Osteonecrosis of the femoral head following trauma is a rare yet severe
disorder in children, occurring after an unstable slipped capital femoral
epiphysis, a femoral neck fracture, or a traumatic hip dislocation. The
reported prevalence of osteonecrosis in patients with a slipped capital
femoral epiphysis is 3% to 58%, depending on the severity and stability of the
slip and the initial
treatment1,2.
There are severe long-term consequences, such as femoral head collapse and
degenerative changes of the hip, but no clear guidelines for
treatment3.
Osteonecrosis of the femoral head has also been shown to occur in up to 40% of
hips with a femoral neck fracture and 12% of those with a dislocation and is
associated with a poor clinical
outcome4-7.
Bone resorption and femoral head collapse can be rapid in children with
femoral head osteonecrosis. Osteoclasts resorb necrotic bone during
revascularization in an attempt at repair; however, this usually occurs before
new bone has been laid down in a sufficient amount to maintain structural
integrity8,9.
Attempts have therefore been made to slow osteoclastic resorption with
nitrogen-containing bisphosphonates, in the hope of maintaining femoral head
sphericity and allowing eventual revascularization and ossification without
collapse of the femoral head.
In a bone-chamber study of rats, Astrand and
Aspenberg10
documented that high-dose systemic alendronate treatment led to the retention
of trabecular bone structure and encouraged new bone formation onto structural
frozen bone grafts, indicating that repair was possible in this model as long
as the bone template was present. One of us (D.G.L.) and
colleagues11
applied the same hypothesis in a rat model of traumatic osteonecrosis of the
femoral head treated with zoledronic acid, a high-potency bisphosphonate.
Parenteral treatment with zoledronic acid resulted in better preservation of
both histological architecture and macroscopic femoral head sphericity
compared with those seen in control animals, and prophylactic treatment led to
a further improvement toward the condition in intact controls. Kim et
al.12 expanded
these concepts to a large-animal (piglet) ischemic osteonecrosis model, in
which parenteral ibandronate treatment was effective in increasing the
retained trabecular number and bone volume of the femoral head, with improved
preservation of the epiphyseal quotient (the height of the femoral head
divided by its diameter). Treatment resulted in an improvement compared with
the status of control ischemic femoral heads, but the results were still not
identical to those in intact controls; prophylactic treatment provided a
further improvement in outcome.
We are aware of three recent human trials of alendronate for the treatment
of femoral head osteonecrosis in adults. In a case series of sixty patients
(100 hips) with osteonecrosis who were treated with alendronate, and in which
twelve patients had a history of trauma, Agarwala et
al.13 concluded
that early surgical intervention could be avoided in most patients. Lai et
al.14 assessed
alendronate for the treatment of nontraumatic osteonecrosis of the femoral
head in a randomized controlled trial with a minimum duration of follow-up of
twenty-four months. Only two of twenty-nine femoral heads in the group treated
with alendronate collapsed (with one requiring a subsequent total hip
arthroplasty) compared with nineteen of twenty-five hips in the control group
(with sixteen needing arthroplasty). Nishii et
al.15 treated
twenty hips with osteonecrosis of the femoral head with alendronate for one
year and compared the results with those in thirteen control hips that had not
received treatment. In the alendronate group, the biochemical markers of bone
resorption (N-telopeptide of Type-I collagen) decreased more than did those of
bone formation (alkaline phosphatase). The alendronate group also showed a
lower frequency of femoral head collapse (5% compared with 46% in the control
group) and reported less hip pain at one year.
We are not aware of any previous reports on the use of bisphosphonates for
the treatment of osteonecrosis of the femoral head after trauma in children.
The goal of this prospective series was to determine whether bisphosphonates
could preserve femoral head sphericity in children with traumatic
osteonecrosis. At our institution, our clinical experience has been focused on
the use of pamidronate and zoledronic acid for pediatric orthopaedic
conditions such as osteogenesis imperfecta and osteopenic disorders.
