Abstract
Background:
The operative management of failed hallux metatarsophalangeal joint surgery can be difficult. There is often substantial shortening of the first ray. Arthrodesis of the first metatarsophalangeal joint is one treatment option, but results in further shortening of the first ray.
Methods:
We present a large retrospective series of patients who had an interposition bone block arthrodesis procedure performed using a nonvascularized tricortical autologous iliac crest bone graft in an attempt to maintain first ray length. Twenty-four patients (25 feet) underwent this salvage procedure over a 10-year period. The mean follow-up period was 62 (range, 11-117) months.
Results:
Indications included failed hallux valgus surgery, Keller’s procedures, various other first metatarsal osteotomies, and failed arthrodesis procedures with shortening of the first ray. All patients had tricortical iliac crest grafts inserted into bony defects of mean length 10.7 (range, 8-15) mm. Fixation was achieved using either multiple wires or a low-profile plate. Mean lengthening was 4.4 (range, 0-8) mm. Three feet went on to nonunion requiring further revision surgery and at latest follow-up 2 had united. All patients underwent removal of a stabilizing Kirschner wire. Median patient-reported outcome score using the Foot and Ankle Disability Index was 84.6 (range, 37.5-97.1).
Conclusion:
These results suggest this procedure was successful in salvaging failed first ray surgery and prevented further shortening. We believe good results can be achieved, but care must be taken to ensure wound closure without excessive tension to permit wound healing.
Level of Evidence:
Level IV, retrospective case series.
Failed first metatarsophalangeal joint (MTPJ) surgery can result in deformities of the first ray which are often painful. Commonly performed procedures include various metatarsal realignment osteotomies for mild to moderate hallux valgus deformity, arthrodesis of the first MTPJ for severe deformity and arthrodesis/replacement of the first MTPJ for the treatment of hallux rigidus. Primary or subsequent procedures can result in shortening/bone loss, which can be severe, making revision surgery increasingly challenging.
A significant proportion of the body’s weight is taken through the forefoot during normal gait and up to 50% is taken by the first metatarsal (MT) during toe-off. 6 With increasing bone loss the first MT shortens and the biomechanics of the forefoot changes resulting in a transfer of the load-bearing function of the first MT to the lesser MTs. This can result in transfer metatarsalgia and intractable plantar keratosis under the lesser MT heads.6,7,20,21 For subsequent surgery to be successful, the surgeon must restore the biomechanics of the medial longitudinal arch, or at the very least avoid further shortening of the first ray. 19 The aim is to achieve a stable, pain-free, and well-aligned first MTPJ while restoring the load-bearing function of the first MT.3,15
Traditionally, MTPJ arthrodesis has been used as a salvage procedure in an attempt to prevent further shortening.6,12,19 However, arthrodesis alone can result in further shortening in cases where there have been failed previous procedures, as a significant amount of bone must be resected to reach healthy bone capable of achieving fusion. The available methods of avoiding transfer metatarsalgia in these circumstances are shortening osteotomies of the lesser MTs at time of arthrodesis of the first MTPJ and distraction arthrodesis of the first MTPJ, achieving lengthening by means of an interpositional bone block.4,5,12,16,17,22
The largest series thus far of bone block arthrodesis was described by Myerson et al in 2000. 16 The authors presented a case series of 24 patients who underwent distraction arthrodesis using either autograft or allograft. There have been a number of smaller case series since then on this topic with promising results.4,5,22 We present our experience using interpositional bone block arthrodesis as a salvage procedure for revision hallux MTPJ surgery. Our aims were to review the clinical, radiological, and functional outcomes in these patients.
Methods
Patient Demographics
A retrospective review was conducted of all cases of interposition bone block arthrodesis (IBBA) performed by the senior author over a 10-year period from 2001 to 2011. Local approval for collection and publication of data was obtained from our hospital’s audit and research department. Twenty-four patients (25 feet) were operated on in total; 18 patients were female and 6 were male, with a mean age of 52.1 (range, 34-69) years. Surgery was performed on the left foot in 14 cases and in the right foot in 11. In total 9 patients were smokers and 3 had rheumatoid disease. None had diabetes or peripheral vascular disease. No patients were excluded. The mean follow-up for all patients was 62 (range, 11-117) months.
