Pathogenesis, early detection, and treatment of keratoconus: an updated narrative review

Introduction

Background

Keratoconus (KC) is a common corneal ectatic disorder hallmarked by progressive paracentral and central corneal thinning and steepening, resulting in high irregular astigmatism (1). The pathogenesis is multi-factorial, involving genetic, environmental, biomechanical, and cellular mechanisms (2). Genetically, numerous loci and gene variants have been implicated, particularly those with known involvement in composing the extracellular matrix (ECM) involved in collagen synthesis, and corneal embryonic development (3). Notable genes include lysyl oxidase (LOX), MIR184, hepatocyte growth factor (HGF), and RAB3GAP1 (2,3). In addition, chronic eye rubbing, low-grade inflammatory states and mechanical trauma contribute to worsening corneal ectasia (1,4). Chronic eye rubbing is a well-known risk factor, through direct mechanical trauma and stress to the posterior corneal stroma; resulting in subsequent anterior corneal changes (4-6). KC biomechanically is also hallmarked by pathological activation of proteolytic enzymes, downregulation of protease inhibitors and increased expression of proinflammatory makers, decreased number of stromal keratocytes, and changes in collagen types XIII, XV, XVIII (1,2,4,6) The combinatory effect results in bilateral, asymmetric corneal stromal thinning, degeneration, and conical protrusion (2).

Clinically, corneal thinning and conical protrusion results in irregular myopic astigmatic shifts. Patients tend to experience worsening visual activity with uncorrectable vision loss. In addition, the pathological conical protrusion and thinning can result in other notable signs: Fleischer ring’s, Munson’s sign, and Rizzotti’s sign. Fleischer rings indicate hemosiderin deposition from the tear film on the side of abnormal conization (7). Munson’s sign represents a V-shaped appearance of the lower eyelid, that is most prominent with inferior gaze (7,8). Rizzuti’s sign referencing a bright nasal limbal reflection when the light is shown in the temporal limbal area. Corneal opacifications can become pronounced due to degeneration within Bowman’s membrane and subsequent corneal scarring (7-9). Biomechanical changes to Descemet’s membrane can result in vertical striae produced by compression of Descemet’s membrane (Vogt’s striae) and acute hydrops (7-9). Corneal hydrops is a feared complication, secondary to breaks in Descemet’s membrane resulting in acute stromal edema and sudden, painful vision loss.

Recent advances in diagnostic imaging studies have dramatically altered the ability to detect subclinical KC and monitor disease progression (8,10). Scheimpflug tomography, anterior segment optical coherence tomography (AS-OCT), and corneal biomechanical assessments have aided in precise quantification of the corneal conical shape, pachymetry thickness, and intrinsic biomechanical structural parameters (8,10).Utilization aided to facilitate earlier disease progression; however, clinical application stills varies.

Concurrently, the therapeutic landscape of KC has continued to expand over the last several decades. Current management has moved away corrective interventions to intrinsic biomechanical stabilization and stromal structural rehabilitation (10,11). Oppositely, emerging adjuvant technologies—including customized corneal collagen crosslinking (CXL) protocols, intracorneal ring segment (ICRS) implantation, aiming to slow disease progression (12).

Rationale and objective

In addition to changes in the therapeutic landscape of KC, the use of artificial intelligence (AI) and machine learning (ML) models is becoming increasingly utilized in screening and diagnostics of KC management. Recent work has demonstrated the potential use of machine and deep learning algorithms to combine indices from corneal topography, tomography and optical coherence tomography to help risk stratify patients and detect earlier KC changes. In addition, models have begun focusing on identifying pre-operative screening markers that correlate to the likelihood of post-refractive surgery ectasia changes, potentially opening an avenue for better post-surgical predictions (13). In recent studies, have begun extending the use of AI to treatment planning, to help guide timing of interventions (corneal cross linking, ICRSs, etc.), however these have lacked data set heterogeneity and standardization; thus limiting widespread real work application (14). Thus, given the rapidly changing landscape of therapeutic management options, it is imperative to have a comprehensive understanding of the underlying pathophysiology to continue optimizing individual patient managements and improve long term outcomes in KC (14,15). We present this article in accordance with the Narrative Review reporting checklist (available at https://aes.amegroups.com/article/view/10.21037/aes-2026-0020/rc).

