DR News: Better Bone Care
By Beth W. Orenstein
Radiology Today
Vol. 26 No. 2 P. 5

About 10 million women (mostly) and men in America have osteoporosis—a disease of the bones that makes them weak and more likely to break, according to the Bone Health & Osteoporosis Foundation. Osteoporosis will cause one in two women and one in four men over the age of 50 to break a bone, possibly from a minor fall or even sneezing or bumping into furniture, the foundation says. Worldwide, every year, osteoporosis leads to 37 million fractures and costs health systems billions, according to a 2021 study in the Lancet. Many cases of osteoporosis are only diagnosed when it’s too late and a fracture has already occurred.

However, timely intervention can prevent bone loss that can lead to what are often life-limiting fractures in people with osteoporosis. “All efforts to diagnose and treat osteoporosis are important,” says Lori Fitton, PhD, ARNP, a nurse practitioner with University of Iowa Health Care (UIHC) Orthopedics and Rehabilitation in Iowa City.

A dual-energy X-ray absorptiometry (DEXA) scan is the gold standard for identifying or ruling out osteoporosis or its precursor, osteopenia. However, a national survey of physicians released in 2023 found that few believe screening for osteoporosis is a priority. The US Preventive Services Task Force recommends DEXA scans for women over 65 but has no recommendations for men. Medicare covers this test once every two years for women over 65 and men over 70, or more often if medically necessary, but many people don’t get the scans they need. Osteoporosis is considered a “silent disease” because most people don’t know they have it until they break a bone.

Discovering Density
Three-quarters of cases of osteoporosis are undiagnosed. To address the need for earlier diagnosis, UIHC and Naitive Technologies Ltd, based in London, England, have announced a joint research project to opportunistically identify people at risk of low bone mineral density using X-rays they may have had for other reasons. The study, which was launched last September, uses Naitive Technologies’ OsteoSight™, an FDA-designated Breakthrough Medical Device designed to opportunistically identify people at risk of low bone mineral density. OsteoSight was developed from an urgent clinical need to mitigate the incidence of fractures associated with osteoporosis, the company says in a press release announcing the research.

The IRB-approved retrospective study will review deidentified X-rays from up to 2,500 patients (men and women) aged 50 and over from Iowa.

Tens of millions of hip and pelvic X-rays are taken each year for a variety of reasons, often to investigate a concern or patient complaint such as chronic hip pain, says John Connell, PhD, a principal clinical scientist at Naitive. OsteoSight automatically reviews these routine X-rays for evidence of osteoporosis in the hip. If it detects signs of bone disease, the patient’s care team will be informed, and a plan for further evaluation and treatment can be devised.

Patients may have had X-rays of their hips or other bones after a fall or to look for arthritis. While the risk factors for osteoporosis are well known, diagnosis rates for osteoporosis remain poor, Connell says. “We can address the poor diagnosis rate through this collaborative study between Naitive and UIHC.”

OsteoSight uses a combination of imaging science, physics, and machine learning to extract information from X-rays that would otherwise not be possible, Connell says. It then integrates with existing workflows to bring this information to providers. OsteoSight, which received its FDA Breakthrough Device Designation in 2023, was developed by Will Briggs, MD, PhD, and colleagues at Naitive. The collaboration with UIHC was instigated by Fitton and Michael Willey, MD, an orthopedic surgeon at UIHC.

This is a retrospective study using deidentified data, so it will not affect the care that anyone receives. Once Osteo- Sight is cleared for use in the United States, it would identify people likely to benefit from a bone health assessment that may not otherwise have been picked up, Connell says.

Addressing Biases
The study also will investigate any potential biases in the machine learning model and evaluate the reasons behind any instances of failed image processing.

Low bone mineral density, while more common in women, occurs in both sexes. It is important, Connell says, that OsteoSight performs equally well for everyone. “We will be looking at how well OsteoSight performs across the whole study population, but also in planned subgroup analyses such as sex, age, and ethnicity,” Connell says.

OsteoSight has been carefully designed to only present a result when there is high confidence, Connell says. “Our image processing pipeline intentionally filters out cases where we know a measurement shouldn’t be made, such as when a patient has a prosthetic hip, the image quality is poor, or the X-ray image has been mislabeled. OsteoSight is a reliable technology, but like all medical imaging technologies, there will be some cases where analysis is not possible due to technical or anatomical issues.”

The purpose of this study is to assess performance in real-world practice. “For that reason, we want to understand as much as possible about how well it rejects unsuitable cases,” Connell says.

The researchers expect to have results in about a year. They hope to publish their results in the Journal of Orthopaedic Trauma. “An abstract is currently under review for the first dissemination of the results. Beyond this, the data will continue to undergo further curation and rigorous analysis, with plans to submit the findings for peer-reviewed publication later this year,” Connell says.