Bath - Wessex Total Ankle Replacement:
Eight-years results

D. BALTAS[1], G.A. MACHERAS[2], G. TERZIS[1], K. KATEROS[3], S. KOUTSOSTATHIS[2]
[1]Royal United Hospital, Bath, United Kingdom
[2]2nd Orthopaedic Department, 1st IKA Hospital, Athens
[3] General Peripheral Hospital of Leivadia

Mailing address:
D. Baltas
1 Zinodotou St, Pangrati
116 34 Athens

ABSTRACT
We report the results of a prospective study of ninety-two total ankle arthroplasties performed using a multiaxial, unconstrained, cemented prosthesis on seventy-nine patients, over a 15-year period. Fifty-two of them (forty-five patients) have been reviewed clinically and radiologically after average follow-up of 8.1 years. Of these thirty-six (69.2%) had satisfactory pain relief. There was substantial improvement in the mobility in 28 (53.9%) of them. The patients' assessment of the operation was satisfactory. Thirty-five replacements (67.3%) were considered as complete success.
The complications encountered in all 92 arthroplasties are discussed in details also. The overall rate of complications in the group as a whole was 31%. These included superficial wound infection, aseptic loosening, infection and talar collapse.
Key words: Ankle, Osteoarthritis, Rheumatoid arthritis, Total ankle arthroplasty.

MATERIALS - METHODS
92 ankle replacements (79 patients) were performed between 1980 and 1990, at Royal United Hospital in Bath. The Bath ankle replacement is an unconstrained, multiaxial prosthesis. It consists of a tibial component made of Ultra High Molecular Weight Polyethylene (UHMWP). It presents a shallow hemi-spherical concavity inferiorly which articulates with the talar component, which is a talar steel dome on a base of variable height (figures 1,2). Both components have been designed for use with cement.
The spherocentric concept was prompted by the "ball and socket" ankle adaptation seen in congenital tarsal fusion and occasionally acquired fusion of Still's disease and long-standing rheumatoid arthritis when the ankle assumes additional sub-taloid function. The key feature of the prosthesis is the ability to move freely in any direction therefore permitting true ankle and pseudo sub-talar movements. This feature reduces the problem of very accurate placement, which is inherent in a uniaxial or strictly anatomical arthroplasty. The ability of the prosthesis to allow inversion, eversion and rotation movements of the ankle, minimizes the shearing forces at the bone-cement prosthesis interface. The variable depth talar component is designed to restore appropriate collateral ligament length and diminish malleolar impingement and pain.

