Pain and radiographic damage in osteoarthritis

A person who has severe knee pain and extensive radiographic evidence of osteoarthritis is thought to have an “obvious” explanation for the pain. Some people, however, may have severe radiographic damage (Kellgren-Lawrence grade 41) with no pain, or vice versa.2 Such discordances, which are also seen in rheumatoid arthritis,3 have been the subject of much interest for clinicians and researchers. Some researchers have suggested that little or no relation exists between the “illness” of joint pain and the “disease” of radiographic osteoarthritis.

In the linked study (doi:10.1136/bmj.b2844), Neogi and colleagues assess the association between knee pain and radiographic osteoarthritis in people whose knees were discordant for pain.4 The study design was a within person, knee matched case-control study that examined 696 people from the Multicenter Osteoarthritis Study (MOST) and 336 from Framingham. Odds ratios were 151 (95% confidence interval 43 to 526) in MOST and 73 (16 to 331) in Framingham for pain in association with Kellgren-Lawrence grade 4 osteoarthritis in one knee and grade 0 in the other knee.4 These data show that pain and radiographic damage are strongly associated.

The data are biologically plausible and reflect clinical experience. However, as is always the case, the study had limitations. The first concerns patient selection. The MOST study enrolled people who were at high risk for knee osteoarthritis as a result of obesity, knee pain, aching, or stiffness, although the Framingham osteoarthritis database comprised unselected subjects. Nonetheless, people with knees discordant for frequent pain were a minority in each cohort—23% in MOST and 24% in Framingham. This finding also illustrates the polyarticular nature of osteoarthritis; in one previous study, 70% of people with osteoarthritis of the hip or knee reported pain in more than one of four hip or knee joints.5

Inclusion of people who already had a painful knee may have selected for a different phenotype than people with structural joint damage but no symptoms. Findings linking radiographic osteoarthritis and pain may therefore not be relevant to many people in the community with structurally damaged joints who do not see a doctor for joint pain. It may have been of interest to study people with a converse pattern, who had discordant radiographic damage in only one knee but not the other, to examine how often they might have been discordant for pain.

A second limitation concerns the report of pain in people with arthritis, which the authors correctly note is multifactorial, being affected by structural abnormalities, psychosocial variables, demographic factors, and situational variables. For example, in a multicultural setting, Latino patients with rheumatoid arthritis had the highest pain scores and Asians the lowest, with black and white patients having intermediate scores.6 The complex sources of painful symptoms in arthritis suggest that fundamental differences may exist between people with symptomatic osteoarthritis and those with similar degrees of joint damage but no symptoms.

A third limitation concerns differences in interpreting data in groups of patients compared with individuals. An odds ratio of 151 indicates a high probability that someone with radiographic grade 4 damage will have pain. However, 17% of people in the MOST study and 14% in the Framingham study with apparently severe radiographic grade 4 osteoarthritis had pain-free knees. In addition, 67% of the patients in MOST and 28% in Framingham with at least moderately severe grade 3 radiographic knee osteoarthritis had a knee that was asymptomatic.

In previous studies in which participants were not selected for discordant pain, rates of definite radiographic osteoarthritis in asymptomatic people were even higher.2 The large number of patients with osteoarthritis who have advanced radiographic osteoarthritis in asymptomatic joints suggests that attempts to define disease severity exclusively according to structural degeneration will remain inadequate, despite strong associations between damage and pain.

Neogi and colleagues have shown a strong association between symptomatic osteoarthritis and radiographic structural degenerative disease. Nonetheless, clinicians must be aware of differences between patterns in patients involved in research and those in clinical settings. This is true for the diagnosis and management of most rheumatic diseases, for which pathognomonic findings are rarely available. For example, although most patients with rheumatoid arthritis have an abnormal test for rheumatoid factor, anti-cyclic citrullinated peptide antibodies, erythrocyte sedimentation rate, or C reactive protein, results are normal in 30-40%.7 8 Of course, group data improve the probability of a correct diagnosis, but managing individual patients requires recognition of patterns outside the usual probabilities, and attention to the statistically improbable outlier may provide clues to pathogenesis.