Patello-femoral osteoarthritis at baseline is predictive for tibio-femoral osteoarthritis five years later and vice versa

Abstract

Purpose: The awareness of the importance of patellofemoral osteoarthritis (PFOA) is increasing in recent years. Suggested is that PFOA is the precursor of tibiofemoral OA. Is PFOA at baseline predictive for TFOA five years later? The other way around might also be possible, so is TFOA at baseline predictive for PFOA five years later. Methods: Of a random subsample (n=337) of the 891 females (aged 45-60) from the nested cohort of the Rotterdam Study, baseline and 5-years follow-up MRIs of both knees were assessed for knee OA with the MRI Osteoarthritis Knee Scoring (MOAKS). Based on these scored features we applied the proposed MRI definition. We distinguished the PFOAMRI-definition from the TFOAMRI-definition. X-rays of both knees were scored with Kellgren and Lawrence classification criteria (K&L). Binomial logistic generalized estimated equations were used to determine the predictive value of PFOAMRI for TFOAMRI at follow-up. For this analysis we excluded the knees with TFOA at baseline. We adjusted the analysis for age and BMI and for K&L-score at X-ray of the TF- joint. In addition, we did the same analysis to test if TFOA in knee without PFOA at baseline was predictive for PFOA at 5 year follow-up. Results: Figure 1 shows the distribution and shift during follow-up of the knees defined with isolated PFOAMRI, isolated TFOAMRI, combined PF&TFOAMRI and knees without OA of the first 337 women of whom MRIs are scored. At baseline mean age is 54.9 years (sd=3.9) and mean BMI is 26.9 kg/m2 (sd=4.3). Mean follow-up time is 5.1 years (sd=0.4). 595 knees of 319 women have no TFOA at baseline (25 women are in the analysis with just one knee). Of these 595 knees 30 knees have PFOA at baseline; 56 knees (9.7%) have PFOAMRI at follow up and 46 knees (7.9%) have TFOAMRI at follow up. PFOA was associated with TFOA at follow-up with an odds ratio (OR) of 7.3 (3.2-17.0; p<0.001) independent of the K&L-score (OR=3.0 (1.6-5.8); p=0.001). Pretest predictive value was 7.7% (46/595); posttest predictive value is 40% (12/30). In the knees without PF OA (n=607) at baseline 42 knees were defined with TFOA at baseline, 73 were defined with TFOA at follow-up (3 knees had a TKR) and 37 knees were defined with PFOA at follow-up. TFOA at baseline was associated with PFOA at follow-up (OR=6.5 (2.5-17.0) p<0.001) independent of age, BMI and K&L score. K&L score was not associated with PFOA at follow-up at all (OR=0.92 (0.4-2.3); p=0.85). Pretest predictive value is 6.3% (38/607) and posttest predictive value is 28.6% (12/42). Conclusions: Structural PFOAMRI at baseline is predictive for structural TFOAMRI at follow-up independent of K&L score. TFOAMRI at baseline is predictive for PFOAMRI at follow-up independent of K&L score. These results will be validated in the second subsample of the cohort.