Furthermore, the preclinical evidence was based on the results of parenteral
administration of either high-dose or high-potency bisphosphonates. We
therefore decided to carry out a prospective trial of these bisphosphonates as
an adjunctive treatment for our patients.
Approvals
The Hospital Pharmacy Committee at our institution considered the protocol
for the treatment of traumatic osteonecrosis with bisphosphonates and granted
approval on an individual basis for patients who had provided written informed
consent. Approval for review of this prospective case series was also obtained
from the local Institutional Ethics Committee. All of the study patients
presented to our institution between October 2000 and December 2003.
Treatment Protocol
Before treatment was commenced, we instituted a protocol for identifying
patients with femoral head osteonecrosis, as the preclinical evidence had
suggested that early treatment would be necessary. Technetium-99m bone scans
have been documented to reliably predict femoral head osteonecrosis in this
patient group16.
Therefore, all children who were at risk for the development of osteonecrosis
underwent bone scans as soon as feasible after surgical treatment.
Technetium-99m conjugated with methylene diphosphonate was injected
intravenously, and a standard three-phase whole-body bone scan was obtained,
including pinhole images of both hips. The delayed-phase pinhole images were
assessed for ischemia of the femoral head by a nuclear medicine physician.
Single-photon-emission computed tomography (SPECT) was used in selected cases
as determined by the nuclear medicine physician. One patient's scan was
equivocal and was repeated three days later. The repeat scan showed no
osteonecrosis, and the patient was not included in the treatment group.
Differentiation between partial and total femoral head osteonecrosis was not
feasible with this technique.
The protocol for commencing bisphosphonate therapy during the study period
is shown in Figure 1. If the
bone scan did not show ischemia (i.e., was "hot"), the patient was
observed and no adjunctive therapy was offered. If the bone scan showed
ischemia (i.e., was "cold"), the patient was offered adjunctive
treatment with bisphosphonates. Patients with osteonecrosis were advised to
remain partially weight-bearing for the first year, but compliance was not
measured. When a patient had a slipped capital femoral epiphysis, the
stability of the slip was classified according to the definitions described by
Loder et al.1. The
slip was classified as unstable if the patient had such severe pain that
walking was not possible, even with crutches, regardless of the duration of
symptoms. The slip was classified as stable if walking and weight-bearing were
still possible with or without crutches. Only patients with an unstable
slipped capital femoral epiphysis were considered to be at risk for the
development of osteonecrosis.
Treatment with pamidronate (ten patients) or, later, zoledronic acid (seven
patients) was begun within three months after the initial traumatic event.
Bisphosphonates were administered intravenously in order to improve absorption
and bioavailability and to ensure patient compliance. The primary rationale
for changing to zoledronic acid when it became available to us was a
theoretical increase in potency and shorter infusion times. Prospective
clinical and radiographic evaluations were carried out every three months
following the initiation of the bisphosphonate therapy.
Assessment
The primary clinical outcome measures were the Harris hip
score17, the Iowa
Hip Rating18, and
the Global Pediatric Outcomes Data Collection Instrument (Global
PODCI)19,
administered at the time of final follow-up by a pediatric orthopaedic fellow.
The sphericity and congruence of the affected hips were assessed on the final
follow-up radiographs with use of the consensus radiographic classification
described by Stulberg et
al.20 (see
Appendix). This scoring system is widely used to assess radiographic outcomes
in hips affected by Legg-Calvé-Perthes disease, and it has acceptable
intraobserver
reliability21. It
is useful for assessing outcomes in adolescents with femoral head
osteonecrosis, as sphericity and congruence are of paramount importance in the
evaluation of long-term outcomes of treatment of any pediatric hip condition,
and we could find no other available validated scoring system. We also applied
the staging system of
Ficat22, as it is
more commonly used to evaluate femoral head osteonecrosis in adults. As our
patients all presented with an acute traumatic event, they all had a Ficat
stage-1 hip at presentation. All radiographs were also reviewed for the
presence of femoral head resorption.
Adverse Effects
Because of concern over theoretical adverse renal effects, including
nephrocalcinosis and renal insufficiency, all patients had a baseline renal
ultrasound examination and renal function blood tests before treatment.
