We appreciate the comments from Jia and associates regarding our statement on the diagnostic value of anterior and posterior corneal elevation relative to each other. We must emphasize that our study was not meant to explore this specific aspect of keratoconus detection, but our data did in fact suggest what the authors questioned. In order to better support this claim, we went back to our dataset, which unfortunately only included front and back elevation at the thinnest corneal location, as provided by the Pentacam’s Belin-Ambrósio display. The thinnest point is usually within the “corneal region near the pupil,” as the authors suggest, although the latter is somewhat complicated to define.
For discriminating between Groups 1 and 2 (see the original article for details ), the area under the receiver operating characteristic curve for front elevation was 0.90 ± 0.03, whereas for back elevation it was 0.91 ± 0.03. The optimal cutoff value (defined as that closest to the perfect classification point in the curve) was >5 μm for front elevation (78% sensitivity, 94% specificity) and >10 μm for back elevation (82% sensitivity, 91% specificity). When comparing Groups 1 and 3, the area under the curve for front and back elevation was the same, 0.97 ± 0.02, and the optimal cutoff values were >6 μm for front elevation (95% sensitivity, 96% specificity) and >17 μm for back elevation (95% sensitivity, 100% specificity). But when dealing with subclinical keratoconic eyes (fellow eyes with unremarkable topography), the area under the curve for front and back elevation was 0.85 ± 0.04 and 0.86 ± 0.04, respectively; and the optimal cutoff values were >4 μm (73% sensitivity, 83% specificity) and >8 μm (76% sensitivity, 86% specificity), respectively.
After the additional analysis, we believe that the argument in question still holds; that is, that back elevation data do not seem to be clinically superior to front elevation for keratoconus detection. This statement on posterior elevation was suggested earlier by de Sanctis and associates, who concluded that “its efficacy is lower for subclinical keratoconus, and thus data concerning posterior elevation should be combined with curvature data in stratifying patients with this condition.” Nonetheless, we must acknowledge that neither our study nor that of de Sanctis and associates was designed to determine whether detectable back corneal changes occur before front abnormalities, and that prospective studies of keratoconus progression should be undertaken to appropriately answer this issue. This was perhaps what the authors questioned about our work, and we agree with them on this aspect.