The study population consisted of seventy-three patients who were
operated on with a structural allograft between January 1995 and December
1998. All patients treated with a structural allograft in the foot or ankle
were eligible for inclusion in this study, although those who had been treated
prior to 1995 and who were the subject of a prior study were
excluded16-19.
We retrospectively reviewed the records of the seventy-three patients who had
undergone a total of seventy-five foot and ankle operations with a structural
fresh-frozen femoral head allograft. There were forty-one female patients and
thirty-two male patients with an average age of forty-six years (range, nine
to eighty-two years). The reasons for the use of a structural graft were to
restore more normal dimensions of the foot and ankle following surgery or
trauma and to treat arthritis or deformity in situations in which conventional
cancellous graft would not provide sufficient mechanical support. Use of these
grafts had been planned preoperatively, and no intraoperative findings changed
the plan. Structural graft was also indicated when we wanted to avoid the
shortening associated with the use of cancellous graft, even if union could be
obtained with a cancellous
graft20. The
alternatives to structural bone graft were presented to each patient
preoperatively, and appropriate informed consent was obtained after a full
discussion regarding the risks and benefits of allograft compared with
autograft. All patients were told during their visit with the surgeon that
smoking was a risk factor for nonunion.
A subtalar arthrodesis was performed for salvage following failed treatment
of a calcaneal fracture in twenty-seven patients (twenty-eight feet)
(Table I). Sixteen patients
with flatfoot deformity were treated with lengthening of the lateral column of
the foot, either with an osteotomy of the calcaneus (eleven feet in ten
patients, with three additional feet in three patients also treated with an
osteotomy of the medial cuneiform) or with a calcaneocuboid arthrodesis (three
feet in three patients). Three other patients were treated with a
calcaneocuboid arthrodesis because of a crush injury of the cuboid and the
distal part of the calcaneus. Eight patients were treated with an arthrodesis
of the first metatarsophalangeal joint following failure of a correction of
hallux valgus with marked shortening or osteonecrosis of the first metatarsal
(Figs. 1-A,
1-B,
1-C and
1-D). Six ankle arthrodeses
were performed in six patients, including four who had sustained an ankle
fracture followed by collapse of the joint with either osteonecrosis or severe
bone loss and two who had diabetic neuroarthropathy (Figs.
2-A, 2-B,
2-C and
2-D). A tibiocalcaneal
arthrodesis was performed in five patients with diabetic neuroarthropathy and
in one patient with rheumatoid arthritis and osteonecrosis of the talus. A
fibular osteotomy was performed in four patients to restore length following
malunion of an ankle fracture. A tarsometatarsal arthrodesis was performed in
two patients following trauma and in one patient with a deformity after an
attempted correction of hallux valgus. The height, weight, activity level,
occupation, history of systemic illness, prior surgical procedures, and
smoking status were determined for each patient.
Of the seventy-five procedures, fifty-eight were performed with the use of
intravenous sedation and a regional ankle or popliteal nerve block, and the
remainder were done with the patient under general anesthesia. Preoperatively,
each patient received a prophylactic dose of cephalosporin antibiotics
intravenously. No tourniquet was used in fifty-six procedures, a thigh
tourniquet was used in fourteen, and an Esmarch bandage was used in five.
Each allograft was taken as a corticocancellous block from a fresh-frozen
cadaveric femoral head and neck specimen and was thawed in normal saline
solution for ten minutes before use. The graft was cut to size according to
the location of the arthrodesis or osteotomy. If possible, the thicker
cortical calcar region of the femoral neck was used, and the size and shape of
the graft were contoured with a reciprocating saw. For example, the graft for
an arthrodesis of the hallux metatarsophalangeal joint was obtained from the
margin of the neck of the femoral head, and appropriate dorsiflexion of the
hallux was maintained by the contour of the harvested graft
(Fig. 3). The size of the
structural bone graft was determined after resection and débridement of
all sclerotic and avascular bone back to bleeding margins. A smooth lamina
spreader was inserted into the defect and was adjusted until the desired
correction was noted fluoroscopically. A ruler or a piece of foil from a
suture pack was then used to measure the size of the graft that was needed.
Once the desired correction was obtained, the graft was impacted into place
and the lamina spreader was withdrawn from the osteotomy or arthrodesis
site.
