Total Prosthetic Device
The semiconstrained Coonrad-Morrey prosthesis (Zimmer, Warsaw, Indiana) with fixation of both stems with cement was used for all patients. This device is the third-generation design and incorporates a flange, a loose hinge, and a porous coating of the distal aspect of the humeral component and the proximal aspect of the ulnar component. The ulnar component comes in two sizes. Standard humeral stems come in two diameters and are also available in lengths of ten, fifteen, and twenty centimeters. Thirty-seven patients had insertion of a fifteen-centimeter stem; two, a ten-centimeter stem; and two, a twenty-centimeter stem. The selection of the diameter and length of the stem was at the discretion of the surgeon. Usually, shorter implants were selected for smaller humeral intramedullary canals. No custom devices were used for any of these patients. The prosthesis is made of a titanium alloy. The articulation used in this study consists of three high-density-polyethylene bushings rotating around a titanium-alloy pin that is secured with a split ring. It allows a so-called play of 7 to 10 degrees of varus and valgus and 7 degrees of rotation25. Since 1993, a cobalt-chromium pin has been used instead of the original titanium pin. The anterior flange at the distal end of the humeral stem is designed to resist posterior displacement and external rotational stresses on the humeral component when incorporated with bone graft. The surfaces of the distal aspect of the humeral component and the proximal aspect of the ulnar component were treated with sintered beads to improve fixation of the cement and to allow bone ingrowth behind the flange. On the basis of observations of late fractures of the ulnar stem, the layer of sintered beads on the ulnar component was replaced with a precoat of methylmethacrylate in 1991 (Fig. 1). Seven patients had a precoated ulnar component.
Selection of the Patients
From 1981 to 1993, forty-one consecutive patients who had post-traumatic osteoarthrosis or dysfunction of the elbow and who met our inclusion criteria all were managed with the same design of joint replacement (Table I). Patients who had an acute injury of the joint of less than three months' duration, had rheumatoid arthritis, had had a previous total joint replacement, or had a non-union were excluded20. Patients who had ankylosis of the elbow with an arc of flexion of less than 30 degrees also were excluded. They are believed to constitute a different subset of patients because of the soft-tissue contracture, which necessitates marked additional dissection, and the typical absence of pain.
There were thirty-one women and ten men with an average age of fifty-seven years (range, thirty-two to eighty-two years) at the time of the operation. The injury involved the dominant extremity of twenty-nine patients. The primary diagnosis was post-traumatic osteoarthrosis in thirty-eight patients and post-traumatic dysfunction due to a flail elbow in three. Of the latter three patients, one had had a complete resection of the joint for débridement at the site of a comminuted fracture; one, for the treatment of an infection at the site of an impending non-union; and one, for salvage of a failed osteosynthesis. The average time from the original fracture to the joint replacement was sixteen years (range, three months to sixty-four years). All but three patients had had at least one previous operation (average, 2.3 procedures; range, zero to seven procedures). The twenty-three patients who were less than sixty years old had had an average of 2.7 previous procedures, and the eighteen who were sixty years old or more had had an average of 1.8 previous procedures. The primary indication for the index procedure was pain for thirty-six patients (88 per cent); a reduced, painful range of motion for two (5 per cent); and dysfunction with a flail elbow for three (7 per cent).
Previous Complications
At the time of presentation, six patients had mild-to-moderate ulnar neuropathy as the result of either the initial injury or a previous operation, one patient had a persistent radial-nerve palsy due to a complete traumatic laceration, and one patient had had an infected non-union that had been treated with resection.
