Prodromal symptoms in knee osteoarthritis: a nested case-control study using data from the osteoarthritis initiative

Abstract

Purpose: A better understanding of the timing of emergent symptoms is needed to complement longitudinal studies of disease pathogenesis and early disease biomarker discovery. This study sought to investigate the existence, duration and nature of a prodromal symptomatic phase preceding incident radiographic knee OA. Methods: Data for these analyses were from the incidence cohort of the Osteoarthritis Initiative (OAI) public use data sets. We imposed a nested case-control design. Cases were knees which had developed new tibiofemoral radiographic OA (Kellgren-Lawrence grade ≥ 2) at any of the subsequent follow-up visits, up to 4 years. For each case, four control knees were selected using incidence density sampling. Cases and controls were assigned a common baseline time-point, t0, corresponding to the time of incident radiographic OA. Candidate prodromal symptoms were WOMAC and KOOS subscale scores and individual items (dichotomised as none/mild vs. moderate/severe/extreme), available up to 4 years prior to t0 (t0-1, t0-2, t0-3, t0-4). Trajectories in WOMAC/KOOS subscale scores in cases and controls over time were fitted by multilevel models, adjusting for age and gender, allowing for repeated measures within individuals. These models were used to estimate differences in subscale scores at each time-point and to estimate the length of prodromal phase for each subscale score (defined as the point at which subscale scores of cases and controls were predicted to cross (i.e. be equal)). Conditional binary logistic regression was used to identify differences in individual items between cases and controls at each time-point prior to t0. Random intercept binary logistic regression models, adjusting for age and gender, were then used to estimate the length of prodromal phase for each item (odds of exposure predicted to be equal for cases and controls). Results: Over 4 years 240 cases of incident radiographic knee OA were recorded in 219 participants. Differences between cases and controls in all WOMAC/KOOS subscale scores were observed at least 2 years prior to incident radiographic knee OA but, due to small numbers of cases, were statistically significant only at t0-1 (e.g. difference between cases and controls at one year before incident radiographic knee OA in predicted KOOS Pain subscale score (0-100) = 2.96 (95% CI: 1.48 to 4.43); difference in predicted WOMAC Physical Function (0-68) = 1.26 (95% CI: 0.37 to 2.14)). The time-point at which WOMAC/KOOS subscale scores for cases were predicted to diverge from those of controls ranged from 25 months (WOMAC Pain; 95% CI: 16 to 34 months) to 34 months (WOMAC Stiffness; 95% CI: 21 to 47 months) prior to incident radiographic knee OA. At t0-1, differences between cases and controls were seen in 16 of 20 individual items with odds ratios ranging from 1.12 (pain on standing; 95% CI: 0.58 to 1.65) to 2.18 (subjective swelling; 95% CI: 1.21 to 3.15). Pain and difficulty on activities associated with higher dynamic loading of the knee appeared to have generally longer prodromal phases (e.g. difficulty standing from sitting (37 months; 95% CI: 1 to 73 months) and pain when twisting/pivoting on the knee (37 months; 95% CI: 13 to 61 months) versus those with less dynamic loading such as pain on standing (14 months; 95% CI: 4 to 25 months) and pain when walking (21 months; 95% CI: 11 to 31 months). Conclusions: Our study found that incident radiographic knee OA is preceded by prodromal symptoms lasting up to 3 years. This has potential implications for understanding phasic development and progression of osteoarthritis and for early recognition and management.