1 2
Picture 1 and Picture 2. Bath-Wessex Total ankle replacement

The prosthesis is inserted after the ankle joint has been exposed through an anterolateral incision. Skeletal traction is used to distract the ankle and to restore the correct tension in the collateral ligaments. The appropriate bone cuts are made using templates (figure 3) and then trial components are inserted (figure 4). The correct tension of the collateral ligaments is maintained using a talar component of appropriate size. Finally the definite components are inserted. The wound is drained and meticulously closed. Three doses of broad-spectrum antibiotics is given, one dose pre-operatively, one dose immediate post operatively and one six hours after. A short posterior leg cast is applied with the foot in neutral position. The patient rest with the leg elevated until the drains are removed. When wound is satisfactory, the cast is removed and ankle exercises as well as weight-bearing using crutches are commenced (figures 5,6).
The average age of the patients was 59.7 years (range 23 to 84 years). Of the total 92 arthroplasties 67 (72.8%) of them had rheumatoid arthritis, 7 arthroplasties (7.6%) had osteoarthritis, 13 arthroplasties (14.1%) had post-traumatic arthritis and 5 arthroplasties (5.4%) had other types of arthritis (3 arthroplasties had psoriatic arthritis and 2 arthroplasties had Still's disease). All patients were grouped into two categories. A general category of inflammatory arthritis (Group A), 72 arthroplasties (78.2%), that included rheumatoid arthritis, psoriatic arthritis and Still's disease, enabled a comparison to be made between the patients who had inflammatory arthritis and those who had degenerative arthritis (Group B), 20 arthroplasties (21.8%).
At review eight (8.7%) of the 92 arthroplasties, had been removed completely and eight more arthroplasties (8.7%) had been revised. These 16 arthroplasties have not been included in the study of surviving arthroplasties but were included in the discussion of the complications. The reason for removal was severe loosening or infection. The reason for revision was aseptic loosening of the prosthesis. Eight patients (ten arthroplasties) had died. One patient had a below-knee amputation as result of an infected total knee replacement. From the remaining 63 arthroplasties we were unable to review a further thirteen (11 patients). All these arthroplasties were functioning at the time of review.
Fifty-two arthroplasties were reviewed. The average age was 61.3 years. Of these 36 (69.2%) had inflammatory arthritis and 16 (30.8%) had degenerative arthritis. The average age in the group with inflammatory arthri-tis was 60 years and 63.1 years in the group with degenerative arthritis. There were 18 (34.6%) men and 34(65.4%) women in total. In the group with degenerative arthritis there were 7 (43.7%) men and 9 (56.3%) women. In the group with inflammatory arthritis there were 11 (30.5%) men and 25 (69.5%) women. The average follow-up time for all the arthroplasties was 8.1 years (2.8-12 years). The group with degenerative arthritis had an average follow-up of 7.5 years (2.8-12 years) whereas the group with inflammatory arthritis had an average follow-up of 8.2 years (3.1-12 years).
All the 52 surviving arthroplasties were reviewed and assessed using a standardised form. The same form had been used as pre-operative assessment at the time the patient had the first operation. Subjective data were collected relative to pain, discomfort during specified activities and patient's satisfaction from the procedure. Objective data were obtained relative to the range of motion, stability, effusion, the need for walking aids and the ability to perform specific tasks such as stair climbing and rising from a chair without assistance
All the reviewed arthroplasties were assessed radiologically with plain radiographs. These included weight-bearing AP and lateral views of the ankle and lower tibia, lateral views of the ankle in plantar-flexion and dorsi-flexion. The position of the components was assessed with radiographs that were taken after the operation and at review time, to determine the accuracy of insertion during the primary arthroplasty as well as the shift that occurred in the position of the components at review time. This was done measuring the degree of valgus or varus position and plantar or dorsiflexion of the components in respect to axis of the tibia. The range of motion and the degree of talar collapse was measured radiographically as well.
Results were considered excellent if the patient evidenced relief of pain, improved ambulatory status, functional range of motion and an overall subject report of improvement. Results were considered good if pain was no worse than mild to and if ambulatory capacity and functional range of motion were improved with or without gain in ankle joint range of motion. Results were fair if pain was mild to moderate, patient evidenced loss of motion, no improvement in ambulation and an overall disability no greater than before surgery. Poor results were failures, that patients were made worse by surgery and patients whose prosthesis were removed.
Forty-two patients (75%) had one more joint at the lower limbs involved also. These were almost exclusively from the inflammatory arthritis group. Thirtyfive patients (97%) from group A had more than one joint involved, as opposed with group B where 7 patients (43.7%) had the disease affecting more than one joint. There was involvement of the hip joint in 31 patients, of the knee in 27 patients, of the joints of the hind foot in 19 patients, of the joints of the forefoot in 24 patients and 22 patients had all the joints mentioned above affected. In thirty patients there was involvement of the joints of the foot. Twenty-four belong to group A and six to group B.

3 4
Picture 3. Intra-operative picture of the exposed ankle joint, showing the bone cuts.
Picture 4. Intra-operative picture of the exposed ankle joint with the trial components in situ.