Because of the risk of hypocalcemia, total serum calcium levels were monitored
on the first and third days following infusion of the
bisphosphonate23.
In addition, the patients were monitored clinically every three months for
growth abnormalities and any other rare bisphosphonate-associated
complications, such as uveitis and osteonecrosis of the jaw.
Statistical Analysis
Means and standard deviations are reported. Unpaired t tests were used to
compare the primary outcome measure scores between the subgroups in the study.
The Fisher exact test was used for tests of proportions. Paired t tests were
employed to compare patient height Z scores at the commencement of
bisphosphonate treatment with those at one and two years following
treatment24. A p
value of <0.05 was selected as the level of significance for this study
before the data analysis.
Of the twenty-eight patients at risk, seventeen (thirteen boys and four
girls with a mean age of 12.7 years at the onset of the disease) were
diagnosed with femoral head osteonecrosis following initial surgical treatment
and subsequently were treated with intravenous bisphosphonates. This group
included twelve of twenty-two patients with a severe unstable slip of the
femoral capital epiphysis, four of four patients with a transcervical hip
fracture, and one of two patients with a hip dislocation seen at our
institution during the study period (Fig.
1). All of the slipped capital femoral epiphyses had been unstable
at the initial presentation to either the referring hospital or our unit, and
an attempt had been made to pin the femoral head in situ. Five of these index
procedures had failed at other institutions, and therefore these patients
underwent a single subsequent operation (repinning in situ and injection of
bone marrow from the ipsilateral iliac crest) at our institution before
starting bisphosphonate treatment. The transcervical hip fractures were all
displaced and were treated with closed reduction and cannulated screw
fixation. The hip dislocation was posterior and treated with closed
reduction.
All patients who were entered into the study had a "cold"
femoral head on the delayed pinhole image produced with the technetium-99m
bone scan. Typical appearances were a variable decrease in isotope uptake on
the blood pool phase and a complete absence of isotope uptake in the femoral
head and growth plate on the affected side on the delayed pinhole images.
During the study period, twenty-two patients with an unstable slipped capital
femoral epiphysis presented to our institution. Ten of them had a
"hot" femoral head on bone scanning and were therefore not offered
bisphosphonate therapy. All had a Stulberg Class-I outcome with no
complications at the time of follow-up, at a minimum of two years. Of the four
patients presenting with a femoral neck fracture, all had femoral head
osteonecrosis, as did one of two patients presenting with a traumatic hip
dislocation.
The average duration of bisphosphonate treatment was 20.3 months (range,
seven to thirty-nine months). The first ten patients in the study received a
dose of 1.0 mg/kg of intravenous pamidronate over three to four hours on
alternate months. Later, as patient responses and safety data were reviewed,
this dose was increased to 1.5 mg/kg on alternate months. The average number
of infusions of pamidronate was 9.6 (range, seven to thirteen), with an
average total dose of 10.8 mg/kg (range, 3.9 to 18.2 mg/kg). The last seven
patients were treated with zoledronic acid at a dose of 0.025 to 0.05 mg/kg
(maximum, 2.0 mg) given over thirty minutes after the initial diagnosis, at
six weeks after the injury, at three months after the injury, and every three
months thereafter. The average number of infusions of zoledronic acid was 8.1
(seven, eight, or nine), and the average total dose was 0.23 mg/kg (range,
0.17 to 0.34 mg/kg).
The clinical and radiographic outcomes are presented in the Appendix. At an
average of 38.7 months (range, 25.6 to 58.3 months) following the start of
bisphosphonate therapy, fourteen of the seventeen patients were completely
pain-free. Thirteen of the seventeen patients walked with a clinically normal
gait. Of these thirteen patients, nine had equal limb lengths and four had a
1-cm limb-length discrepancy. Three patients had a short-leg gait (2 cm of
limb-shortening in two patients and 2.5 cm in one), and one had a
Trendelenburg gait due to hip abductor muscle dysfunction. The mean Harris hip
score was 91.2 ± 11.2 points (range, 65 to 100 points), the mean Iowa
Hip Rating was 92.1 ± 8.9 points (range, 66 to 100 points), and the
mean Global PODCI score was 91.5 ± 8.3 points (range, 77 to 100
points).
