Basic Science Tip #516

Detecting rapidly progressive knee osteoarthritis: a cloudy crystal ball

Knee osteoarthritis typically develops over a decade or more, However one of five people typically have more pain and disability at onset and have an accelerated radiographic knee osteoarthritis (AKOA) with end stage disease within 4 years. The use of demographics and clinical findings has resulting in only a 40% rate of identification of AKOA relative to longer term KOA. Investigators used longitudinal data from a case–control study in the osteoarthritis initiative (OAI), a longitudinal multicenter observational cohort study of adults with or at risk of developing symptomatic KOA. This included demographic, clinical, and biochemical data along with radiographic and magnetic resonance imaging. Investigators hypothesized that the addition of an MRI scoring system would more accurately identify patients at risk for AKOA. They used classification and regression tree (CART) models to both assess the ability of baseline MR features to classify participants who will develop AKOA and to determine if adding baseline MR features to a model that includes clinically accessible measures of symptoms, function, radiographic features, clinical findings, and demographic characteristics improved classification of adults who will develop AKOA. CART is a non‐parametric method allowing for complex interactions and nonlinear associations. It identifies the most discriminating characteristic, with its associated cut point, to differentiate the two groups such as AKOA or non-accelerated KOA). After the initial characteristic split based on the cut point, CART finds the next best characteristic with their corresponding associated cut point. CART analysis is easily interpretable

Clinical data included accessible pain, function, clinical knee exam, and quality of life measures. This included WOMAC pain and function along with walking speed. Demographic data included age, sex, and body mass index collected at baseline. Biochemical data included high sensitivity cross reactive protein and serum blood sugar. MRI scoring included bone marrow lesion volume, effusion-synovitis volume, cartilage damage index, meniscal extrusion and degeneration, cruciate ligament degeneration, and patellar fat pad changes.

Contrary to the hypothesis, the CART models with and without MR features each explained approximately 40% of the variability. Adding MR-based features to the model improved specificity (0.90 vs. 0.82) but lowered sensitivity (0.62 vs. 0.70). Of interest, serum glucose, effusion-synovitis volume, and cruciate ligament degeneration were statistically important variables in classifying AKOA individuals.

The clinical take home is that early MRIs may be useful in sorting out mechanical complaints but are not useful in determining who will develop AKOA. In contrast, in later stages of the disease, MRI may reveal far greater damage than can be detected on radiographs.

Reference
Price LL, Harkey MS, Ward RJ, MacKay JW, Zhang M, Pang J, Davis JE, McAlindon TE, Lo GH, Amin M, Eaton CB, Lu B, Duryea J, Barbe MF, Driban JB. Role of Magnetic Resonance Imaging in Classifying Individuals Who Will Develop Accelerated Radiographic Knee. J Orthop Res. 2019 Nov;37(11):2420-2428. doi: 10.1002/jor.24413. Epub 2019 Jul 29. PMID: 31297900