Methods

This narrative review was conducted using a literature search in PubMed, as can be referenced in Figure 1 as below. Articles were published in English between 2023 and 2025, with an emphasis on recent studies focusing on KC pathogenesis, diagnosis, and treatment. Studies were excluded if they had non-relevant information, and if had outdated clinical management recommendations (Table 1).

Figure 1 PRISMA flowchart. Identification of studies via PubMed using Rayyan. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Discussion

Pathophysiology

At the cellular and molecular level, KC is characterized by the disruption of stromal collagen lamellae, a reduced collagen content and decreased natural cross-linking, likely due in part to reduced activity of lysyl oxidase (LOX) (16). Thus, resulting to progressive ectatic corneal changes and thinning (16,17). In parallel, increased activity of matrix metalloproteinases (MMPs), particularly MMP-1, MMP-2, and MMP-9, contributes to the breakdown of collagen types I and V within the ECM (Figure 2) (16,18).

Figure 2 Overview of KC pathogenesis, including risk factors, clinical presentation and associated mechanisms underlying the pathological changes appreciated on physical exam. KC, keratoconus; UV, ultraviolet.

Oxidative stress plays a central mechanistic role in KC pathogenesis. Corneal tissues in KC demonstrate increased levels of reactive oxygen species (ROS) alongside reduced antioxidant defenses. Specifically increased activity of superoxide dismutase (SOD), catalase, and glutathione peroxidase. Excess ROS induces mitochondrial dysfunction and DNA damage in keratocytes, triggering apoptosis. Importantly, oxidative stress also activates redox-sensitive transcription factors such as NF-κB and AP-1, which upregulate the expression of proinflammatory cytokines and MMPs. This causes a feed-forward loop in which oxidative stress amplifies ECM degradation both directly through cellular damage, and indirectly via MMP induction and inflammatory signaling (16,17). Chronic inflammation further amplifies this degenerative cascade. Proinflammatory cytokines, including interleukin (IL)-1β, IL-6, IL-17A, tumor necrosis factor alpha (TNF-α), are elevated in KC tear film and corneal tissue and play numerous roles in the inflammatory state within KC pathogenesis. IL-1β is released from damaged epithelial cells, and acts as a mediator by stimulating keratocytes to produce IL-6 and TNF-α. In addition, IL-1β is a keratocyte apoptosis regulator. TNF-α further enhances NF-κB activation, reinforcing the transcription of additional cytokines and MMOs. IL-6 contributes by aiding in IL-17A differentiation, which secrete IL-17A. In turn, IL-17A works in conjunction with TNF-α to markedly increase MMP expression and suppress tissue inhibitors of metalloproteinases (TIMPs). This imbalance between MMPs and TIMPs accelerates collagen breakdown and stromal ectasia (16,17).

In addition to cytokine signaling, immune cells—including neutrophils, natural killer (NK) cells, and γδ T cells—contribute to the inflammatory state. Neutrophils release MMP-9 and ROS, further kindling inflammation and oxidative stress that mediate tissue damage. The combined effects of cytokine-driven signaling, immune cell activity, and oxidative stress result in progressive keratocyte loss and ECM breakdown (16,17).

Hormonal influences, particularly involving the hypothalamic-pituitary-adrenal axis, have also been implicated in KC pathogenesis. Corneal cells express receptors for glucocorticoids, estrogens, and androgens, suggesting a hormonal responsiveness. Dysregulation of these pathways may influence corneal homeostasis by modulating collagen synthesis, inflammatory signaling, and oxidative stress responses. For example, altered cortisol dynamics may affect keratocyte metabolism and immune activity, while estrogen is known to influence corneal biomechanics and MMP expression. However, the precise mechanisms and their contribution to disease progression remain incompletely understood and still warrant further investigation.

Genetic associations

The MIR184 gene is one of the rare variants that has been identified to be a potential risk factor for KC development and disease progression. Specifically, point mutations in the miR-184 (+57 C>T) region was found to be involved in familial KC development, early-onset development of anterior polar cataracts, and Fuchs endothelial corneal dystrophy (2). Although this variant is incredibly rare, with reported frequencies below 1% worldwide, it supports a mechanistic link between disruptions in microRNA regulation and corneal pathologies (2,19). In contrast, other research supports the polygenic background of KC development, that pathogenesis may be more commonly influenced by multiple small-effect variants to collectively alter corneal biomechanical makeup (20-22).