All patients had failed prior surgery on their first MT, and our indication for IBBA was an acquired shortened first ray and transfer metatarsalgia. Shortening was determined radiologically and clinically, by comparison with the other side. All patients in this series had a degree of transfer metatarsalgia. In 11 cases the primary procedure was a Keller’s procedure, 5 of which developed a “cock-up” deformity. In 7 cases the primary procedure was a failed first MTPJ arthrodesis. Five cases had other failed first MT osteotomies. One patient had a symptomatic first MTPJ arthroplasty (MOJE ceramic press fit arthroplasty) and 1 developed severe avascular necrosis of the first MT head following hallux valgus correction. Four patients had failed fusions after their primary procedure prior to IBBA being performed, and 1 patient who had a failed arthrodesis had a subsequent failed excision prior to IBBA. Eleven patients had a concurrent procedure performed, mostly on the second toe. These data are summarized in Table 1.
Patient Demographics.
Abbreviations: MT, metatarsal; MTPJ, metatarsophalangeal joint; PIPJ, proximal interphalangeal joint. Patient numbers are the same as in Table 2. Procedures marked * had subsequent fusions; procedure marked ** had a subsequent Keller’s excision with interposition of capsule. Patients are listed in chronological order.
Operative Technique
All procedures were performed by the senior author. Patients were supine with an upper thigh tourniquet, and procedures were performed with fluoroscopic guidance. All patients were treated using tricortical autograft from the ipsilateral iliac crest. Allograft was not used in any cases.
The incision was over the dorsomedial aspect of the hallux; when possible the old incision was used. During the approach the extensor hallucis longus tendon was preserved with a cuff of fascia and retracted. Old scar tissue was meticulously excised, and the MT head prepared by removing articular cartilage if present. Bone was resected until good quality cancellous bone was found. Multiple drill holes were then made into both the MT and proximal phalanx.
Skin retraction was removed and gentle traction was applied to the hallux. The resulting gap was then measured to determine the bone block dimensions. Although the desired goal was to restore the normal length of the first ray, this was often impractical. Particular care was taken not to overstretch any soft tissues, which could lead to skin ischaemia and wound breakdown as we believe that tension free closure of the soft tissue envelope is vital. Tricortical and cancellous graft was then harvested from the ipsilateral iliac crest and contoured to ensure a secure fit. Multiple through and through drill holes were made in the bone block with a 2 mm Kirschner wire before it was slotted into the fusion site. This was performed through the cancellous surfaces that faced the prepared surfaces of the MT and the proximal phalanx.
The toe was placed in the desired position, and the graft was temporarily secured with 2 Kirschner wires. We aimed to correct any residual hallux valgus and adjusted the dorsiflexion angle by placing the foot in a plantigrade position on a metal foot plate placed on the operating table. Our preferred MTPJ fusion position allowed the pulp of the hallux to rest just above the surface of the foot plate.
In 8 cases fixation was carried out using a 5-hole dorsal plate aiming to fix 2 screws through the MT, 1 screw through the proximal phalanx and the fourth screw through the bone graft where possible. This fixation was augmented by means of a further Kirschner wire placed obliquely through the proximal phalanx, graft, and MT to gain more secure purchase. In the other 17 cases a plate was not used and fixation was achieved by means of a total of 4 or 5 Kirschner wires. Our choice of fixation was decided by the remaining length of the proximal phalanx after resection of bone. Where this fragment was too short for plate fixation, wires only were used. Where plates were used, we used Hallu®-Fix C & S plates (Newdeal®, Integra™ LifeSciences Corporation, Plainsboro, New Jersey) for our fixation as they were low-profile, precontoured, nonlocking plates, providing the required rigidity without being too prominent. The Hallu-Fix C plate was used where there was sufficient length and bone stock in the proximal phalanx. We found the dorsiflexion built into the Hallu-Fix C plate to be excessive, and so we straightened it prior to insertion. In cases where we used the Hallu-Fix S plate, the distal phalangeal end needed to be slightly flattened.
The wires were then cut short beneath the skin, close to bone to avoid skin irritation. Gaps on either side of the fusion were packed with additional cancellous bone from the iliac crest. The wound was closed in layers (subcutaneous and skin). Any concurrent procedures were then performed.
