There is some controversy regarding whether total knee arthroplasty is best performed with or without insertion of a patellar prosthesis. Pain may develop in the patellofemoral joint postoperatively if the arthroplasty is performed without a patellar replacement, whereas complications such as patellar fracture, osteonecrosis, and loosening may occur after a patellar replacement2,3,10. We performed a patellar replacement in one knee of patients with rheumatoid arthritis who had simultaneous bilateral total knee arthroplasty, and we followed the postoperative course to determine whether a patellar prosthesis is of clinical value.
This prospective study was designed to allow a direct comparison between the results of total knee arthroplasty with and without patellar replacement. The patellar replacement was randomly assigned to one knee of each patient so that the two operations could be compared in the same patient. With this design, we were able to control the influence of operative factors, the variability in pain threshold among individuals, and the bias arising from the selection of a knee by the surgeon on the basis of radiographic and clinical findings.
The present investigation was an extension of our previous study11. In that study, thirty-five patients were followed for at least two years after total knee arthroplasty. Because assessment with the knee score of The Hospital for Special Surgery4 did not reveal an advantage to combining a patellar replacement with total knee arthroplasty, we initially concluded that routine insertion of a prosthesis was not advisable. However, our results were somewhat different after we followed most of the patients for at least six years.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.
†Department of Joint Disease and Rheumatism, Nippon Medical School, 3-5-5, Iidabashi, Chiyoda-ku, Tokyo 102, Japan.
Of the thirty-five patients enrolled in our previous study, six were not followed at our hospital and three others died during the follow-up period, leaving twenty-six patients for inclusion in the present study. The mean age of the patients at the time of the operation was 56.1 years (range, forty-two to seventy-three years). There were two men and twenty-four women. According to the classification of Steinbrocker et al., the anatomical changes were stage III (advanced) in six patients and stage IV (late) in twenty, while the physical function was class II (the patient is able to conduct usual activities despite pain in the joint with, and restriction of, physical movement) in two patients, class III (the patient is practically incapable of performing activities of daily living) in eighteen, and class IV (the patient is disabled or bedridden) in six.
All twenty-six patients had simultaneous bilateral total knee arthroplasty, and a patellar replacement was performed in one knee of each patient. The side of the patellar replacement was selected randomly with a coin toss just before the operation. The knee was opened, the patella was inverted, and a lateral retinacular release was done, irrespective of patellar tracking, before the bone was cut. The synovial membrane was resected as completely as possible. A modification of the original total condylar knee prosthesis (Yoshino-Shoji total knee prosthesis; Biomet, Warsaw, Indiana) was used for all knees. Unlike the original total condylar knee prosthesis, this model has a tibial component that is flattened posteriorly12,15,16.
All of the operations were done in our department, and care was taken to avoid rotation of the femoral1 and tibial components. The tibia was cut at a right angle to the longitudinal axis in order to remove the articular surface of the medial and lateral tibial condyles. For the femur, the posterior portion of the medial and lateral condyles was cut first, and equal amounts of bone were removed from each condyle with use of the posterior condylar axis as a landmark. The condyles were not heavily damaged posteriorly in any patient. The anterior portion of the femoral condyles was cut along a line parallel to the line of the osteotomy. The mediolateral soft-tissue balance was confirmed, with the knee in 90 degrees of flexion, with use of a knee distractor. If the soft-tissue tension was uneven, the tighter side was released until balance was obtained12,15,16. After the operation, the patients were followed in our department or at affiliated hospitals.
The knees were evaluated before and after the total arthroplasty with the knee score of The Hospital for Special Surgery4. In addition, pain on standing and on ascending or descending stairs as well as tenderness of the patellofemoral joint were assessed. Three blinded observers, who were not the surgeons, analyzed the data. We used the method of Laurin et al. to evaluate the so-called sunrise radiographs, which were made annually with the method described by Merchant et al. In brief, patellar alignment was classified as type I (normal), type II (shifted), type III (tilted), or type IV (shifted and tilted).
The Student t test was used to compare the preoperative and postoperative knee scores of the knees that had had arthroplasty with a patellar replacement with those of the knees that had had arthroplasty without a patellar replacement.
The mean duration of follow-up was 6.6 years, (range, 6.0 to 7.5 years).
Knee Score
With the numbers available, the two groups did not differ significantly with regard to pain, function, range of motion, muscle strength, flexion contracture, instability, or over-all score4 before or after the operation (Table I). Three patients (five knees) had a flexion contracture of 11 degrees or more after the operation. Of these three patients, two had a flexion contracture of both knees and one had a contracture of the knee that had had a patellar replacement. All five knees had had a flexion contractures of 30 degrees or more before the operation.
Other Clinical Parameters
Pain on standing, pain on ascending and descending stairs, and tenderness of the patellofemoral joint occurred in four, eight, and nine knees that had not had a patellar replacement, respectively. In contrast, none of these symptoms occurred in the knees that had had a patellar replacement.
