This article is more than 5 years old. Information may no longer be current.
Corneal elevation measurement methods comparable for evaluating subclinical keratoconus
Study finds sensitivity and specificity rates were slightly higher for the standard best-fit sphere method.
Click Here to Manage Email Alerts
Measurements of posterior corneal elevation using the standard best-fit sphere method and the enhanced best-fit sphere method were similarly efficacious in discriminating subclinical keratoconus from normal corneas, according to a study.
This is important because detecting keratoconus in candidates for refractive surgery reduces the risk of poor outcomes and serious postoperative complications.
“To our knowledge, no study had previously assessed the accuracy of these methods in discriminating the early stages of the disease,” lead author Ugo de Sanctis, MD, PhD, said.
Study
The study, published in Cornea, enrolled 30 consecutive patients with subclinical keratoconus and 37 candidates for refractive surgery with normal corneas.
Posterior corneal elevation was measured using the Pentacam rotating Scheimpflug camera (Oculus).
“This camera provides accurate measurements of posterior corneal elevation and can measure this parameter using different methods,” de Sanctis said. “The standard method employs the BFS as reference shape above which to measure corneal elevation. This is in contrast to the E-BFS method, which relies on the difference between elevation values measured above a reference shape, after excluding a 3.5-mm circular area centered on the thinnest point of the cornea, and those measured above the standard BFS.”
The study found that mean posterior elevation values were statistically greater (P < .001) in subclinical keratoconus than in normal corneas for both detection methods: 38 µm vs. 20.3 µm for the standard best-fit sphere (BFS) method and 15 µm vs. 7.8 µm for the enhanced best-fit sphere (E-BFS) method. However, sensitivity and specificity rates were slightly higher for the BFS method than for the E-BFS method: 73.3% vs. 60% and 86.5% vs. 83.8%, respectively.
Neither difference was statistically significant, and the overall accuracy of the two tests were similar.
“Their accuracy was good because the area under the receiver operator characteristic (ROC) curve was above 0.7 for both methods: 0.8 for the standard and 0.72 for the E-BFS,” de Sanctis said.
The correlation between posterior elevation measurements obtained with the two methods was statistically significant but weak, de Sanctis said.
The disease was discriminated by both methods in 50% of the patients with subclinical keratoconus. It was discriminated in seven patients by the BFS method alone and in three patients by the E-BFS method alone.
Combined methods
“Since these two methods measure corneal elevation in different ways, they might advantageously be used in combination,” de Sanctis said. “If positivity for one test had been taken as sufficient to discriminate the disease, the sensitivity would have risen to 83.3%. However, the specificity would have decreased to 78.4%. This means that roughly 20% of candidates for refractive surgery would have been excluded.”
“However, their specificity can be improved by distinguishing the abnormal elevation that occurs in corneas with congenital astigmatism on the basis of elevation patterns,” he said.
According to the ROC curve analysis, the test performance could have been marginally increased by using slightly different cutoff points instead of 29 µm for the BFS method and 12 µm for the E-BFS method.
“The cutoff points with the widest area under the curve were 27 µm for standard and 14 µm for enhanced,” de Sanctis said. “If we had used these rectified cutoff points, the sensitivity, specificity and accuracy for the standard method would have changed to 80%, 83.8% and 0.82, respectively, and to 56.7%, 89% and 0.74 for the E-BFS method. However, clinical judgment regarding test efficacy would have remained the same.”
“Although in many cases this parameter enables clinicians to diagnose subclinical keratoconus on the basis of a single datum, it should be used together with other indices to improve disease discrimination,” he said. – by Bob Kronemyer