In 4 patients with previously failed fusions, Osigraft (eptotermin alfa; Howmedica, Limerick, Ireland) recombinant human BMP-7 was used intraoperatively. These patients were also given a pulsed electromagnetic field bone stimulator (PEMF) to use postoperatively. In 1 patient with a previously failed fusion we used a PEMF bone stimulator postoperatively, but no Osigraft intraoperatively.
Postoperatively patients were allowed to mobilize heel weight-bearing in a wedged shoe for 6 weeks. They were seen at 10 days in the clinic for a wound check and dressing changes and subsequently seen at 6 weeks with radiographs of their foot when weight-bearing status was decided. If the swelling had subsided and there were no adverse radiological features, patients were allowed to mobilize fully weight-bearing in their normal footwear. If there were any concerns, they were kept weight-bearing in the wedge shoe for a few weeks longer. Due to local irritation, all patients required removal of the stabilizing Kirschner wire at 3 to 4 months postoperatively, which was performed either under a general anesthetic in the operating room (21 patients) or under local anesthetic in the clinic (4 patients). This was considered a planned procedure, and patients were informed of this prior to the initial surgery. The key steps of our procedure are summarized in Figure 1.
The key stages in our procedure: (A) Joint exposed by dorsomedial incision and retracting EHL; (B) bone resected to healthy margins and gap measured; (C) suitably sized tricortical bone block harvested from ipsilateral iliac crest; (D) bone block placed in situ and toe positioned; (E) fixation achieved.
Radiological Evaluation
All patients had preoperative weight-bearing anteroposterior (AP) and lateral radiographs of the foot and a series of postoperative radiographs to determine union. From 2007 onward radiographs were stored digitally. For radiographs older than 2007, hard copy films were obtained. All radiographic measurements were adjusted for a magnification of 110% as is standard in our department. Radiographs could not be obtained for patients operated in 2001 and 2002 as radiographs prior to 2003 had been destroyed.
For the 21 patients with adequate radiographs, we determined time to radiological union and lengthening achieved. Lengthening of the first ray was determined by comparing the perpendicular distances between the proximal end of the distal phalanx of the first toe and the proximal end of the proximal phalanx of the second toe on the pre- and postoperative AP radiographs (Figure 2). These points were chosen as they were unaltered by surgery in most of our patients. In 2 patients with concurrent second ray surgery, the proximal phalanx on the third toe was used instead. The postoperative radiograph used for measurement was one in which radiological union had occurred (the most recent radiograph prior to revision was used in cases of nonunion). For radiographs available digitally, measurements were confirmed by a second observer blinded to the graft length and measurements of the first observer. The differences between measurements taken by the first and second observer were within 1 mm in all cases. For the 4 cases we could not obtain radiographs for, time to union was determined from documentation of radiographic and clinical findings in the clinic annotations from the original follow-up.
Lengthening was determined as the difference between (b) and (a)—the perpendicular distances between the fixed points of the proximal ends of the proximal phalanx of the second toe and the distal phalanx of the first toe in the pre- and postoperative AP radiographs.
Patient-Reported Outcomes
We used a validated patient-reported outcome tool known as Foot and Ankle Disability Index (FADI).9,14 This is a subjective assessment of our patients’ daily function (activities that can be performed) and levels of pain. In addition we also asked patients whether they felt they were satisfied with the results of the procedure overall. Patients previously discharged were contacted for follow-up questionnaires. We were unable to collect survey data on 2 patients: 1 died of an unrelated cause and the other did not wish to complete any questionnaires.
Results
Clinical and radiological fusion occurred in 22 out of 25 feet (88%) with a median time to fusion of 4 (range, 2-24) months. The mean tricortical graft length used was 10.7 (range, 8-15) mm. The graft lengths were obtained from the operative notes. For those in whom fusion was achieved and radiographs were available (21 patients), the mean lengthening was 4.4 (range, 0-8) mm. Figure 3 demonstrates the pre- and postoperative radiographs of a patient who successfully fused.
A series of AP radiographs demonstrating the pre- and postoperative appearances of a patient who successfully fused.
