Accurate measurements of corneal biometric parameters are crucial for myopia refractive surgery planning, and expecially important for diagnosis and management of keratoconus. This study analyzed keratoconus corneal tomography parameters using CASIA2, Sirius, and Pentacam. The Ks and Kf measured using Sirius were the highest (p < 0.0001). The thinnest corneal thickness measured using Pentacam was the thickest (p < 0.0001), while the significance disappears in the Kmax ≤ 48D keratoconus group. In keratoconus group, Pentacam and CASIA2 showed the best consistency in the measurement of TCRP and fKm, whereas Pentacam and Sirius showed the best consistency in the measurement of pKm. To our knowledge, this is the first comparison of these three instruments, with different principles, in both normal and keratoconus populations.
Interchangeability in diverse clinical settings among different devices were indispensable in clinical practice.The fKm of the control group measured using Pentacam, Sirius, and CASIA2 were 43.52 ± 1.36 D, 43.94 ± 1.37 D, and 43.65 ± 1.34 D, respectively. These values are similar to fKm measured in previous studies using Pentacam (43.27 ± 0.09 D), Sirius (43.11 ± 0.16 D) [20] and CASIA2 (43.49 ± 1.5 D) [19]. For keratoconus, the fKm measured using Pentacam, Sirius, and CASIA2 were 47.61 ± 5.03 D, 48.7 ± 5.62 D, and 47.55 ± 4.52 D, respectively, which are slightly higher than those reported in previous studies (Pentacam, 46.69 ± 4.45 D; Sirius, 46.64 ± 4.34 D [21]; CASIA2, 46.66 ± 2.65 D [17]). The possible reason is that eyes with pre-clinical and AK1-3 stage KC were included in the previous studies, whereas only eyes with clinical stage KC were included in the present study, possible bias may exist due to patient inclusion criteria. Regarding the measurement of the corneal characteristics, the three instruments showed no significant difference in the measurement of astigmatism in the control group. Pérez Bartolomé et al. [22] reached a similar conclusion after comparing Pentacam and Fourier-domain OCT (FD-OCT). This may be because the instruments show higher consistency among patients with regular astigmatism than in keratoconus patients [23].
In relation to the assessment of TCT, the TCT in control group measured using Pentacam was 9.7 µm thicker than that measured using CASIA2. The TCT of the KC group measured using Pentacam was 11.4 µm thicker than that measured using CASIA2, and the difference was more significant in eyes with higher corneal refractive power (Kmax > 48 D). Flockerzi et al. [19] suggested that CASIA2 tends to underestimate the TCT of normal eyes and eyes with keratoconus compared with Pentacam, and that the degree of underestimation increases with the TKC stage. Pentacam included tear film in the corneal thickness measurement make contribution to the difference [24]. The observed difference means the agreement between devices for TCT varied significantly depending on the corneal pachymetry thickness value; also, this may be due to inaccurate TCT positioning caused by the deformed corneal surface, which further increased the difference between the measurements [25].
In terms of corneal elevation, which was compared based on BFS of CASIA2, the anterior and posterior elevations measured using Sirius were the highest in the KC group and the lowest in the control group. Hernández-Camarena et al. [26] reported that in normal eyes, the maximum anterior and posterior corneal elevation measured using Sirius is always lower than those measured using Pentacam, however, this phenomenon is not observed in eyes with keratoconus. Elevation difference may attributed to a difference in image acquisition method, especially for the posterior surface imaging, curvature analysis and process. Our findings reinforce that Sirius is less likely to detect abnormal posterior corneal morphology than Pentacam, which should be concerned by clinical doctors when planning for corneal refractive surgery.
