Keratectasia severity staging and progression assessment based on the biomechanical E-staging

Corvis Biomechanical and Tomographic Biomechanical Index

One of the main CST indices is the Corvis Biomechanical Index (CBI, Fig. 4), which is a logistic regression algorithm that combines several biomechanical parameters to differentiate between KC and healthy corneas [29, 31]. In a second step, the CST imports tomographic data from the Pentacam and generates the Tomographic Biomechanical Index (TBI), which combines information from the Pentacam and the CST. Both parameters behave in a quasi-binary manner. Analogous to a yes-or-no decision, the output may assume the value “0” in green for a healthy cornea and the value “1” in red for KC [31,32,33]. The study that introduced the CBI reported a 100% specificity and 94.1% sensitivity of the CBI in diagnosing KC within the training dataset and a 98.4% specificity and 100% sensitivity in the validation dataset [31]. The imported Pentacam display indicates topographic and tomographic KC severity based on TKC and BAD values, but from a biomechanical view, neither the CBI nor the TBI provide information about KC severity (Fig. 4). This raised the question about a biomechanical KC severity staging [21, 34,35,36] and one possible answer to that question is the BEST, which was developed based on the linearized CBI and thus comprises the following biomechanical parameters, that are reported to form the CBI: (1) A1 velocity (the velocity of the corneal apex at inward applanation A1), (2) deformation amplitude ratio of 2 mm (DA ratio 2 mm, the ratio between central deformation and deformation 2 mm from the corneal center), (3) integrated radius (the sum of the inverse radii of the concave state between the first and second applanation), (4) Ambrósio relational thickness in horizontal (ARTh, the ratio of corneal thickness at thinnest corneal thickness to the pachymetric index of progression) and (5) stiffness parameter A1 (SP-A1, calculated from the ratio of the force of the air applied to the corneal displacement).

Fig. 4figure 4

“Biomechanical/tomographic assessment” Corvis ST® (Oculus, Wetzlar, Germany) examination of a right eye of a 39-year-old male keratoconus patient. Upper left: “Corvis ST – biomechanical assessment”: individual biomechanical parameters are displayed including stiffness parameter A1 (stiffness parameter on inward applanation), integrated radius, ARTh (Ambrósio relational thickness in horizontal), and DA ratio 2 mm (deformation amplitude ratio of 2 mm). Keratoconus population highlighted in red, normal population in green. Black vertical line marking the patient's value, indicating the group the patient belongs to. Lower left: deformation curve of the cornea after application of an air puff indentation. The Corvis Biomechanical Index (CBI) makes a quasi-binary decision between a healthy cornea (green, value zero) and keratoconus (red, value one) – in this example, the value is 0.99. Right: “Pentacam – tomographic assessment ” showing anterior radius of curvature, corneal thickness, pachymetric progression, maximum keratometry value (Kmax), inferior-superior value, Pentacam random forest index (PRFI), topographic keratoconus classification (TKC) and final Belin-Ambrósio enhanced ectasia display deviation index (BAD-D). The combination of the data from the Corvis ST® and the Pentacam® in the Tomographic Biomechanical Index (TBI) is also considered quasi-binary

Reliability and reproducibility of the CST measurements

In our previous study, these biomechanical parameters were examined in a cohort of 173 eyes consisting of: (1) 15 healthy eyes from 15 healthy volunteers and (2) 158 KC corneas from 112 KC patients with TKC stages 1, 1–2, 2, 2–3, 3, 3–4, 4 and n = 26, 16, 36, 18, 31, 26, 5 eyes, respectively [37]. It has been reported that the reliability of successive Scheimpflug imaging and anterior segment optical coherence tomography measurements decreases with increasing KC severity [24], which led to the question whether similar observations could be found when analyzing biomechanical measurements in KC. Two studies reported that CST measurements were of excellent reliability in mild to moderate KC [38] and in three different KC severity stages [39]. To determine reliability and reproducibility at every stage of TKC, each eye was examined five times consecutively with the CST. Additionally, each eye was examined with the Pentacam prior to and after the five CST measurements with the intention to record the TKC stage and to investigate whether the repeated mechanical stress in the form of the CST measurements influenced the Pentacam measurements – and if so, to what extent. A1 velocity, DA ratio 2 mm and integrated radius increased, whereas ARTh and SP-A1 decreased with increasing KC stages. Cronbach’s alpha was calculated as a measure of internal consistency and showed a very good to excellent reliability of the measurements with values ≥ 0.834. When comparing the initial and final tomographic measurements, there was no consistent pattern of significant differences, indicating that repeated CST measurements did not affect the Pentacam’s tomographic measurement.