RESULTS
In all patients reviewed the major symptom prior to surgery was pain. 34 arthroplasties (65.4%) had severe pain pre-operatively. At review only 4 arthroplasties (7.7%) had severe pain, whereas 36 arthroplasties (69.2%) had no or little pain that did not required the use of analgesics (table 1).
The clinically measured range of movements of dorsiflexion, plantarflexion and combined movements in both groups is shown in table 5. The range of clinically measured combined movement was from 045° (mean 22°). This was less than before the arthroplasty (28°). The radiologically measured movement was from 249° (mean 18°). The patients in both groups had no improvement in the range of motion after the operation. Both groups presented loss of movements. In the group with degenerative arthritis the mean range after the arthroplasty was reduced from 31° to 22°. In the group with inflammatory arthritis the combined mean range was reduced from 27° to 22°. It seems that all the loss of motion was mainly in plantar flexion whereas in dorsiflexion there was no change.
All 52 arthroplasties were had very restricted mobility pre-operatively. None of them could walk more than 400 meters. At review 28 arthroplasties (53.9%) could walk more than 400yds and 15 of these (28.8%) could walk more than 1000 meters (table 2). In 24 arthroplasties (46.1%) ambulation was possible with no external support (table 3) and 35 (67.3%) could arise from a chair without assistance. Fourteen (27%) could climb stairs un-aided and 29 (55.7%) needed the use of one banister.
The results were excellent in 13 arthroplasties, good in 22, fair in 13 and poor in 4 arthroplasties (table 6). Patients in group A performed better than those in group B. Twenty-six patients (72.2%) in group A had excellent or good result, as opposed to 9 patients (56.2%) that belong to group B. The radiographic range of motion in all 52 arthroplasties was 2-39° (mean 18°) combined dorsi-flexion and plantar-flexion.
Analysis of the positioning of the components showed that the tibial components were inserted at a mean of 3° in valgus (range (-3°)-12°) and at neutral position in the lateral plane (mean 0°, range (-23°)-26°). The talar components were inserted at a mean 4° in valgus (range (-8°)-16°) and were plantar-flexed 2° (range (-23°)-26°). At review time these values were maintained (table 4.)
Radiolucent lines wider than 2mm were present in more than 75% of the arthroplasties reviewed. The amount of talar collapse was 0-21mm (mean 5mm). In the group with inflammatory disease it was 0-21mm (mean 5mm). In the group with degenerative disease it was 0-9mm (mean 4mm). The complications encountered were wound breakage, wound infection, infection of the arthroplasty, loosening of the components and talar collapse. At maximal review time eight arthroplasties (8.7%) had been removed and eight more arthroplasties (8.7%) had been revised, giving an overall failure rate of 17.4% at 7.9 years. Of these arthroplasties all eight that were removed belong to the group with inflammatory arthritis. Six of the arthroplasties that were revised belong to the group with inflammatory arthritis and two were from the group with degenerative arthritis. The failure rate was 21% for the group with inflammatory arthritis and 10% for the group with degenerative arthritis.

The cause of all the revisions was aseptic loosening of one or both components with severe pain. There was loosening in both components in seven of the eight arthroplasties that were revised. In one ankle there was loosening only of the talar component. The revision procedure was carried out 2-5 years after the primary procedure.

Picture 5. Severe arthritis of the ankle joint in a patient with rheumatoid arthritis.


Picture 6. The same patient as in Pictureς 5 after having the ankle replaced with Bath-Wessex
Total ankle replacement.

7
Picture 7. Lateral radiographs of the ankle joint in flexion and extension.


Picture 8. Loosening of the talar component, due to severe
collapse of the dome and body of the talus.

Severe symptomatic loosening of the prosthesis with talar collapse was the main reason for removing the prosthesis in six of the eight arthroplasties (figure 4). All these six arthroplasties had been previously revised for loosening. The time of the final surgery to remove the components was 3-6 years after the primary procedure and 1-3 years after the revision procedure. Two arthroplasties had the components removed as result of infection. None of them underwent any further surgery to reconstruct the ankle joint.
Another complication that occurred as the result of talar collapse and subsequent sinkage of the prosthes-is was malleolar impingement (figure 7). It occurred in 7 arthroplasties (7.6%). All 7 patients underwent further surgery.
There were problems with the healing of the wound in 4 arthroplasties (4.3%). Three of them developed superficial wound infection with wound breakage. These were treated with bed rest and antibiotic therapy. Three healed satisfactory and one required split skin graft. Three of them were from the group with inflammatory disease and one from the group with degenerative disease. The final result of the arthroplasty was not compromised in by this complication and the range of movements and pain relief were the same in these four arthroplasties as in the group as a whole.
Two arthroplasties developed deep infection and had to be removed (2.1%). Both belonged in the group with inflammatory arthritis. In one of them the patient developed chronic osteomyelitis of the os calcis with discharging sinus and required further surgery.

