Localized dead bone, a potential hint to an underlying condition

In this issue of the Journal, Hill and Khodaee¹ present radiographs showing a bone infarct, which after evaluation seems unrelated to the patient’s pain syndrome. Finding asymptomatic bone infarcts on radiographs is not particularly rare, but as I thought about the clinical context, it seems that I am now seeing fewer patients referred or ultimately diagnosed with osteonecrosis than I used to. If true, this is a good thing, and it may reflect reduced prescribing of long-term corticosteroids.

The history of how osteonecrosis came to be recognized as a unique clinical entity is an interesting tale, although the pathophysiology of osteonecrosis remains incompletely understood. McCarthy² in 1982 wrote a perspective in which he reviewed many of the early descriptions of bone infarction, placing them within the historical context of the growing knowledge of microbiology, pathobiology, and radiographic imaging in the 19th and early 20th centuries.

Hippocrates apparently first described localized bone death in the heel following improperly bandaged foot fractures, noting that “the malady may last the patient’s whole life.”² Descriptions of bone necrosis in the 18th and 19th centuries can retrospectively be interpreted as most likely being osteomyelitis (suppurative necrosis). The concept of aseptic vs infectious bone necrosis could not be clarified until Pasteur and others were able to identify the microbiologic basis of infectious inflammation. Axhausen in 1907 was apparently the first to write (in German) specifically about aseptic necrosis as he described the fate of bone grafts using the terminology “anemic infarct,” indicating his recognition of lack of adequate arterial blood supply contributing to the evolving avascular necrosis.²

With the advent of radiography in the very early 1900s came a flurry of descriptions of localized bone infarct syndromes that still bear eponymous diagnostic titles: Legg-Calvé-Perthes’ hip, Köhler’s tarsus, Kienböck’s lunate, and Freiberg’s metatarsal head. Phemister³ in Chicago linked the radiographic findings of the osteonecrotic hip to detailed pathologic descriptions and introduced the concept of surgical drilling into the femoral head to reduce pressure and improve blood flow, a practice that with various iterations continues still.

Mankin⁴ published a review of nontraumatic osteonecrosis in 1992 in which he discussed 4 pathophysiologic mechanisms by which osteonecrosis could develop. He reviewed several clinical syndromes, highlighting corticosteroid-induced osteonecrosis specifically. Notably, then, as now, this may be the most common identified causative association with osteonecrosis, and yet its pathogenesis remains, in my mind, incompletely understood.

Despite the recognized association of corticosteroid use with osteonecrosis, it is still not clear why the relationship between dose and duration of therapy is so variable between individual patients who develop, or don’t develop, osteonecrosis. When I used to see more patients with corticosteroid-associated osteonecrosis, I was struck by the high prevalence of multifocal involvement: hips, knees, as well as shoulders, and often asymptomatic long-bone infarcts discovered on radiography. Knee effusions were common, with noninflammatory synovial fluid only very rarely containing observable lipid droplets.

Although most of us still prescribe a lot of corticosteroids for short-term use, probably still at higher doses and more often than warranted, there has been a major successful movement within rheumatology and transplant medicine to limit the dose and duration of corticosteroids to treat systemic inflammatory diseases and prevent rejection. We now try to rely on alternative immunosuppressive and anti-inflammatory medications. Cheng et al⁵ published what is likely a testament to the benefit of this approach, demonstrating a reduction in the incidence of osteonecrosis of the femoral head in patients who received a kidney transplant in the era after a corticosteroid-minimization policy was instituted.

Counter to my impression and the above, the COVID-19 pandemic has introduced a new risk factor for osteonecrosis. A US study found that the incidence of osteonecrosis as an indication for total hip arthroplasty increased from 1.4% in the years before the pandemic (2016–2019) to 1.6% in 2020 to 2021, and a previous COVID-19 diagnosis was independently associated with a higher osteonecrosis rate (3.9% vs 3.0%).⁶ Other studies suggest that there may be an amplified risk for patients diagnosed with COVID-19 who received corticosteroids, similar to the clinical impression of some that patients with systemic lupus erythematosus also have an exaggerated risk for osteonecrosis if treated with corticosteroids.

The discovery of osteonecrosis by imaging, whether the patient is symptomatic or not, warrants consideration for the presence of an underlying condition, including chronic excessive alcohol use, Gaucher disease, endogenous or iatrogenic hypercortisolism, hemoglobinopathies, human immunodeficiency virus infection, and others. The preferred treatment of early osteonecrosis to prevent progression remains undefined—there is no panacea. Arthroplasty is an effective treatment for severe symptomatic osteonecrosis of the hip,⁷ knee, and shoulder, although the need for revision of knee and shoulder arthroplasties seems higher than desirable.

• Copyright © 2026 The Cleveland Clinic Foundation. All Rights Reserved.