Corvis Biomechanical Factor as linearized Corvis Biomechanical Index

The previous study revealed that the biomechanical parameters included within the CBI vary depending on KC stage. The first step towards a biomechanical severity staging for keratectasia was the conversion of the CBI into its linear term, which was first referred to as the “CBI beta” in 2020 [21]. The CBI beta was calculated for a representative cohort of 448 KC corneas from the HKC aiming to reflect different tomographic KC severity stages. The control group consisted of 112 healthy corneas from healthy volunteers presenting for refractive surgery evaluation [22]. A Pearson correlation between the CBI beta and the tomographic parameters (“A, B, C”) was performed and the CBI beta correlated best with “B” (posterior radius of curvature, R2 = 0.796 [22]). Due to the lack of an own scale and the increasing importance of posterior corneal curvature in keratectasia screening and diagnosis [13,14,15,16,17,18,19], the CBI beta was converted in a linear manner to the scale of posterior corneal curvature resulting in the Corvis Biomechanical Factor (CBiF) with the unit millimeters (CBiF = −0.24294226 × CBI beta + 6.02). The CBI beta is independent of posterior radius of curvature, the two parameters CBiF and PRC only share the same scale, but are not replaceable [22].

The biomechanical E-staging (BEST)

In our previous study, for each of the 448 KC corneas, a CBiF value was obtained. Based on the 2.5 and 97.5 percentiles and an interval spacing of 0.58, the five E-stages were defined for the KC group [40]. The distribution of the E-stages was analyzed and found to be normal in this study group. For independent validation, the BEST was thereafter applied onto an independent data set of 860 KC corneas from study groups in Milano and Rio de Janeiro [22, 31, 41], which also revealed a normal distribution. Therefore, it could be concluded that the BEST achieved comparable and reproducible results in two different KC cohorts. It was shown that the most severe KC stage was set by parameter “B” followed by “E”, “A” and “C” [40]. As posterior elevation data and thinnest corneal thickness were diagnostic criteria in the selection of KC corneas for the study population used to create the BEST and also in Sideroudi's baseline HKC study [27], these are only frequency data and one cannot infer from this in principle that stages of “B” are generally more advanced than “E” or “A” in every KC [40]. The resulting BEST is meant to be an addition to the tomographic ABCD KC staging system that was published previously as it supplements biomechanical severity information for keratectasia. This recently introduced biomechanical staging parameter thus enables a combination of tomographic and biomechanical severity assessment in daily practice. In the following, the clinical application of the BEST shall be discussed and evaluated, whether it adds information to tomographic findings.

The BEST-Display

The BEST has been integrated into the CST software since November 2022. The “Homburg Biomechanical E-Staging” display enables the simultaneous presentation of up to four differently colored CST examinations (Fig. 5). The baseline examination is usually shown in red color, followed by up to three follow-up examinations in green, yellow and blue color. The upper left part of the chart visualizes the corneal deformation response for the first (red) and the last (blue) of the four displayed examinations and a superimposed view of these two deformation videos. The inferior part of the chart shows the absolute values and the number of standard deviations from the mean in healthy patients for the individual biomechanical parameters DA ratio 2 mm, integrated radius, ARTh, SP-A1 and stress–strain index (SSI). The SSI shows the biomechanical strength of a healthy, 50-years-old patient’s cornea in a gray colored curve (value 1.0). A SSI > 1.0 and a left shift of the curve indicates a higher corneal stiffness, a SSI < 1.0 and a right shift of the curve, a lower corneal stiffness [42]. In each follow-up examination, the CST uses the baseline examination as a reference and evaluates each parameter, whether the cornea became “stiffer”, “softer” or whether the difference was “not significant”.

