Between 2000 and 2006 we performed salvage tibiotalar arthrodesis in 17 diabetic patients (17 ankles) with grossly unstable ankles caused by bimalleolar fractures complicated by Charcot neuro-arthropathy. There were ten women and seven men with a mean age of 61.6 years (57 to 69). A crossed-screw technique was used. Two screws were used in eight patients and three screws in nine. Additional graft from the malleoli was used in all patients. The mean follow-up was 26 months (12 to 48) and the mean time to union was 5.8 months (4 to 8). A stable ankle was achieved in 14 patients (82.4%), nine of whom had bony fusion and five had a stiff fibrous union. The results were significantly better in underweight patients, in those in whom surgery had been performed three to six months after the onset of acute Charcot arthropathy, in those who had received anti-resorptive medication during the acute stage, in those without extensive peripheral neuropathy, and in those with adequate peripheral oxygen saturation (> 95%). The arthrodesis failed because of avascular necrosis of the talus in only three patients (17.6%), who developed grossly unstable, ulcerated hindfeet, and required below-knee amputation.
Diabetic neuropathic arthropathy is a destructive process affecting the bony components of a denervated joint. Although it was originally associated with tabes dorsalis, diabetes mellitus is now the main cause of Charcot neuro-arthropathy in both the developed and developing world.1 It has been estimated to affect at least 1 in 680 diabetics and may occur in up to 29% of diabetics with peripheral neuropathy.1,2 Its incidence is suspected to be rising with the increasing prevalence of diabetes.1,2
The aetiology and pathophysiology of neuropathic destruction of bones and joints are poorly understood. However, Eichenholtz3 described three phases as follows: stage I or dissolution, a phase of bone and joint destruction; stage II or coalescence, fracture healing, and stage III or resolution, remodelling.
A Charcot fracture/dislocation of the ankle or foot is often caused by a minor acute injury or by chronic repetitive minor injuries. However, it may also be the result of acute major trauma such as a fall from a height or a motor-vehicle accident.4 Although conservative treatment with the use of a total contact cast, followed by appropriate bracing and use of footwear is the treatment of choice of most neuropathic fractures and dislocations of the ankle or foot, encouraging results after ankle arthrodesis with internal fixation have recently been reported.5,6
The aim of this study was to assess the outcome of tibiotalar arthrodesis in diabetic patients with ankle fractures complicated by Charcot arthropathy, nonunion and instability, and to evaluate the factors which could cause failure of the fusion.
Patients and Methods
Between 2000 and 2006 17 patients (17 ankles) with diabetic neuropathic arthropathy and instability of an ankle were treated by tibiotalar arthrodesis using cannulated screws. They all presented initially with hypermobile deformed ankles with positive anterior drawer and talartilt tests, diminished sensation distally and a longstanding history of diabetes. Early in the acute stage, according to the referral notes, the local redness, warmth and swelling subsided after elevation of the foot for five to ten minutes. The clinical details are given in Table I⇓. There were ten women and seven men with a mean age of 61.6 years (57 to 69). All had a history of a twisting injury resulting in a displaced bimalleolar fracture. They had all been treated conservatively initially in local community hospitals before their referral. Conservative treatment consisted of a mean period in a cast of 7.5 weeks (5 to 10) before acute Charcot neuro-arthropathy was diagnosed. This was followed by treatment in a total contact cast for a mean period of 11.3 weeks (3 to 23.4) before referral. A total of 11 patients had a short period of non-weight-bearing, followed by partial weight-bearing, while six followed a strictly non-weight-bearing protocol throughout. After referral the patients continued to use the total contact cast which was replaced bi-weekly, for a mean period of 6.8 weeks (5 to 8.6) before surgery, with strict non-weight-bearing.
