- Peripheral corneal thinning
- Endothelial diseases
- ‘Eye to AI: Transforming Microbial Keratitis Detection with Deep Learning’
Peripheral corneal thinning
Sanjay Kedhar, MD, discussed “Refining the Diagnostic Algorithm for Peripheral Corneal Thinning.” He said that this matters because thinning may signal sight-threatening disease, and there can be overlap among degenerative, ectatic, and ulcerative disorders. Early recognition helps prevent perforation.
He offered several steps for corneal thinning:
- Step 1: Confirm true thinning. Dr. Kedhar said to rule out “pseudo thinning.” Tools to determine thinning can be anything from cheap techniques like corneal light wedge or optical section, pachymetry, up to anterior segment OCT.
- Step 2: Evaluate the epithelium and inflammation. Dr. Kedhar said this is one of the more critical aspects. An intact epithelium indicates non-inflammatory/degeneration/ectasia; an epithelial defect indicates inflammatory or infectious; and pain and redness indicate inflammation.
- Step 3: Key diagnostic tests. Dr. Kedhar listed several key tests like anterior segment OCT, which can quantify thinning; topography, which can detect ectasia; sensation testing; cultures/PCR, which can test for infection; and autoimmune labs for a vasculitis workup.
Editors’ note: Dr. Kedhar has no relevant financial interests.
Endothelial diseases
Kathryn Colby, MD, PhD, presented “Update on Descemet Stripping Only: Trials and Future Directions.” DSO, she said, involves the removal of the central endothelium without replacement in Fuchs dystrophy patients. It is straightforward, though it requires a little patience, but is usually easily accomplished. She noted that her index patient is now 11 years out and still doing well.
Ripasudil 0.4% is used for DSO. This is a rho kinase inhibitor approved for glaucoma in Japan, which Dr. Colby said reduces IOP by increasing outflow facility. It promotes proliferation of human cultured corneal endothelial cells in vitro. It also increased endothelial healing/corneal clearance in two in vivo endothelial injury rabbit models (cryotherapy and mechanical scraping). However, Dr. Colby noted that this is not approved in the U.S.
She went on to mention the Phase 2 study of ripasudil after DSO, which was a randomized, placebo-controlled, multicenter, multinational trial, sponsored by Kowa. It enrolled 65 patients at 38 sites in five countries. The primary endpoint was endothelial cell density at 12 weeks read by an independent reading center.
From the study, ripasudil was shown to speed cornea clearance after DSO. It also reduced the DSO failure rate. In early studies, there was about a 25% failure rate. Dr. Colby said the failure rate is around 10% now. The Phase 2 trial showed, without supplemental ripasudil, about 27% failure; it was around 10% with ripasudil.
A Phase 3 study is now underway, Dr. Colby said. There are two studies, with 100 patients each, one of DSO with cataract and one of DSO without cataract (and QID ripasudil in both).
The treatment phase in the study with DSO and cataract was completed in March 2025. The second, without cataract, completed the treatment phase in June 2025. The FDA required a safety study as well, which was completed.
There has been much progress with DSO since the first modern day cases in 2014, Dr. Colby said. Ripasudil appears to improve the speed of healing and visual acuity recovery, while reducing failure rate.
Dr. Colby also mentioned some advanced imaging techniques that could potentially help understanding in this area. She particularly discussed imaging the peripheral endothelium with a scanning specular microscopy, though this is only approved for research (and not clinical use). It’s very difficult to image the peripheral cornea. Dr. Colby hopes that in the future, a heat map of endothelial density can be generated, as well as other parameters, similar to a pachymetric map with corneal topography.
During a panel discussion, Dr. Colby and W. Barry Lee, MD, commented on barriers to adoption of corneal cell injection therapy. Dr. Colby noted that, in the U.S., there are good technique options that work well, with known costs. For something to be adopted practically, the insurance companies need to be willing to cover it. Dr. Lee agreed that patients need to get technology and be able to afford it. Another question is if the visual acuity will come back as quickly as with DMEK. Even if it’s not as good as DMEK, the question may be if the safety profile will give it the advantage. The safety profile is vast compared to endothelial transplants, Dr. Lee said.