At the latest radiographic follow-up evaluation, nine hips were rated as
Stulberg Class I or II; six, as Stulberg Class III; and two, as Stulberg Class
IV (see Appendix). No areas of femoral head resorption were seen on the
radiographs in nine of the seventeen patients. The absence of noticeable areas
of resorption in the femoral head on radiographs was associated with an
increased number of femoral heads being classified as Stulberg Class I or II.
Eight of nine femoral heads without resorption had a Stulberg Class-I or II
outcome. In contrast, only one of eight heads displaying resorption had a
Stulberg Class-I or II outcome (p < 0.01). The absence of femoral head
resorption was also associated with an improvement in the mean Harris hip
score (96.0 points compared with 85.8 points for the hips with resorption; p
< 0.05) and the mean Global PODCI score (96.0 compared with 86.4 points; p
< 0.01), although, with the numbers studied, the difference in the mean
Iowa Hip Rating was not significant (94.0 compared with 89.9 points; p =
0.18). Case examples are shown in Figures
2 and
3.
Four patients with a slipped capital femoral epiphysis eventually underwent
a flexion/valgus femoral osteotomy for the treatment of posterior angulation
and impingement in flexion, and one additional patient was a candidate for
such intervention at the time of writing. In two patients with a slipped
capital femoral epiphysis, a stable slip of the contralateral hip developed
during the study period, one at six months and the other at seven months
following the initial slip. Both cases were treated with pinning in situ
without complications.
After the initial bisphosphonate infusion, most patients had an acute-phase
response, including fever (nine patients), headache (ten), nausea or vomiting
(thirteen), or general malaise (thirteen), which lasted less than seventy-two
hours and had no apparent long-term effects. Hypocalcemia (<2.1 mmol/L)
occurred after the initial dose of bisphosphonates in thirteen of the
patients, but it did not occur following subsequent doses. There were no
adverse renal effects and no clinically symptomatic cases of uveitis or
osteonecrosis of the jaw.
Follow-up bone scans were performed on all patients, typically ordered
between four to six months following the injury and repeated until photopenia
was no longer seen on the pinhole views (Figs.
2 and
3). In one patient with a
basicervical femoral neck fracture, photopenia was still present at seventeen
months after the injury. For the remaining sixteen patients, the mean time of
documented revascularization was 6.3 ± 3.1 months. In one of these
patients, the growth plate signal partially recovered, but the remaining
patients showed increased uptake in the femoral head, no growth plate signal,
and variable changes attributable to remodeling in the femoral head and
neck.
Analysis of the growth data for the study patients revealed no adverse
effects. The height Z scores actually increased from a baseline mean of 0.09
± 1.97 at the initiation of treatment to a mean of 0.43 ± 1.62
at one year (p < 0.05) and a mean of 0.90 ± 1.63 at two years (p
< 0.01) following treatment.
The rationale for adjunctive bisphosphonate therapy for adolescents with
femoral head osteonecrosis after an injury is that it may slow resorption of
necrotic bone (catabolism) and allow revascularization and subsequent bone
formation (anabolism) to occur before there is collapse of the femoral head.
Investigations of novel treatment strategies such as administration of
bisphosphonates are of paramount importance as there is currently no effective
treatment for traumatic femoral head osteonecrosis in adolescents.
It is widely accepted that the long-term prognosis for traumatic femoral
head osteonecrosis in this age group is commonly collapse and deformity of the
femoral head with secondary degenerative changes in the hip and a poor
clinical
outcome3,25-27.
For example, Krahn et
al.27, in a
thirty-one-year retrospective study of thirty-six patients with slipped
capital femoral epiphysis that led to femoral head osteonecrosis, reported
that 75% of the patients had evidence of marked degenerative change on
radiographs of the hip. Salvage surgery, such as proximal femoral realignment
osteotomy or hip arthrodesis, may be required at an early age to prevent or
treat progression of hip
osteoarthritis28,
although these secondary procedures may increase the technical difficulty of
any future joint
arthroplasty29.
