Diabetes is the most common cause of neuropathic arthropathy of the foot, affecting between 0.1% and 2.5% of the sixteen million patients with diabetes mellitus in the United States1. It most commonly involves the midfoot2, and deformities in this region as well as in the hindfoot can result in substantial morbidity3.
The ideal treatment protocol remains an issue of debate as can be seen from a survey of American Orthopaedic Foot and Ankle Society members who listed treatment of Charcot foot deformities as one of the two most controversial problem areas in foot and ankle surgery4. Nonsurgical treatment remains the standard of care for most patients with neuropathic changes5-7, although some recent articles have questioned such an approach8-11. Since these deformities are progressive, patients may fail nonoperative treatment and thus have development of a severe, fixed, rocker-bottom foot deformity with plantar ulceration as a result of increased pressure and decreased plantar sensation7. Resection of osseous plantar prominences has been proposed, but the results are generally poor in patients with ulcers involving the lateral column associated with instability of the midfoot or hindfoot12. Others13 have suggested that simple exostectomy is a less morbid procedure with quicker ulcer healing. In our opinion, this is only possible in patients with a stable Charcot foot deformity and therefore with little likelihood for progression of the deformity. If exostectomy is performed in an unstable Charcot foot deformity, the collapse progresses with recurrence of plantar osseous prominences and ulceration. We believe that this progression may lead to soft-tissue compromise, infection, and osteomyelitis, perhaps resulting in amputation.
Another option is surgical realignment and stabilization in order to restore a plantigrade foot2,14. Surgical reconstruction of a neuropathic foot deformity is difficult, requires prolonged treatment, and has potential complications. However, a successful arthrodesis is superior to amputation. The total potential economic benefit of strategies that reduce amputation risk has been shown to range from $2900 to $4442 per person with a history of a foot ulcer over a three-year period15. In addition, it has been shown that energy expenditure is higher in patients with a below-the-knee amputation16. This is particularly important since patients with diabetes have a high rate of associated cardiac and vascular disease17. Most importantly, the contralateral foot may have development of Charcot changes with collapse and infection, possibly requiring amputation. It has been estimated that 9% to 20% of patients with diabetes undergo a second amputation within one year after an initial amputation, and that rate increases to 28% to 51% within five years1. Patients with diabetes and only one lower extremity have a 50% to 66% incidence of contralateral amputation within five years of undergoing the first amputation18. Reiber et al. reported that, by five years following an initial amputation of a lower extremity, 28% to 51% of amputees with diabetes had undergone amputation of the contralateral leg19.
We believe that reconstruction can prevent amputation in patients who have unbraceable or unstable deformities or those with recurrent ulceration. After approval by our Hospital Ethics Committee, we studied a series of patients to evaluate whether operative realignment of severe Charcot foot deformities is possible with an acceptable risk of complications.
From July 2001 through July 2005, we performed reconstructive surgery on fifteen patients with diabetes mellitus who had severe neuropathic midfoot deformities consisting of a collapsed plantar arch with a rocker-bottom foot deformity (Table I). There were nine men and six women with a mean age of fifty-five years (range, thirty-four to seventy years), and the mean duration of preoperative care was two years. The primary indication for surgery was a midfoot deformity with an associated nonhealing plantar ulceration in patients who had failed management with a Charcot restraint orthotic walker (CROW; Leserre et Lenoir, Geneva, Switzerland). A CROW is a customized bivalved total contact ankle-foot orthosis. It is made from a positive plaster mold of the involved limb and consists of a rigid polymer shell that is lined with specific foams of different densities. A rocker-bottom sole is placed on the orthosis. A secondary indication for surgery was a patient without a plantar ulcer but with deformity and midfoot instability that was not amenable to brace treatment. While there is no clear definition of this type of instability, we assessed it by applying stress to the forefoot in the sagittal plane with the ankle locked in dorsiflexion. In this manner, one can objectively assess abnormal midfoot motion. We believe that, if this instability cannot be controlled with an orthosis, there will be progressive collapse of the midfoot and eventual ulcer formation in most patients. This can be seen clinically by progressive changes in the appearance of the foot, particularly with an increase in the rocker-bottom deformity. Similar indications have been outlined by others20,21. The presence of an open plantar ulcer, which had failed nonoperative treatment (wound care, total contact casting, and antibiotics), was not considered a contraindication to surgery.
