MANAGEMENT OF MIDFOOT DIABETIC NEUROARTHROPATHY
September 27th, 1993
Mark S. Myerson, M.D.; Michael R. Henderson, M.D.; Terence Saxby, F.R.A.C.S.; and Kelly Wilson Short, P.T.
Between 1986 and 1990, we treated 68 patients with diabetes and neuro-arthropathy of the midfoot, 21 of whom had bilateral involvement. Patients were managed according to a strict protocol defined by activity of the neuroarthropathy, instability, ulceration, infection, and ischemia. Initial management of acute neuroarthropathy (18 feet) was open reduction and arthrodesis (8), a total-contact cast or brace (9), and amputation (1). All patients with subacute neuroarthropathy (30 feet) were initially treated in a total contact cast. Four of these feet subsequently required amputation, two required arthrodesis, and one required exostectomy. For chronic neuroarthropathy (41 feet), a total-contact cast or a molded orthotic insert with or without bracing was used initially in all feet. Subsequent surgical salvage for this group included arthrodesis (9), plantar exostectomy (6), amputation (2), and abscess drainage (2). Four patients died during this treatment period, and 64 patients (85 feet) were evaluated at a mean interval of 3 years (range, 1 to 6 years) after initiation of treatment. This treatment program was found to be successful in 82 of 85 feet treated.
Advances in the treatment of patients with diabetes in the past decade have resulted in both increased longevity and improved quality of life for the more than ten million with this disease in the United States. Of the orthopaedic manifestations of diabetes, neuroarthropathy of the foot and ankle is common, and the outcome dependent on proper management. The incidence of neuroarthropathy in diabetes ranges from 0.1 to 2.5%.4,7,9 The most common areas of neuroarthropathy are the midfoot and hindfoot.1,16 Although a clear correlation between the level of diabetic control and the onset or progression of neuroarthropathy has not been demonstrated, most patients are more than 40 years old at presentation.16 Due to the increasing longevity of diabetic patients, it is anticipated that many orthopaedic surgeons will have an opportunity to manage this condition.
We present a retrospective review of the results of treatment of a large clinical series of diabetic patients with midfoot neuroarthropathy. Our purpose is to highlight those factors we believe to be essential in the successful management of this condition.
Materials and Methods
Between 1986 and 1990, we treated 68 patients with diabetes and neuroarthropathy of the midfoot, including 21 patients who had bilateral involvement. These 89 feet formed the basis for this retrospective review of their clinical course while under treatment. The right foot was involved in 37 cases (42%); the left foot, in 52 cases (58%). The average age at presentation of the 40 female and 28 male patients was 54 years (range, 25 to 73 years).
A thorough history, including management and complications of diabetes, previous forefoot surgery, and any antecedent trauma to either foot, was obtained. The activity of the neuroarthropathy at presentation was staged according to Eichenholtz2: stage-I feet exhibit hyperemia, swelling, and radiographic evidence of joint dislocation or bone fragmentation; stage-II feet demonstrate radiographic evidence of new bone formation; and stage-III feet present with deformity and no warmth or swelling. The location of any foot ulcer was noted and graded according to Wagner17. Anteroposterior and lateral weight-bearing radiographs of both feet and ankles were obtained, and the area of involvement was classified according to Brodsky et al.1 For the purpose of this review, only Brodsky Type 1 (tarsometatarsal and naviculocuneiform) and Type 2 (transverse tarsal and subtalar, but without any involvement of the talocalcaneal joint) were included.
Of the 89 feet, 18 (20%) presented in Eichenholtz Stage I, 30 (34%) in Stage II, and 41 (46%) in Stage III. Ulceration was noted at presentation in 38 feet; it was located over the medial midfoot in 21 (24%), the lateral midfoot in 6 (7%), and the forefoot in 11 (12%). Ulcers were present in four Stage-I feet, four Stage-II feet, and 30 Stage-III feet. There were 65 examples of Type 1 (tarsometatarsal and naviculocuneiform) involvement (73%) and 24 cases (27%) of Type 2 (talonavicular and calcaneocuboid) involvement.