Operative Technique
The operative technique has been described elsewhere19; however, some features should be emphasized. The ulnar nerve was always transposed anteriorly in a subcutaneous pocket if this had not been performed previously. The Mayo triceps-sparing approach was used in all but two patients, who had a triceps-splitting approach at the discretion of the surgeon. The Mayo exposure includes the release and lateral reflection of the triceps from the olecranon in continuity with the ulnar periosteum and the fascia of the forearm along with the anconeus. The collateral ligaments were detached from the epicondyles, which allowed complete exposure of the elbow joint. An intramedullary injection system was used for optimum insertion of the cement, to which one gram of tobramycin was added from 1983 on. An important element was the placement of a bone graft between the anterior flange and the distal part of the humerus to resist posterior displacement and rotational stresses on the humeral component after ingrowth. In one patient, the deformed distal aspect of the humerus was in contact with the flange, making insertion of a bone graft unnecessary. Beginning in 1985, no effort to repair the collateral ligaments was made. Marked deficiencies in bone stock with absence of one epicondyle or both did not change or complicate the implantation of the humeral component. Only the humeral diaphysis is required to obtain secure fixation of this device. Therefore, no attempt was made to reconstruct the condyles. The semiconstrained, hinged prosthesis yielded immediate stability, making repair of the collateral ligaments unnecessary. At the end of the procedure, the extensor mechanism was reattached to the olecranon with two number-5 non-absorbable sutures, allowing immediate use of the joint. The average duration of the operation was two hours and twenty-three minutes (range, one hour and fifteen minutes to seven hours). This consecutive series consisted of the patients of five staff surgeons.
Clinical Evaluation
The results were rated with use of the Mayo elbow performance score18, with a maximum of 100 points. The components of this score include 45 points for no pain, 20 points for flexion of more than 100 degrees23, 10 points for joint stability, and 25 points for the ability to perform selected daily activities. The results were defined as excellent (90 to 100 points), good (70 to 89 points), fair (50 to 69 points), and poor (less than 50 points). Patients who had mild pain could not be considered to have an excellent result, and those who had moderate pain could be considered to have, at most, a good result, even if the maximum score was given for all of the other components. Excellent and good results were considered satisfactory, and fair and poor results were considered unsatisfactory.
Follow-up
All forty-one patients were followed clinically for at least two years, with an average duration of five years and eight months (range, two to twelve years). Two patients died of unrelated causes nine years and two years and four months after the index operation. One of these patients had been examined clinically and radiographically at five years and six months and the other, at two years. Radiographs made at least two years postoperatively (average, five years and one month; range, two to twelve years) were available for thirty-nine of the forty-one patients. Sixteen patients returned to our institution for examination at the time of the study, fourteen were assessed by a local orthopaedic surgeon, and eleven responded to a questionnaire and sent radiographs made at a local radiological institute. The questionnaire allowed the patient to grade the result as much better, better, the same, or worse. The range of motion of the elbow of the eleven patients who were assessed with use of a questionnaire was determined from measurements made at previous examinations at our institution or by a local orthopaedic surgeon (at an average of four years and four months before completion of the questionnaire) as well as on the basis of a diagram that allowed the patient to document his or her perceived arc of motion. The range of motion was not known for three of these eleven patients. The most recent range of motion after a minimum of two years of follow-up was used to obtain a Mayo elbow performance score for these eleven patients.
Radiographic Evaluation
The osseous loss seen preoperatively and immediately postoperatively and the preoperative osseous and articular deformities were recorded. Nine (22 per cent) of the forty-one patients had a substantial articular deformity resulting in a carrying angle of more than 30 degrees. Eight patients (20 per cent) had a marked osseous deformity (Figs. 2-A, 2-B, 3-A, 3-B, 3-C, 4-A, 4-B). Over-all, fourteen patients (34 per cent) had a severe articular or osseous deformity. Ten (24 per cent) of the forty-one elbows were subluxated preoperatively, and seven (17 per cent) were dislocated. Twelve patients (29 per cent) had extensive loss of bone stock with loss of both condyles preoperatively (Figs. 3-A, 3-B and 3-C). Resection of one condyle or both for implantation of the prosthesis into the deformed humerus was necessary in three other patients.
Postoperatively, the integrity of the prosthesis and signs of loosening were evaluated. Type-0 radiographic loosening indicates a radiolucent line less than one millimeter thick and involving less than 50 per cent of the interface; type-I, a radiolucent line at least one millimeter thick and involving less than 50 per cent of the interface; type-II, a radiolucent line more than one millimeter thick and involving more than 50 per cent of the interface; type-III, a radiolucent line more than two millimeters thick and around the entire interface; and type-IV, gross loosening24.
Statistical Analysis
The relationship between discrete variables was determined with the Wilcoxon signed-rank test. Factors were tested statistically for association with a satisfactory or unsatisfactory result with use of univariate analysis with the chi-square test. A p value of less than 0.05 was considered significant.