Four patients with active soft-tissue or bone infection, as defined by fever, an uncontrollable serum glucose level, purulent drainage, or lymphangitis, did not undergo reconstructive surgery until resolution of the infection. In our experience, it is very difficult to differentiate between deep soft-tissue infection and bone infection in the foot in the context of Charcot changes, and we have not found magnetic resonance imaging or other diagnostic tests to be helpful in this regard. Therefore, we treat all deep soft-tissue infections as osteomyelitis. These four patients were admitted to the hospital for irrigation and débridement as an urgent procedure, at which time cultures were obtained and amoxicillin (Augmentin; GlaxoSmithKline, Münchenbuchsee, Switzerland) was administered. Once the culture results were available, we adapted the antibiotic therapy on the basis of these results. The wounds were checked daily, and patients were discharged usually after one week for further outpatient care and were continued on antibiotics until the active infection resolved, as determined by a return of the laboratory parameters to normal along with cessation of drainage. Cultures revealed that one patient (Case 1) had Pseudomonas species, Enterobacter cloacae, and Pasteurella multocida; one (Case 2), methicillin-sensitive Staphylococcus aureus; one (Case 3), methicillin-resistant Staphylococcus aureus and Serratia ficaria; and one (Case 4), Proteus mirabilis and Enterobacter cloacae. All patients received six weeks of organism-specific intravenous antibiotics before the index procedure and for twenty-four hours postoperatively.
Thirteen patients presented with symptoms of a nonhealing midfoot plantar ulcer, which was classified with use of the Wagner classification system22 (Table I). Since these patients had insensate feet, as determined with use of the 5.07/10-g Semmes-Weinstein monofilament, they did not complain of pain. The average duration (and standard deviation) of the midfoot plantar ulcer was 9.7 ± 5.1 months (range, five to twenty-three months), and the average size of the ulcer was 2.8 × 2.9 cm (range, 1 × 2 cm to 4 × 6 cm). In order to be selected for foot reconstruction, the patient had to have continued ulceration despite twice weekly dressing changes and optimized shoe or orthotic fitting. Two patients with progressive radiographic changes had no ulcers, and, as stated above, the indication for surgery was instability that could not be controlled by an orthosis leading to further collapse. We performed transcutaneous oximetry in all patients once the edema had resolved, and no patient had a pressure of <30 mm Hg.
A radiographic assessment included weight-bearing radiographs made preoperatively, non-weight-bearing radiographs made immediately postoperatively, and anteroposterior and lateral weight-bearing radiographs of the foot made at the time of long-term follow-up. The talus-first metatarsal angle was measured on both the anteroposterior and lateral radiographs. A positive value was recorded for the apex of a medial deformity on the anteroposterior radiograph and the apex of a dorsal deformity on the lateral radiograph. The calcaneus-fifth metatarsal angle was measured on all of the lateral radiographs.
We evaluated our patients using a radiographic classification, the acquired midtarsus deformity classification system23, which divides Charcot midfoot deformities into four types (I through IV) according to the major anatomic location of the pathology. Additionally, the severity is classified as either alpha or beta, with the latter implying a greater severity of disease as manifested by the following specific radiographic criteria: (1) a dislocation is present; (2) the lateral talus-first metatarsal angle is =30°; (3) the lateral calcaneus-fifth metatarsal angle is 0° or negative; or (4) the anteroposterior talus-first metatarsal angle is =35°. On the basis of this system, the classification for each of our fifteen patients (Table I) was determined at the time of the index procedure and not over the period of time that the patients were followed preoperatively. The severity was classified as beta in twelve feet because of angular deformity and as alpha in three feet.
The patients were treated surgically with realignment and arthrodesis by the same surgeon (M.A.). All patients were assessed weekly until surgical wound and ulcer healing had occurred and then at six-month intervals. The time to ulcer healing was defined as the length of time needed for complete closure of the skin (epithelialization) with no evidence of drainage or sinus tract formation. Outcome measures included soft-tissue assessment, radiographic evidence of union of the arthrodesis, complications, and need for amputation. Nonunion of the arthrodesis was defined as a lack of radiographic signs of healing six months postoperatively.
Operative Procedures
At the time of induction of anesthesia, all patients receive a second-generation cephalosporin (Zinacef; GlaxoSmithKline) as antibiotic prophylaxis. A thigh tourniquet is used for all procedures, and an occlusive dressing is placed over any open ulcers prior to the start of the operation. The incision is made along the medial aspect of the foot in line with the talus-first metatarsal axis. Blunt dissection is carried down to the bone with careful attention paid to the tibialis anterior tendon and the posterior tibial tendon as it inserts on the navicular bone. Since the anatomy is frequently markedly disturbed, it is important to avoid cutting these tendons. They are not detached from their insertions but are mobilized as needed to gain access to the medial column joints. At times, they are inserting onto a fragment of bone that has become detached from its normal position, in which case it is reattached to the reconstructed medial column.
