Screening for Subclinical Keratoconus

Screening for subclinical keratoconus is often framed as uncovering a hidden diagnosis. That view is too narrow. Benjamin Franklin's maxim that "an ounce of prevention is worth a pound of cure" is a clinical mandate in refractive surgery. Screening is individualized prevention that helps to protect vision, optimize outcomes, and avoid complications, and it allows intervention before a therapeutic window closes.

Beyond a Binary Diagnosis

Keratoconus and ectatic corneal disease are not binary entities. Rather, they represent a spectrum of biomechanical, structural, optical, epithelial, endothelial, genetic, molecular, inflammatory, and behavioral factors. The question is not only whether the patient has keratoconus but also what the cornea reveals today about its condition, risks, needs, and tolerance for intervention.

Defining the Clinical Question

Two complementary domains can be distinguished. The term corneal ectasia diagnostics refers to the comprehensive evaluation of suspected or established disease. This evaluation includes screening, confirmation, classification, staging, prognosis, individualized management, and follow-up. The term ectasia risk assessment refers to the preoperative evaluation of whether a cornea can tolerate the biomechanical impact of elective refractive surgery. It also informs phakic IOL and refractive cataract surgery planning, where IOL selection, toric strategy, enhancements, and counseling depend on the same susceptibility judgment. Predictive AI can enhance both domains.¹

The Limits of Current Labels

The Global Consensus process is important, but the resulting documents should not be read as immutable doctrine. Consensus methodology requires coordinators to reflect the panel, not their individual opinions. The first and second Global Consensus documents should therefore be understood as frameworks for practice that identify open questions and help move the field from definitions toward decision-making.^2-4

Nomenclature, for example, remains an open question. The terms suspicious keratoconus, keratoconus suspect, subclinical keratoconus, and forme fruste keratoconus have been used inconsistently. Henriquez et al highlighted the heterogeneity of definitions, which reinforces that these terms often reflect different levels of diagnostic certainty rather than sharply separated biologic entities.⁵ Definitions that lean too heavily on future follow-up warrant caution. A first-visit patient rarely has prior images, and tomorrow's outcome is not yet written. Intellectual humility entails making the most responsible decision today while accepting that some questions resolve only with time.

Recognizing Ectasia Susceptibility

Very Asymmetric Ectasia

Keratoconus is fundamentally bilateral, whereas secondary ectasia can occur unilaterally after a mechanical insult. A topographically normal fellow eye with very asymmetric ectasia (VAE) should not be presumed to be normal because the map is quiet. Nor should it be presumed to be diseased because the other eye is affected. Its tomographic, biomechanical, epithelial, and intereye profile defines its position along the susceptibility continuum. The VAE-NT (very asymmetric ectasia with normal topography in the fellow eye) model has helped develop and validate enhanced diagnostic methods.⁶

A case series illustrates this continuum. A patient had unilateral ectasia from rubbing one eye, and the fellow eye was normal and stable. Another patient had clinical keratoconus in one eye and a fellow eye with normal Placido topography but abnormal tomography and biomechanics, consistent with forme fruste susceptibility. A third patient had normal topography and no symptoms in either eye, yet tomography and biomechanics revealed bilateral subclinical disease. The lesson is to recognize distinct biologic realities rather than to force all patients into a single label.⁷

The Two-Hit Model

The two-hit theory is a practical model for clinical reasoning.^8-10 The first hit is intrinsic susceptibility arising from factors such as genetic predisposition, stromal architecture, biomechanical weakness, atopy, inflammation, epithelial compensation, or molecular vulnerability. The second hit is mechanical or surgical stress from eye rubbing, chronic allergy, sleep-related compression, trauma, or tissue alteration from laser vision correction. Susceptibility plus load determines decompensation.

Building a Multimodal Risk Profile

Beyond Anterior Topography

The diagnostic journey began with the anterior surface. Rabinowitz established quantitative topographic criteria, and Klyce, with Maeda and Smolek, pioneered neural network interpretation of Placido-based topography.^11,12 Placido disk–based topography remains essential for anterior curvature, irregular astigmatism, and tear film quality. Still, anterior curvature is only one expression of a deeper condition.

The Multimodal Intelligence Framework is intended to go beyond, not over, any single technology.¹³ It includes the patient history and examination, Placido topography, Scheimpflug or OCT tomography, pachymetric and epithelial mapping, corneal biomechanics, aberrometry, axial length measurements, specular or confocal microscopy when endothelial or Descemet membrane abnormalities are possible, and genetic testing and molecular biology. AI does not replace these facets; it integrates them.