At the latest follow-up examination, sixteen patients (39 per cent) had an excellent result, eighteen (44 per cent) had a good result, five (12 per cent) had a fair result, and two (5 per cent) had a poor result, according to the Mayo elbow performance score (Table II). The average score was 82 points (range, 20 to 100 points) postoperatively compared with 44 points (range, 15 to 75 points) preoperatively (p < 0.0005). Over-all, thirty-four patients (83 per cent) had a satisfactory objective result, and seven (17 per cent) had an unsatisfactory objective result. Every patient who had a satisfactory objective result also described the subjective outcome as satisfactory (much better or better). In addition, two patients who had an unsatisfactory objective result rated the elbow as satisfactory because of improvement compared with the previous status. Both patients indicated that they had moderate pain. Over-all, of the thirty-eight patients who had a functioning implant (two patients had had a resection arthroplasty, and one patient had an unrevised broken ulnar component), thirty-six (95 per cent) had a satisfactory subjective result at the latest follow-up examination. One patient, who had an infection, and one patient, who had a broken ulnar component, were dissatisfied with their status.
Pain
Preoperatively, thirty-seven patients (90 per cent) had moderate or severe pain. Postoperatively, twenty-nine (76 per cent) of the thirty-eight patients who had a functioning implant had no or only mild pain. The difference between the preoperative and postoperative levels of pain was significant (p < 0.0005).
Motion
Preoperatively, the average arc of flexion was 40 degrees (range, -8 to 85 degrees) to 118 degrees (range, 70 to 150 degrees), and the average rotation was 61 degrees (range, -5 to 90 degrees) of pronation to 62 degrees (range, -10 to 90 degrees) of supination. Postoperatively, the average arc of flexion improved to 27 degrees (range, 0 to 70 degrees) (p < 0.001) to 131 degrees (range, 105 to 150 degrees) (p < 0.0005). The average rotation also improved, to 67 degrees (range, 7.5 to 90 degrees) of pronation (p = 0.05) to 67 degrees (range, 3 to 90 degrees) of supination.
Preoperatively, nine patients had a moderately limited arc of flexion of 50 degrees or less (average and standard deviation, 36 ± 8 degrees). These nine patients had improvement to an average arc of 97 ± 21 degrees (range, 60 to 120 degrees) (p = 0.008). A range of 0 to 140 degrees of flexion was attained intraoperatively in one patient, but adhesions developed without explanation and limited the arcs of flexion and rotation of the forearm to 40 degrees each. The adhesions between the triceps muscle and the distal aspect of the humerus were freed, and the radial head was resected five months after implantation of the prosthesis. Flexion improved to an arc of 35 to 105 degrees and rotation improved to an arc of 75 degrees of pronation to 35 degrees of supination.
Instability
Only sixteen patients (39 per cent) had a stable elbow preoperatively; thirteen had moderate varus-valgus laxity of 5 to 20 degrees; and twelve, including the three patients who had a flail elbow, had a grossly unstable elbow (Table I). Postoperatively, all thirty-eight of the functioning elbow prostheses were stable.
Function
Preoperatively, the patients were able to perform an average of 2.8 ± 2.0 of the five daily activities assessed22. After implantation of the prosthesis, the function returned almost to normal: the patients were able to perform an average of 4.8 ± 0.4 of the five daily activities (p < 0.0005).
Radiographic Findings
All grafts behind the flanges were found to have incorporated by the time of the latest follow-up. There was no aseptic loosening. Aseptic non-progressive radiolucent lines at least one millimeter thick and involving less than 50 per cent of the interface (type I) were observed on the humeral side in two (5 per cent) of the thirty-eight patients who had a functioning implant and on the ulnar side in five patients (13 per cent). No type-II, III, or IV aseptic radiolucent lines were observed. Two patients had a periprosthetic infection: the implant in one was measurably loose and the implant in the other was found to be well fixed at the time of removal. Slight periarticular ossifications (less than one centimeter) were observed in only four patients, with no clinical implications or effect on the outcome.
Complications and Reoperations
Eleven patients (27 per cent) had at least one major complication (Table III), and nine of them (22 per cent of the series) had an additional operation. One patient (Case 14) (2 per cent) had three complications, all of which were treated operatively.