DISCUSSION
Arthrodesis is currently recognised as the accepted procedure for treatment of severe arthritis of the ankle joint[1,2,7,15,19]. It has been a commonly performed procedure since it was described by Alberth in 1878 (as cited by Fjermeros and Hagen)6. However the results are not uniformly good. More than 25 different operating techniques have been described, none of which provide a satisfactory success rate. In reported series there were good or excellent results in 59% to 75% of the patients[11,13]. In other reported large review series the results were rather disappointing with unsatisfactory results at 31% of the patients[3,9,12,13]. Complication rates have been reported between 40% and 60%[3,12,14]. The problems of arthrodesis include pseudarthrosis (15%-20%), increased stress on subtalar and ChopartΥs joints, which may lead to painful arthrosis of the involved joints, gait disturbance and inability to walk[12]. Furthermore in patients with rheumatoid arthritis there is involvement and damage of other joints. In these patients arthrodesis of the ankle may stress the adjacent joints, which are already compromised by the disease. For these patients ankle replacement may offer significant advantage.
The reports published to date regarding the results of total ankle replacements have been rather disappointing. Pain relief ranges from nil to 47.6%[4,10,18]. In our series however, 12 arthroplasties had complete pain relief and 38 arthroplasties (73%) had good or excellent pain relief with no or little pain requiring no analgesics. This compares favourably with the best-reported results for ankle arthrodesis where pain relief is reported in 59% to 75% of the patients[2,15,17,19]. The pain relief was equally satisfactory in both groups. The group with inflammatory arthritis had better rate of total pain relief, 11 patients (31.4%) as opposed with the group with degenerative disease, 1 patient (5.8%) in the group with degenerative disease.
Despite satisfactory pain relief none of the groups experienced improved range of movements. On the contrary the range of clinically measured combined movement was less after the arthroplasty, with the mean range of 28° mostly plantarflexion. However it is similar with reported results of ankle arthroplasties of different design[8,10,14,18]. The radiologically measured movement was from 239° (mean 18°). This is the same as the best result reported so far for radiologically measured movement7. All the loss of movement was in plantar-flexion.
Despite the loss of movements the overall improvement in the mobility of the patients is significant. Twenty-eight patients (53.9%) could walk more than 400yds, as opposed to none before the operation. The patients in group B performed better in this subject. This can be explained by the fact that those patients had fewer joints involved, they had better bone-stock and their muscular strength was much better than those in patients in group A. In twenty-four arthroplasties (46.1%) ambulation was possible without external support after the procedure as opposed to 18 arthroplasties (34.6%) before the arthroplasty. We believe that the mobility is improved mainly because the pain has been reduced in the ankle joint, allowing the foot to move more freely. This is especially true in rheumatoid patients.
The overall functional result obtained by the various prostheses is not encouraging at present. It has been shown that the functional result, when measured by gait analysis, was much worse than expected[12]. Other workers report a generalized but slight and unpredictable improvement in function[8,10,18]. In our series however we had 35 (67.3%) patients with excellent or good result. From those 26 arthroplasties (72.2%) belong in group A and 9 (56.2%) in the group B.
One major disadvantage of ankle arthroplasty is the high percentage of complications. Most of the complications occurred in the group with inflammatory arthritis, which was anticipated in view of the poor bone and skin quality these patients have.
The overall failure rate was 17.4%. However only 2 of the 16 arthroplasties that were removed or revised belonged to the group with degenerative arthritis (failure rate 10%). The other 14 arthroplasties were from the group with inflammatory arthritis (failure rate 21.7%). Malleolar impingement occurred in 7 cases (7.6%). This is the same reported in other large series[17]. It occurred exclusively in the group with inflammatory arthritis. In all these cases the impingement was the result of talar collapse and valgus shift of the talar component. It was treated with excision of the impingement.
In predicting the long-term survival of ankle prosthesis, radiographic lucency of the bone cement interface is significant. Significant radiolucent lines (>2mm) have been reported to be as high 85%[5,8,10]. Our results agree with those workers, having a 75% incidence of radiolucent lines around the prosthesis. Talar collapse and sinkage of the prosthesis still remains a major problem. It has been reported to be as high as 50%[4]. In the reviewed arthroplasties in our series however 14 arthroplasties (26.9%) had talar sinkage more than 4mm. This problem is also almost exclusive to the patients of the group with inflammatory arthritis. Thirteen arthroplasties (37.1%) were from this group and only 1 (5.8%) was in the group with degenerative arthritis. This was anticipated in view of the poor quality of the bone in rheumatoid patients and accounts for the high failure rate seen in this group.
Although that poor wound healing has been reported in 25% to 40.3% of the arthroplasties in other published series[15-18], in our experience this problem was encountered in 4 arthroplasties (4.3%) only. This is the result of meticulous operating technique and postoperative rest until the wound has healed satisfactorily. Two arthroplasties (2.1%) developed deep sepsis and had to be removed.