Bisphosphonates are metabolically stable analogs of inorganic pyrophosphate
in which the P-O-P bond has been replaced with a nonhydrolyzable P-C-P
bond30. They are
versatile and potent antiresorptive (anticatabolic) agents, exerting their
effects by inhibiting components of the intracellular mevalonate pathway and
preventing prenylation of intracellular proteins in osteoclasts.
Bisphosphonates have been shown to improve bone density in a variety of
conditions, such as osteogenesis
imperfecta31 and
Gaucher disease32,
and bone strength in preclinical models of distraction osteogenesis and
fracture
repair33-35.
In patients with femoral head osteonecrosis, because osteoclastic resorption
of necrotic subchondral bone may lead to mechanical weakening and subsequent
collapse of the femoral head, inhibition of osteoclastic activity by
bisphosphonates may attenuate or prevent the progression of the collapse
associated with this repair process.
During the time course of this study, twenty-two patients presented with an
unstable slipped capital femoral epiphysis. Twelve of these twenty-two
patients had a "cold" bone scan consistent with femoral head
osteonecrosis. Published rates of osteonecrosis in patients presenting with
unstable slipped capital femoral epiphysis range from 47% to
58%1,36,37.
The authors of those reports assumed that all of the femoral heads with
osteonecrosis went on to collapse and those that did not collapse did not have
osteonecrosis, but it is not possible to be certain of this on the basis of
the methodologies employed in those studies. Technetium-99m bone scanning has
been shown to be a reliable predictor, with excellent sensitivity and
predictive value, of femoral heads at risk for
collapse16,38.
In one study16,
osteonecrosis developed in five of six hips that had had "cold"
pretreatment bone scans. The same outcome was seen in three "cold"
hips scanned following surgical fixation in another
study38.
Our results require careful interpretation in the light of this literature.
We did not treat the ten patients who presented with an unstable slipped
capital femoral epiphysis and a "hot" bone scan during the period
of our study. Had we reported our outcomes simply in terms of the rate of
femoral head collapse in all patients with an unstable slipped capital femoral
epiphysis, we would have stated that seventeen (77%) of twenty-two femoral
heads remained spherical and five (23%) of the twenty-two collapsed. The rate
of femoral head collapse in this raw evaluation is lower than the 47% to 58%
rates reported in the literature on unstable slipped capital femoral
epiphysis. It should be noted that the duration of follow-up in the current
investigation was shorter than that in some of the other studies, so caution
in the interpretation of these comparisons is needed.
However, because we used bone scanning, we are able to state more precisely
how the patients fared depending on whether or not they had evidence of
osteonecrosis on the bone scan. None of the ten patients with an unstable
slipped capital femoral epiphysis and "hot" bone scans showed any
resorption or collapse. Of the twelve patients with an unstable slipped
capital femoral epiphysis and "cold" bone scans, seven had a
Stulberg Class-I or II outcome, which is possibly an improvement over the
natural history.
We can thus confirm that children who present with an unstable slipped
capital femoral epiphysis and have normal bone scans following surgical
fixation do not require further adjunctive treatment such as administration of
bisphosphonates. We treated all patients with an unstable slipped capital
femoral epiphysis, a femoral neck fracture, or a hip dislocation and a
"cold" bone scan with bisphosphonate therapy. Restrictions were
placed on weight-bearing for the first year following the injury. Despite the
common belief that necrotic bone is strong, Pringle et
al.9 showed, in a
piglet model of ischemic necrosis, that the indentation stiffness of the
femoral head is reduced by 74% compared with controls by four weeks after the
induction of osteonecrosis and considerable deformity is present by eight
weeks. Ideally, we would have insisted on weight-bearing restrictions for
eighteen months, as radiographic evidence of bone resorption was observed in
this study group up to this time. This length of time was, however,
impractical in an adolescent population, and we became concerned about the
general effects of inactivity. The desirability of preventing radiographically
obvious resorption in the femoral head was reinforced by our findings of
increased rates of sphericity and higher Harris hip and Global PODCI scores in
patients without resorption.