Fig. 5figure 5

The “Homburg Biomechanical E-Staging” Corvis ST® (Oculus, Wetzlar, Germany) display of a male, 27-year-old patient with keratoconus and annual follow-up examinations from 2019 to 2022. Top left: data from the four separate examinations presented at the same time (red, green, yellow, and blue), visualization of the baseline corneal deformation (red), and corneal deformation in the final examination (blue). Upper center: superimposed display of corneal deformation visualizations. Below: absolute values and number of standard deviations from the mean value in healthy patients for (1) deformation amplitude ratio 2 mm (DA ratio), (2) integrated radius, (3) Ambrósio relational thickness in horizontal (ARTh), and (4) stiffness parameter SP-A1. (5) stress–strain index (SSI); left-shift of the curve corresponds to a higher material stiffness, a right-shift to a lower material stiffness. Right part: biomechanical E-staging for the four examination time points. Linear progression from red (Nov 22, 2019) to green (Nov 13, 2020) to yellow (Nov 12, 2021) to blue (Jul 1, 2022). Progression assessment displayed below the individual biomechanical parameters with DA ratio and SP-A1 indicating progression from the first follow-up. Progression assessment also displayed below the biomechanical E-staging indicating progression at the last follow-up examination (blue). Tomographic values for 2019, 2020, 2021, 2022: anterior simulated keratometer readings, steep meridian (K2): 44.5 D, 44.4 D, 44.4 D, 44.9 D; maximal anterior keratometry values (Kmax): 47.6 D, 46.8 D, 47.1 D, 48.4 D; thinnest corneal thickness (TCT): 525 μm, 521 μm, 513 μm, 511 μm and Belin-Ambrósio deviation (BAD) index: 3.35, 3.85, 3.58, 3.93

On the right side, the CBiF and the resulting BEST are displayed for each of the four single examinations. While the BEST increases with increasing KC severity, it is important to note that the CBiF decreases by definition with increasing KC severity. The CBiF and BEST are therefore not individual parameters, but combine the information derived from several biomechanical parameters in a similar way to the tomographic BAD index, which is composed of several deviation values. Based on this information, the CST states below the displayed CBiF and BEST values, whether the keratectasia is progressive or not. This assessment is based on the actual values of the BEST without defined confidence intervals as opposed to the Belin ABCD progression display of the Pentacam.

Biomechanical analysis of A0B0C0 corneas

Besides healthy corneas without ectatic regions that are classified as A0B0C0 according to the ABCD staging system, there exist also early KC stages with an A0B0C0 staging result (Fig. 6). The Pentacam automatically provides the ABC stage in the “Topometric/KC Staging” display. The example of a 34-year-old male patient wearing glasses shows an inferior steepening typical of KC, although the Pentacam staging result is A0B0C0. The table “Belin ABCD Keratoconus Staging” shows the actual “A, B, C” values together with the classification result in decimals (A0.0 B0.0 C0.9 in the example). The values are always rounded down and thus it is possible for KC corneas to be classified as A0B0C0. Consequently, A0B0C0 as a staging result does not exclude the diagnosis of KC and the ABC grading must be considered as a supplement to the ophthalmologist’s assessment of the Pentacam and not as a replacement [20]. Among the HKC patients, there were 200 (100 right and 100 left) eyes classified A0B0C0 with manifested KC in the partner eye. The mean BAD index was 1.97 ± 1.31 (mean ± standard deviation) in the A0B0C0 group and 7.62 ± 6.57 in the fellow eyes (P < 0.0001, Wilcoxon matched-pairs test). The mean CBI was 0.48 (A0B0C0, and thus approached 0.5 – not providing a clear decision between zero/healthy versus one/ectatic) and 0.84 (approaching one/ectatic) in the fellow eyes with manifested KC. The BEST was 0.83 (approaching stage E1) in A0B0C0 corneas and 2.20 in the fellow eyes with manifested KC, which revealed abnormal biomechanics in A0B0C0 corneas more intuitively than the CBI.

