Background/Purpose: Radiographic patellofemoral osteoarthritis (OA) is prevalent in approximately 25% of population-based samples and 39% of individuals with knee pain. Discordance between knee OA features and pain is common yet not well understood. It may in part be explained by mechanical factors, such as patellofemoral alignment or bony morphology, that could cause pain by altering joint load. It may also be due to factors that influence pain perception, such as psychosocial or central mechanisms. We aimed to compare within-person, between-knee alignment, morphology and presence of patellofemoral OA in individuals with unilateral frequent anterior knee pain, to determine if these variables explain pain discordance.

Methods: We evaluated a subsample from the Multicenter Osteoarthritis Study (MOST), a cohort of 3026 individuals aged 50 – 79 years who had, or were at risk of, knee OA at baseline. We identified individuals who presented at the 60-month clinic visit with discordant patellofemoral pain. We defined this as: (i) a response of ‘yes’ to the question “During the past 30 days, have you had pain, aching, or stiffness in your knee on most days?” for one knee, but ‘no’ to the same question regarding the contralateral knee; and (ii) this pain was reported to be isolated to the anterior knee region using a knee pain map. For those not meeting inclusion criteria at the 60-month visit, we also included participants who met the same criteria at 84-months.

We measured alignment and morphology using MR images (patellar tilt angle, bisect offset, sulcus angle, lateral trochlear inclination) and lateral radiographs (Insall-Salvati ratio). Patellofemoral OA was defined on lateral view radiographs as any osteophyte ≥ grade 2, or joint space narrowing ≥ 1 plus any osteophyte, sclerosis or cyst ≥1.

We modelled the odds of having pain using conditional logistic regression, which enables within-person, between-knee comparisons. Within-person matched pairs (i.e. knees) removes the effect of person-level risk factors that may contribute to pain (e.g. age, sex, BMI, kinesiophobia, depression, genetics, and other known/ unknown factors) that could otherwise explain pain variability among people.

Results: Inclusion criteria were met in 136 participants, 115 with history of surgery/ injury data, 97 with bilateral radiographs and 71 with MR images. The mean age (n=71) was 69 (8) years, BMI 30.2 (5.3) kg/m², 47 (66%) women. Alignment and morphology measures did not differ between knees. Odds of having patellofemoral OA was significantly increased in knees with pain compared to those without (OR 6.6 [95% CI 1.5, 29.7], as was history of surgery or injury (OR 2.1 [1.2, 3.6]).

Conclusion: In individuals with unilateral frequent anterior knee pain, odds of having patellofemoral OA were higher in knees with pain than without. Alignment and morphology measures did not differ by knee. Dynamic alignment may be more relevant to pain outcomes.