LOX is one of the better-replicated candidate genes suspected to contribute to KC, which is responsible for encoding an enzyme responsible collagen and elastin cross-linking (Figures 2,3) (23,24). Specifically, variants such as rs2956540 and rs10519694 have been associated with KC susceptibility in certain populations, suggesting the pathogenic mechanism of impaired ECM stability and reduced corneal tensile strength contributing to overall pathogenesis (3,22-25). These variants are common polymorphisms in human populations, with minor allele frequencies (MAFs) ranging from 0.30–0.45 for rs2956540 (G>C) and around 0.2 for rs10519694 (C>T) across studied populations (24). Similarly, other polymorphisms in COL5A1, identified as a common genetic variant locus, have been associated with both central corneal thickness (CCT) aberrancies and KC risk. Both identifiable risk factors support the notion that the ECM and corneal stroma play an integral role in identifying those at high risk for KC development and progression (Figures 2,3) (22). Other genes (e.g., ZNF469) previously identified to play a role in other corneal pathologies, such brittle corneal syndrome, may also contribute to KC pathogenesis due to intrinsic ECM stromal instability, however, more direct research is needed to confirm the relationship (24,25).

Figure 3 Visual outlining suspected genetic association in KC and their associated pathophysiological effects. ↑, upregulated; ↓, downregulated. 25-HD, 25-hydroxyvitamin D; DHEA, dehydroepiandrosterone; IgE, immunoglobulin E; IL-1, interleukin-1; KC, keratoconus; MMP-2, matrix metalloproteinase-2; mRNA, messenger RNA; ROS, reactive oxygen species; SOD, superoxide dismutase; TNF, tumor necrosis factor.

Important to note, VSX1 was historically believed to be a potential causative gene for KC, but limited clear evidence has been uncovered to support clear contribution (Figure 2) (19,24,25). In addition, other candidate genes involved with systemic oxidative stress, wound healing, and protease regulation—including SOD1, TGFB1, etc.—may be promising future avenues but continue to have variable replication across studies, emphasizing the genetic heterogeneity of KC (2,24).

Genome-wide association studies (GWAS) also support the polygenic nature of KC pathogenesis. These studies have identified certain common loci—FOXO1, FNDC3B, and RAB3GAP1—that may influence the development of KC directly, or through endophenotypes such as CCT and corneal resistance factors (2,3,24,25). Many of these genes that have been identified directly overlap with other loci known to regulate corneal structural integrity and ECM biomechanics. Recently, polygenic risk scores have been developed to combine variants as listed above with CCT, corneal ECM biomechanics, and KC itself to provide a potential avenue for improving clinical risk stratification and aid with early KC detection in certain genetically susceptible individuals (Figure 2) (2,19).

Thus, the emerging understanding is that KC is due to dysregulation of biological processes related to collagen cross-linking, ECM remodeling, and intrinsic biomechanical integrity. The current understanding of disease pathology aligns with the current therapeutic strategies utilized in practice, including corneal collagen cross-linking which acts to biomechanically strengthen the cornea (3). Currently, genetic testing is not routinely recommended for most sporadic KC cases given the likely polygenic nature of the disease. Occasionally in familial KC presentations, testing for MIR184 may be of some benefit, but clinical application varies institution to intuition (2,3,19).

Association to ECM dysregulation

In combination with genetic predisposition, ECM dysregulation is a central component underlying the pathogenesis of KC and is secondary to several interrelated molecular and structural abnormalities. Intrinsic collagen abnormalities, impaired collagen cross-linking, aberrant proteoglycan and glycoprotein functionality, MMP upregulation, and suppression of ECM gene expression have all been identified to potentially contribute in varying levels to the pathogenesis.

In KC eyes, there is a global reduction in the quantity and organization of collagen types 1, 3, 5, and 6 collagen within the corneal stroma (11). The overall reduction results in decreased intrinsic biomechanical strength and increased susceptibility to ectatic corneal changes. Collagen fibrils in KC are also irregular in diameter and spacing; with degeneration of microfibrils and aberrant proteoglycan attachment within the central and peripheral corneal stroma (1,11).

