IDENTIFICATION OF FACTORS WHICH AFFECT CLINICAL OUTCOME OF AUTOLOGOUS CHONDROCYTE IMPLANTATION USING Z-TRANSFORMATION AND MULTIPLE REGRESSION ANALYSIS

Abstract

Introduction: Autologous Chondrocyte Implantation (ACI) was first described in 1994(1) and has become an increasingly widely used treatment for chondral defects in the knee. The intention of this study was to identify which patient and/or surgical factors affect clinical outcome. In order to do this, a multicentre database of patients treated with ACI was established.

Methods: Four European centres collaborating in the EuroCell project (2) contributed data. These centres have historically used different outcome measures to follow up their patients. In order to analyse this data, a method of z-transformation was used to standardise the clinical scores. This has allowed a large number of patients to be investigated even when different scores have been used. A panel of predictor variables was agreed relating to patient factors and operative technique. Linear multiple regression analysis was performed to determine which predictor variables significantly influenced clinical outcome.

Results: A total of 284 patient datasets from four centres were investigated with 1 to 10 year follow-up. In 213 datasets the Modified Cincinnati (Noyes) clinician evaluation was used (3). The remaining 73 patients had outcome data measured with the modified Lysholm score (4). Outcome was defined as the change in score to latest follow-up. Z-transformation (z-change) was performed for each score type. The regression model was: z-change = − 0.11 − 0.5*z-preop − 0.43*R4 + 0 .30*OC + 0.20*FC (R2=0.30) The regression analysis showed that the factors which affected outcome were one centre (R4), pre-operative score (z-preop), osteochondral defects (OC), and lesions of the femoral condyle (FC). Factors which were found not to affect outcome included the age of the patient, size of the defect treated, number of defects treated and time to follow-up. Variations in operative technique, including the location of the cartilage harvest, the use of fibrin sealant and the timing of patch placement, were not found to have an effect on clinical outcome.

Conclusions: The method of z-transformation is a useful way of compiling multicentre data where different outcome measures have been used. This has allowed a large dataset to be compiled and factors which influence clinical outcome to be identified.