Of the four knees that were painful when the patient stood, one was type I, two were type II, and one was type IV, according to the classification of Laurin et al. Of the eight knees that were painful when the patient used stairs, two were type I, four were type II, and two were type IV. Of the nine knees that had tenderness of the patellofemoral joint, two were type I, four were type II, and three were type IV. Thus, some knees that had excellent congruity of the patellofemoral joint were painful, whereas some knees that had poor congruity were not. However, all of the painful knees had had arthroplasty without a patellar replacement; pain did not develop in any of the knees that had had a patellar replacement, even when there was poor congruity of the patellofemoral joint.
Patellar Alignment
Of the knees that had arthroplasty with a patellar replacement, thirteen were type I, one was type II, one was type III, and eleven were type IV preoperatively and ten were type I, five were type II, seven were type III, and four were type IV postoperatively. Of the knees that had arthroplasty without a patellar replacement, ten were type I, one was type II, four were type III, and eleven were type IV preoperatively and seven were type I, ten were type II, four were type III, and five were type IV postoperatively. With use of chi-square analysis, there was no significant difference, with the numbers available, between the groups with regard to either the preoperative or the postoperative type. In six knees that had not had a patellar replacement, the patella was so displaced laterally that the lateral edge protruded beyond the line representing the lateral limit of the femur on a postoperative sunrise radiograph. All six of these knees had radiographic evidence of erosion of the lateral articular surface of the patella.
Complications
No patient had a patellar fracture or loosening of the patellar component. In addition, none of the patients had persistent postoperative synovitis that was resistant to treatment.
A patient with rheumatoid arthritis, which was controlled satisfactorily, had a reoperation for pain in the knee that had not had the patellar replacement. Patellar alignment had been excellent, without any erosion of the articular surface, two years after the index total knee arthroplasty (Fig. 1-A). At 6.5 years postoperatively, however, the patella had a slight lateral shift with erosion of the lateral and medial articular surfaces (Fig. 1-B). At the reoperation, there was obvious recurrence of synovitis on the articular surfaces of the patella. There was peripatellar synovial proliferation and pannus formation on the patellar surface, but the synovitis was not generalized and had not caused joint effusion. The case of this patient demonstrates that synovitis may recur even if the rheumatoid arthritis is well controlled and that a preserved patella is liable to become secondarily involved after total knee arthroplasty.
Another patient had patellar tilt, six years postoperatively, on the side that had had a patellar replacement, even though the alignment was relatively good after one year (Fig. 2-A); this knee did not have pain (Fig. 2-B). The patella that was not replaced had evidence of erosion, with narrowing of the joint space, after six years; however, the excellent alignment that had been evident at one year (Fig. 2-C) had been maintained (Fig. 2-D). The patient had pain in this knee on standing and on ascending or descending stairs as well as tenderness of the patellofemoral joint, even though the rheumatoid arthritis was under control. This patient had not had a reoperation at the time of the latest follow-up evaluation.
Our present and previous studies differ from others2-4,5,8-10 in that the results of total knee arthroplasty with and without a patellar replacement were compared prospectively in the same patients. Our study design minimized differences between the two knees being compared, allowing differences between the operative results to be revealed.
Neither our previous investigation11 nor the present one revealed a significant difference, with the numbers available, between the knee scores of The Hospital for Special Surgery after total knee arthroplasty with a patellar replacement and those after such an arthroplasty without a patellar replacement. It was considered possible that the lack of influence of a patellar replacement on the results of total knee arthroplasty was attributable to the fact that the knee score of The Hospital for Special Surgery largely assesses pain at rest and with walking. In addition, the pain score is a prominent portion (as much as 30 points) of the total score. Thus, this score was apparently not sensitive enough to detect the effect of patellar replacement in patients who have rheumatoid arthritis.
Consequently, in the present study, we also assessed pain on standing and on using stairs as well as tenderness of the patellofemoral joint. Unlike the knee score of The Hospital for Special Surgery, these tests revealed abnormalities after total knee arthroplasty without a patellar replacement but not after arthroplasty with a patellar replacement. Accordingly, the present, more detailed evaluation provided evidence of a difference in clinical outcome between total knee arthroplasty with and without a patellar replacement. Patellofemoral pain after total knee arthroplasty without a patellar replacement has been reported by other authors2,5 and supports our findings.
Erosive changes were noted on the articular surfaces of the patella that had not been replaced, irrespective of whether there was good alignment of the patella; this finding is consistent with that of Steinberg et al. In the present series, however, the synovitis was not associated with joint effusion and was not resistant to postoperative treatment. In addition, the recurrence of synovitis was confirmed in the patient who had a reoperation. In the knees that had a patellar replacement, a lack of patellar alignment was not accompanied by articular pain, suggesting the clinical usefulness of a patellar prosthesis.
We found that relatively good patellar alignment was maintained postoperatively if the preoperative alignment had been excellent. However, if the alignment had been poor before the operation, it gradually worsened even if excellent alignment had been achieved immediately after the operation. Thus, lateral retinacular release was not very useful for maintaining patellar alignment, although it may be better to perform this procedure than none at all. In addition, even if excellent patellar alignment had been maintained, pain sometimes developed when arthroplasty had been performed without a patellar replacement. Complications related to patellar replacement have been reported2,3,10, but none were associated with the procedure in the present series.
In conclusion, these findings suggest that, in order to alleviate pain and to prevent erosive changes of the patella, it is advisable to perform a patellar replacement as part of total knee arthroplasty in patients who have rheumatoid arthritis.