Five patients were underweight (BMI < 18.5 kg/m2), six had normal weight (BMI, 18.5 kg/m2 to 24.9 kg/m2), and six were overweight (BMI, 25 kg/m2 to 29.9 kg/m2). Typical diabetic chronic sensorimotor distal polyneuropathy was present in nine patients. Atypical diabetic polyneuropathy, chracterised by loss of vibration sense and decreased pin-prick sensation, but satisfactory sensation to heat and response to 5.07 (10 g) Semmes-Weinsten monofilament testing7 on their plantar surface, was present in eight patients. Six men were heavy smokers. All the patients had an active lifestyle.
In all the patients, arthrodesis was undertaken in stages II or III of Charcot arthropathy according to the classification of Eichenholtz.3 Resolution of the acute dissolution stage was determined clinically by the lack of local swelling, erythema and elevated skin temperature, and by comparison with the contralateral ankle and foot. Anti-resorptive treatment, either in the form of calcitonin administered intranasally (200 IU daily) in five patients with impaired renal functions, or oral alendronate (70 mg once weekly) was started during the first (dissolution) stage in all patients except for five who could not afford it. The treatment was initiated at a mean of 14.6 weeks (5.2 to 23.4) after the onset of the acute attack and was continued for a mean of 6.4 weeks (5 to 8) pre-operatively and for up to 12 months post-operatively, or until a diagnosis of nonunion had been confirmed.
The distal circulation was checked using Doppler ultrasound (Fukuda Denshi Co. Ltd., Tokyo, Japan). The presence of a pulsatile flow in the posterior tibial artery and an ankle-brachial index of 0.65 or more signified adequate perfusion.6 The mean ankle-brachial index of the patients was 0.9 (0.75 to 1.1).
Plain anteroposterior and lateral radiographs, supplemented by MRI when necessary, were performed pre-operatively to evaluate the severity of the joint pathology, and to exclude the presence of avascular necrosis of the talus or occult infection. In addition to the ununited bimalleolar fractures, all patients had radiological evidence of neuropathic arthropathy, namely sclerosis, fragmentation, and intra-articular derangement with tibiotalar sub-luxation or dislocation.
Cessation of smoking was a prerequisite for surgery. Oxygen saturation was measured using a pulse oximeter (NPB-190; Nellcor Puritan Bennett LLC, Pleasanton, California) on the toes of the ipsilateral foot just before the induction of general anaesthesia in ten patients. Spinal anaesthesia was used in seven. A tourniquet was not used.
An image intensifier was available for the duration of surgery. With the patient in the supine position the fractured medial and lateral malleoli were initially excised through separate incisions. The distal fibula was divided obliquely 2.5 cm proximal to the joint line with an oscillating saw from the proximolateral to the distal-medial in order to preserve the distal tibiofibular syndesmosis. This was followed by thorough debridement of the ankle and resection of the articular surfaces of the distal tibia and talus, which was as conservative as possible, in order to preserve bone stock while providing healthy, bleeding opposing surfaces. Resection was completed with osteotomes in order to avoid thermal necrosis. A residual bony defect was uniformly observed on the lateral aspect of the distal tibia. This was filled with cancellous autograft harvested from the excised medial or lateral malleolus (Fig. 1⇓).
Temporary tibiotalar fixation was then obtained by crossed guide-wires with the ankle and foot in 5° of valgus with slight external rotation, neutral dorsal/plantar flexion, and, wherever possible, with the talus in a posterior position relative to the tibia in order to reduce the mechanical lever arm of the foot during gait. In nine patients in whom the lateral tibial defect was small a third guide-wire was inserted and the length and location of the wires were checked fluoroscopically. Cannulated 4.0 mm or 7.3 mm screws (Synthes USA, Monument, Colorado) with or without washers, were used for definitive fixation. In the other eight patients, the big lateral tibial defect required transverse screw fixation for the graft followed by two-screws for the arthrodesis (Fig. 2⇓).
Post-operative care and evaluation.
