First In-Office Screening for Diabetic Retinopathy Cleared by FDA

The ability to assess eye health in patients with diabetes can begin with an in-office screening using artificial intelligence software, accelerating referral to an ophthalmologist for earlier treatment.

Written by Kathleen Doheny

Interviews with Michael Abramoff, MD, Elena Christofides, MD, FACE, and Lawrence Singerman, MD, FACS

A new, artificial intelligence-based screening software for diabetic retinopathy will make a quick assessment for diabetic retinopathy possible by support staff in endocrinologists' and primary care providers' offices.

The  Food and Drug Administration cleared this first of its kind screening device for diabetic retinopathy—IDx-DR— following an expedited review under its Breakthrough Device Program. The IDx-DR is currently the only autonomous, diagnostic system to employ artificial intelligence (AI) intended for the front lines of health care, according to the company.

“While 30% or more of those with diabetes have some degree of diabetic retinopathy,2 few receive the recommended regular eye exams,” Dr. Abramoff told EndocrineWeb. Assessing for diabetic retinopathy has been hampered by limits on visit time since patients with diabetes seek care for more urgent medical concerns, and no easy way to assess changes in vision during a visit to the PCP or endocrinologist.

Potential to Catch Diabetic Retinopathy Much Earlier

“The new test takes only a few minutes, is about 90% accurate, and produces results in a matter of minutes,” filling a current void at the primary level care, Dr. Abramoff said, and “takes up about the space of a closet.”

Should the company gain approval to market IDx-DR, “this device will fill a crucial need, identifying diabetic retinopathy early when it is most treatable, preventing vision loss,” said Michael Abramoff, MD, the Robert C. Watzke professor of ophthalmology and visual sciences at the University of Iowa, Iowa City. He is founder and president of IDrx, based in Iowa City. The company is focused on developing clinically-aligned autonomous algorithms that identify the condition in medical images.

Robust Clinical Study Offers Favorable Support for Approval

Dr. Abramoff presented results of his clinical trial of the new AI test at the Macular Society meeting in Beverly Hills, California.1

A multi-center, intent-to-screen design study was conducted at 10 primary care sites across the US to evaluate the accuracy of IDx-DR in detecting moderate to severe diabetic retinopathy, including diabetic macular edema (DME).

The system was evaluated in comparison to standardized imaging and grading protocols by the Wisconsin Fundus Photograph Reading Center.3 https://www.ophth.wisc.edu/research/fprc/ This grading included Early Treatment Diabetic Retinopathy Study Severity Scale (ETDRS) and DME determinations from wide-field stereoscopic photographs and macular Optical Coherence Tomography. More than mild retinopathy was defined as ETDRS level 35 or more, and/or DME.

In all, 900 men and women were enrolled in the study, with 819 people with diabetes completing a full evaluation by both the new system and the standardized protocol. Of that total, 198 (23.8%) had mild-to-moderate diabetic retinopathy; the AI system detected the disordered vision with a sensitivity of 87.2% (95% CI, 81.8-91.2%) and a specificity of 90.7% (95% CI, 88.3-92.7%).

The participants had either type 1 or type 2 diabetes, but the researchers did not separate them out by type. Participants were age 22 years and older. Some participants were newly diagnosed, while others had had diabetes for years.1

How the IDx-DR Assesses for Diabetic Retinopathy

A sophisticated retinal camera takes images of the eye; then the IDx-DR analyzes the findings captured by the camera for the presence of lesions and other features associated with diabetic retinopathy.

The results are run through an algorithm capable of interpreting the images for any signs of diabetic retinopathy. In four hours, office support staff may be trained to operate the system, Dr. Abramoff said, "[The device] guides you to take the images."

Depending on the AI test results, PCPs and endocrinologists will assess whether the patient would benefit from referral to an ophthalmologist.

Dr. Abramoff declined to discuss costs, citing the need to await the pending review by the FDA. However, he indicated that the system has been operational in Europe, having been used successfully on many patients. Also, other systems with similar capabilities are under evaluation.4

How Does the IDx-DR Play Among Clinicians?

Provided the IDx-DR gains clearance from the FDA, ''the device will really be a step forward," said Elena Christofides, MD, FACE, an endocrinologist in Columbus, Ohio, upon reviewing the findings of the clinical trial for EndocrineWeb.

The ability to screen patients in our own offices will be very helpful, she said, “especially in rural areas where ophthalmologists may be hours away. In some cases, it will prevent a long trip for a patient to get screened by an ophthalmologist only to find out no diabetic retinopathy is present.”

However, she told EndocrineWeb, “primary care physicians are already shouldering a lot of the burden for the care for those with diabetes.” While clinicians treating for diabetes are a good place to provide this machine, she suggests offering it to pharmacies, as well. Dr. Abramoff also speaks about that possibility in some press reports.

“It’s evident that the system has the capacity to identify diabetic retinopathy at certain levels that mandate seeing a retinal specialist,” said the founder of the Macula Society, Lawrence Singerman, MD, FACS, clinical professor of ophthalmology at Case Western Reserve University in Cleveland Ohio, and the Bascom Palmer Eye Institute in Miami. He attended the research presentation and also weighed in on the findings.

The device, if cleared, ''can help sort out those who have enough disease to send to a retinal specialist'' as well as identify those who don't need to be seen, saving these patients both cost and time," he told EndocrineWeb.

There are no relevant financial disclosures by any of these physicians.

Sources

  1. Abramoff M. A multicenter study to evaluate performance of an automated device for the detection of diabetic retinopathy. Presented at: Macular Society Meeting February 21-24, 2018 in Los Angeles, California.
  2. American Diabetes Association: A Position Statement on Diabetic Retinopathy. Diabetes Care. 2017; 40(3):412-418.  
  3. University of Wisconsin. Fundus Photograph Reading Center. Available at: www.ophth.wisc.edu/research/fprc/. Accessed February 26, 2018.
  4. Ting DSW, Yim-Lui C, Lim G, et al. Development and validation of a deep learning system for diabetic retinopathy and related eye diseases using retinal images from multiethnic populations with diabetes. JAMA. 2017;318(22):2211-2223.

Continue Reading

Oral Antioxidant Supplementation for Diabetic Retinopathy