal strategy.

Central corneal thickness was measured with an optical low coherence reflectometer, operating at a scanning speed of 0.5 m/s and a repetition rate of 15 Hz, and used in conjunction with a superluminescent diode (λ = 850 nm). The mean and standard deviation of three determinations, each consisting of 20 scans, was calculated for each eye. All measurements were made by the same examiner.

Details pertaining to the experimental design of the instrument used have been documented elsewhere21-25; only a few relevant features are presented here. The design of the optical reflectometer is based on a Michelson interferometer. A superluminescent diode serves as the broad band light source, this having an output power of 8 mV at 850 nm and a spectral width of 18 nm. It is coupled to a fibre optic Michelson interferometer, the light being split between the two arms of this instrument in a 50/50 coupler. One arm runs to the cornea and the second to a rotating cube. An aiming diode laser is integrated into the interferometer with a 95/5 coupler. Longitudinal scanning is achieved by means of the rotating cube. The detection system consists of a silicon photodiode, an amplifier and an oscilloscope for signal display. Large signals are detected at interfaces where there is a pronounced change in refractive index—that is, at the air/tear and endothelial/aqueous humour junctions. The longitudinal scanning system of the instrument, which permits a scanning speed of more than 20 m/s and a repetition rate of about 400 Hz, is based on a rotating glass cube. The whole instrument is attached to a clinical slit lamp (Haag Streit International, Switzerland). The superluminescent diode and the fixation laser signal are positioned before the front lens of the slit lamp, thereby allowing the reflectometer's server to direct the instrument with biomicroscopic stereopsis. The pilot lasers are attached horizontally and converge on the slit lamp's focal point, thereby facilitating the alignment of the instrument with the x-y-z control of the slit lamp, and that of the latter with a chin and head rest. With the aid of two converging pilot lasers, the centre of the longitudinal scan, which is adjusted to coincide with the focal point of the slit lamp, is aligned with the cornea at the appropriate distance. The optical centre of the cornea is located by the subject's fixation of a low power laser beam, aligned coaxially with that of the 850 nm superluminescent diode. The diameter of the measuring beam emitted from the superluminescent diode and impinging on the surface of the central cornea is about 20 μm. The instrument was calibrated with a BK7 glass window (Melles Griot, USA) with a thickness of 1 mm and a refractive index of 1.5100 at 850 nm. For a given position of the glass, measurements were found to be reproducible to 1 μm. The thickness of the cornea was calculated from the optical distance using an average corneal refractive index value of 1.376.

A specific strategy for data collection and evaluation was adopted. From a series of 20 longitudinal scans, the upper and lower five values were deleted. From the remaining 10, the mean was calculated as the current thickness value, and the range and standard deviation determined. The results were recorded online in a database, which stores the identity of the subject, the date and time of the measurement, and the individual scan values, including technical information on their generation from each series. The computed mean from three scan series was recorded manually together with the range and standard deviation. A set of 10 ultrasound pachymetric measurements was also obtained to confirm the magnitude of the optical low coherence reflectometric ones.

The statistical analysis was performed using ANOVA, the Student-Newman-Keuls test being then implemented for multiple comparisons.

Results

The central corneal thickness of patients with ocular hypertension (mean 563 (SD 29) μm; range 515–611 μm) was significantly higher (p ⩽0.001) than in normal (control) individuals (mean 524 (25) μm; range 483–570 μm) or in subjects with normal tension glaucoma (mean 518 (0.5) μm, range 469–564 μm), open angle glaucoma (mean 515 (35) μm; range 454–581 μm), or pseudoexfoliation glaucoma (mean 507 (25) μm; range 470–567 μm), there being no significant differences between the latter four groups (Table 1). Patients with ocular hypertension were also significantly younger than those within any of the three glaucomatous groups (p ⩽0.003).

Discussion

The influence of scleral rigidity and central corneal thickness on IOP readings was first discussed by Goldmann and Schmidt in 1957.1 But a further 18 years elapsed before these intimations were put on a firm footing by Ehlers et al, who postulated that the applanation tonometry yielded accurate measurements only at a central corneal thickness of 520 μm.10 They calculated the average error evoked by a thicker or thinner cornea to be 0.7 mm Hg per 10 μm deviation from the “normal” value of 520 μm. This estimate was later corroborated in a case report of a patient with a 900 μm thick cornea, in which instance a tonometric error of 0.63 mm Hg per 10 μm corneal thickness deviation was calculated.14 On the basis of these calculations, Copt et al19 and Shah et al20 reclassified 36% and 44%, respectively, of their normal tension glaucomatous patients as having open angle glaucoma, and 56% and 35%, respectively, of individuals with ocular hypertension as being normal.

On the other hand, more recent cannulation studies have revealed average corrections per 10 μm deviation in corneal thickness to be considerably lower—namely, 0.18 to 0.23 mm Hg1213 and 0.19 mm Hg.15

Optical low coherence reflectometry represents the most precise pachymetric method now available. Using these technique, we confirmed that patients with ocular hypertension had thicker than normal central corneas. This circumstance is responsible for the artificially high estimates of IOP within this group of individuals. Using Ehlerset al's formula of 0.7 mm Hg per 10 μm deviation from the normal corneal thickness value, six of our 12 patients with ocular hypertension had to be reclassified as normal. The highest overestimation of IOP in this group was 8.5 mm Hg. And even when using the more conservative formula based upon cannulation studies (0.2 mm Hg per 10 μm deviation) five of the 12 individuals had to be reclassified as normal.

In our study, patients with normal tension glaucoma did not have thinner central corneas than either normal subjects or individuals with primary open angle glaucoma or pseudoexfoliation glaucoma, although we cannot rule out the possibility that if our sample size had been larger a difference would have become manifest. Using Ehlerset al's formula, the greatest underestimation of IOP in the normal tension glaucomatous group was only 3.57 mm Hg; and, accordingly, none of these patients had to be reclassified as having primary open angle glaucoma.

When one takes into account all relevant factors—the inaccuracy of both applanation tonometry24101115 and ultrasonic pachymetry,21 the existence of various formulas for adjusting estimates of IOP according to central corneal thickness, and high interindividual variability in corneal thickness—it becomes abundantly clear that the reclassification of patients is not a straightforward business—that it cannot be conducted in a decisive and unequivocal manner.2627 Given this situation, there would appear to be little justification for monitoring central corneal thickness on a routine basis for diagnostic purposes.2628

However, in chronic conditions, a deviation of only 10% from the normal central corneal thickness has a measurable impact on tonometry, as was confirmed in a recently published meta-analysis.26Hence, the measurement of central corneal thickness may be useful in selected cases. This issue is likely to become of increasing importance in the near future, as widespread performance of refractive surgery—with ensuing changes not only in corneal thickness but also in the structure of the cornea—affect the reliability and reproducibility of applanation tonometry in a manner as yet not fully appreciated.235-729

Current efforts to develop a new tonometric head—combining optical low coherence reflectometric pachymetry with Goldmann applanation tonometry—are thus well worthwhile, the accurate measurement of central corneal thickness being important not only for individual patient care, in permitting more precise estimations of IOP, but also for clinical studies, in assuring a more reliable classification of subjects.