Background:The aim was to assess the intra-session, inter-session and inter-observer repeatability of curvature and aberrometric measurements of Pentacam-derived posterior corneal surface in normal control (CG), keratoconic (KCG) and after corneal collagen cross-linked (CXLG) groups. Methods: This was a prospective, clinic-based, observational study that was conducted at the Eye Institute of Thrace (EIT) in Alexandroupolis, Greece. Sixty-two normal eyes formed the control group, 33 were in the keratoconic group, while 34 eyes, which already had received corneal collagen cross-linking formed the CXLG. Curvature and aberrometric measurements were obtained using the Pentacam by two operators within each of the two sessions. Precision (Pre), repeatability (COR) and the intra-class correlation coefficient (ICC) were calculated for evaluating intra-session and inter-session repeatability. Bland-Altman analysis was used for assessing inter-observer repeatability. Results: Simulated keratometric parameters K1 and K2, asphericity at the central 8 mm zone (Q8) and central steepest radius of best fit toric ellipsoid reference body (Rs), achieved an acceptable level of accuracy. Aberrometric parameters were repeatable, reliable and reproducible only in terms of the root mean square (RMS) errors (coma, coma-like and the RMS of higher order aberrations or HOA RMS). For both observers, best precision and lowest range of agreement (ROA) was found for HOA RMS parameters in all groups (less than 0.27 and less than 0.29 for the control group, less than 0.95 and less than 0.72 for the keratoconic group and less than 0.67 and 0.53 for the collagen cross-linked group).
Conclusion:The Pentacam system provides precise measurements of corneal posterior keratometry, reference body radius and asphericity for large diameter analysis. Although Zernike coefficients for individuals presented poor repeatability, the coma, coma-like and HOA RMS seem to be accurate indices in characterisation of back corneal aberrations.