Corneal collagen crosslinking for the treatment of progressive keratoconus

Meegan Vandepeer, Deanne Forel, JonHenry Jacobsen, Akwasi Ampofo, Ning Ma, Thomas Vreugdenburg
Record ID 32018001203
Authors' objectives: This report investigates the clinical efficacy, safety, costs and cost-utility associated with corneal collagen crosslinking (CXL) for treating progressive keratoconus. In addition, legal, social, ethical and organisational issues associated with the use of CXL are explored.
Authors' results and conclusions: Clinical evaluation Systematic literature searches were conducted in four biomedical databases (PubMed, Embase, Cochrane Library, CINAHL) to identify existing systematic review (SR) and randomised controlled trial (RCT) evidence addressing the research questions. The results in this executive summary are presented at longest follow-up due to the slow-progressing nature of the disease, and the fact that CXL aims to slow disease progression rather than improve symptoms. CXL compared to sham or no treatment The clinical efficacy of CXL compared to sham or no treatment was informed by RCTs. The safety of CXL was informed by 4 RCTs plus a SR of lower levels of evidence. There are risk of bias concerns in the RCTs relevant to both safety and efficacy related to incomplete data, randomisation, conflicts of interest and between-eye correlation. The systematic review of CXL safety was deemed to be of very low quality. The overall strength of evidence for the clinical efficacy outcomes was very low. Compared to sham or no treatment, a significant difference was observed for uncorrected visual acuity (UCVA) (SMD -0.71, 95% CI -1.28, -0.14) and Kmax (the maximum curvature in the central cornea) (MD -1.85 D, 95% CI -3.44, -0.25) in favour of CXL at longest follow-up (36 months). The weighted mean absolute changes indicate that Kmax decreased over time relative to baseline for CXL, but slightly increased in the sham and no treatment group. Meta-analyses of best-corrected visual acuity (BCVA), Kmin (the minimum curvature in the central cornea), Kmean (the average curvature in the central cornea), cylinder (the amount of astigmatism) and spherical equivalent (an estimate of the eyes refractive error) showed no significant difference between CXL and sham or no treatment at longest follow-up. One RCT reported on corneal transplant rates with none reported in the CXL arm (compared with 10% in the control arm), no RCT reported on quality of life. Adverse events that can occur following CXL include keratitis, haze, sterile infiltrates, stromal oedema, golden striae, corneal scars, photophobia, increased lacrimation, dry eye, ocular irritation, blurred vision, ocular pain, corneal erosions, epithelial defects and corneal vascularisation. Haze, whilst generally only temporary, appears to be the most commonly reported event following CXL, with incidence rates of up to 100%; however, as safety was poorly reported no comment can be made as to how common these events are with any level of certainty. In general, there is uncertainty with the evidence owing to the small size of many of the included studies, their variable definition of keratoconus progression – used as inclusion criteria in the trials – and the lack of validated parameters for measuring keratoconus progression. All of which make it difficult to draw conclusions to any degree of certainty. Costs and cost-effectiveness A Markov model was constructed to evaluate the comparative cost-utility of CXL over no treatment. Deterministic and probabilistic sensitivity analyses were used to quantify the impact of parameter uncertainties and to elicit key drivers of the model. The economic model was based on an existing model from the UK by Salmon (2015) The UK model was adapted to best reflect appropriate clinical findings and Swiss population; however, some of the input values and clinical assumptions from the Salmon model were used directly in the current evaluation due the lack of available information in the clinical evidence base. Over a 25-year time horizon, the base-case ICER was CHF 25,841 per QALY gained. Sensitivity analyses suggest the ICER is most sensitive to the efficacy of CXL, which was represented by the rate of CXL failure requiring repeat treatment. It is also sensitive to the utility difference between early disease status (AK stage 1) and a more advance status (AK stage 2). Most ICERs produced from all scenarios were below the hypothetical ICER threshold of CHF 100,000. Probabilistic sensitivity analyses showed 99.4% chance of CXL being cost-effective against the hypothetical threshold. A budget impact analysis was undertaken to estimate the financial impact of publicly reimbursing CXL. In the base-case analysis, it was assumed that only a proportion (50%) of newly diagnosed keratoconus patients would receive CXL after reimbursement becomes available. It was estimated that CXL would cost CHF 573,346 in the first year (2021) and increase to CHF 584,865 (2025) as the total cost in Switzerland for all patients. However, when considering existing diagnosed patients who have not been able to receive CXL, the worst-case scenario would have the CXL costing over CHF 4.5 million in Switzerland in the first year (2021). This should be considered the ceiling of the financial impact subject to significant overestimation. The cost will taper down to the base-case level after 2 to 3 years of initial treatment backlog. When the initial patient loadings are cleared, the fourth and the fifth-year projections are in line with the base-case. Legal, social, ethical and organisational issues Children with keratoconus represent a vulnerable patient group, as the disease progresses faster, they are more likely to progress to advanced disease stages compared to adults, and have difficulties adhering to conservative treatments (such as spectacles and contact lenses); CXL may, from an ethical/social perspective, be more beneficial in this cohort. It is unclear whether utilisation of healthcare resources (e.g. related to corneal transplants), will change if CXL is reimbursed. Conclusion CXL appears to have a beneficial effect on slowing keratoconus progression with respect to UCVA and Kmax. A range of adverse events can occur following CXL based on moderate to low quality evidence. The most frequently reported adverse event was temporary corneal haze with incidences up to 100%; however, these results are subject to a high degree of uncertainty. CXL is likely to be a cost-effective treatment for keratoconus; however, there are still uncertainties regarding the applicability of the economic evaluation result to the Swiss context due to the use of inputs in the model that were not specific to the Swiss population.
Project Status: Completed
Year Published: 2021
English language abstract: An English language summary is available
Publication Type: Mini HTA
Country: Switzerland
MeSH Terms
  • Keratoconus
  • Collagen
  • Cross-Linking Reagents
  • Photosensitizing Agents
  • Riboflavin
  • corneal collagen crosslinking
  • crosslinking
  • keratoconus
  • PROMs
  • efficacy
  • effectiveness
  • safety
  • costs
  • economics
  • cost-effectiveness
  • budget impact
  • legal
  • social
  • ethical
  • organisational
  • ophthalmology
  • eye
Organisation Name: Swiss Federal Office of Public Health (FOPH)
Contact Address: Federal Office of Public Health, Schwarzenburgstrasse 157, CH-3003 Berne, Switzerland
Contact Name: Stephanie Vollenweider
Contact Email:
Copyright: Swiss Federal Office of Public Health
This is a bibliographic record of a published health technology assessment from a member of INAHTA or other HTA producer. No evaluation of the quality of this assessment has been made for the HTA database.