Background/Purpose: Knee replacement (KR) represents a clinically important and cost-effective endpoint of knee osteoarthritis (OA). We have shown previously that, in a year prior to KR, medial cartilage loss (measured quantitatively with MRI) was greater in knees with KR than in controls, who displayed the same (baseline) radiographic disease stage. The best discrimination was observed in the central medial tibia (cMT). However, the trajectory of the more long-term cartilage thickness change prior to KR is unknown. The purpose of this study therefore was to examine rates of cartilage loss up to 4 years prior to KR.

Methods: We studied knees from the Osteoarthritis Initiative (OAI) who received a KR between 12-60 month follow-up (12M-60M). Each knee with KR was matched with one control who did not receive a KR through 60M by sex, baseline KLG (0-1, 2, 3, 4 strata), and age (±5y). Sagittal 3T DESSwe MR images were used for segmenting the weight-bearing femoro-tibial cartilages (Chondrometrics GmbH), with blinding to acquisition order. The visit directly prior to KR (T0) and those 12, 24, 36, and 48 months prior to T0 (i.e. T-1 through T-4) were analyzed as available. Cartilage thickness in cMT was selected as the primary and that in the total medial compartment (MFTC) as the secondary endpoint. Comparisons included paired t-tests between case/control pairs, and conditional logistic regression model, with and without adjustment for pain and BMI at the beginning of the respective observation interval.

Results: 220 knees of 190 OAI participants received a KR up to 60M (37@24M, 60@36M, 58@48M, and 65 @60M). Of these, 189 (58 baseline KLG 0-2, 69 KLG3, and 62 KLG4) had MRIs for at least one longitudinal interval, and a matched control (age 64±8.7; BMI 29±4.5; 85 male; 104 female). Analysis of slopes of annual change from all available time points revealed significantly greater rates of cMT cartilage loss in KRs than in controls (n=189: 94±137  vs. 55±104µm; p=0.0017; odds ratio [OR] = 1.36 [95% CI: 1.11;1.66] without, and 1.34 [1.09;1.66] with adjustment for pain and BMI). Between T-1 and T0, cMT cartilage loss was 2.9x that in controls (n=152: 114±209 vs. 39±159µm; p=0.0007, OR 1.42 [1.14;1.75] without, and 1.48 [1.18;1.86] with adjustment). The difference was less in previous time intervals (≤1.7x), but stronger for the T-2 to T0 period (n=127: 209±281 vs. 61±156µm [3.4x]; p<0.0001, OR=1.61 [1.29;2.01] without, and 1.64 [1.30;2.06] with adjustment). However, no significant difference was observed during T-4 to T-2 (n=60: 119±255 vs. 125±175µm; p=0.86, OR=0.97 [0.71;1.33] without, and 1.21 [0.81;1.81] with adjustment). Results for MFTC were similar to those for cMT.

Conclusion: The rate of medial cartilage loss was substantially greater in knees with KRs than in those without during a two-year period prior to T0 (i.e. within 2 years proximate of the timing of the KR), even after matching for baseline radiographic disease status. During earlier time intervals (i.e. 4 to 2 years prior to T0) the rates of medial cartilage loss were not different between KRs and controls. This trajectory of cartilage thickness change resembles that observed for change in symptoms prior to KR.