The initial step is to reduce and realign the medial column and perform a medial column fusion. This fusion involves the talonavicular, navicular-medial cuneiform, and medial cuneiform-first metatarsal joints. All joints to be fused are débrided of cartilage and sclerotic bone, and a reduction is performed by decompression, appropriate wedge osteotomy, and bone removal (either the navicular or medial cuneiform) as necessary. In four patients with the most collapse, a small external fixator was used as a temporary aid to obtain reduction and it was removed before closing the wound. In all patients, internal fixation is performed with a cannulated intramedullary screw (8 mm in diameter and 150 mm in length; Stryker, Selzach, Switzerland) that is placed from the posterior part of the talus into the first metatarsal to stabilize the medial column. This is accomplished by first drilling a guidewire (from the cannulated screw set) in a retrograde fashion parallel to the subtalar joint through the head of the talus exiting posterolateral to the Achilles tendon. In order to control the position of the talus and ensure that the guidewire is centered in the medial column, a Schanz screw can be placed in the talus and used as a joystick. A small incision is then made to reach the tip of the guidewire and to bring it out in order to drill it antegrade. The guidewire is drilled through the head of the talus, the navicular, and the medial cuneiform into the medullary canal of the first metatarsal. The farther it can be placed in the first metatarsal, the better. Currently, the longest screw length available is 150 mm. In our patient series, we calculated the percentage of the length of the first metatarsal containing screw threads. The average length of the first metatarsal was 60.4 mm, and the average amount of screw thread in the first metatarsal was 30.5 mm. Thus, on the average, the screws were inserted into approximately 50% of the length of the first metatarsal. Ideally, even longer screws could be used.
After drilling over the guidewire, the cannulated screw is inserted. It remains in an intramedullary position and does not broach the first metatarsal cortex. Additionally, with the use of a medial column intramedullary screw, we use two 3.5-mm screws placed on the medial side in order to fix the frequently fragmented navicular and medial cuneiform. These screws usually pass from the medial cuneiform to the navicular (or talus) and into the first metatarsal. After stabilizing the medial column as described, we proceed from medial to lateral and assess clinically and radiographically whether there is a need to reduce and stabilize the navicular-second cuneiform joint and the tarsometatarsal joints. If this is the case, we use a dorsal 3.5-mm plate that extends from the talus to the second metatarsal. Then, the lateral column is addressed, and the subtalar and calcaneocuboid joints are approached either through a lateral longitudinal incision or an Ollier-type incision centered over the sinus tarsi. If there is lateral column collapse, a 6.5-mm intramedullary lateral column screw (from the calcaneus to the fourth or fifth metatarsal) is introduced. This is combined with a subtalar fusion with use of a single 6.5-mm screw. Figure 1 shows reconstruction with a medial column screw and additional internal fixation for medial and lateral column collapse and severe deformity through the Lisfranc joint. The specific joints fused and the fixation techniques used in each patient in this series are given in Table II.
Bone graft is used in all patients and consists either of local graft, morselized pieces from the foot (seven patients), or cancellous bone from the proximal part of the tibia (eight patients). No allograft or other bone graft substitute was used. If an equinus deformity (<5° of dorsiflexion) is present, either an Achilles tendon lengthening (eight patients) or a gastrocnemius recession (five patients) is performed, depending on the origin of the deformity. If the equinus deformity can be reduced with the knee in flexion, then a gastrocnemius recession is performed. Otherwise, the Achilles tendon lengthening is performed as an open procedure through a medial longitudinal incision parallel to the tendon. In some patients, an additional soft-tissue release of the posterior subtalar and ankle joints is performed in order to reduce a plantar-flexed talus. As a final step, the occlusive ulcer dressings are removed, the tourniquet is released, and the ulcer is débrided. Necrotic tissue and callus are removed as an ellipse of full-thickness skin around the ulcer to expose bleeding tissue.
Postoperatively, all patients are placed in a non-weight-bearing below-the-knee splint, which is changed at two and six weeks postoperatively and monthly thereafter for a total of four months. The ulcers are managed with wet-to-dry dressing changes twice weekly. After four months, patients are allowed progressive weight-bearing in a removable cast. When the patients are bearing full weight without difficulty in the removable cast, they are placed into an extra-depth wide-toed shoe.
