Effectiveness of biomechanical evaluation in early diagnosis of keratoconus

Objective: To compare different analyses considering biomechanical properties in addition to three-dimensional evaluation of the cornea in the early diagnosis of keratoconus (KC) Materials and Methods: Sixty-eight eyes of 68 patients and 45 eyes of 45 healthy volunteers were included in the study. The mild eye with ectasia of patients with bilateral KC was group 1 (45 eyes of 45 patients), the fellow eye with normal tomographic findings of patients with KC in one eye was group 2 (23 eyes of 23 patients) and the right eye of healthy volunteers was group 3 (45 eyes of 45 control subjects). Demographic characteristics, best spectacle lens corrected visual acuities and detailed biomicroscopic examination findings were recorded. Corneal tomographic measurements were performed with Sirius, Pentacam HR and Corvis ST devices. The parameters obtained from the devices and the combination models created with the parameters were compared in terms of their effectiveness in discriminating between the groups. Areas under the curve (AUROC) were compared using the Delong test. Results: Tomographic biomechanical index (TBI)(AUROC=0.999), Corvis biomechanical index (CBI)(AUROC=0.958), Belin-Ambrosio advanced ectasia screen total deviation value (BAD-D) (AUROC=1.000), Pentacam random forest index (PRFI)(AUROC=1.000) and Baiocchi-Calossi-Versaci index (BCV)(AUROC=1.000) parameters showed the highest diagnostic power, but were not superior to each other in differentiating the eyes of KC (group 1) and healthy (group 3) eyes. Logistic regression model 1 (LRM1) (AUROC=0.970), which was created from the parameters of Ambrosio relational thickness (ARTh) and deformation amplitude ratio max 2 mm (DAR max 2 mm) on the horizontal meridian of Corvis, showed similar diagnostic power to CBI(AUROC=0.958) with 91% sensitivity and 89% selectivity in differentiating groups 1 and 3. TBI(AUROC=0.863), BAD-D (AUROC=0.814), PRFI (AUROC=0.863), BCV (AUROC=0.845), posterior surface BCV index (BCVb)(AUROC=0.885) and posterior surface asymmetry index (Slb)(AUROC=0.883) parameters showed the highest diagnostic power and were not superior to each other in differentiating groups 2 and 3. However, the diagnostic efficiency of these parameters was superior to CBI (AUROC=0,719) in differentiating groups 2 and 3. Logistic regression model 2 (LRM2)(AUROC=0.849), which was constructed from the posterior surface elevation value of Pentacam at the thinnest point of the cornea (B.Ele.Th) and the stiffness parameter of Corvis at the time of first applanation (SP-A1), was superior to CBI (AUROC=0.719) in terms of diagnostic power with 83% sensitivity and 80% selectivity. Conclusion: There was no diagnostic superiority between TBI and CBI in the discrimination of corneas with KC and healthy cases. However, TBI was superior to CBI in terms of diagnostic power in differentiating group 2 with corneas suspected of ectasia from healthy cases. Indices including morphological-biomechanical parameter combinations such as TBI and LRM2 may be more effective in the early diagnosis of keratoconus. Therefore, parameter combinations obtained by using different devices can be utilised in the early detection of suspected KC. Key words: Ectasia, corneal biomechanics, keratoconus