CONCLUSION
Ankle replacement using the Bath-Wessex spherocentric prosthesis is a viable option in the treatment of arthritis of the ankle. Our results are better than any other currently available in such a long-term review. Our results showed a favourable result in 73% of the patients and an overall survival rate of 82.6% at 7.9 years follow-up. It provides satisfactory pain relief, especially in patients with rheumatoid arthritis. This is we believe, the result of the unconstrained design, which eliminates stresses to the neighbouring joints of the foot.
We feel that results after total ankle arthroplasty are better than those of the early days. The complication rate however remains quite high, although much less than other published series, compared with joint replacement procedures in other joint of the body (hip, knee). Further more the functional outcome is not as good as expected and although it provides satisfactory pain relief in certain patients, further improvements need to be made to the design of the prosthesis and to the methods of fixation to guarantee satisfactory long-term results.
At present we feel that an unconstrained type of ankle prosthesis, such as the Bath and Wessex total ankle replacement, has distinct advantages over other designs and is a good option for an elderly rheumatoid patient, especially if there is involvement of other joints in the area.

REFERENCES
1. Alexakis P., Smith R.C., Wellish M. Indications of ankle fusion versus total ankle replacement versus pantalar fusion. Orthop trans. 1977; 1, 87.
2. Adam W., Ranawat C. Arthrodesis of the hindfoot in rheumatoid arthritis. Orthop Clin North Am. 1976; 7, 4, 827.
3. Ahlberg A., Henricson A.S. Late results of ankle fusion. Acta Orthop Scan. 1981; 52, 103.
4. Bolton-Maggs M.A., Sudlow R.A., Freeman M.A.R. Total ankle arthroplasty; a long-term review of the London hospital experience. J Bone Joint Surg. [Br] 1985; 67-B, 785.
5. Demottaz J.D., Mazur J.M., Thomas W.H., Sledge C.B., Simon S.R. Clinical study of total ankle replacement with gait analysis; a preliminary report. J Bone Joint Surg. [Am] 1979; 61-A, 976.
6. Dini A.A., Bassett F.H. Evaluation of the early result of Smith total ankle replacement. Clin Orthop. 1980; 180, 228.
7. Fjermeros H., Hagen R. Posttraumatic arthrosis in the ankle and foot treated with arthrodesis. Acta Chir Scan. 1967; 133, 527.
8. Herberts P., Goldie I.F., Korner L., Larssom U., Linborg J., Zachrisson B.E. Endoprosthetic arthroplasty of the ankle joint; a clinical and radiological follow-up. Acta Orthop Scan. 1982; 53, 687.
9. Johnson E.W., Baseker E.H. Arhodesis of the ankle. Arch Surg. 1968; 97, 766.
10. Lachiewicz P.F., Inglis A.E., Ranawat C.S. Total ankle replacement in rheumatoid arthritis. J Bone Joint Surg. [Am] 1984; 66-A, 340, 3.
11. Lance E., Pavel A., Patterson R.L., Fries I., Larsen I. Arthodesis of the ankle: a follow-up study. J Bone Joint Surg. [Am] 1971; 53-A, 1030.
12. Mazur J.M., Schartz E., Simon S.R. Ankle arthodesis:long term-folow-up with gait analysis. J Bone Joint Surg. [Am] 1979; 61-A, 964.
13. Morrey B.F., Windeman G.P. Complications and long term results of ankle arthodesis following trauma. J Bone Joint Surg. [Am] 1980; 62-A, 77.
14. Newton S.A. Total ankle arthroplasty: clinical study of fifty cases. J Bone Joint Surg. [Am] 1982; 64-A, 104.
15. Said E., Hunka L., Siller T.N. Where ankle fusion stands today. J Bone Joint Surg. [Br] 1978; 60-B, 211.
16. Samuelson K.M., Freeman M.A.R., Tuke M.A. Development and evolution of the ICLH ankle replacement. Foot Ankle. 1982; 3, 32.
17. Stauffer R.N. Salvage of painful total ankle arthroplasty. Clin Orthop. 1982; 170, 184.
18. Stauffer R.N., Segal N.M. Total ankle arthroplasty: four years experience. Clin Orthop. 1981; 160, 217.
19. White A.A. A precision posterior ankle fusion. Clin Orthop. 1974; 98, 239.