During his presentation, Dr. Lee mentioned some of the corneal endothelial cell therapies in development, notably from Aurion, Emmecell, and OcuCell. The Aurion technology has the most experience so far, with more than 200 procedures. Dr. Lee noted the experience and recent publication from Shigeru Kinoshita, MD, PhD, and colleagues in Japan, of 10-year data with corneal endothelial cell therapy with success in maintaining corneal transparency and a favorable safety profile. Dr. Lee also mentioned the CLARA Phase I/II study from the U.S./Canada, which featured five arms and showed no graft rejections and no serious ocular adverse events. There was a dose-dependent improvement in the three Aurion cell dose groups for the primary endpoint of proportion of responders with a ≥15-letter improvement from baseline in BCVA, with a statistically significant improvement in the high-dose group.
Corneal cell therapy has the potential to transform cornea care, he said, where one donor can fix 1,000 patients instead of one patient.
Editors’ note: Dr. Colby has no relevant financial interests. Dr. Lee has financial interests with Aurion and Emmecell.
‘Eye to AI: Transforming Microbial Keratitis Detection with Deep Learning’
Maria Woodward, MD, spoke about using AI for microbial keratitis, highlighting some of the gaps in care, possibilities with AI tools, and future directions.
She first noted that one issue with treatment for microbial keratitis now is that patients cannot afford current care. She cited patients asking which medications are most important to use if they had to prioritize just one or two of them, issues with transportation, and patients not understanding the seriousness or possibility of losing the eye due to not using drops or refrigerating them.
The other factor of this is the state of medicine, she said. There are high volumes of information, and it’s complex. Demands on time are unparalleled, and there is unequal and costly care.
Dr. Woodward went on to discuss diagnosis, noting that clinicians lack a reliable, rapid, and effective strategy to diagnose microbial keratitis. Clinical phenotypes have overlapping history, symptoms, and morphology features. Additionally, you often don’t get cultures back in a timely fashion. But Dr. Woodward said that tissue-based diagnostic innovations that are being developed are promising.
The bottom line is when we ask cornea doctors if they’re confident if they know the organism class, they’re not, Dr. Woodward said. She noted two parts to diagnosis. The first is if it’s infectious or not. AI algorithms have been developed to differentiate microbial keratitis from other anterior segment etiologies. The second part of diagnosis is what the organism class is. Many people struggle with this, she said, adding that there has been good progress for AI tools here as well, using photography or confocal to differentiate between the major classes. There are many promising tools in preliminary stages, but most have been tested in single centers and not across a wide geography of patients.
Dr. Woodward moved on to discuss prognosis. She said there is a complex spectrum of phenotypes. But a standard, precise quantification of microbial keratitis features is lacking, and she also noted the lack of severity scales. Most clinicians rely on experience to choose treatment intensity.
In terms of effectiveness of treatments in this area, Dr. Woodward said that we rely on memory and clinical notes to evaluate morphology and symptom changes. There are no quantified disease trajectories, and this impacts ongoing management.
Another issue is knowing clinical endpoints, and it’s not always clear when the patient is healed. There are not clear, quantified measures of clinical outcomes.
Looking forward to what’s needed in this space, Dr. Woodward said data sharing is important. There is a need for datasets and algorithms that are findable, accessible, interoperable, and reproducible. It will also be important to improve datasets to represent diverse populations, as well as testing efficacy and effectiveness of algorithms.
Editors’ note: Dr. Woodward has no relevant financial interests.
EyeWorld Onsite is a digital publication of the American Society of Cataract and Refractive Surgery (ASCRS).
For sponsorship opportunities or membership information, contact: ASCRS • 12587 Fair Lakes Circle • Suite 348 • Fairfax, VA 22033 • Phone: 703-591-2220 • Email: ascrs@ascrs.org
Opinions expressed in EyeWorld Onsite do not necessarily reflect those of ASCRS. Mention of products or services does not constitute an endorsement by ASCRS.
Click here to view our Legal Notice.
Copyright 2025. All rights reserved.
