Of the 132 subjects who took part in our study, 23 did not have an MRI from which subchondral bone cysts could be assessed (MRI not available or image unclear). The 109 subjects analyzed had a mean age of 63.2 (SD ± 10.3) years, and a mean BMI of 29.3 (SD ± 5.1) kg/m2. Demographics were not different between those who were included in the study and those who were not (data not shown). Eighty-eight (81%) subjects completed the follow-up; 21 were lost to follow-up for reasons including knee surgery, severe illness, loss of interest, death, and unclear MRI images from which cysts could not be assessed. Those who completed the follow-up had a lower mean BMI than did those who did not (mean ± SD, 28.8 ± 5.0 and 31.3 ± 5.4, respectively; P = 0.05).
Fifty-two (47.7%) subjects had at least one subchondral bone cyst at baseline. They were more likely to be male subjects, although no significant difference was found in age, weight, height, or BMI. Those with cysts had less lateral tibial cartilage volume and greater tibial plateau bone area compared with those who did not have a cyst (Table ). Of subjects with a cyst at baseline, 98.0% also had a BML (Table ). Furthermore, those with subchondral bone cysts were more likely to have large BMLs (grade ≥ 3). In contrast, those with a BML but no cyst at baseline tended to have small BMLs (grade 1).
Comparison of characteristics between subjects
Twenty-one (23.9%) subjects had a cyst that increased in score over a 2-year period (cyst progression), including 6 (13.0%) in whom one or more subchondral bone cysts developed (Table ). All had a coexisting BML at baseline. Of those with a cyst at baseline, cyst progression was observed in 15 (35.7%) subjects, whereas a decrease in cyst score (cyst regression) was observed in 10 (23.8%) subjects, with 6 (14.3%) resolving completely (Table ). No change in cyst (stable) was observed in the remaining 17 (40.5%) subjects.
Natural history of subchondral bone cysts
The mean cartilage volume was lower in both compartments in those with cysts, compared with those with BMLs only or neither cyst nor BML present (Table ). In the medial compartment, those with cysts present had a mean medial cartilage volume of 1,589 mm3 compared with a mean of 1,809 mm3 in those with BMLs only and 1,923 mm3 in those with neither (P for trend, 0.004). Similarly those with cysts also had the least amount of lateral tibial cartilage volume compared with those with BMLs only or neither (mean, 1,607, 1,962, and 2,131 mm3, respectively; P for trend, <0.001). In the longitudinal analyses (Table ), those with cysts had the highest rate of cartilage loss (9.3%) compared with the other two groups (6.3% and 2.6%) (P for trend, <0.001). Similar results were obtained when the subject with a cyst but no BML was excluded.
Relation between increasing grade of severity of subchondral bone abnormality and tibial cartilage volume
We extended our observation by examining the effect of increasing grade of severity of subchondral bone abnormality (grade 1, normal; 2, BMLs only; 3, BML and cyst present) on risk of knee-joint replacement over a 4-year period (Table ). For every one grade increase in severity of bone abnormality in the medial compartment, the risk of joint replacement was increased (odds ratio, 1.99; 95% CI, 1.01 to 3.90; P = 0.05) when adjusted for age, gender, and K-L grade. No significant association was found in the lateral compartment. Again, similar results were obtained when excluding the subject with a cyst but no BML.
Effect of increasing grade of severity of subchondral bone abnormality on joint replacement
When we examined the effect of change in subchondral bone cyst on cartilage, we found that those who had cyst regression in the lateral compartment had significant reduction in lateral tibial cartilage loss (regression coefficient, -11.81; 95% CI, -16.64 to -6.98; P < 0.001) compared with those who were stable or progressed. However, those who had cyst progression tended to have greater medial cartilage loss (regression coefficient, 3.51; 95% CI, -0.35 to 7.37; P = 0.07) than did those who were stable or regressed, although the results did not reach significance. Sixteen (33.3%) subjects had a knee-joint replacement over a 4-year period (Table ). Because of the low numbers of progression and regression (one and three subjects, respectively) in this group, we could not examine the relationship between cyst change and risk of joint replacement.