Autologous chondrocyte implantation: multilevel modelling of 20-year longitudinal outcome data

The objectives of the study were to investigate demographic, injury and surgery/treatment-associated factors that could influence clinical outcome, following Autologous Chondrocyte Implantation (ACI) in a large, “real-world”, 20 year longitudinally collected clinical data set. Multilevel modelling was conducted using R and 363 ACI procedures were suitable for model inclusion. All longitudinal post-operative Lysholm scores collected after ACI treatment and before a second procedure (such as knee arthroplasty but excluding minor procedures such as arthroscopy) were included. Any patients requiring a bone graft at the time of ACI were excluded. Potential predictors of ACI outcome explored were age at the time of ACI, gender, smoker status, pre-operative Lysholm score, time from surgery, defect location, number of defects, patch type, previous operations, undergoing parallel procedure(s) at the time of ACI, cell count prior to implantation and cell passage number. The best fit model demonstrated that for every yearly increase in age at the time of surgery, Lysholm scores decreased by 0.2 at 1-year post-surgery. Additionally, for every point increase in pre-operative Lysholm score, post-operative Lysholm score at 1 year increased by 0.5. The number of cells implanted also impacted on Lysholm score at 1-year post-op with every point increase in log cell number resulting in a 5.3 lower score. In addition, those patients with a defect on the lateral femoral condyle (LFC), had on average Lysholm scores that were 6.3 points higher one year after surgery compared to medial femoral condyle (MFC) defects. Defect grade and location was shown to affect long term Lysholm scores, those with grade 3 and patella defects having on average higher scores compared to patients with grade 4 or trochlea defects. Some of the predictors identified agree with previous reports, particularly that increased age, poorer pre-operative function and worse defect grades predicted poorer outcomes. Other findings were more novel, such as that a lower cell number implanted and that LFC defects were predicted to have higher Lysholm scores at 1 year and that patella lesions are associated with improved long-term outcomes cf. trochlea lesions.