New study finds limits of keratometry in keratoconus

Reproducibility of keratometry is better for early keratoconus than for advanced keratoconus, a new study showed.The study could help clinicians decide when to use corneal cross-linking in their efforts to stop the progression of the disease, wrote Tom H. Flynn, PhD, and his colleagues from the Corneal Service, Moorfields Eye Hospital, St. George’s Hospital, London. They published their finding in the British Journal of Ophthalmology.

Reproducibility of keratometry is better for early keratoconus than for advanced keratoconus, a new study showed.

The study could help clinicians decide when to use corneal cross-linking in their efforts to stop the progression of the disease, wrote Tom H. Flynn, PhD, and his colleagues from the Corneal Service, Moorfields Eye Hospital, St. George’s Hospital, London. They published their finding in the British Journal of Ophthalmology.

Cross-linking comes with some risks, including reported cases of persistent stromal haze, sterile infiltrates, corneal melt, and microbial keratitis, so clinicians would like to reserve the procedure for patients whose keratoconus is most likely to progress, they wrote. The challenge is identifying those patients.

Current or recent progressions provide some of the best indications of future progression. Serial Scheimpflug corneal tomography is the standard practice to monitor progress of ecstasia. It works well for gross or moderate progression, but cannot always detect subtle progression.

It is hard to tell which changes in parameters show real change in the patient and which are measurement error.

The authors routinely look for the curvature measured at the steepest part of the cornea (Kmax), corneal thickness at the thinnest location (TCT) and K1 and K2.

Previous studies on interobserver and intraobserver agreement in keratoconus have produced conflicting results. As a result, the authors undertook a new study to see how this type of variability might affect measuring the progression of keratoconus.

They recruited 100 patients with keratoconus and used a Pentacam (Oculus Optikgeräte GmbH, Wetzlar, Germany).

All had Pentacam-derived Ansler-Krumeich stage 1 or greater. The researchers excluded patients under 18 years of age, patients who previously had surgery, and patients with corneal scarring.

Each of two observers scanned the same eye of each subject. The eye to be scanned and the sequence of scans were randomly determined. Both observers were able to get high-quality scans of 93 patients.

Results

The differences between measurements by the same observers when measuring Kmax tended to increase with increasing values of the parameters assessed. A clear difference in the repeatability emerged between mild and more advanced cases of keratoconus when patients were stratified into two groups according to their Amsler-Krumeich stage. But this difference did not show up for K1, K2, or TCT.

For stages 1-2, the 95% limits of repeatability in Kmax were -0.84D to 0.94D, and for stages above 2, they were -2.04D to 1.97D.

The difference between measurements by different observers tended to increase as the keratometry values increased for all parameters. And a clear difference in reproducibility emerged between mild and more advanced cases. For Amsler-Krumeich stage 1-2, 95% limits of reproducibility in Kmax were -0.90D to 1.01D, and for stages above 2, -3.71D to 3.86D.

The limits of agreement were greater than those reported in previous research with Pentacam measurements in normal eyes, and a little larger than for keratoconic eyes, the researchers wrote.

Dr. Flynn and his colleagues suggested that their findings could allow doctors to judge whether differences between scans are likely due to variation between or within observers. “If differences greater than our limits are seen, this is suggestive that real change has occurred,” they wrote.

Fortunately, they found good reproducibility of measurements for patients with early keratoconus, and these are the patients who will benefit most from cross-linking, the wrote. Based on their findings, the researchers consider a change of 1D in K1, K2, and Kmax to indicate real changes in corneal shape.

Since intraobserver variability was lower than interobserver variability, clinicians may reduce the measurement error if the same person performs the scan each time, they wrote.

“These data describe the precision of important tomographic measures with Penacam in early and moderate/advanced keratoconus and will help clinicians to more accurately identify tomographic progression of keratoconus,” they concluded.

Tom H. Flynn, PhD

E: Tom.flynn@moorfields.nhs.uk