Low-molecular weight heparin was started on the night of surgery, and continued until the patients were walking comfortably. The patients began to walk, aided, at a mean of 3.4 months (3 to 4). A well-moulded full-contact below-knee cast was applied when the wounds had healed. The patients were followed clinically and radiologically at three-week intervals. When radiological signs of union first appeared, partial weight-bearing was allowed until there was complete union. The cast was then replaced by a permanent custom-made ankle-foot orthosis, and the patients were subsequently reviewed every three months.
The mean follow-up was 26 months (12 to 48) and no patient was lost to follow-up. Nonunion was defined as the presence of movement clinically or the absence of osseous bridging radiologically at 12 months post-operatively according to the criteria of Stuart and Morrey.4 The method of Mazur, Schwartz and Simon8 was used to obtain standardised results. This used a 90-point scale, 50 points being allocated for absence of pain and 40 for full function. A score of 80 to 90 was considered to be excellent, 70 to 79, good, and 60 to 69, fair. Any score less than 60 was classified as poor.
This was performed using SPSS version 11.0.1 for Windows (SPSS Inc., Chicago, Illinois). One-way analysis of variance (ANOVA) and its non-parametric equivalent, the Kruskal-Wallis test were used for variables which were small and not normally distributed. A p-value ≤ 0.05 was considered to be statistically significant.
The adjunctive use of calcitonin in five patients, started at a mean of 14 weeks (9 to 20) after development of the acute stage (Eichenholtz stage I) resulted in resolution after seven to eight weeks. Similarly, resolution occurred in seven patients after adjunctive treatment with alendronate after five to six weeks. They had developed acute neuro-arthropathy at a mean of 15 weeks (8 to 23) before initiation of alendronate treatment. Resolution occurred in the five patients who did not receive anti-resorptive medication after a mean of 11.1 weeks (10.4 to 11.6) from the onset of the acute stage.
Progression to Eichenholtz stage II was observed in six patients at a mean of 2.8 months (2.6 to 2.9) after the development of the acute stage. Only one of these had received anti-resorptive treatment (alendronate), and all were strictly non-weight-bearing. The remaining patients passed the acute stage at a mean of 5.5 months (3.5 to 7.1). All were partially weight-bearing using a total contact cast before their referral.
Nine patients (53%) had satisfactory results with union at a mean of 5.8 months (4 to 8). There was a persistent pseudarthrosis for the remaining eight patients (47%). Assessment scores were excellent (82 points) in one patient (6%), good (mean 74.2 points (71 to 78)) in eight (47%), fair (mean 63.4 (61 to 67)) in five (29%) and poor (mean 56 points (54 to 57)) in three patients (18%). A stable ankle joint with bony union or a stiff nonunion was present in 14 patients (82%). Moreover, there was no widening of the ankle mortise, shortening of > 2.5 cm, or varus malalignment. Compromised function, with or without associated instability, was the main reason for the fair or poor scores. The fibrous nonunion was stable and painless in five patients (29%), but unstable with associated avascular necrosis of the talus in three (18%). Decreased performance of everyday activities, with limping and increased energy consumption were the main causes of unsatisfactory results, without, however, associated pain. No major soft-tissue complications were noted.
Three patients (18%), all with union, developed radiological evidence of degenerative changes of the subtalar and midtarsal joints after two years without associated symptoms. Male gender (Kruskal-Wallis, p = 0.067), smoking (ANOVA, p = 0.411), and a history of previous foot ulcers (Kruskal-Wallis, p = 0.419) did not adversely affect the outcome significantly (Table II⇓). Moreover, a trend towards better fusion rates was obtained using three screws for the fusion (nine patients) rather than using two screws with graft augmentation (eight patients) (Kruskal-Wallis, p = 0.067).
Fusion rates were higher when the oxygen saturation was above 95% (ANOVA, p = 0.003), the patient was not overweight (ANOVA, p = 0.02), without pre-operative clinical evidence of a peripheral neuropathy (Kruskal-Wallis, p = 0.038), anti-resorptive medication had been administered during the acute phase (ANOVA, p = 0.003) and surgery was performed between four and six months after the development of acute neuro-arthropathy (ANOVA, p = 0.02) (Table II⇑). There was no significant difference between the use of alendronate and calcitonin (Kruskal-Wallis, p = 0.162).
