An exploratory analysis of forme fruste keratoconus sensitivity diagnostic parameters

Area of Science:

• Ophthalmology
• Biomedical Engineering
• Corneal Imaging

Background:

• Keratoconus (CKC) and forme fruste keratoconus (FFKC) are progressive corneal diseases.
• Accurate diagnosis relies on identifying sensitive biomarkers from advanced imaging technologies like Pentacam and Corvis ST.
• Previous studies have analyzed numerous parameters, necessitating a comprehensive statistical evaluation to pinpoint the most effective diagnostic indicators.

Purpose of the Study:

• To conduct a secondary statistical analysis of Pentacam and Corvis ST parameters.
• To identify more sensitive diagnostic parameters for clinical keratoconus (CKC) and forme fruste keratoconus (FFKC).

Main Methods:

• Extracted and screened parameters and AUC data from existing literature to create databases for CKC and FFKC.
• Employed random forest methods (%IncMSE and IncNodePurity) for preliminary parameter selection.
• Utilized partial dependency analysis to identify sensitive parameters that enhance diagnostic performance, followed by a diagnostic test meta-analysis for validation.

Main Results:

• Analyzed 88 parameters (766 records) for CKC and 57 parameters (346 records) for FFKC.
• Identified 60 important parameters, with 20 selected as sensitive for keratoconus, primarily related to the thinnest point of the cornea.
• The stiffness parameter at first applanation (SPA1) emerged as a sensitive Corvis ST parameter for FFKC, alongside the Tomographic and Biomechanical Index and Corvis Biomechanical Parameter (CBI).
• Meta-analysis of SPA1 showed an AUC of 0.87 (95% CI, 0.84-0.90) with no significant heterogeneity.

Conclusions:

• SPA1 demonstrates comparable sensitivity to CBI for FFKC diagnosis and may be an early indicator of corneal biomechanical changes in keratoconus progression.
• Elevation parameters, particularly those related to the thinnest point of corneal thickness, are highly significant for diagnosing keratoconus.
• The study validates the use of advanced statistical methods for identifying sensitive diagnostic parameters in corneal imaging.