Mediolateral oversizing influences pain, function, and flexion after TKA

Abstract

Purpose: Manufacturers of total knee arthroplasty (TKA) have introduced narrower femurs to improve bone-implant fit. However, few studies have reported the clinical consequences of mediolateral oversizing. Our hypothesis was that component oversizing negatively influences the results after TKA.

Methods: One hundred and twelve prospectively followed patients with 114 consecutive TKA (64 females and 50 males) were retrospectively assessed. The mean age of the patients was 72 years (range, 56 to 85 years). The dimensions of the femur and tibia were measured on a preoperative CT-scan and were compared with those of the implanted TKA. The influence of size variation on the clinical outcomes 1 year after surgery was assessed.

Results: Mediolateral overhang was observed in at least one area in 66 % of the femurs (84 % in females and 54 % in males) and 61 % of the tibia (81 % in females and 40 % in males). Twenty-two patients presented no overhang in any area and 16 had overhang in all studied zones. The increase in the Pain and KOOS scores were 43 ± 21 and 36 ± 18 in the patients without overhang and 31 ± 19 and 25 ± 13 in patients with overhang (p = 0.033; p = 0.032). Knee flexion was 127° ± 7 and 121° ± 11, respectively. Regression and latent class analysis showed a significant negative correlation between overall oversizing and overall outcome.

Conclusions: This study confirms that oversizing may lead to worse clinical results in TKA. The clinical consequences are that surgeons should pay attention not to oversize implants during implantation nd that oversizing should be ruled out in case of so called unexplained pain.

Level of evidence: IV.

Fig. 1

Three reference zones were defined on the femoral implant a zone 1, corresponding to the posterior part of the anterior chamfer, was located at a variable distance from the posterior bicondylar line (BCL) depending on the implant size (39.4–48.5 mm; see Appendix in ESM). Zone 2 was located at a variable distance from the posterior bicondylar line (BCL) depending on the implant size (26–36 mm; see Appendix in ESM), but was directly posterior to the point where the implant began to narrow. Zone 3 corresponds to the posterior condylar bone cut, situated 10 mm from the BCL. On the CT scan b, the analysis was done on the axial cut located at the level of the distal femoral cut made at the time of surgery (10 mm from the most distal point of the medial condyle). The bone dimensions corresponding to the three zones defined were measured: zone 1, 10 mm from the BCL, zone 2 and zone 3, at the distance corresponding to the size of the implanted prosthesis. On the tibia, the mediolateral dimension (zone 4) was used as the reference (c). On the CT scan, the measurement was taken on the axial cut located at the bone cut made at the time of surgery (d). The transverse, mediolateral dimension was measured

Fig. 2

These histograms represent the distribution of the size variation (X axis) in the four zones studied in females (blue columns) and males (red columns)

Fig. 3

These figures represent the increase in the pain score (Y axis) in relation to the size variation (X axis) for the four zones studied. No threshold value was found on these curves

Fig. 4

These figures show the flexion angle (Y axis) in relationship to the size variation (X axis) for the four zones studied. No threshold value was found on these curves

Fig. 5

In the Latent Class Analysis, the first latent variable was defined as the “prosthetic fit”. It was obtained with the structural equation model from the measured variation of size in the four defined zones. The second latent variable was defined as the “post-operative outcome”. It was obtained with the structural equation model from the postoperative pain score and the MPF. The relationship between the two latent variables was explored through a Spearman correlation. In this structural equation model, the rectangles represent the observed variables while the circles represent the latent variables. The two latent variables, “prosthetic fit” and “post-operative outcome”, were found to be negatively correlated (r = −0.26 with a p = 0.005)