Two overweight patients developed a deep-vein thrombosis despite thromboembolic prophylaxis. Both had a history of dyslipidaemia and varicose veins. Four patients (23.5%), who had a history of previous foot ulcers and had poor peripheral oxygen saturation (< 90%), had a superficial wound infection. Two patients (11.8%) had approximately 10° of valgus malunion due to mild graft collapse, however both had good functional scores. The three patients with avascular necrosis of the talus had grossly unstable ankles and were severely handicapped. They later developed ulceration of the lateral hindfoot although no acute neuropathic exacerbation was noticed at follow-up. The use of a modified ankle-foot orthosis was not helpful and therefore they opted for a below-knee amputation rather than waiting for the ulcers to heal and a subsequent tibiocal-caneal fusion. All had had diabetes for more than 20 years and were older than 60 years. Two were overweight. Table III⇓ summarises the complications encountered.
The mainstay of treatment for Charcot neuro-arthropathy of the ankle and foot is prolonged immobilisation in a plaster cast or a brace. Occasionally, a patient will present with severely disabling instability and failure of immobilisation in a brace or cast. This study gives the results of an attempt to salvage the limbs of 17 patients with this severe clinical problem using cannulated screws to obtain tibiotalar fusion after the resolution of the acute stage of neuro-arthropathy. A solid fusion was achieved in nine patients (53%) and a stiff fibrous union was obtained in five (29.4%). Only three patients (18%) developed unstable pseudarthroses which led to below-knee amputations. These results compare favourably with those of previous studies which reported rates of fusion of between 33% and 40%.4,9 Technical factors contributing to the successful outcome in our patients may include the use of full-thickness skin incisions, gentle handling of soft tissues, the use of osteotomes rather than a saw to resect the articular surfaces in order to avoid thermal necrosis, bone grafting, and a three-screw fixation construct. The adjunctive use of anti-resorptive medication in the acute stage, prolonged post-operative immobilisation and the fact that most patients were non-smokers, also probably played a role.
The presence of an intact peripheral circulation pre-operatively, as assessed by pulse oximetry, had a statistically significant effect on the outcome. This can be explained by adequate vascularity with a good ankle-brachial index, good healing potential and haemoglobin concentration and soft-tissue viability. It is recommended that pulse oximetry be used routinely before any type of arthrodesis. A pulse oximeter is small and easy to use and, contrary to measuring devices for transcutaneous oxygen pressure does not require any special precautions or frequent calibration. Joyce et al10 found that the peripheral oxygen saturation measured by pulse oximetry correlated better with a standard sonographic scoring system than the transcutaneous oxygen saturation as measured by the ankle/brachial Doppler index. They considered it to be the best non-invasive method for assessing the peripheral circulation.10 This was not appreciated during the early stages of this study and no effort was made to improve the peripheral oxygen saturation before surgery. However, we believe that if the saturation measures < 95% pre-operatively, it is possible to improve this by stopping smoking, strict control of dyslipidaemia, correction of anaemia, the use of anticoagulants and a course of pentoxifylline for four weeks before surgery. Campbell11 found that pentotoxifylline (800 mg per day) and its metabolites effectively decreased the viscosity of blood, thereby improving the perfusion and oxygenation of tissues in diabetics.
Smoking releases toxic agents, such as nicotine, carbon monoxide and hydrogen cyanide, which interfere with metabolism. Animal studies have demonstrated the adverse effect of nicotine on bony healing, with increased rates of delayed union, nonunion and inhibition of graft revascularisation after spinal fusions.12,13 A poorer outcome was found in this study in the patients who smoked, but this difference was not statistically significant (p = 0.411). Likewise, Cobb et al14 in a study on ankle arthrodesis, found the relative risk of nonunion to be 3.75 times higher for smokers. Ishikawa, Murphy and Richardson15 also found that smokers undergoing hindfoot fusion had a significantly higher rate of nonunion than non-smokers (18.6% vs 7.1%, respectively). The relative risk of developing nonunion was 2.7 times higher for smokers.