Infection developed in two patients (5 per cent), as mentioned previously. Treatment consisted of a resection arthroplasty. Two patients were seen for particulate synovitis and osteolysis due to debris associated with worn bushings (Figs. 3-A, 3-B and 3-C). Both were managed with synovectomy, débridement, and exchange of the bushings. One patient (2 per cent) had impingement between the radial head and the prosthesis and was managed successfully with resection of the radial head. A ruptured triceps also was noted in this patient after a triceps-splitting approach and was repaired. All reoperations performed for a mechanical failure, a fracture, or worn bushings resulted in a satisfactory outcome.
The ulnar component broke in five patients after two to nine years (average, four years and four months) (Figs. 4-A and 4-B). Before breakage, all five patients had an excellent result with an asymptomatic, essentially normal elbow. The cause of failure was regular lifting of weights of more than fifty kilograms (two patients), landing on the outstretched arm after a fall downstairs (one), and striking a golf club into the turf (one). No specific event or excessive activity could be identified for the breakage in one patient. By the time of the most recent follow-up examination, four of the five ulnar components had been exchanged, resulting in two good and two excellent outcomes (Table IV) at four, twenty-seven, twenty-two, and thirty-nine months after the replacement. At each of the four revisions, the well fixed ulnar component was removed, enlarging the envelope of the well fixed cement, and the new component was reinserted into the previously stable cement cavity. The revisions involved an average of less than seventy minutes of tourniquet time. There were no operative or delayed complications from the reoperation. It is interesting to note that all ulnar components that broke were in patients who were excessively active. While restoration of function typically represents a successful outcome, in the present study the high degree of functional restoration was the cause of failure.
Factors Influencing the Results
Mechanical complications, such as breakage of the ulnar component and wear of the bushings, were attributed to overuse of the elbow in three patients, who performed strenuous, repetitive labor, and to considerable traumatic injury in two. It is of interest to note that five of the seven patients who had a mechanical failure had had marked preoperative deformity of the elbow. Over-all, ten patients were considered to have had marked angular deformity before the operation. Thus, the rate of complications (five of ten) for the elbows that had had a preoperative deformity was significantly higher than the rate for the other elbows (6 per cent [two of thirty-one]; p = 0.02). The deformity was corrected with implantation of the prosthesis. In both patients who had worn bushings, almost the same deformity recurred, possibly as a result of the loads imparted by the distorted and shortened soft tissues. Eight (35 per cent) of the twenty-three patients who were less than sixty years old had a complication. Accordingly, the proportion of satisfactory results in this age-group (eighteen [78 per cent] of twenty-three) was lower than that in the patients who were sixty years old or more (sixteen of eighteen). However, we could not detect a significant difference because of the small sample sizes.
All of the patients in this series had marked destruction of the elbow joint. A high proportion of them also had considerable loss of bone, instability, and deformity. In addition, twenty-three patients (56 per cent) were less than sixty years old at the time of the operation. All but three of the forty-one had had at least one previous operative procedure, and most had had several previous procedures. Management of this patient population is very challenging, with only a few options even theoretically available for operative treatment. Arthrodesis reliably relieves pain16 and restores a strong extremity. However, because it results in great functional impairment26, arthrodesis of the elbow rarely is considered a viable option6,21. Interposition arthroplasty may be considered for a young patient, particularly one who has stiffness. Restoration of motion and relief of pain can be achieved with a reasonable but unpredictable rate of success8,13,17,29. However, this procedure is technically demanding, with an even higher rate of complications than that associated with semiconstrained total elbow replacement17. Interposition arthroplasty also is not considered suitable for patients who perform strenuous physical labor13. In addition, marked loss of bone is a contraindication to this procedure8. Urbaniak and Black30 reported the results of allograft replacement of the entire elbow joint after six months to six years. Seven of ten patients had a satisfactory result. The rate of pain relief was high, with only minimum or mild symptoms in all patients. Complications occurred in three of the ten patients: two patients had a non-union, and one had extensive resorption of the graft associated with chronic dislocation of the elbow. Continued degenerative changes and fragmentation of the allograft were seen in some of the elbows after two years. However, other authors reported less favorable results after other total elbow replacements performed with allografts5.
