Unlocking precision: How corneal cell area analysis revolutionises post-transplant stem cell monitoring

Abstract

The corneal epithelium is maintained by limbal stem cells (LSCs). Dysfunction of the LSCs, resulting from chemical and thermal burns, contact lens-related disease, congenial disorders, among other conditions, leads to limbal stem cell deficiency (LSCD), a sight-threatening condition. An effective treatment of LSCD, with 76% of patients reporting regained sight up to 24 months after the operation, consists of transplanting ex-vivo cultured LSCs from the patients other healthy eye (i.e. autologous) or donor (i.e. allogeneic) to the affected eye. The post-operative assessment of corneal recovery is crucial but relies on ponderous and generally subjective visual inspection of a large number of microscopic images of the corneal epithelial cells, relying on the personal experience of the practitioner to interpret imprecise, qualitative diagnostic criteria. From a unique library of 100,000 cornea cell images from 34 patients, we have randomly selected 10 individuals (3,668 images) to demonstrate that the frequency distribution of the epithelial cell areas is a sensitive diagnostic tool of the corneal epithelium status. After a successful operation the distribution of cell areas is rather flat, reflecting an anomalously wide range of cell areas. As the cornea recovers, the frequency distribution becomes narrower with high statistical confidence and eventually approaches that of the healthy cornea. The corneal epithelial cell shape is independent of the cornea status despite a widespread expectation that healthy cells have a hexagonal shape. We also show that the corneal epithelial cell area distribution and its variation with the depth within the cornea are specific to each patient. Significance StatementChemical and thermal cornea burns, contact-lens damage and hereditary factors, among other conditions, cause limbal stem cell deficiency (LSCD), a widespread sight-threatening condition. An efficient LSCD treatment involves a stem cell transplant from the patients other healthy eye, in unilateral cases, or a donor, in bilateral cases. Traditional post-operative cornea monitoring is laborious and often subjective as it relies on visual inspection of microscopy corneal epithelial images. We show that the distribution of epithelial cell areas is a sensitive LSCD diagnostic, evolving systematically to a healthy form after a successful treatment. We have developed computer algorithms to implement this quantitative, sensitive and precise approach which can radically improve the quality of both cornea monitoring in disease and response to treatment.