The lack of autonomic control presents clinically as hyperaemia and elevated temperature in the area of the Charcot joint. Elevated alkaline phosphatase levels are used to assess the extent of the subsequent osteoclastic resorption. This process appears to be analogous to the early destructive or osteolytic phase of Paget’s disease.16 Recently, treatment has been based on this analogy, using alendronate or calcitonin, with moderate success.17,18 Our application of anti-resorptive treatment in the acute phase significantly affected the rate of fusion. This may be due to an indirect effect of treatment by inhibition of osteoclasts, leading to improved bone stock and allowing good bone purchase of screws. Selby, Young and Boulton19 proposed that bisphosphonates may be a useful adjunctive treatment for neuropathic ankles before surgery.
Although the non-weight-bearing total contact cast is the cornerstone of rapid healing in the acute phase of Charcot arthropathy, as seen in this study phase and reported by others,20–23 the adjunctive use of alendronate or calcitonin in the acute phase in our study led to rapid resolution of the otherwise slowly improving acute Charcot reaction at a mean of 6.4 weeks (5 to 8). The study by Jude et al24 in which a single intravenous dose of pamidronate caused a significant reduction in the skin temperature, activity of the disease and markers of bone tumours within only four weeks lends support to this finding. Tan et al25 suggested that bisphosphonates not only target the osteoclasts but may also have direct anti-inflammatory properties which could be therapeutically beneficial at the site of activity of disease. Rajbhandari et al,26 reported their experience of marked symptomatic relief in patients with acute Charcot disease treated by two infusions of 90 mg of pamidronate. Bem et al18 suggested that treatment of acute Charcot arthropathy with intranasal calcitonin, including patients with renal failure, could prevent resorption of bone and progression of the condition.
Rates of fusion were significantly higher when surgery was performed between four and six months after the acute phase, i.e., during the coalescence stage (Eichenholtz stage II). Arthrodesis surgery during that stage avoids the severe osteopenia and the active destructive process which often lead to a high rate of complications and failure of the implant associated with the acute dissolution stage (Eichenholtz stage I).6,27 However, Simon et al28 successfully performed forefoot fusions in 40 patients during the acute stage, but none of their patients had involvement of the ankle. In this study, six patients progressed to stage II in less than three months after the application of the total contact cast in the acute stage. This finding is not unusual, since in the study by Chantelau20 the mean time from application of the total contact cast to healing was reported to be three months (2 to 9) in patients with incipient Charcot neuro-arthropathy compared with 5.5 months (2 to 12) in patients with overt Charcot neuro-arthropathy. Also, Sinacore29 reported that all of his patients with an acute Charcot fracture, subluxation or dislocation progressed to healing at a mean of 86 days (SD 45). Armstrong et al30 reviewed the rate of healing of 55 patients with type-II diabetes and an acute Charcot arthropathy treated with a total contact cast and reported a mean time in a cast of 130 days (SD 74).
A high rate of complications is reported in patients with Charcot arthropathy of the hindfoot treated by tibiotalo-calcaneal arthrodesis. Stone and Daniels31 had five patients (71%) with a fibrous nonunion, one of whom required a below-knee amputation. Thordarson and Chang32 identified an area of radiolucency at the proximal tip of standard retrograde ankle fusion nails which correlated with symptoms. Pinzur33 reported five patients who developed displaced ‘stress’ fractures which required further fixation. Caravaggi et al34 described four patients (28.6%) with deep infection, ulcers, and fibrous union after ankle arthrodesis with a compressive intramedullary nail. On the basis of these findings and the results of this study, it appears that tibiotalocalcaneal fusion for Charcot arthropathy of the ankle is indicated when there is additional involvement of the subtalar joint, fragmentation or avascular necrosis of the talus, a history of diabetes for more than 20 years, or obesity.