Stromal integrity is further compromised due to dysregulated activity of LOX and related enzymes. Dysregulation results in weakened covalent bonds between the collagen and elastin fibers within the corneal stroma. Expression and position of proteoglycans (including lumican, keratocan, and decorin) and glycoproteins [including fibronectin and cartilage oligomeric matrix protein (COMP)] further disrupt the normal interfibrillar matrix and enhance corneal biomechanical instability (5,11,18,20,26). MMP upregulation, especially MMP-1, MMP-3, and MMP-9, drives excessive degradation of collagen and other ECM exacerbated (4,6,18,22,27,28).

Transcriptomic and proteomic studies also have demonstrated downregulation of genes and proteins involved in maintained of the ECM organization also underlie KC. Downregulation of COL1A1, PCOLCE, ADAMTS2, and BMP1 contribute to intrinsic ECM dysregulation, and contribute to the subsequent protein synthesis of structural proteins by keratocytes due to stress responses (i.e., p-eif2 and atf4) (6,20,29). Pathologic crosstalk between LGALS9⁺ epithelial cells and COMP⁺ stromal cells via CD44/CD45 and thrombospondin signaling results in shifting to a pro-fibrotic and pro-inflammatory ECM remodeling (7,27).

New research has emphasized the potential to test for and monitor ECM-related molecular markers in hopes to detect early KC, stratify risk of disease severity and progression, and monitor progression with greater sensitivity than traditional tomographic or visual metrics (30). LOX, MMP-9, and procollagen C proteinase enhancers (PCPE-1 and PCPE-2)—may reflect the important balance of collagen synthesis and degradation within the cornea in eyes with KC (30,31). In addition, low LOX expression has been linked to poor outcomes following corneal cross-linking. This emphasizes that measurement of LOX activity may be of incredible prognostic utility, as indicators of impaired ECM cross-linking functionality early in disease may provide a route to stratify disease severity (30,32).

Oppositely, elevations in levels of MMP-9—potentially measured in the corneal epithelium, cone apex, or tears—are indicative of inflammatory activity and intrinsic ECM degradation (28,33).Supporting it may act as a sensitive indicator of impending inflammatory conical KC progression. Similarly, reductions in PCPE-1 and PCPE-2 may additionally correlate with progressing disease, representative of impaired collagen and ECM remodeling (28,33,34).

Thus, integration of ECM biomarkers into clinical practice may help to improve risk stratification strategies and individualize clinical management. For example, utilization of standardized immunoassays, proteomic tear analysis and tissue level profiling can complement current keratometry and pachymetry imaging; in hopes to improve identification of early clinical progression to refer for collagen-crosslinking interventions.

AI and ML applied to KC

Recent radiomics and AI research has primarily emphasized imaging-based feature extraction, with limited incorporation of underlying biological data (35,36). This limitation is most relevant in the context of the preceding sections on genetic associations and ECM alterations, as current imaging-driven models often fail to account for the molecular mechanisms that underpin KC pathogenesis. This gap highlights an important opportunity: combining radiomic signatures with genetic and molecular datasets may improve biological interpretability and enhance the clinical utility of predictive models. Moving beyond prediction toward biologically informed AI has the potential to uncover mechanistic relationships and advance the precision of findings in this domain.

ML has been applied to corneal topography and biomechanical data for KC detection, spanning a variety of algorithms (36-38).For example, prior studies have reported that random forest models trained on Pentacam-derived parameters achieved a high diagnostic accuracy, with area under the receiver operating characteristic curve (AUC) values frequently exceeding 0.90. Thus, enabling discrimination between KC and normal corneas in young adults (38). These models incorporate data from corneal topography, pachymetry, corneal density, and anterior chamber dynamics (37,38).However, many of these studies rely on retrospective data sets derived from single centers, and external validation across diverse populations and imaging platforms remains limited, thus potentially limiting external generalizability (37,38).

