Abstract
Background: Revision total hip arthroplasty is indicated for most
periprosthetic fractures that occur around the stem of the femoral implant.
The purpose of the present study was to assess the results and complications
of revision total hip arthroplasty for the treatment of periprosthetic femoral
fractures.
Methods: We evaluated 118 hips in 116 patients who underwent
revision total hip arthroplasty because of an acute Vancouver type-B
periprosthetic femoral fracture. The femoral implant used for the revision was
a cemented stem in forty-two hips, a proximally porous-coated uncemented stem
in twenty-eight, an extensively porous-coated stem in thirty, and an
allograft-prosthesis composite or tumor prosthesis in eighteen. The mean
duration of follow-up was 5.4 years.
Results: Kaplan-Meier analysis demonstrated that the probability of
survival was 90% at five years and 79.2% at ten years with revision or removal
of the femoral implant for any reason as the end point. Sixteen femoral
components were rerevised: ten were rerevised because of loosening; three,
because of loosening in association with a fracture nonunion; two, because of
recurrent dislocation; and one, because of a new periprosthetic fracture.
Additionally, six femoral implants were resected because of deep infection
(five) or prosthetic loosening (one). Radiographs of the ninety-six hips with
a surviving implant showed that twenty-one had evidence of loosening of the
femoral implant, four had a nonunion of the femoral fracture, and two had both
a nonunion and loosening of the femoral implant.
Conclusions: Revision total hip arthroplasty for the treatment of a
periprosthetic fracture around the stem of the femoral implant successfully
restored function for most patients. The greatest long-term problems were
prosthetic loosening and fracture nonunion. Better results were seen when an
uncemented, extensively porous-coated stem was used.
Level of Evidence: Therapeutic study, Level IV (case
series [no, or historical, control group]). See Instructions to Authors for a
complete description of levels of evidence.
Periprosthetic femoral fractures are recognized as a challenging and
difficult problem to treat after total hip arthroplasty, and the development
of effective management strategies for this complication is
important1-4.
Periprosthetic fractures are not rare. In a series of more than 30,000 hip
arthroplasties from one institution, the prevalence of postoperative femoral
fractures was 1.1% after primary hip arthroplasty and 4.0% after revision
arthroplasty5.
The treatment of these fractures has progressed markedly as a result of a
better understanding of the indications for operative treatment, advances in
prosthetic components, and improved operative
techniques6.
Classification systems have provided an algorithm for, and have aided in
decision-making about, periprosthetic fracture
treatment7-9.
For a periprosthetic fracture near the tip of the stem of a well-functioning
and well-fixed femoral component, open reduction and internal fixation of the
fracture with retention of the femoral component usually is recommended,
although revision may be considered in unusual
circumstances10-15.
However, if the fracture is associated with a loose or failed implant, then
revision is
preferred16,17.
The primary purpose of the present study was to evaluate a large number of
patients who had been treated with revision of the femoral component because
of a periprosthetic femoral fracture in order to determine the results and
complications of this procedure. A secondary goal was to evaluate different
methods of femoral revision in order to determine the likelihood of success of
each technique for this particular indication.
Between January 1980 and July 1998, 136 consecutive hips in 134 patients
(seventy-one men and sixty-three women with a mean age of 66.5 years) were
treated with femoral component revision at a single institution because of a
periprosthetic femoral fracture at the site of a total hip arthroplasty.
Eleven patients died and seven patients were lost to follow-up within two
years after the index procedure, leaving 118 hips in 116 patients (fifty-nine
men and fifty-seven women) who had had a mean age of 65.3 years (range,
thirty-seven to ninety-one years) at the time of surgery. None of the patients
who died had had removal or revision of the implant. Three patients were
treated with a revision or resection within two years after the index
procedure and are included in the results. Twenty-one different surgeons
performed the index procedures. Sixty fractures involved the left hip, and
fifty-eight involved the right hip. The average duration of follow-up was 64.9
months (range, two to 185 months). The only patients with less than two years
of follow-up were those who had a revision within twenty-four months after the
index operation.
Fractures were classified with use of the Vancouver system described by
Duncan and Masri18
(Table I). This classification
system has been validated and shown to have high intraobserver and
interobserver
reliability19.
Fractures involving the greater or lesser trochanter (type A) and those
markedly distal to the prosthesis (type C) were excluded. Seven hips had a
type-B1 fracture (a fracture around the stem with a stable prosthesis),
seventy-six had a type-B2 fracture (a fracture around the stem with a loose
prosthesis), and thirty-five had a type-B3 fracture (a fracture around the
stem with a loose prosthesis and marked proximal bone loss or severe proximal
bone damage). Fractures that occurred intraoperatively were excluded, as were
patients who had an established periprosthetic fracture nonunion before
revision.
The underlying diagnosis that had necessitated the initial arthroplasty was
degenerative joint disease in seventy-nine patients (eighty hips),
developmental dysplasia in thirteen patients (fourteen hips), rheumatoid
arthritis in nine patients (nine hips), osteonecrosis in eight patients (eight
hips), and previous femoral neck fracture in seven patients (seven hips). The
average time between the arthroplasty and the fracture was 8.1 years (range,
one month to 17.5 years). The implant that was in place when the fracture
occurred was a primary prosthesis in forty-eight hips and a revision
prosthesis in seventy. The implants included a cemented stem in ninety-nine
hips, a proximally coated uncemented implant in fifteen, a tumor prosthesis or
an allograft-prosthesis composite in three, and an extensively porous-coated
uncemented stem in one.