Huang et al35 suggested that tibiotalocalcaneal fusion in neuropathic ankles should be reserved for combined ankle and subtalar arthritis or severe deformities and instability which could not be corrected by fusion of the ankle alone. According to Gruen and Mears36 and Beaudion et al37 fusion of the subtalar and ankle joints may potentially alter the biomechanics of the foot and ankle, since it may also alter forefoot movement. Since movement occurs, after hindfoot fusion, at the Chopart and Lisfranc joints, the subtalar joint must be fused in eversion to allow for normal movement of the Chopart joint. If this requirement is not met, the Chopart joint will become stiff resulting in a rigid forefoot.
In the present study, eight patients had atypical diabetic polyneuropathy since they had loss of vibration and decreased pinprick sensation but retained sensation to heat and to 5.07 (10 g) Semmes-Weinstein Monofilament testing. Dyck et al38 evaluated the periosteal nociception of the tibia in healthy subjects and in patients with neuro-arthropathy without overt neurological findings. They concluded that neuro-arthropathy could still occur in patients with subclinical neuropathy and preservation of superficial and deep nociception. Stevens et al39 have suggested that patients with diabetic Charcot neuro-arthropathy have a variant of neuropathy which is different from the usual distal chronic sensorimotor neuropathy, characterised by preservation of some sensory elements (light touch, warmth). Sinacore40 suggested that severe sensory neuropathy does not always precede or always cause Charcot arthropathy of the foot in diabetic patients. Absence of full-blown peripheral neuropathy was a statistically significantly prognostic factor for successful fusion in the current study. Conversely, Stuart and Morrey4 reported a complication rate of 62% in their series of 13 patients, six of whom had a typical peripheral neuropathy pre-operatively.
A history of foot ulcers did not significantly affect the outcome in our patients. Ulcers could be due to occult infection or a compromised tissue perfusion. Caravaggi et al34 found that all patients who developed complications initially presented with an ulcer which had been present for more than six months. They postulated that the reason for failure could be chronic osteomyelitis not suggested by the clinical and laboratory findings.
Using finite-element analysis models, it has been shown that bone stock significantly affects the initial stability at the site of the arthrodesis. With a two-screw fixation construct, the poorer the bone stock is, the larger are the micro-movements. The addition of a third screw significantly reduced the peak micromovements in osteoporotic bone by up to 2.4 times.41 In our patients higher fusion rates were achieved with three screws although no statistically significant difference could be reached.
Although better initial stability has been predicted for ankle arthrodesis when contour of the articular surfaces was preserved rather than resected,41,44 the patients in this study first presented with bimalleolar fractures from which fragments were resected during arthrodesis. This offered some advantages. First, it provided a source of autograft. Secondly, it avoided widening of the fused ankle, eliminating the associated cosmetic and functional difficulties. Thirdly, the absence of the lateral malleolus minimised the possibility of varus malunion and allowed for adequate eversion of the subtalar joint.
In summary, early recognition and prevention of collapse are still the best options for the management of patients with diabetic Charcot arthropathy. Appropriate education, improved clinical evaluation and early intervention are increasingly used to this end. Once collapse is present, the use of an off-loading total contact cast and anti-resorptive medication are recommended in the acute stage. In the ensuing stages, salvage of the affected joint by tibiotalar arthrodesis is preferable and is most likely to succeed in patients of normal body-weight, with good peripheral oxygen saturation, who have begun anti-resorptive medical treatment in the acute stage, have a strong motivation to stop smoking and do not have extensive peripheral neuropathy. Nevertheless, this salvage surgery is not intended to substitute the lifetime of appropriate footwear or a brace. Finally, regular observation and protected weight-bearing of the contralateral, uninvolved foot and ankle are recommended.
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.
- Received August 13, 2007.
- Accepted March 19, 2008.
- © 2008 British Editorial Society of Bone and Joint Surgery