Fig. 6figure 6

The “Topometric/KC-staging” Pentacam® display (Oculus, Wetzlar, Germany) of a 34-year-old male patient with glasses. Typical inferior steepening but resulting stage is A0B0C0. Actual values of ABC staging in the table “Belin ABCD Keratoconus Staging”: A0.0B0.0C0.9

“Unilateral keratectasia”

Sometimes, ophthalmologists may question the statement of the “Global Consensus on Keratoconus and Ectatic Diseases” [1] that unilateral KC does not exist because there exist indeed extremely asymmetrical KC cases that may show a normal tomography in one eye. Studies analyzing asymmetry in KC severity stages found that: (1) the progression of KC in one eye seems to be independent of the fellow eye [43] and that (2) among 350 KC patients, 30% had the same TKC stage in both eyes, whereas 34% showed a difference of one stage, 25% of two stages and the remaining 11% of more than two stages [44]. Another study analyzed 26 tomographically unremarkable fellow eyes of manifested KC in the other eye and found biomechanical abnormalities that helped to distinguish the tomographically unremarkable fellow eyes from healthy controls [45]. In view of these findings, it seems appropriate to carefully reflect on the definitions of the terms “subclinical KC” versus “forme fruste KC”, that have often been used interchangeably in literature. Marc Amsler examined eyes without clinical KC signs and manifested KC in the fellow eye using Placido photokeratoscopy and defined “forme fruste” as an incomplete, abortive or unusual form of a syndrome or disease [46, 47]. Consequently, a “forme fruste KC” should be a cornea with inconspicuous topographic and tomographic parameters contralateral to a cornea with manifested or advanced KC. “Subclinical KC”, however, may nowadays be diagnosed in both eyes of an individual at a very early stage and this does not necessarily mean that the disease does or will follow an asymmetrical course. One could hypothesize that the increasingly widespread use of advanced corneal tomographic and biomechanical diagnostics will achieve an earlier detection of KC than in the past, leading to increasing incidence values and earlier initiation of a stage-appropriate therapy [7].

An example of a 21-year-old male patient with glasses and a bilateral visual acuity of 20/20 (Fig. 7) illustrates the importance of biomechanical analysis in “unilateral KC”. In the right eye, there is a with-the-rule-astigmatism and unremarkable KC indices, whereas the Pentacam detects TKC1 KC in the left eye. The Belin/Ambrósio enhanced ectasia display shows unremarkable elevation data for the right eye and a borderline BAD index (1.96). The left eye shows clearly abnormal findings in elevation data and BAD index (3.9). The BEST, however, displays E1.4 for the right and E1.9 for the left eye, thus revealing abnormal corneal biomechanics (E1, rounded down) in both eyes. This example highlighted that abnormal corneal biomechanics can be uncovered even in tomographically unremarkable eyes and that “unilateral KC” may be a snapshot of a highly asymmetrical keratectasia.

Fig. 7figure 7

The “Topometric” Pentacam® (Oculus, Wetzlar, Germany) display of a 21-year-old male wearing glasses. Regular with-the-rule astigmatism on the right, inferior steepening on the left, with classification as keratoconus in stage TKC1

Regular astigmatism or keratoconus?

Regular astigmatism is a differential diagnosis of KC, especially in patients presenting to the ophthalmologist because of fluctuating high cylinder values. Despite thorough examination, an incipient subclinical KC cannot be ruled out with certainty in some cases. This is illustrated by the example of a 17-year-old male patient (Fig. 8) achieving a spectacle-corrected visual acuity of 20/25 with suspect of KC. The topometric Pentacam analysis showed a regular with-the-rule astigmatism in both eyes with inferior steepening and classified the right eye TKC1-2 and the left eye TKC1. The BEST showed slightly abnormal corneal biomechanics with E1.0 in the right and E1.1 in the left eye.