Source of Funding
There was no external funding for this study.
No patient was lost to follow-up. The mean duration of follow-up was forty-two months (range, two to six years) (Table III). The mean time in the removable weight-bearing cast was eight weeks (range, six to ten weeks). Thirteen patients were able to walk with custom-made extra-depth wide-toed shoes with molded inserts. One patient with a contralateral transtibial amputation required the use of a Charcot restraint orthotic walker for balance purposes. This patient continued to be followed at six-month intervals. She routinely walked with a below-the-knee prosthesis on one side and used an orthosis on the side that had been managed operatively. She stated that she was more comfortable with the orthosis.
The mean time for ulcer healing was three months (range, one to four months). In no patient was there a recurrence of plantar ulceration at the time of the latest follow-up. Preoperative, immediate postoperative, and follow-up radiographs were reviewed. The talus-first metatarsal angle reflects the longitudinal arch on the lateral radiograph and forefoot abduction on the anteroposterior radiograph. While the latter was not a major deformity among our patients, the longitudinal arch collapse was severe in all patients. The mean preoperative talus-first metatarsal angle on the lateral radiograph was 31.1°, was reduced to a mean of 1.4° immediately postoperatively, and was a mean of 7.2° at the time of final follow-up. The only patients with >2° of loss of correction were the four with a nonunion. With regard to the lateral column, the mean preoperative calcaneus-fifth metatarsal angle measured 4.1°. It corrected to a mean of 30° on the immediate postoperative radiographs and measured a mean of 28.1° at the time of final follow-up in the group of patients in whom the fusion healed. In the four patients with a nonunion, the final calcaneus-fifth metatarsal angle was a mean of 13°. The mean time to radiographic union of the involved joints was four months (range, three to six months) for the patients who had union of the fusion. Radiographs of the ankle were made routinely throughout the follow-up period, and no patient had Charcot changes develop at the level of the ankle at the time of the last follow-up.
Complications
One patient without a previous ulcer had a surgical wound infection and osteomyelitis develop and became septic despite repeated surgical débridement and organism-specific antibiotic therapy. Radiographs made at four months postoperatively revealed a solid union. An amputation was performed for uncontrollable infection six months after the index surgery. In all other patients, wound-healing was uneventful. Four patients had a nonunion develop, as determined radiographically at six months after surgery. The location of the nonunion was at the tarsometatarsal joint in three feet and at the talonavicular joint in one foot. This was accompanied in each foot by breakage of the 8-mm screw. Of the four patients with a nonunion, three maintained a plantigrade foot and did not have a recurrence of the rocker-bottom deformity. One patient (Case 4) with a nonunion had a substantial loss of correction and a recurrence of the rocker-bottom deformity with an impending midfoot plantar ulceration. For this reason, he underwent a second reconstructive procedure. The only broken screw was the one used for the medial column. It was removed, the deformity was corrected, and a new medial column screw was inserted. Iliac crest bone graft was placed around the nonunion site. Radiographs revealed a solid union after the revision surgery. Another patient underwent a second procedure to correct a severe claw-toe deformity of the first ray that developed postoperatively.
We were able to demonstrate good results with the use of a medial column screw (with adjunctive fixation as necessary) to achieve a well-aligned, stable, plantigrade foot in a group of patients with Charcot arthropathy who had failed nonoperative treatment for recurrent plantar ulceration and/or progressive midfoot collapse. Furthermore, the goal of preventing recurrent ulceration was realized in all patients except one who had a deep postoperative infection develop and required a transtibial amputation six months after the index procedure.
External fixation has been reported to be effective in the correction and stabilization of Charcot foot deformities24-27, with a recent emphasis on the use of neutral ring fixation9,10. In one study of fifty-one adults with nonplantigrade feet at high risk for ulceration10, the procedure reduced the mean lateral talus-first metatarsal angle from 27.6° to 6.4°. At a minimum follow-up of one year, forty-four of fifty-one patients had the desired outcome. In another study of twenty-six high-risk patients with diabetes and multiple comorbidities, fourteen patients had wounds of >2 cm in diameter9. Correction of the deformity was maintained with a neutrally applied three-level ring fixator. At a minimum follow-up of one year, twenty-four of the twenty-six patients were free of ulcers and infections and were able to walk with depth-inlay shoes and custom orthoses. One patient had a below-the-knee amputation for unresolved osteomyelitis, four patients had plantar ulcers develop over osseous prominences, two patients had tibial stress fractures develop, and one patient died two months postoperatively.