Integrating Structure and Biomechanics

Scheimpflug tomography added anterior and posterior elevation, pachymetry, and relational thickness. The Belin/Ambrósio Enhanced Ectasia Display (BAD) moved screening toward integrated structural analysis, and BAD-Dv4 is its AI-optimized evolution.^14,15 Biomechanical assessment added the recognition that the cornea is a load-bearing structure, not merely a shape. The stiffness parameter at first applanation (SP-A1) and Corvis biomechanical index (CBI) capture this behavior from dynamic Scheimpflug images, whereas the tomographic biomechanical index (TBI) and TBIv2 integrate tomography and biomechanics for challenging cases, including VAE-NT.^16-18

Combining Susceptibility With Procedural Impact

The BrAIN Enhanced Ectasia Software (BEES; BrAIN, Brazilian Artificial Intelligence Network in Medicine) operationalizes the aforementioned reasoning in real-world practice. It is not a treatment recommendation, a guarantee of safety, or permission to operate. Instead, it is physician-led clinical decision support. BEES integrates validated diagnostics with a structured risk assessment by reading intrinsic susceptibility and planned procedural impact together. Data from the Pentacam (Oculus Optikgeräte) support BAD-Dv4, with its normalized index, the Dv4I expression from zero to one (0.5 equivalent to 1.45), and tomographic architecture. The Corvis ST (Oculus Optikgeräte) adds CBI, SP-A1, TBIv2, and dynamic biomechanical behavior when available.

The planned procedure is then entered into the model through treatment type, refractive correction, optical zone, flap or cap parameters, ablation depth, and patient age. The relational tissue altered index estimates the procedure-related second hit, allowing plan comparison beyond static residual stromal bed or percent tissue altered.¹⁹

The level of susceptibility and the relational procedural impact may support treatment if the clinical context is consistent. For example, a high impact may require reducing treatment ablation depth, changing the optical zone, or considering a lens-based alternative. High susceptibility, even with a low impact, merits caution. High susceptibility plus a high impact generally argues against elective corneal laser vision correction. BEES organizes this reasoning, but the physician makes the decisions.

The Figure illustrates an enhanced-specificity case in which inferior steepening was not overcalled as ectasia, because a multimodal assessment and BEES demonstrated reassuring tomography, preserved biomechanical stiffness, low BrAIN TBI, a minimal predicted surgical impact, and stable outcomes 10 months after Presbyond Laser Blended Vision (Carl Zeiss Meditec).

From Screening to Prevention and Care

Identifying Risk Before Disease Becomes Obvious

Active screening has a public health dimension. In a prospective Gaffrée and Guinle University Hospital study, 521 previously undiagnosed adults underwent tomography and corneal biomechanics.²⁰ A subjective anterior topometric analysis identified 5.8% of eyes as ectatic or suspicious for ectasia, whereas multimodal indices revealed broader susceptibility. A BAD-Dv3 score of 1.9 or higher was present in 10.9% of eyes, a CBI score of 0.5 or higher was present in 26.9% of eyes, and a TBIv2 score of 0.43 or higher was present in 32.7% of eyes. These numbers do not represent the prevalence of clinical keratoconus. Rather, they reveal hidden vulnerability beyond what is obvious. Eye rubbing was strongly associated with disease, reinforcing the idea that screening must be paired with education.

This is the spirit of Violet June, the global keratoconus awareness campaign, and its message that awareness is prevention.²¹ Screening should identify risks, educate patients about the risks of eye rubbing, guide allergy control, and establish active follow-up when certainty is impossible. The pathway begins with the patient history and examination, continues through multimodal imaging, and ends with integration. No single index is a green light. Discordance matters.

Turning Risk Into a Plan

Patient care must be individualized. Not every abnormality requires surgery, but every meaningful abnormality deserves an explanation and a plan. CXL is considered when disease progression is documented or the risk is high, especially in children. Visual rehabilitation is considered when glasses or contact lenses fail. Endothelial status is relevant when keratoplasty or anterior segment anomalies are present. Refractive surgery is avoided when susceptibility and impact make the risk unacceptable.

AI as a Second Reader

The future is not AI's replacing the surgeon. It is (A2I)² or applied AI integrated with applied ancient intelligence, the clinical intuition passed down through training.¹ AI should function as a disciplined second reader that is transparent, explainable, population-aware, externally validated, and clinically supervised.

Screening for subclinical keratoconus is therefore not merely diagnostic but also ethical. It calls for pausing before weakening the cornea and integrating technology with humility. Patients are examined before action is taken. Characterization occurs before cutting. Integration comes before deciding.

An AI language model (ChatGPT, GPT-5.5 Thinking, OpenAI) and Grammarly (Grammarly) were used to assist with writing, editorial organization, English language refinement, reference formatting, and figure preparation. BEES is discussed in the article as an AI-assisted clinical decision support platform for ectasia susceptibility and refractive surgery ectasia risk assessment. In the case shown in the Figure, BEES supported physician-led decision-making by integrating tomographic, biomechanical, and procedure-impact data. BEES is not a treatment recommendation, a guarantee of safety, or a substitute for clinical judgment. All scientific content, clinical interpretations, references, and conclusions were reviewed, verified, and revised by the authors, who assume full responsibility for the accuracy, originality, and integrity of the manuscript. This disclosure follows the principle that AI may support scientific writing and clinical reasoning but cannot replace authorial responsibility, transparency, accountability, or physician judgment.²²