Hip osteoarthritis is an increasingly important cause of morbidity in the ageing population causing pain, disability and loss of function through joint failure. It has an estimated lifetime risk of 25%. While the contribution of skeletal morphology to disease is now established, the lack of imaging biomarkers that can reliably predict osteoarthritis development is limiting. Novel biomarkers for clinically relevant hip osteoarthritis are needed particularly methods to identify patients who are destined for total hip replacement (THR) surgery and who might benefit from novel interventions. The chosen techniques studied here involved 2D and 3D measures of hip cortical bone structure from routine clinical computed tomography (CT) scans of patients to predict clinically-relevant hip osteoarthritis. Outside the research environment, clinical practice and trials still use simple 2D outcome measures from radio- graphs. Clinical diagnosis and surgical decision making therefore rely on combinations of poorly-defined disease signs and symptoms and controversially, 2D radiographic imaging. Such methods overlook key 3D disease features. In many osteoarthritis guidelines, clinicians are now advised to avoid imaging altogether.

This work asked three important research questions concerning hip osteoarthritis.

1. Does imaging have clinical utility in managing patients with hip pain, and in particular are hip radiographs effective in predicting THR?
2. Does 3D imaging using clinical CT have clinical utility in predicting THR, and is there a 3D phenotype of bone thickening associated with eventual THR?
3. Is the 3D bone thickening associated with hip osteoarthritis an inevitable consequence of ageing?

In order to answer these, CT cortical thickness data was acquired from two population based studies;​ one in only older men and women from the Age, Gene/Environment Susceptibility-Reykjavik (AGES) Study in Iceland and the other from the Mindways study acquired from 11 healthcare centres around the USA, of women aged 20-97 years.

CT scans were coupled with new image analysis techniques called 3D cortical bone thickness mapping, statistical parametric mapping and disease feature distribution mapping. Recently it was hypothesised that focal bone thickening in regions of the hip cortex would be associated with eventual hip joint replacement and radiological progression of osteoarthritis features.

Combining hip pain, radiographic osteoarthritis grade and CT gave excellent discrimination of THR (0.90;​0.85,0.95). Conversely, hip pain was a poor to marginal predictor (AUC=0.70;​ 95%CI 0.62,0.78). The AUC for radiographic Kellgren and Lawrence (K&L) osteoarthritis grade alone was 0.87(0.81,0.92), irrespective of hip pain, with single mJSW being a rea- sonable discriminator (0.80;​0.73,0.87). Osteoarthritis was also associated with a CT-defined crescent of femoral head surface bone thickening (Odds Ratio for THR;​ 5 per SD thicker;​ 5.0(3.2,7.7), 0.85(0.79,0.91)). In a typical clinical presentation with hip pain, all of the imaging parameters measured ranged from good to excellent in terms of clinical utility.

Imaging unequivocally predicted THR for osteoarthritis, whether or not pain had become apparent. Contrary to current guidelines, images of patients with hip pain have good to excellent clinical utility for selecting surgically-relevant presentations. If patients with definite hip pain had definite radiological osteoarthritis, 85% went on to THR within 3 years.