Twenty-two out of 24 patients returned surveys (23 out of 25 feet). The median score of patients was 84.6 out of a maximum of 100 (range, 37.5-97.1). Six patients scored below 70; 3 of them had limitations due to painful nonunions, 2 had painful neuromas, and 1 had symptomatic hardware. Eighteen of 22 patients (82%) could perform activities of daily living with little or no pain. Recreational activities were limited in most patients, and all had the expected difficulty with standing on their toes. Eighteen of 23 operated cases (78%) were satisfied. No patient complained of persistent transfer metatarsalgia. These results are summarized in Table 2.
Results and Complications.
Abbreviation: FADI, Foot and Ankle Disability Index. Patient numbers are the same as in Table 1.
Three patients went on to nonunion and eventually required revision fusion at a mean time of 16 (range, 6-23) months. One died of unrelated causes prior to fusion; the other 2 successfully fused. All 3 had either a primary or subsequent failed fusion of their first MTPJ. Two of these had been using a PEMF bone stimulator, 1 was a smoker, and 1 was a smoker and had rheumatoid arthritis. In all 3 cases the nonunion occurred at the distal interface of the fusion. Two patients had delayed union. They were initially thought to have painless nonunions but subsequently required removal of symptomatic hardware. On radiographs taken at between 18 and 24 months after their fusion surgery, they were found to have progressed to solid fusion radiologically, and this was confirmed at surgery.
There were no cases of superficial or deep infection. Two out of 24 patients (8%) developed painful neuromas related to the operation site (requiring localized denervation of the dorsomedial cutaneous branch of the hallux) and 1 out of 24 patients (4%) developed chronic pain at the graft harvesting site. Six patients had some persistent discomfort at the graft harvesting site, which did not affect their daily activities or sleep. Three patients had symptomatic hardware and required plate removal. All patients required removal of the stabilizing Kirschner wire or wires either under general or local anesthetic.
Discussion
Arthrodesis of the first MTPJ is a well-documented treatment for hallux rigidus but has also been successfully used to treat severe hallux valgus and failed hallux valgus surgery such as Keller’s excision arthroplasty and corrective MT osteotomies.2,8,22 Arthrodesis has less morbidity than IBBA and may be well tolerated despite a moderate amount of shortening of the first ray. However, Schuh and Trnka have suggested that the result is less satisfactory in cases with severe bone loss, particularly when shortening is associated with transfer metatarsalgia. 19
IBBA aims to achieve arthrodesis while preserving/restoring length in these cases and has been used successfully as a salvage procedure by a number of authors.4,5,16,22 Myerson et al reported on 24 cases where fixation was undertaken with a combination of wires, screw and plates. The authors used autograft initially and subsequently changed to femoral head allograft. 16 Brodsky et al and Bhosale et al reported 12 and 10 cases, respectively, and used plates to achieve fixation in all cases.4,5 We used a stabilizing wire in all cases and either a plate or multiple wires to achieve fixation. Several methods of fixation for arthrodesis have been used and include dorsal plates with a lag screw, a single lag screw, dorsal plates only, and Kirschner wires. Biomechanical studies comparing the relative strengths of these types of fixation have found that dorsal plates in combination with lag screws provide the most stable construct. Lag screws alone provided a less rigid construct, followed by dorsal plates alone and finally Kirschner wires.10,11,18 McKeever and Bennett et al noted that the clinical outcome depends on successful fusion occurring and a satisfactory final position of the joint, rather than the method by which it is stabilized.3,15 Our series certainly lends support to this observation. The majority of our patients had fixation by wires alone. We believe that this method of fixation was necessary due to poor bone stock and was the least likely to cause soft tissue tension.