The average heights and lengths of the graft were 14 mm (range, 9 to 20 mm)
and 21 mm (range, 16 to 30 mm) for the subtalar arthrodeses, 10 mm (range, 8
to 14 mm) and 19 mm (range, 16 to 24 mm) for the calcaneal osteotomies, 11 mm
(range, 9 to 12 mm) and 18 mm (range, 17 to 20 mm) for the calcaneocuboid
arthrodeses, 16 mm (range, 8 to 43 mm) and 14 mm (range, 7 to 24 mm) for the
hallux metatarsophalangeal arthrodeses, 24 mm (range, 19 to 36 mm) and 20 mm
(range, 18 to 29 mm) for the tibiocalcaneal arthrodeses, 14 mm (range, 10 to
36 mm) and 22 mm (range, 18 to 28 mm) for the ankle arthrodeses, 16 mm (range,
15 to 19 mm) and 20 mm (range, 18 to 24 mm) for the tarsometatarsal
arthrodeses, 12 mm (range, 9 to 14 mm) and 10 mm (range, 9 to 12 mm) for the
fibular osteotomies, and 7 mm (range, 6 to 8 mm) and 16 mm (range, 15 to 20
mm) for the medial cuneiform osteotomies.
A variety of internal fixation devices were used, depending on the size and
shape of the bone graft, the adjacent bone structure, and the requirements for
obtaining stability in each case. Rigid internal fixation was attempted for
every patient, and a plate and/or screw fixation of some sort was always used,
although the actual construct varied.
The postoperative course was standardized according to the procedure
performed. The limb was immobilized postoperatively in a posterior plaster
splint, and sutures were removed once wound healing had occurred, between ten
and twenty days after the operation. Patients who had undergone surgery of the
midfoot or forefoot were then treated with immobilization in a below-the-knee
cast, with weight-bearing of up to 20 kg for six weeks followed by gradual
progression of weight-bearing in a removable boot or cast until union was
noted. Those who had undergone hindfoot or ankle surgery were not permitted to
bear weight for eight weeks. Patients with neuropathy did not begin bearing
weight until twelve weeks postoperatively, after which they wore a
below-the-knee weight-bearing cast for an additional twelve weeks and then a
weight-bearing boot until fusion was noted radiographically.
The goal of the study was to investigate whether structural allografts
could be used to achieve union. Because it is sometimes difficult to determine
union radiographically, we also used parameters such as a lack of swelling or
warmth in the limb as guides indicating osseous union.
Prevalence of Union
Sixty-nine (92%) of the seventy-five foot and ankle procedures were
followed by successful healing (Table
II). Non-union occurred after two arthrodeses of the
calcaneocuboid joint as well as after one subtalar, one hallux
metatarsophalangeal, and one tarsometatarsal joint arthrodesis and one fibular
osteotomy. Four of the six nonunions were treated with revision, with ultimate
success. One of the two patients with a nonunion after a calcaneocuboid
arthrodesis was minimally symptomatic and had no resorption or collapse of the
graft; no additional surgery was performed. The other patient with a
calcaneocuboid nonunion had fusion between the graft and the calcaneus; the
arthrodesis was revised successfully with cancellous autograft. The patient
with subtalar nonunion smoked two packs of cigarettes a day, had had two
previous surgical procedures on the hindfoot, and was noted to have hard,
dense sclerotic and avascular bone at the time of the arthrodesis. Partial
collapse with resorption of the graft occurred, and revision was attempted
unsuccessfully with cancellous autograft. Ultimately, a triple arthrodesis was
achieved with cancellous autograft and a direct-current internal bone
stimulator (EBI Medical Systems, Parsippany, New Jersey). The patient with the
nonunion of the hallux metatarsophalangeal joint had undergone three prior
operations and smoked ten cigarettes a day. The nonunion occurred at the
distal host-graft interface. The length of the hallux was maintained, and
revision arthrodesis was performed with cancellous autograft harvested from
the ipsilateral
calcaneus21. The
failure of the tarsometatarsal arthrodesis occurred at the proximal host-graft
interface and was successfully revised with cancellous autograft harvested
from the calcaneus. That patient had had two previous surgical procedures on
the foot and smoked one pack of cigarettes a day. In the patient with the
nonunion following the fibular osteotomy, a deep infection developed one month
postoperatively and was treated with débridement and removal of the
allograft. That patient smoked two packs of cigarettes daily and had had one
prior operation. Once the infection was successfully treated, the patient did
not wish to have any additional intervention.