Moreover, KeratoEL, a novel ensemble-based ML model, was developed for KC detection using SS-1000 CASIA OCT images and Electronic Health Record (EHR)-derived parameters. This model demonstrated improved classification performance compared to individual algorithms, highlighting the potential of different approaches. Nonetheless, like other models in this domain, its validation has largely been confined to internal or single-cohort data sets. In parallel, genetic studies have identified multiple loci that have been associated with susceptibility to KC, with 15 genes appearing multiple times across independent analyses, emphasizing the biological heterogeneity of the disease (39). However, integration of genetic markers in ML platforms remains uncommon, limiting the ability of current models to link imaging phenotypes with underlying molecular drivers discussed in earlier sections. While recent ML models have clearly shown predictive success, systematic integration of multi-modal biological and genetic information remains underexplored. Future work may combine corneal topography, biomechanical data, and patient-specific genetic profiles within a single interpretable ML framework to enhance predictive accuracy and provide mechanistic insights into KC pathogenesis. Together, these findings highlight a promising avenue for integrating multi-modal biological data with imaging-derived features in AI models, potentially enabling earlier diagnosis, deeper mechanistic understanding, and more personalized management of KC.

While recent ML models have clearly shown predictive success, systematic integration of multi-modal biological and genetic information remains underexplored. Importantly, multiple barriers hinder clinical implementation of these technologies, including the variability in imaging devices (i.e., different software platforms), lack of standardized input features, cost considerations, and challenges in integrating AI systems into clinical practice. In addition, regulation and prospective, multi-center validation studies are still needed prior to widespread adoption. Future work may combine corneal topography, biomechanical data, and patient-specific genetic profiles within a single interpretable ML framework to enhance predictive accuracy and provide mechanistic insights into KC pathogenesis. Such integrative approaches would directly bridge the molecular and ECM alterations, as described above, with imaging phenotypes. This would allow for a better understanding of the biological nature of the disease. Together, these findings highlight a promising avenue for integrating multi-modal biological data with imaging-derived features in AI models, potentially enabling earlier diagnosis, deeper mechanistic understanding, and more personalized management of KC.

Therapeutic landscape and emerging strategies

KC management has undergone significant changes from historical management and now focuses on surgical interventions aimed at directly halting progression to minimize visual decline. Despite surgical management options, glasses and contact lenses remain in the first line for visual optimization but cannot prevent progression. CXL is the current gold standard to prevent KC progression, with the Dresden epithelium-off protocol demonstrating the most evidence to stabilize keratometry to improve visual outcomes (40,41).However, some data have supported less corneal flattening as a result.

Like epi-off CXL, recent efforts have focused on transepithelial (epi-on) CXL techniques. Epi-on CXL is advantageous as it avoids epithelial removal prior to cross-linking, which decreases infection risk and post-operative pain. Historically, epi-on CXL has provided less biomechanical stiffening due to the difficulty of riboflavin and oxygen to diffuse with intact epithelium. However, with recent advances in iontophoresis-assisted delivery techniques, oxygen supplementation, and pulsed light CXL protocols, the stromal penetration has improved, narrowing the gap between epi-on and epi-off CXL outcomes (42,43).

Adjunctive and surgical options are also available in selected cases to CXL. Topographic guided photorefractive keratectomy (PRK) used in conjunction to CXL (i.e., Athen’s protocol) may improve corneal regularity and visual outcomes in mild KC. ICRSs may also be useful to provide biomechanical and visual benefits in patients who remain intolerant to contact lenses. In advanced KC, Bowman’s layer transplantation may be utilized as a minimally invasive option prior to deep anterior lamellar keratoplasty (DALK) (43). Overall, current management techniques are continuing to evolve rapidly, but continue to emphasize personalized, stage-specific care.

Strengths and limitations

This is a comprehensive review of KC, integrating aspects including KC pathophysiology, genetic susceptibility, ECM remodeling, AI applications, and emerging therapeutic approaches. The manuscript utilizes a broad range of recent and relevant literature to synthesize a coherent overview of the molecular, genetic, and biomechanic mechanisms that underlie KC pathophysiology. Both rare and high-impact genetic mutations and common susceptibility loci are mentioned to provide a more balanced overview of the polygenic architecture underlying KC, highlighting potential targets for future investigation of genetic pathways and ECM-related biomarkers. Additionally, this review ties molecular mechanisms to emerging state-of-the-art imaging modalities and AI-based diagnostic tools, emphasizing a translational perspective to bridge ongoing discovery to clinical practice.