Our choice of graft was tricortical autologous graft from the ipsilateral iliac crest. Arguments against the use of an autograft would be the need for a further procedure with its associated risks. However in our series there was just 1 case of chronic graft donor site pain, requiring management by a specialist pain team. Autograft is cost-effective and has osteogenic, osteoinductive, and osteoconductive properties. 13 It is easy to harvest and provides an excellent mechanical strut allowing fusion to occur while maintaining length and has remained our preferred choice for this procedure. We supplemented our graft with Osigraft/PEMF bone stimulation in patients with a failed primary or secondary MTPJ arthrodesis in whom there was no apparent mechanical cause for initial nonunion. These patients were also noted to have poor bone quality at the time of surgery. The patient who had PEMF bone stimulation only was operated on for a failed fusion with no obvious mechanical cause for failure; however at surgery the quality of bone stock was good, and so there was no indication for Osigraft. These were patients in whom we felt the biology could be improved with the use of such adjuncts. 1
The union rate for IBBA reported in the literature has ranged from 79% to 91%, whereas the union rate for a standard arthrodesis has been variously reported but is generally accepted as 85% or greater.2,3,6,11 Our union rate was 88%. Apart from nonunion several complications have been reported in the literature with this technique including infection, symptomatic hardware, wound breakdown/necrosis, need for a flap to cover the wound, and scar sensitivity.4,5,16 It has been speculated that excessive stretching of the soft tissues may result in ischemia and increase the incidence of wound complications. 5 The degree of lengthening achieved in the literature has varied. Brodsky et al achieved a mean lengthening of 7.5 mm, 5 whereas Myerson et al achieved a mean lengthening of 13 mm using a mean graft length of 22 mm. 16 To accommodate this lengthening Myerson et al performed lengthening of the extensor hallucis longus tendon. 16 Skin quality is often very poor in patients with multiple previous first ray procedures and scaring of soft tissue. We downsized our bone block as necessary to allow easy approximation of skin edges and so used a mean graft length of 10.7 mm, achieving a mean lengthening of 4.2 mm. The discrepancy in the graft length and actual lengthening achieved is due to the amount of bone resected during the course of surgery which would have otherwise caused shortening. There was no change in length between the immediate postoperative radiographs and the final follow-up radiographs. None of our patients were shortened. The complication rate in the literature (excluding nonunions) has been reported as 20 to 25%.4,5,16 Our comparable complication rate is 12%, and notably we had no cases of wound break down.
In our series we did not aim for specific hallux valgus (HVA) and dorsiflexion angles (DFA). Some authors have suggested that the ideal HVA is between 5 and 10 degrees, and the ideal DFA is between 20 and 25 degrees.2,5,6 Other articles suggest that achieving a good functional position is more important. 3 We adjusted the HVA to ensure that the space between the first and second toes was as near normal as possible, and determined the DFA as described above. We found the DFA built into the precontoured plates we used to be excessive and so routinely straightened them slightly to achieve the desired DFA prior to its application.
One drawback of our technique was that all patients required removal of the stabilizing wire once union had occurred. This was done under local anesthesia in some cases in clinic, but often required a further general anesthetic. We cut the wires close to the bone so as to avoid local irritation, which was not a problem until the wire or wires began to back out 3 to 4 months after surgery. In the 4 patients who had removal of wires in clinic, the stabilizing wire had backed out significantly and was close to the skin. When we used multiple wires for fixation, we removed all the wires if they had backed out sufficiently to cause irritation (8 cases). In addition, despite our use of a low-profile plate, 3 patients required removal due to symptoms once fusion had occurred.
Despite our relatively lesser degree of lengthening, 78% of our patients were satisfied with their surgery and 82% could perform their activities of daily living with little or no pain. A number of patients even went back to recreational activity. A number of previous studies have used objective scoring systems to measure patient outcome. We opted to use an exclusively patient-reported outcome measure, the FADI. This has been validated against other measures and was felt to be appropriate for use in this group of patients.9,14 We acknowledge that a patient-reported outcome measure is more useful when compared to a preoperative score. However, due to the retrospective nature of our series we were not able to obtain these data in a significant proportion of patients. Nevertheless, we feel that the score may still be useful as a stand-alone value for future research to compare to. As ours was a partly retrospective series, we were not able to obtain old radiographs and all data for every patient, but we feel that we are still able to draw valid conclusions. Having used a variety of fixation methods, we appreciate that it is difficult to directly compare our series with smaller series that have used a single method of fixation. We believe our series demonstrates that satisfactory outcomes can be achieved with different fixation methods.
In conclusion, IBBA was a salvage procedure for painful forefoot conditions with severe bone loss. It has been successfully used by a number of authors and our experience supports this. We have found it to have fusion rates and satisfaction rates comparable to primary arthrodesis procedures. 3 Our experience would suggest that significant lengthening is probably not required to achieve satisfactory results.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.