Prospective comparative case series, conducted at a university eye hospital outpatient clinic.
Twenty-three eyes (group 1) were newly diagnosed with KC; 10 eyes (group 2) were asymptomatic fellow eyes that showed neither major topographic anomaly nor clinical signs of KC; 127 healthy eyes of 74 patients served as negative controls (group 3). A seventh-order Zernike decomposition of first-surface aberrations was performed. Single Zernike coefficients, higher-order aberration root mean square (HOA RMS) values, the Z3 index and the output values of discriminant analysis D13 (with input from groups 1 and 3) and D23 (groups 2 and 3) were assessed for their usefulness to discriminate between clinically normal fellow eyes, KC eyes and controls by plotting receiver-operating characteristic (ROC) curves.
There were significant differences between group 1 and group 3 for 11 Zernike coefficients, Z3, total HOA RMS, coma RMS and third-order RMS. Group 2 and 3 showed significant differences only for the coefficients Z3−1 and Z5−1. Z3−1, D13, and D23 discriminated between groups 1 and 3 with maximum sensitivity and specificity. For discrimination between groups 2 and 3, D23 turned out to be the best parameter (Az ROC = 0.98), followed by Z3−1 (Az ROC = 0.96).
Clinically normal fellow eyes of eyes with early KC showed significant differences of first-surface aberrations compared to normal eyes and could therefore be considered as eyes with subclinical KC.