Diagnostic Performance of Integrated Tomographic–Biomechanical Indices Across Keratoconus Stages in a Chinese Population

Purpose: To evaluate the diagnostic performance of integrated tomographic-biomechanical indices (specifically TBI, TBI for Chinese populations [cTBI], Corvis biomechanical index [CBI], and CBI for Chinese populations [cCBI]) across the different stages of keratoconus (KC) in a Chinese population and to explore a two-step approach for improving detection of early KC. Methods: This cross-sectional study included 642 eyes of 642 subjects (one eye per subject) classified into six groups: normal controls, forme fruste keratoconus (FFKC), subclinical keratoconus (SKC), and three clinical keratoconus (CKC) subgroups. A two-step workflow combining TBI and cTBI was additionally implemented in FFKC and SKC. Diagnostic performance metrics, including receiver operating characteristic curves and area under the curve (AUC), were calculated for each index across KC stages using MATLAB. Results: All indices showed limited discrimination for FFKC (AUC < 0.8) but improved markedly with increasing KC severity. For SKC and CKC, these integrated indices demonstrated reliable diagnostic performance. At fixed cutoffs, cCBI and cTBI outperformed CBI and TBI, with higher specificity (91.27% and 88.89% vs. 45.24% and 57.14%) but lower sensitivity in SKC (64.61% and 79.66% vs. 81.36% and 94.92%). The two-step workflow yielded higher accuracy (89.73%) in SKC than any single parameter. Conclusions: Integrated tomographic-biomechanical indices demonstrate reliable diagnostic performance for SKC and CKC in a Chinese population, although accurate discrimination of FFKC remains challenging. Population-adjusted metrics showed better overall performance at the specific cutoff value, driven by higher specificity despite lower sensitivity. A two-step workflow achieved higher accuracy in SKC, offering a balanced, clinically applicable approach for early screening of subclinical cases. Translational Relevance: Integrating population-adjusted tomographic-biomechanical indices and a two-step diagnostic workflow enhances the accuracy of early keratoconus detection, providing a robust, clinically applicable framework for improving diagnostic precision and patient management in Chinese populations.