Background Ultrasound (US) is a useful imaging modality that allows the evaluation of features suggestive of inflammation in osteoarthritis (OA). US features are more common in those with painful knee OA but the association between them remains inconclusive.

Objectives To determine: [1] whether the frequency of ultrasound (US) features of inflammation differs between people with (a) normal asymptomatic knees (controls); (b) knee pain (KP) with no radiographic osteoarthritis (ROA); (c) asymptomatic ROA; and (d) symptomatic ROA (SROA); [2] whether US features associate with knee pain, OA change on x-ray, or symptoms and signs of inflammation; and [3] whether US features change in parallel with temporal change in knee pain.

Methods A multiple group case-control study design was used. Community participants were recruited and divided into 4 groups based on presence or absence of (1) knee pain and (2) ROA. Standardised knee radiographs and assessments for pain, stiffness and function were carried out. All underwent US examination for effusion, synovial hypertrophy, popliteal cysts and power Doppler signal (PD) within the synovium. Features were dichotomised as present/absent, and grey-scale features were measured directly in mm. Follow-up US and pain assessments were collected in some OA and non-OA participants after 3 months.

Comparisons between groups were examined using Χ² tests, one-way ANOVA and post-hoc Bonferroni tests. Logistic regression was used to examine association between US features, knee pain and ROA adjusting for age, sex and BMI.

Results 243 participants were recruited: 90 controls, 59 with KP, 32 with ROA and 62 with SROA. Greyscale US features of inflammation were more common in SROA (effusion 92%, synovial hypertrophy 82% and popliteal cysts 39%) than control and KP participants (p<0.05). US effusion, synovial hypertrophy and popliteal cysts were strongly associated with ROA (aOR(95%CI) =13.39 (6.14, 29.02); 14.39 (6.28, 32.94); 3.19 (1.42, 7.17) respectively) but only synovial hypertrophy was independently associated with pain (aOR=6.56 (2.85, 15.11)). No independent associations were found between PD signal, and pain or ROA. Follow-up of 65 non-OA and 51 OA participants found no relationship between change in pain severity and change in US features.

Conclusions US features of inflammation are most strongly associated with ROA but the relationship with symptoms is less clear. This supports synovial changes as predominantly a secondary rather than primary feature of OA. Further studies are warranted to confirm these findings.

Acknowledgements We are grateful to Arthritis Research UK for funding this work (AHP Training Fellowship Grant no.18861).

Disclosure of Interest None Declared