Time to Union
The average time to union was four months (range, two to ten months)
(Table II), with union of the
tibiocalcaneal, ankle, and hallux metatarsophalangeal joints taking longer
(average, 5.6 months; range, three to ten months). Healing was relatively
rapid following the osteotomies of the calcaneus, medial cuneiform, and fibula
and the subtalar arthrodeses (average, three months; range, two to five
months). Healing was judged by the absence of swelling and pain and by
radiographic evidence of trabeculation across the arthrodesis or osteotomy
site on both sides of the graft. Once the graft was integrated, there was no
evidence of graft resorption or collapse at a mean of 3.5 years (range, two to
five years) following the surgery.
Delayed Union
Union was delayed more than four months following twenty-two (29%) of the
seventy-five procedures (Table
II). This figure does not include the six nonunions described
above. Immobilization was continued for each of the patients with a delayed
union, and an external bone stimulator (EBI Medical Systems) was applied for
sixteen of the twenty patients at a mean of 4.5 months (range, four to 5.5
months) after the surgery. The mean time to union in the twenty patients who
had delayed bone-healing was fifteen weeks. Once union had occurred, there was
no evidence of graft resorption or subsidence at a mean of 3.5 years (range,
2.5 to five years) following the surgery. No fracture of the graft was noted
during the study period.
Infection
Seven patients (seven procedures) had wound problems after a subtalar
arthrodesis; these included one deep infection (osteomyelitis), four
superficial infections, and two cases of minor wound dehiscence in the absence
of infection (Table II). The
other procedures were associated with fewer complications: there was one case
of osteomyelitis after a fibular osteotomy, one superficial infection after an
ankle arthrodesis, and one uninfected wound dehiscence after a tibiocalcaneal
arthrodesis. Of the two patients in whom a deep infection developed, one was a
sixty-one-year-old smoker who had had a fibular osteotomy for a prior malunion
after open reduction and internal fixation of an ankle fracture. One month
following the surgery, a wound dehiscence and deep infection with
Staphylococcus aureus developed, and the patient was treated with
débridement of the wound and removal of the allograft and six weeks of
intravenous cephalosporin antibiotics. The second deep infection developed at
two weeks following a subtalar arthrodesis. There was a wound dehiscence,
which was treated with irrigation and débridement of devitalized deep
tissue and coverage with a local rotation abductor digiti minimi muscle flap.
Staphylococcus aureus was grown on culture, and the patient was
treated with intravenous cephalosporin antibiotics for six weeks. Successful
union occurred at four months following the arthrodesis.
Risk Factors
Smoking, diabetes, osteonecrosis, and prior surgical treatment were the
main possible risk factors in the current
study22-24.
Of the seven patients with diabetes, two had a superficial infection and all
seven had problems with osseous consolidation (either a delayed union or a
nonunion). Seventeen patients had osteonecrosis, involving the distal part of
the tibia, the talus, the dorsal surface of the posterior part of the
calcaneus, or the first metatarsal. Three of those patients underwent an ankle
arthrodesis; six, a tibiocalcaneal arthrodesis; five, a subtalar arthrodesis;
and three, a hallux metatarsophalangeal joint arthrodesis. Of the patients
with osteonecrosis, one had a nonunion following a subtalar arthrodesis and an
additional eleven had delayed union following a subtalar (three patients),
ankle (three), or tibiocalcaneal (five) arthrodesis. Of the twenty-eight
operations (37%) followed by nonunion or delayed union, eighteen (64%) had
been performed in smokers. Additionally, of the eleven patients with wound
problems (superficial or deep infection), six were smokers.
Thirty-five patients had had previous foot and ankle procedures (average,
2.4 procedures; range, one to seven procedures). Of the six patients in this
study who had a nonunion, five had undergone multiple procedures (average, 2.5
procedures; range, one to seven procedures). Of the eleven patients with wound
problems, eight, who had undergone various arthrodesis procedures, had had
multiple previous operations (average, 3.6 procedures; range, two to seven
procedures).