Fig. 8figure 8

The “Topometric” Pentacam (Oculus, Wetzlar, Germany) image of a 17-year-old male wearing glasses. Regular with-the-rule astigmatism in both eyes but with distinct inferior steepening. Classification as keratoconus stage TKC1-2 on the right (Belin-Ambrósio deviation index: 2.58) and TKC1 on the left (Belin-Ambrósio deviation index: 1.56). Biomechanical E-Staging for the right eye: E1 (E1.0), and for the left eye: E1 (E1.1)

BEST progression assessment

Tomographic progression assessment can be based on several parameters. An increase of the BAD index of more than 0.42 was supposed to be indicative of keratectasia progression [48]. The Save Sight Keratoconus Registry Study stated in 2021, that the Kmax increase of more than one diopter or the decrease of thinnest corneal thickness of more than 20 µm per year still remain up-to-date in daily practice [49]. A third study stated that especially in children, an initially advanced KC should be considered a risk factor for further keratectasia progression and recommended tomography changes beyond the limits of repeatability as criteria for determining progression [50]. The limits of repeatability have to be defined to determine how much of a deviation from normal indicates the change required to be considered KC progression rather than deviation owing to test variance [51]. Unfortunately, the “Global Consensus on Keratoconus and Ectatic Diseases” [1] lacked a clear definition of keratectasia progression. In Germany, the Kmax increase of more than one diopter is a legally defined progression parameter.

To evaluate how corneal biomechanics change in progressive KC corneas, a group of KC patients that underwent accelerated epithelium-off corneal crosslinking (9 mW/cm2, 5.4 J/cm2, 10 min) was examined (169 KC corneas of 169 KC patients). The baseline examination and the first follow-up examination that revealed progression, leading to the decision to perform corneal crosslinking were compared. The delta was + 1.405 for Kmax, −0.118 for “A”, −0.371 for “B”, −6.231 for “C” of the tomographic ABCD staging system, + 0.296 for DA ratio 2 mm, + 0.378 for integrated radius, −14.78 for ARTh, −2.485 for SP-A1 and + 0.2817 for BEST, which indicates, that both single biomechanical parameters and the combination of several biomechanical parameters in the BEST may also be used for keratectasia progression assessment (Fig. 9).

Fig. 9figure 9

The “Homburg Biomechanical E-Staging” Corvis ST® (Oculus, Wetzlar, Germany) display of a male 24-year-old patient with keratoconus and follow-up examinations from 2019 to 2021. Red, baseline examination (Nov 21, 2019). Green, first follow-up with same biomechanical E-staging (Nov 26, 2020). Yellow, second follow-up with biomechanical progression detected by the Corvis ST® (May 10, 2021). Blue, third follow-up with further progression (Nov 11, 2021)