All feet were examined in a standardized manner with specific attention to the neurovascular status and the presence and extent of ulceration and infection. Semmes Weinstein monofilaments were used to assess sensory loss. Of the 68 patients, 84% did not have sufficient sensation to protect against injury. Three patients were able to perceive the 4.17 monofilament; six, the 5.07 monofilament, 28, the 6.10 monofilament; and 20, the 6.45 monofilament; 11 were not able to perceive any pressure or sensation at all. Vascular evaluation was performed with Doppler ultrasound (Medtronics) and the foot brachial index was obtained. The Doppler index ranged between 0.8 and 1.0 in 35 patients, between 0.6 and 0.8 in 16 patients, between 0.4 and 0.6 in 11 patients, and was less than 0.4 in six patients. In 17 of these patients, a poor wave form was obtained, and the vessels could not be adequately occluded to obtain a sufficiently accurate recording. The range of motion of the ankle, hindfoot, midfoot, and forefoot joints were noted. Instability of the midfoot was determined manually, noting the presence of crepitus and the potential for reducing an acute dislocation. Chronic instability was judged to be present if, on attempted dorsiflexion of the ankle, motion occurred in the midfoot. This was typically associated with an Achilles tendon contracture, hindfoot equinus, and a mobile "rocker bottom" deformity.
All patients were treated according to a strict protocol determined by the activity of the neuroarthropathy and by the presence of ulceration or infection.
Stage-I acute neuroarthropathy without ulceration was managed with either cast immobilization or surgery. The indication for surgery was specific and performed only for a severe dislocation that was unstable and manually reducible (Figs. 1 and 2). Bone fragmentation was a contraindication to surgery. Eight patients with acute neuroarthropathy underwent open reduction and arthrodesis. These patients were hospitalized, a foot pump (AV Impulse System, Kendall, Mansfield, MA) was used to decrease swelling, and surgery was delayed until skin wrinkling occurred. Rigid internal fixation was gained by interfragmentary screw compression. A short leg cast was then applied, but weight-bearing on the involved extremity was not allowed for 8 to 10 weeks. The cast was changed weekly for 3 weeks, and then every 2 weeks to prevent complications from the initial rapid reduction in swelling. When the clinical signs of warmth and swelling had subsided (usually between 8 and 10 weeks), patients commenced bearing weight in a total-contact short leg cast. The cast was then changed every 2 weeks for 6 weeks, and then at monthly intervals. Casting was discontinued when no swelling was evident and the temperature of the foot was stable on several examinations with a skin thermistor. The total duration of cast treatment in these eight patients treated operatively averaged 5 months (range, 4 to 9 months). Cast treatment alone was used as the primary form of treatment in nine patients. One foot in this group presented with a forefoot that was treated with a Chopart amputation.
All 30 patients who presented with subacute neuroarthropathy were initially managed nonoperatively. These patients had radiographic evidence of bone fragmentation and resorption, which made surgical treatment inadvisable except where infection was present (Fig. 3). A similar program of total-contact cast immobilization was followed until the signs of neuroarthropathy dissipated. Custom-molded shoe inserts with appropriately modified shoes were then prescribed, and combined with bracing if instability was present. During the transition from cast to insert and/or brace, the foot was closely monitored for recurrent warmth or swelling and returned to a cast if either occurred.
Management of ulceration was determined by the grade of the ulcer11,17 All patients with Grade-I or -II ulceration were initially treated with the total-contact cast. Grade-III ulceration associated with deep or systemic infection (five patients) was treated with hospital admission, intravenous antibiotic therapy, and surgical debridement or amputation. After debridement, a total-contact cast was used if the ulcer was clean with granulation tissue (Grade I or II). Postoperatively, patients were allowed to bear weight in a total-contact cast until healing of the ulcer occurred. Once healed, these feet were treated with an appropriate molded orthosis with or without a brace.