Significant differences in corneal refractive power measurements were found across the three devices. Scheimpflug cameras yielded steeper TRCP and Q values for both control and KC groups. In the KC group, TCRP differences vanished in eyes with steeper curvature, and posterior Q value differences disappeared in eyes with faltter curvature. Asawaworrit et al. [17] found that in their study, the TCRP of normal eyes measured using Pentacam and Casis-2 was 43.49 ± 1.41 D and 43.09 ± 1.31 D, respectively, indicating a significantly steeper TCRP measurement by Pentacam; the TCRP of eyes with keratoconus measured using Pentacam and CASIA2 was 45.98 ± 2.9 D and 45.39 ± 2.47 D, respectively, and no significant difference was found between two instruments, which is similar to the results of our study. Seiler et al. [27] reported that the anterior Q value in eyes with keratoconus measured using Pentacam was 0.09 D steeper than that measured using FD-OCT. Schiano-Lomoriello et al. [28] reported that the anterior Q-value in keratoconus eyes measured using Sirius was 0.26 D steeper than that measured using FD-OCT, which is consistent with the results of our study. The differences between the measurement of corneal refractive power parameters mentioned above is due to different measurement range and calculation principle. The TCRP parameters selected in the present study include both anterior and posterior corneal surface curvatures and corneal thickness. TCRP in Pentacam and MPP in Sirius are calculated by ray tracing of anterior and posterior corneal surfaces according to Snell's law [29], whereas RP in CASIA2 is calculated using the Gauss formula [30]. TCRP and MMP in Pentacam and Sirius are calculated within the 4 mm area in the center of the cornea, whereas CASIA2's RP defaults to the 3 mm area in the central cornea, so that any deviation in the gaze or head position of the subject can lead to differences in results between devices [30]. FD-OCT has a higher resolution and repeatability, as well as a faster scanning speed, which can prevent measurement bias caused by tear film break-up due to prolonged eye opening during image capture[31].
In the control group, the 95% LOA for pKm between Pentacam and Sirius was − 0.24 D to -0.04 D, within the clinically acceptable range. However, both instruments had a greater consistency interval than ± 0.5 D for TCRP and fKm. Among the three devices, Pentacam and Sirius showed the best consistency for pKm in keratoconus eyes (-0.39 D to 0.58 D). The 95% LOA for fKm between Pentacam and Sirius was − 2.6 D to 0.6 D, wider than between Pentacam and CASIA2. Shetty et al. [21] reported that the 95% LOA of Pentacam and Sirius for fKm and pKm measurements in eyes with keratoconus is -1.55 D to 1.29 D and − 0.40 D to 0.98 D, respectively, which is similar to the results of the present study. The potential reason for this result is that Pentacam and Sirius both use a rotating Scheimpflug camera for posterior corneal surface data, while Sirius combines Scheimpflug and Placido disc for anterior data. This difference in image acquisition might explain the higher consistency in posterior corneal power compared to anterior and total corneal power between two devices [32].
The consistency interval of refractive power parameters between pairs of instruments was wider in the KC group than in the control group, and the consistency tended to decrease with increase in the severity of KC, which is consistent with the results of the study by Asawaworrit et al. [17]. The possible reason for this is that as the severity of keratoconus increases, the irregularity of the anterior and posterior corneal surfaces reduces the reproducibility of the measurements [19].
The TKC staging of Pentacam only includes parameters of the anterior corneal surface, whereas the keratoconus indicators of Sirius and CASIA2 integrate data on both anterior and posterior corneal morphology [33, 34]. The sensitivity of Pentacam, Sirius, and CASIA2 in the present study was 83.61%, 59.02%, and 73.33%, respectively. The plausible reason for this result is that only eyes that showed “keratoconus” screening results were included, while eyes with “suspected” or “abnormal” results were excluded. Considering that “suspected keratoconus” is a very broad concept, it was not taken into account in the present study [35]. However, if the abovementioned classification is regarded as a positive diagnosis of keratoconus, the sensitivity of Pentacam, Sirius, and CASIA2 would be 90.16%, 91.80%, and 98.36%, respectively. Spira et al. [33] showed that the ESI staging of Caisa-2 and the TKC staging of Pentacam have similar diagnostic abilities for keratoconus. Multiple studies have reported that by integrating both anterior and posterior corneal parameters and constructing a screening index using logistic regression, the sensitivity of Pentacam for diagnosing keratoconus can reach as high as 97.3% [15], while the sensitivity of Sirius can reach up to 96.8% [34], indicating that a comprehensive evaluation of both anterior and posterior corneal parameters is necessary for diagnosing keratoconus..
Limitations exist in our study. First, the sample size was relatively small. Second, only eyes with clinical primary keratoconus were included. Thus, further studies that include subclinical keratoconus and post-LASIK subgroups are needed to corroborate the findings of the present study.