The Stulberg classification was useful for defining which femoral heads
were spherical and which were not. A limitation of using this classification
in this patient cohort is that it could not account for areas of resorption
seen in some femoral heads. We are uncertain that there will be a significant
difference in the long-term outcomes between hips graded as Stulberg Class III
and those graded as Stulberg Class IV in our patient cohort. However, we
assume that the hips graded as Stulberg Class I or II will have a better
prognosis because of the maintenance of sphericity, as these hips were nearly
all graded as Ficat stage I. Although our cohort had very good hip function
overall, hip function in these patients with osteonecrosis of the femoral head
may decrease over time, even when the femoral head has remained spherical.
The choice of bisphosphonate was limited to those with which we were
familiar. Pamidronate has been used commonly for patients with osteogenesis
imperfecta31, and
many centers would be more familiar with this drug than with the newer, more
potent zoledronic acid. The advantages of using zoledronic acid at our
institution were a shorter infusion time and less frequent dosing than with
pamidronate, but these advantages may not be relevant in all centers.
There were no major adverse effects of the bisphosphonate therapy in this
study. The frequency of acute flu-like effects was similar to that previously
reported in the
literature23. In
particular, no renal effects, uveitis, or osteonecrosis of the
jaw39,40
was noted in our small series of subjects. We are not aware of any reports of
bisphosphonate-associated osteonecrosis of the jaw in children; however,
information about this risk should be provided to patients and their families.
It remains possible that subclinical events did occur and were not noted by
our surveillance.
Animal studies have shown diminution of growth with the use of potent
bisphosphonates41.
However, human studies have shown that this growth disturbance is not
clinically
measurable42,43.
While our study revealed no negative effects on growth and provided further
evidence that this problem does not seem to translate from animals to
children, the continued monitoring of growth of children being treated with
bisphosphonates remains advisable.
It was difficult to ascertain the exact dose and duration of bisphosphonate
treatment required for these patients, as this question is not answered by the
available preclinical data. We saw radiographic evidence of resorption up to
eighteen months after the initial injury. A higher dose of bisphosphonates may
have provided stronger inhibitory effects on osteoclast activity and enhanced
the ability to protect against progression of femoral head collapse. For
example, clinical trials of osteoporosis treatment with alendronate have shown
a dose-related increase in bone mineral density in the lumbar
spine44. However, a
higher dose or longer duration of treatment must be balanced against a higher
risk of adverse effects. Bisphosphonates do not distribute in high
concentration to necrotic bone and only penetrate the revascularizing femoral
head45. This would
appear to mandate continued administration of the drug until revascularization
has occurred. Although the follow-up bone scans documented apparent
revascularization by six months, bone scans do not have sufficient resolution
to show if revascularization is fully complete. However, the findings on these
scans confirm that systemically administered bisphosphonate is usually being
distributed to the femoral head by six to nine months.
Alternative future approaches that require investigation include local
bisphosphonate delivery with or without follow-up systemic treatment.
Furthermore, rather than relying on the natural bone-forming (anabolic)
response, which took up to eighteen months to repair the femoral head
according to our radiographic observations, the addition of a properly timed
anabolic therapy may be useful. A few patients in this study had bone marrow
injections in concert with additional necessary surgery for screw placement,
but we cannot determine if these interventions had useful effects on the basis
of this small number of patients.
More than half of the hips with femoral head osteonecrosis in this study
had a Stulberg Class-I or II outcome, which we believe may be better than the
described natural history of this severe disorder. The positive nature of this
observational series supports the need for larger multicenter randomized
trials in the future.
A table showing clinical details on all study patients and a figure
depicting the Stulberg classification are available with the electronic
versions of this article, on our web site at
(go to
the article citation and click on "Supplementary Material") and on
our quarterly CD-ROM (call our subscription department, at 781-449-9780, to
order the CD-ROM). ?