The initial treatment for chronic neuroarthropathy (41 feet) was a brace or shoe modification (11/41) or a total-contact cast (30/41). Stage-I and -II feet that reached a chronic quiescent phase were similarly treated. For rigid, deformed feet, especially those with a plantar bony prominence, insert modifications were sometimes insufficient to prevent recurrent skin breakdown after healing in a total-contact cast. The indication for exostectomy was the presence of recurrent ulceration refractory to treatment with a total contact cast, brace, or molded shoe (Fig. 4). If the midfoot was unstable, realignment and arthrodesis or amputation, not ostectomy, was performed. The indication for arthrodesis was severe deformity associated with recurrent ulceration, particularly if the midfoot was unstable and associated with a fixed hindfoot equinus contracture. In these patients, an Achilles tendon lengthening was routinely performed with the midfoot arthrodesis. Nine patients with chronic neuroarthropathy underwent arthrodesis. Rigid internal screw fixation was our preferred method of stabilization. In three of nine patients, severe bone resorption was present, and it was found easier to use percutaneous threaded Steinmann pins. The postoperative protocol for these patients treated with arthrodesis was no different from that of patients in the acute neuroarthropathy group treated with arthrodesis. The duration of cast treatment in these patients, however, was slightly longer; the mean length of cast immobilization was 6.5 months (range, 5 to 10 months). Amputation was indicated in the presence of instability and recurrent ulceration when associated with infection, or with ischemia not amenable to or not improved by revascularization.
The initial recommended treatment was recorded. If a change in the treatment method during the patient's subsequent course was necessary, the factors involved in this decision were outlined. All surgical procedures performed on these feet were noted, including the date and outcome of each procedure. Cases in which structural instability of the midfoot influenced the treatment decision were listed, and examples of reactivation of the neuroarthropathy were sought. The status of each of these feet at their most recent visit was determined.
Of the 68 patients, 51 (75%) required insulin for control of their diabetes, whereas 17 were managed by diet or oral agents. Of the 89 feet, 31 (35%) had been operated upon previously, usually in the form of lesser toe procedures or toe amputations. In 19 patients (21%), a history of trauma preceded the onset of neuroarthropathy. The majority of these were minor events that occurred during ambulation, such as a twisting injury. Eight patients were under treatment for complications of diabetic nephropathy. Four required dialysis, and four had undergone renal transplantation and were maintained on immunosuppressive agents.
In 22 patients (25%), the correct diagnosis had not been made at the time of referral to our facility. In nine patients, no diagnosis had been made, and the findings were attributed to infection in five, gout in two, fracture in two, arthritis in two, venous insufficiency in one, and a tumor in one patient.
The initial treatment was total-contact casting for 58 feet (65%), a brace or molded orthotic device for 22 feet (25%), and surgery for the remaining nine feet (10%), with eight procedures performed for acute dislocation and one for infection.
Of the 18 feet that presented with acute neuroarthropathy (Eichenholtz Stage I), eight demonstrated manually reducible tarsometatarsal dislocation and were treated with open reduction and arthrodesis. These feet remained stable at an average follow-up of 28 months, with one deceased patient not included. Nine feet with acute neuroarthropathy were managed nonoperatively, and eight were stable at an average of 32 months follow-up. Forty months after an episode of tarsometatarsal involvement, one of these nine feet developed a new focus of neuroarthropathy, which ultimately required a triple arthrodesis for salvage. No other patient in this group manifested any findings of instability at follow-up examination.
Thirty feet presented with subacute neuroarthropathy (Eichenholtz Stage II). After management with a total-contact cast or brace alone, 21 were stable at an average of 26 months follow-up. One foot was not considered stable, and one patient died without sufficient follow-up to evaluate the foot. Three feet were treated surgically after they reached the chronic stage. Two of these were stable at 37 months after arthrodesis; the third was stable 18 months after an exostectomy. Four of the feet in this group failed initial total-contact cast treatment. Three developed severe infections (one after an exostectomy) and required amputation. The fourth had a complex deformity felt not to be amenable to realignment arthrodesis and was treated with a Chopart amputation.
Forty-one feet presented with chronic deformity (Eichenholtz Stage III). Twenty-two were successfully treated by nonoperative means and were stable at an average of 33 months follow-up. Six feet required exostectomy, and nine underwent arthrodesis. Two feet were returned to molded inserts after drainage of an abscess. There were two amputations in this group. A Chopart amputation was required to treat a deep midfoot infection, and a below-the-knee amputation was performed after an unsuccessful revascularization procedure. Of the