This review is limited by a reliance on previously published literature, which restricts sample size, demographics, and methodologies of data collection, introducing some variability and uncertainty to conclusions drawn within the paper. Some genetic association studies did not consistently replicate results across populations, limiting the ability to definitively associate or eliminate certain loci as risk factors for KC development. Heterogeneity within KC diagnostic techniques may also limit comparability of findings. Lastly, clinical application and generalizability of AI and ML diagnostic techniques remain largely unexplored, limiting access to current data for validation of these methods.

Conclusions

KC is a multifactorial corneal ectatic disorder driven by the combination of genetic, biomechanical, and environmental factors leading to progressive corneal stroma thinning and biomechanical aberration. Current literature supports dysregulation of ECM remodeling, abnormal response to oxidative stress, and increased inflammatory signaling all contribute to KC disease initiation and conical progression. Genetic studies highlight the polygenic and multifactorial nature of KC, with variants in LOX, MIR184, COL5A1, RAB3GAP1 likely contributing to abnormal collagen cross linking and ECM structural integrity.

Advances in corneal imaging techniques—including Scheimplug tomography, anterior segment OCT, and biomechanical evaluations—have all helped to improve the detection of subclinical KC disease and helped to refine disease monitoring. Emerging molecular biomarkers—LOX, MMP-9, and PCPE—alongside ML-based analytical frameworks may help to enhance early diagnosis; risk stratify patients based on stage at diagnosis and individualize patient management.

CXL remains the gold standard for halting KC progression. However, innovations continue to expand treatment options, with transepithelial approaches, topography-guided PRK, and intracorneal ring implantations.

Integration of genomic, proteomic, and computational approaches will continue to redefine KC management and diagnostics, fostering a shift toward biologically informed, precision-based patient to patient-centered interventions that will improve visual outcomes.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the Narrative Review reporting checklist. Available at https://aes.amegroups.com/article/view/10.21037/aes-2026-0020/rc