The use of allograft bone dates back to the early
1900s25,26.
The first long-term follow-up evaluation showed that these grafts were
partially replaced and incorporated by the host and that joints could be
preserved for as long as twenty years after
surgery27.
Allografts provide the form and matrix of bone tissue, but no viable cells are
transplanted. In addition, bone allografts are incorporated into the host more
slowly than are bone autografts, and they can induce an immune response, which
may delay the osteoinductive phase of bone-graft
incorporation28,29.
Although structural autograft is widely used, it is not without problems.
There may not be a sufficient volume of bone to span a structural defect. More
importantly, there is a substantial risk of morbidity at the donor iliac crest
(fracture, hemorrhage, pain, nerve or arterial injury, and cosmetic
deformity)7,30.
One study4 showed
that use of autograft, in comparison with use of demineralized bone matrix
allograft, resulted in a longer operative time, substantially greater blood
loss, and an overall higher cost to patients.
Concerns with Allograft Use
Structural allografts are weakest during revascularization, and the
mechanical properties of the bone graft may be affected by preservation
techniques3,31-33.
However, no graft fractured during the follow-up period of our study.
Another concern with the use of structural allografts, although rare, is
the transmission of infection. An audit from a bone bank in Leicester,
England34, showed
contamination of femoral head grafts from both live and cadaver donors, and
one of nine large allografts that were implanted was complicated by a clinical
infection. In a review of 303 procedures, there was one case in which a
contaminated allograft was responsible for a clinical
infection35. A few
case reports have noted viral transmission through allografts, with the risks
apparently related to the type of
allograft36,37.
The deep infections that developed in two patients in our study appeared to be
unrelated to the use of the femoral head allograft because no infection
developed in other patients treated with allografts from the same two donors.
Most of the wound problems and infections in our series were associated with
dissection of poor-quality skin and soft tissues in patients who smoked and
who had been treated with multiple prior operations. The prevalence of wound
problems and infections was particularly high after the subtalar fusions in
this series.
It was thought that structural rather than cancellous graft was necessary,
to maintain length and alignment, in the surgical procedures in this study.
The use of structural allograft has been supported in the spine, hand, and
tumor surgery literature, and others have concluded that there is no
significant difference in the prevalence of nonunion between patients treated
with allograft and those treated with
autograft5,11.
In a review of the use of allogenic bone in thirty-six primary joint
arthrodeses,
Zasacki38 reported
that, of ten subtalar joint arthrodeses involving an interpositional
allograft, three did not result in fusion. One of us (M.S.M.) and
colleagues39
reported the results of arthrodesis of the hallux metatarsophalangeal joint
with use of structural bone graft to restore length in twenty-four patients;
there were five nonunions, four of which were treated with iliac crest
autograft and one of which was treated with structural allograft. In a study
by Brodsky et
al.40, fusion was
achieved in eleven of twelve patients treated with an interposition iliac
crest autograft to salvage the hallux metatarsophalangeal joint. The results
of interposition of a structural iliac crest autograft for distraction
subtalar arthrodesis were reported by Carr et
al.17 (with fusion
achieved in thirteen of sixteen patients), by Amendola and
Lammens41 (with
fusion achieved in all of fifteen patients), by Easley et
al.16 (with fusion
achieved in all of twenty-four patients), and by Trnka et
al.42 (with fusion
achieved in thirty-two of thirty-seven patients).
Although there was a limited number of nonunions in the current study, the
prevalence of delayed union was high. It is not clear whether these delayed
unions were the result of the relatively high-risk patient population or a
factor inherent to the structural allograft. These are complex procedures,
and, as noted previously, reports in the literature have not shown
substantially greater success with structural autograft harvested from the
iliac
crest16,17,40-43.
In conclusion, our results suggest that structural allograft can be used to
correct deformity and to fill bone voids in reconstructive surgery on the foot
and ankle. This technique avoids the potential for complications associated
with harvesting autograft, reduces operative time and cost, and can fill large
voids that often cannot be filled with autograft. This approach should be used
cautiously in high-risk patients such as smokers and those with compromised
soft tissues at the surgical site due to multiple previous surgical
procedures. ?