The results of total joint replacements with highly constrained designs in the 1970s were disappointing because of high rates of loosening4,9,24. Although a decrease in the rates of loosening was reported after the introduction of semiconstrained and unconstrained replacement devices3,6,10,20,27, those reports dealt almost exclusively with the treatment of rheumatoid arthritis. There is very little information regarding total joint replacement with an unconstrained or semiconstrained device for the treatment of post-traumatic osteoarthrosis. Inglis and Pellicci11, in 1980, reported little improvement in nine patients who had been managed with a semiconstrained Pritchard-Walker and triaxial implant. Lowe et al.15 reported a satisfactory result in only one of seven patients who had an unconstrained device: four implants were unstable, one was associated with an infection, and one was loose. In 1984, Soni and Cavendish28 reported a good or excellent result for only three of eight patients who had an unconstrained implant. Figgie et al.7 were the first of whom we are aware to report a positive experience with a semiconstrained triaxial prosthesis with custom-designed stems for the treatment of post-traumatic conditions of the elbow: eight of their nine patients had a satisfactory result. However, the follow-up was limited to 7.6 years, and different designs with variable stems as well as fixation with or without cement were used7. From the same institution, Kraay et al.14, in 1994, reported their experience with a linked semiconstrained implant. Of 113 patients, eighteen had been managed for post-traumatic osteoarthrosis, non-union, or fracture. The results for these patients were disappointing, with a five-year rate of survival of the implants of only 53 per cent. There were five loose humeral components and two infections among the eighteen patients.
The present study represents our experience with a single, non-customized, semiconstrained implant. The follow-up of forty-one patients for as long as twelve years is the most comprehensive in the literature, to our knowledge. The results show that severe post-traumatic osteoarthrosis or dysfunction can be treated reliably with a semiconstrained device. Although some patients were followed for as long as twelve years, we must note that the average duration of follow-up was a little more than five years. In light of this fact, the satisfaction expressed by the patients was particularly gratifying. Thirty-six (88 per cent) of the forty-one patients considered the condition of the elbow to be improved. All thirty-eight elbows in which the implant had been retained were stable, and function had been reliably restored. Thirty (73 per cent) of the forty-one patients had relief of pain. However, even though the rate of pain relief was considered rather high, it must be recognized that it was not as high as that reported after the treatment of rheumatoid arthritis18; in that study, 93 per cent of fifty-four patients reported no or only mild pain. This may be explained in part by the severe preoperative condition of the elbows as a result of the initial injury and the number of previous operations in the current study. Only two patients in the present study had non-progressive radiolucent lines, which did not exceed more than 50 per cent of the prosthesis-bone interface. No component was found to be loose after a maximum duration of follow-up of twelve years. This represents a marked improvement compared with the previous constrained implants, which had high rates of loosening4,24,28.
The rate of complications was notable and should serve to outline the limitations of total joint replacement. Breakage of the ulnar component and wear of the bushings were, in part, associated with patients' severe non-compliance with the postoperative protocol and with marked preoperative deformity. The non-compliance included working as a lumberjack, lifting in excess of 100 kilograms, lifting forty-kilogram bags of horse feed, and the like. Participation in strenuous physical activity and anticipated non-compliance by patients who are less than sixty years old are therefore, in the opinion of the senior one of us (B. F. M.), relative contraindications for this procedure. Currently, patients are advised, as before, not to lift an object that weighs more than 4.5 kilograms and not to lift, on a repetitive basis, an object that weighs more than one kilogram. Caution against excessive use that may lead to fracture is specifically noted. It might also be noted that all fractures occurred through the sintered beads and there were no fractures of the precoated ulnar components. However, in the final analysis, the surgeon must reconcile the needs of a patient who is markedly limited by dysfunction of the elbow that could be dramatically improved by replacement with the realization that this decrease in limitations sometimes results in excessive activity and failure of the implant.
One major advantage of this implant is its ability to correct deformity. However, our experience suggests that it does so at the expense of increased rates of wear. Nevertheless, the 78 per cent rate of successful results for the twenty-three patients who were less than sixty years old suggests that joint replacement can be a viable option for these patients, particularly if they have low demands and a low level of activity. Despite these favorable results, the indication for total joint replacement should still be very restrictive for patients who are less than sixty years old. Unfortunately, other reconstructive procedures are limited. Careful attention to operative technique and experience with the procedure enhance the likelihood of a satisfactory outcome in this challenging patient population.