BEST post-crosslinking assessment

Corneal crosslinking is effective in halting KC progression both in adolescents [52] and adult KC patients [53]. A tomographic pseudoprogression can be observed within the first six postoperative weeks, which is not indicative of the long term effect [53]. To investigate how the BEST behaves after crosslinking, a retrospective study was performed based on 49 KC corneas of 41 patients who underwent at least one preoperative CST measurement and one postoperative measurement about one year after corneal crosslinking [54]. For some patients, there existed additional measurements at different follow-up time points, so that 22 KC corneas could be evaluated at 5 months, 49 at 11 months and 21 at 23 months after corneal crosslinking. The tomographic evaluation of Kmax and anterior corneal curvature (“A”) showed significantly lower values after crosslinking, demonstrating tomographic stabilization. The postoperative BEST value was significantly higher than preoperatively at 5 months (P = 0.035, paired t-test), but not significantly different at 11 months (P = 0.0578) and remained fairly similar to the preoperative baseline examination at 23 months (P = 0.9683). Therefore, the BEST cannot be used to assess progression or stability in keratectasia within the first year after corneal crosslinking, whereas it indicates stabilization at preoperative values in the long term after more than one year (Fig. 10). The answer to the question why the BEST increases within the first months after corneal crosslinking seems to lie in the Scheimpflug-based corneal thickness measurement. The majority of KC corneas (> 75%) had a postoperatively lower thinnest corneal thickness measurement and all of them showed a higher postoperative BEST value. Some KC corneas (≤ 25%) had a postoperatively higher thinnest corneal thickness measurement and showed an equal or lower BEST value. Corneal thickness influences the BEST via the biomechanical ARTh value. This raises the question, whether changed thickness values after corneal crosslinking are real or whether they represent measurement artifacts. Some studies reported an underestimation of corneal thickness after corneal crosslinking by Scheimpflug cameras [55, 56] because of stromal hyperdensities within the first year after crosslinking, causing a pachymetric artifact. Consequently, the postoperative BEST value increase could be interpreted as another artifact related to that Scheimpflug camera-related pachymetric artifact. One could assume based on these results that it requires the human cornea one to two years to approach its preoperative biomechanical status after corneal crosslinking. The postoperative stiffening effect of corneal crosslinking could be observed earlier. Individual biomechanical parameters such as the integrated radius and SSI indicated an increase in corneal stiffness as early as one month after corneal crosslinking [57, 58].

Fig. 10figure 10

Representative course of the “Homburg Biomechanical E-Staging” after corneal crosslinking as displayed by the Corvis ST® (Oculus, Wetzlar, Germany). Female, 30-year-old patient with keratoconus and follow-up examinations from 2019 to 2021. Red, baseline examination (Oct 22, 2019). Accelerated corneal crosslinking was performed one month later (Nov 18, 2019). Green, first follow-up seven months later with biomechanical (pseudo-)progression (Jun 15, 2020). Yellow, second follow-up with same biomechanical E-staging ten months after crosslinking (Sep 14, 2020). Blue, third follow-up (Sep 27, 2021) with biomechanical stabilization near preoperative value about two years after corneal crosslinking

BEST assessment after intracorneal ring segment implantation

Intracorneal ring segments were initially developed for the correction of myopia and later on used to improve visual acuity in advanced KC corneas delaying the need of keratoplasty. A recent meta-analysis reported that the femtosecond-laser assisted intracorneal ring segment implantation is effective in achieving good visual, keratometric and refractive results [59], especially in KC patients with unsuccessful contact lens fitting and initially poor visual acuity results. A reliability study revealed a reliable assessment of corneal refractive power in KC after intracorneal ring segment implantation with both Scheimpflug-imaging and anterior segment optical coherence tomography [60], which indicates that the quality and reliability of Scheimpflug-imaging, as used also by the CST, is most likely to be not impaired by and after intracorneal ring segment implantation.

Furthermore, for this treatment option, 49 KC corneas treated with intracorneal ring segment implantation were retrospectively evaluated if they had at least one preoperative and one postoperative CST measurement about one year after surgery. Similar to the post-crosslinking evaluation, there existed additional measurements at different follow-up time points for some eyes. There were significantly lower Kmax and “A” values postoperatively when compared to preoperatively. In comparison to preoperative values, ARTh was the only biomechanical parameter that showed consistently significantly lower values postoperatively. Lower ARTh values are indicative of thinned central cornea with a steep increase of corneal thickness towards the periphery [31]. As the thinnest corneal thickness measurement did not change postoperatively, it can be assumed that the ring segments implanted in the 6 mm zone are the actual cause of the steeper thickness increase towards the corneal periphery. This results in a relatively thinner central corneal thickness when compared to the implantation zone. Since these thickness changes in the midperiphery of the cornea persist as long as the ring segments stay in place, and because ARTh influences the BEST, it can be assumed that this leads to a permanent influence on the BEST, that showed higher values than preoperatively in all follow-up examinations (Fig. 11). Consequently, despite not affecting purely biomechanical measurements in early follow-up examinations [61], the implantation of intracorneal ring segments may be detected by the CST and the parameters ARTh and BEST.

Fig. 11

留言 (0)

沒有登入
gif