Les auteurs rapportent huit cas d'échec de prothèse totale de cheville repris par arthrodèse. La moyenne d'âge était de 57 ans (36 à 76 ans). Le diagnostic initial était une arthrose post-traumatique dans quatre cas, une polyarthrite rhumatoïde dans trois cas et une arthrose idiopathique dans un cas. La révision a été pratiquée en moyenne 36 mois (4 à 108 mois) après la mise en place de la prothèse, pour traiter sept descellements aseptiques et un descellement septique. La perte de substance osseuse liée à l'ablation de la prothèse a été le plus souvent compensée par une greffe iliaque. Une ostéosynthèse interne a été utilisée dans sept cas. Dans le cas septique, la fusion a été obtenue sans greffe osseuse avec un fixateur externe. Le recul moyen était de 56 mois (10 à 114 mois). La fusion a été obtenue dans sept cas avec un délai moyen de 3,1 mois (2,5 à 6 mois). Le score de Kitaoka a été amélioré dans tous les cas passant en moyenne de 19,1 à 54. En cas d'échec d'arthroplastie totale de cheville, l'arthrodèse a été obtenue pour 87 % des cas dans notre expérience. Le taux de complication est faible, mais le résultat fonctionnel est modeste et reste inférieur à une arthrodèse de première intention.

We report outcome in eight cases of ankle arthrodesis after failure of a total ankle prosthesis.

This series included eight patients, mean age 57 years (range 36-76) who had initialy: post-traumatic talocrural joint degeneration (n = 4), rheumatoid polyarthritis (n = 3), idiopathic talocrural degeneration (n = 1). Three patients had a New Jersey^® (DePuy) cemented prosthesis, four had a Star^® (Link) uncemented prosthesis and one, whose implantation was performed in another institution, had a spherical cemented prosthesis. The preoperative Kitaoka score was 19.1 (0-32). Seven patients had subtalar joint degeneration; one patient had a subtalar arthrodesis. The tibial component was cemented alone in two patients, the talar piece in three (two with talar fracture). Failure resulted from loosening, talus fracture or deep infection. Bipolar loosening was observed in two patients. Time to revision was 36 months (range 4-108). Arthrodesis was associated with an iliac graft for seven patients: several tricortical grafts (vertical alignment of the corticals) and cancelous grafts for filling. The height of the graft was adjusted to the substance loss. A bone graft could not be used in one patient who had a deep infection. The arthrodesis was fixed with an anterior plate bridging the talocrural space in six patients, with an external fixator in infected patient, with a conventional centromedullary tibial nail transfixing the talocrural joint and planted in the talus and the calcaneus in one. Outcome was assessed with the Kitaoka score. Mean follow-up was 56 months (range 10-114).

The overall Kitaoka score improved to 54/100 (range 42-70) at last follow-up. The arthrodesis provided improvement in all patients although the final outcome was still considered poor in three patients. Radiographic healing was obtained in seven patients at a mean 3.1 months (range 2.5-6). Wound healing was slow in two patients. One patient developed a deep infection early.

The rate of fusion was 87%. This is in the general range reported in the literature; use of an iliac graft allows preserving joint height but because of the poor bone quality often encountered, residual bone stock may be insufficient to achieve complete fixation with screwing. Plate fixation appears to be a better way of achieving fixation. This provides a rate of fusion comparable with earlier series where external fixation was generally employed. For us, external fixation should be reserved for infected cases. Use of a conventional anterograde nail can be another solution in the event of poor bone quality. The overall result remains relatively modest although all the patients achieved a functional gain with arthrodesis. The results obtained are less satisfactory than after first-intention ankle arthrodesis.