No abstract

HIV clinical care now involves prevention and treatment of age-associated comorbidity. Although physical function is an established correlate to comorbidity in older adults without HIV infection, its role in aging of HIVinfected adults is not well understood. To investigate this question we conducted cross-sectional analyses including linear regression models of physical function in 3227 HIV-infected and 3240 uninfected patients

influenced by the highly anisotropic organization of the collagen matrix and regional variation in cartilage water content. We hypothesize that loss of the normal spatial distribution of cartilage MRI T2 values, as quantified by texture features, will identify distinctive signatures of OA prior to symptomatic or radiographic evidence of OA. This change in T2 signal homogeneity could be used to predict symptomatic disease progression. Methods: Patients were selected from the Osteoarthritis Initiative (OAI) database based on the change in total WOMAC score from baseline to three year follow-up (79 non-OA progression and 103 OA progression patients). The incidence cohort was used to identify a rapid progression population of OA defined as asymptomatic with no radiographic changes at the initial time point (WOMAC<3; KL<2) that had a large increase in WOMAC score (greater than 10) at the three year time point. For each patient, at baseline, 725 image texture features were measured from each knee T2 map. A linear discriminant function and feature reduction method was then trained to quantify a texture metric, the T2 texture index of cartilage (TIC), based on 20 image features, to identify early OA changes on MRI. Results: Statistically significant differences were seen in the T2 TIC between the OA non-progression and progression populations at the initial baseline measurement. This image biomarker differentiates subjects that develop OA prior to clinically significant OA symptoms or radiographic findings with an accuracy of 80.3%AE 4.3%, a sensitivity of 83.4% AE 5.2% and a specificity of 77.4AE 4.8%. These image features used to quantify signal homogeneity through a texture metric can be considered "image biomarkers". These features included run length, cooccurrence matrix, entropy, and local energy measures. When the knee is separated into the patella, medial, and lateral compartments, in more than 80% of knees, the texture metric that identified a knee with early signs of OA were primarily located in only one of these compartments. These signal changes were observed in a dominant knee compartment that correlated with the mechanical axis of the knee. Conclusion: We demonstrate that the T2 TIC, derived from local variation in cartilage T2 values, is able to differentiate and predict individuals that will develop symptomatic knee OA from those that remain asymptomatic. The observation that these image features are localized to knee compartments associated with deviation of the mechanical axis suggests this index reflects early structural cartilage change related to biomechanical properties of the knee. The ability to differentiate patients at risk for OA progression prior to symptomatic or radiographic presentation based on qMRI signal changes would be valuable in clinical and epidemiological studies for disease-modifying OA drugs (DMOADs), joint preservation surgical interventions, and the development of posttraumatic arthritis in ACL and meniscal injuries.