Reports on the use of internal fixation for midfoot reconstruction are few, and generally the numbers of patients and length of follow-up are limited28-31. The importance of the type of fixation cannot be overemphasized in the reconstruction of Charcot foot deformities28. Neuropathic bone tends to be hyperemic and osteopenic32, and patients may be noncompliant since they are insensate and do not recognize the harmful effects of bearing weight. The necessity for secure fixation for extensive and multiple joint fusions has led surgeons to seek methods superior to single joint fixation with smaller screws. In vitro studies have shown that a medial plate applied to the plantar bone surface is biomechanically superior to oblique or transverse screw fixation for midfoot stabilization33. However, those were not long intramedullary screws as we used in our study. Furthermore, axial screw fixation was shown to be biomechanically superior to oblique screw fixation in a study examining fusion of the calcaneocuboid joint34, but we have not found similar data for fusion of the medial column. Horton and Olney35 reported that a successful tarsometatarsal fusion was achieved with use of a medial plate in nine patients with tarsometatarsal collapse and degenerative joint disease. However, only one patient had Charcot arthropathy. Because the dissection needed for plate fixation may compromise the blood supply to the bone and increase the risk of wound complications, others have sought effective screw fixation techniques.
Rooney et al.20 introduced a technique of axial intramedullary fixation of the medial column (talus to first metatarsal) and lateral column (calcaneus to fifth metatarsal) using long intramedullary screws, and Sammarco et al. recently reported their results with use of a similar technique36. In this study, we employed an 8-mm screw to provide stronger fixation. A key advantage to such axial intramedullary fixation is that much less dissection is required to insert the screw than is needed to place a medial plate on the tension side of the fusion site.
Radiographic analysis revealed that some degree of arch collapse does occur with time. In our study, the talus-first metatarsal angle on the lateral radiograph increased by a mean of 5.8°. There was no increase in forefoot abduction. Sammarco and Conti also noted medial arch collapse of 7°, in addition to an increase of 5° in forefoot abduction, at the time of long-term follow-up29. In addition, they stated that overcorrection of the deformity at the time of surgery can allow for some postoperative collapse. Johnson2 suggested indefinite bracing when the fusion is limited to joints at or proximal to the talonavicular joint because of the likelihood of Charcot collapse at the midfoot. However, our data show that longitudinal arch collapse occurred only in the patients who had a nonunion develop and had breakage of the screw. In three of our four patients with a nonunion, however, the collapse did not result in a return to a rocker-bottom deformity. In only one patient was a revision procedure necessary because of loss of a plantigrade foot. Our union rate of 73% (eleven of fifteen patients) compares with an 83% union rate (ten of twelve patients) reported by Sammarco et al.36. In the report by Rooney et al.20, regarding forty-three feet in thirty-six patients, there was no specific mention of nonunion.
There are several limitations to our study. First, the study had a small number of patients. Second, our interpretation of instability and a foot that cannot be managed with an orthosis is based on our personal experience as there is no clear definition of this type of instability. Third, we did not use an outcome score based on general patient function and overall health. This might have given our work even more validity since a plantigrade foot and ulcer healing were achieved in fourteen of our fifteen patients, which theoretically should have improved patient function and had a positive effect on their overall health; however, this information was not obtained. Last, it has been suggested that failure of such a reconstructive procedure may well necessitate an amputation; however, in our small series, failure occurred in only one of the fifteen patients. In addition, three of our patients already had an amputation on the contralateral side and, for these patients, a reconstructive procedure that can preserve the foot provides a better alternative to amputation.
Our patients are instructed to remain strictly non-weight-bearing for the first four months postoperatively; this is followed by progressive weight-bearing. This is a lengthy but not unusual amount of time in a setting where recommendations for even longer periods of immobilization have been given21. We realize that such a protocol is difficult as far as patient compliance is concerned, particularly in patients with insensate feet, but all of our patients volunteered that they followed our guidelines closely.
In conclusion, surgical reconstruction of midfoot collapse in a neuropathic foot is a sound alternative to amputation in most patients with diabetes who present with an unbraceable deformity or recalcitrant ulceration. While amputation provides an immediate solution, it markedly increases the energy required for walking in these patients who have the added risk of having a contralateral amputation. Although the procedure is technically demanding, our results show that appropriate, well-planned reconstructive surgery can create a stable plantigrade foot that remains free ofulceration. 