Peer Review File: Available at https://aes.amegroups.com/article/view/10.21037/aes-2026-0020/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://aes.amegroups.com/article/view/10.21037/aes-2026-0020/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Alkanaan A, Barsotti R, Kirat O, et al. Collagen fibrils and proteoglycans of peripheral and central stroma of the keratoconus cornea - Ultrastructure and 3D transmission electron tomography. Sci Rep 2019;9:19963. [Crossref] [PubMed]
2. Bykhovskaya Y, Margines B, Rabinowitz YS. Genetics in Keratoconus: where are we?. Eye Vis (Lond) 2016;3:16. [Crossref] [PubMed]
3. Cerván-Martín M, Higueras-Serrano I, González-Muñoz S, et al. Comprehensive Evaluation of the Genetic Basis of Keratoconus: New Perspectives for Clinical Translation. Invest Ophthalmol Vis Sci 2024;65:32. [Crossref] [PubMed]
4. Chen Y, Hughes G, Chen X, et al. Genetic Variants Associated With Different Risks for High Tension Glaucoma and Normal Tension Glaucoma in a Chinese Population. Invest Ophthalmol Vis Sci 2015;56:2595-600. [Crossref] [PubMed]
5. Cheng AMS, Gupta SK, Steiner T, et al. Repeat Intravitreal Bevacizumab in a Very Low Birth Weight Infant with Recurrent ROP: 3-Year Follow-Up and Literature Review. Case Rep Ophthalmol 2025;16:521-7. [Crossref] [PubMed]
6. Cheng Z, Hao J, Cai S, et al. A novel combined oxidative stress and extracellular matrix related predictive gene signature for keratoconus. Biochem Biophys Res Commun 2025;742:151144. [Crossref] [PubMed]
7. Singh RB, Koh S, Sharma N, et al. Keratoconus. Nat Rev Dis Primers 2024;10:81. [Crossref] [PubMed]
8. Piñero DP, Nieto JC, Lopez-Miguel A. Characterization of corneal structure in keratoconus. J Cataract Refract Surg 2012;38:2167-83. [Crossref] [PubMed]
9. Hayat UK, Shiwani HA, Memon D, et al. Intracorneal Ring Segments in Keratoconus: A Narrative Literature Review. Clin Ophthalmol 2025;19:2249-57. [Crossref] [PubMed]
10. Vinciguerra R, Elsheikh A, Roberts CJ, et al. Influence of Pachymetry and Intraocular Pressure on Dynamic Corneal Response Parameters in Healthy Patients. J Refract Surg 2016;32:550-61. [Crossref] [PubMed]
11. Wollensak G, Spoerl E, Seiler T. Riboflavin/ultraviolet-a-induced collagen crosslinking for the treatment of keratoconus. Am J Ophthalmol 2003;135:620-7. [Crossref] [PubMed]
12. Godefrooij DA, de Wit GA, Uiterwaal CS, et al. Age-specific Incidence and Prevalence of Keratoconus: A Nationwide Registration Study. Am J Ophthalmol 2017;175:169-72. [Crossref] [PubMed]
13. Niazi S, Gatzioufas Z, Doroodgar F, et al. Keratoconus: exploring fundamentals and future perspectives - a comprehensive systematic review. Ther Adv Ophthalmol 2024;16:25158414241232258. [Crossref] [PubMed]
14. Niazi S, Jiménez-García M, Findl O, et al. Keratoconus Diagnosis: From Fundamentals to Artificial Intelligence: A Systematic Narrative Review. Diagnostics (Basel) 2023;13:2715. [Crossref] [PubMed]
15. Shetty R, Rajiv Kumar N, Pahuja N, et al. Outcomes of Corneal Cross-Linking Correlate With Cone-Specific Lysyl Oxidase Expression in Patients With Keratoconus. Cornea 2018;37:369-74. [Crossref] [PubMed]
16. Wojcik KA, Blasiak J, Szaflik J, et al. Role of biochemical factors in the pathogenesis of keratoconus. Acta Biochim Pol 2014;61:55-62. [PubMed]
17. Monteiro de Barros MR, Chakravarti S. Pathogenesis of keratoconus: NRF2-antioxidant, extracellular matrix and cellular dysfunctions. Exp Eye Res 2022;219:109062. [Crossref] [PubMed]
18. Vought R, Greenstein SA, Gelles J, et al. The Pathophysiology of Keratoconus. Cornea 2025;44:137-43. [Crossref] [PubMed]
19. Sahebjada S, Schache M, Richardson AJ, et al. Evaluating the association between keratoconus and the corneal thickness genes in an independent Australian population. Invest Ophthalmol Vis Sci 2013;54:8224-8. [Crossref] [PubMed]
20. Foster JW, Shinde V, Soiberman US, et al. Integrated Stress Response and Decreased ECM in Cultured Stromal Cells From Keratoconus Corneas. Invest Ophthalmol Vis Sci 2018;59:2977-86. [Crossref] [PubMed]
21. Habroosh F, Filippini P, Al Saadi A, et al. Expression of extracellular matrix protein in advanced keratoconus and the normal human cornea. Front Med (Lausanne) 2025;12:1612452. [Crossref] [PubMed]
22. Hughes AE, Bradley DT, Campbell M, et al. Mutation altering the miR-184 seed region causes familial keratoconus with cataract. Am J Hum Genet 2011;89:628-33. [Crossref] [PubMed]
23. He W, Han X, Ong JS, et al. Association of Novel Loci With Keratoconus Susceptibility in a Multitrait Genome-Wide Association Study of the UK Biobank Database and Canadian Longitudinal Study on Aging. JAMA Ophthalmol 2022;140:568-76. [Crossref] [PubMed]
24. Zhang J, Zhang L, Hong J, et al. Association of Common Variants in LOX with Keratoconus: A Meta-Analysis. PLoS One 2015;10:e0145815. [Crossref] [PubMed]
25. Loukovitis E, Sfakianakis K, Syrmakesi P, et al. Genetic Aspects of Keratoconus: A Literature Review Exploring Potential Genetic Contributions and Possible Genetic Relationships with Comorbidities. Ophthalmol Ther 2018;7:263-92. [Crossref] [PubMed]
26. Han T, Yu Y, Wang Y, et al. Single-cell RNA sequencing reveals LGALS9+ epithelial and COMP+ stromal cell crosstalk driving keratoconus pathogenesis. Int J Biol Macromol 2025;329:147969. [Crossref] [PubMed]
27. Kannan R, Shetty R, Panigrahi T, et al. Untargeted Tear Proteomics in a Large South-Asian Cohort Reveals Inflammatory Signaling, ECM Remodeling, and Altered Metabolism in Keratoconus. Invest Ophthalmol Vis Sci 2025;66:60. [Crossref] [PubMed]
28. di Martino E, Ali M, Inglehearn CF. Matrix metalloproteinases in keratoconus - Too much of a good thing?. Exp Eye Res 2019;182:137-43. [Crossref] [PubMed]
29. Du K, Peng R, Xiao G, et al. Transcriptomic profiling reveals neural-immune-stromal dysregulation and risk-associated gene signatures in advanced keratoconus. Sci Rep 2025;15:37262. [Crossref] [PubMed]
30. Dudakova L, Jirsova K. The impairment of lysyl oxidase in keratoconus and in keratoconus-associated disorders. J Neural Transm (Vienna) 2013;120:977-82. [Crossref] [PubMed]
31. Pahuja N, Kumar NR, Shroff R, et al. Differential Molecular Expression of Extracellular Matrix and Inflammatory Genes at the Corneal Cone Apex Drives Focal Weakening in Keratoconus. Invest Ophthalmol Vis Sci 2016;57:5372-82. [Crossref] [PubMed]
32. Shetty R, Sathyanarayanamoorthy A, Ramachandra RA, et al. Attenuation of lysyl oxidase and collagen gene expression in keratoconus patient corneal epithelium corresponds to disease severity. Mol Vis 2015;21:12-25. [PubMed]
33. Song T, Song J, Li J, et al. The candidate proteins associated with keratoconus: A meta-analysis and bioinformatic analysis. PLoS One 2024;19:e0299739. [Crossref] [PubMed]
34. Sağlam ZG, Yildirim M, Laloğlu E. Evaluating serum biomarkers in keratoconus: the role of PCPE-1 and PCPE-2 in collagen metabolism and disease progression. Graefes Arch Clin Exp Ophthalmol 2025;263:2841-8. [Crossref] [PubMed]
35. Li G, Feng P, Lin Y, et al. Integrating radiomics, artificial intelligence, and molecular signatures in bone and soft tissue tumors: advances in diagnosis and prognostication. Front Oncol 2025;15:1613133. [Crossref] [PubMed]
36. Paul PK, Hossan A, Ullah SMA. KeratoEL: Detection of keratoconus using corneal parameters with ensemble learning. Health Sci Rep 2024;7:e2202. [Crossref] [PubMed]
37. Muhsin ZJ, Qahwaji R, Ghafir I, et al. Advances in machine learning for keratoconus diagnosis. Int Ophthalmol 2025;45:128. [Crossref] [PubMed]
38. Syta A, Podkowiński A, Chorągiewicz T, et al. Machine learning-assisted early detection of keratoconus: a comparative analysis of corneal topography and biomechanical data. Sci Rep 2025;15:24399. [Crossref] [PubMed]
39. Chang SH, Yeh LK, Hung KH, et al. Machine Learning-Driven Transcriptome Analysis of Keratoconus for Predictive Biomarker Identification. Biomedicines 2025;13:1032. [Crossref] [PubMed]
40. Ferdi A, Nguyen V, Kandel H, et al. Predictors of progression in untreated keratoconus: a Save Sight Keratoconus Registry study. Br J Ophthalmol 2022;106:1206-11. [Crossref] [PubMed]
41. Ferdi AC, Nguyen V, Gore DM, et al. Keratoconus Natural Progression: A Systematic Review and Meta-analysis of 11 529 Eyes. Ophthalmology 2019;126:935-45. [Crossref] [PubMed]
42. Cifariello F, Minicucci M, Di Renzo F, et al. Epi-Off versus Epi-On Corneal Collagen Cross-Linking in Keratoconus Patients: A Comparative Study through 2-Year Follow-Up. J Ophthalmol 2018;2018:4947983. [Crossref] [PubMed]
43. Pellegrini M, Yu AC, Busin M. Deep anterior lamellar keratoplasty for keratoconus: Elements for success. Saudi J Ophthalmol 2022;36:36-41. [Crossref] [PubMed]