The femoral implant that was used at the time of the index procedure was a
cemented stem in forty-two hips, an extensively porous-coated uncemented stem
in thirty, a proximally porous-coated uncemented stem in twenty-eight, an
allograft-prosthesis composite in fourteen, and a tumor prosthesis in four
(see Appendix). The type of stem and the form of fixation were at the
discretion of the surgeon. Early in the series, most hips received a cemented
implant; midway through the series, most received a proximally porous-coated
uncemented implant; and recently, most received an extensively porous-coated
uncemented implant.
At the time of the index procedure, all patients had revision of the
femoral component and fifty-five patients had concomitant revision of the
acetabular component. Forty-six hips (39%) had allograft struts placed around
the fracture and secured with either cables or wires, and fifteen of these
hips had additional cancellous allograft or autograft placed at the fracture
site. Forty-six hips (39%) had supplemental fixation with cables or wires
alone, and fourteen of these hips had additional cancellous autograft or
allograft placed at the fracture site. Twelve hips (10%) had only cancellous
allograft or autograft placed at the fracture site, with no additional
supplemental fixation. Two hips (2%) that were treated with an
allograft-prosthesis component had plate fixation of the allograft to host
bone. Twelve hips (10%) had no additional fixation or bone graft placed at the
fracture site.
The patients were evaluated and standardized data were collected at two
months after the operation; at one, two, five, and ten years after the
operation; and every five years thereafter. The data that were collected and
the radiographs that were made at these intervals were reviewed
retrospectively. At the time of the most recent follow-up, clinical end point
data were collected with regard to the level of pain and the current use of
walking aids.
The fixation of cemented stems was assessed according to the criteria of
Harris et al.20,
and the fixation of uncemented stems was assessed according to the method of
Engh and Massin21.
Fracture-healing was defined as bridging callus across the fracture site or
sites and was evaluated by both a resident orthopaedist and one of the senior
authors (D.J.B.).
Kaplan-Meier survival analysis was used to calculate survival rates (with
95% confidence intervals).
Revisions
Kaplan-Meier analysis demonstrated that the probability of survival was 90%
at five years, 79.2% at ten years, and 58% at fifteen years with revision or
removal of the femoral implant for any reason as the end point
(Fig. 1) and 91.9%, 82.2%, and
60.2% at the same time-intervals with revision of the femoral implant because
of loosening or nonunion as the end point
(Fig. 2).
At the time of the most recent follow-up, twenty-two femoral components had
been rerevised (sixteen hips) or resected (six hips). Ten rerevisions were
performed because of loosening of the femoral component; three, because of
femoral loosening in association with a fracture nonunion; two, because of
recurrent dislocation; and one, because of a second, separate periprosthetic
femoral fracture. Five of the six resection arthroplasties were performed
because of a deep infection. All of the fractures in these five hips had
healed, and none of the femoral components was loose. The sixth resection
arthroplasty was performed because of aseptic loosening of an
allograft-prosthesis composite; no reimplantation was performed because of
multiple medical comorbidities.
Ninety-six hips had an intact implant and were evaluated at an average of
5.4 years (range, two to 15.4 years). Clinically, seventy-six hips (79%) had
no pain (fifty-seven) or mild pain (nineteen) at the most recent follow-up.
Thirteen hips (13.5%) had moderate pain, and seven (7.3%) had severe pain.
Twenty-nine patients (twenty-nine hips; 30%) did not use any walking aids,
thirty-nine patients (forty hips; 42%) used a cane or one crutch, twenty-one
patients (twenty-one hips; 22%) used a walker or two crutches, and five
patients (six hips; 6%) were unable to walk secondary to pain or multiple
medical comorbidities.
Radiographs were available at the time of the most recent follow-up for
ninety-five of the ninety-six hips with an intact prosthesis. Twenty-one hips
had radiographic evidence of femoral implant loosening, four had a fracture
nonunion, and two had both a nonunion and femoral component loosening. At the
time of the latest follow-up (performed at an average of 79.4 months [range,
twenty-four to 141 months] postoperatively), twenty-one of these twenty-seven
hips had no or mild pain, four had moderate pain, and two had severe pain.
Eight patients (eight hips) used no walking aids, nine patients (nine hips)
used one cane or one crutch, six patients (six hips) used a walker, and four
patients (four hips) were unable to walk.
In addition to the reoperations mentioned above, complications occurred in
nineteen hips (16%), six of which required another surgical procedure as a
result. Six hips had pain related to the wires or cables that had been used at
the time of the index procedure; three of these hips required removal of the
hardware. Five hips had more than one dislocation, and one of them required an
open reduction. Three hips had a superficial wound infection; one was treated
with débridement with retention of the prosthesis, and the others were
treated with antibiotics. One hip required excision of heterotopic
ossification. Two patients (two hips) had an injury of the peroneal division
of the sciatic nerve, one patient (one hip) had a deep venous thrombosis, and
one patient (one hip) had a vascular injury.
Results According to the Type of Implant
Thirty hips received an uncemented extensively porous-coated stem at the
time of the index procedure (Figs. 3-A and
3-B). After a mean duration of follow-up of forty-two months
(range, twenty to eighty-three months), one implant had been revised because
of aseptic loosening, one had been revised because of multiple dislocations,
and two had been removed because of a deep infection
(Table II). Radiographically,
three (12%) of the twenty-six surviving implants were loose and there were no
fracture nonunions. Overall, twenty-three (77%) of the thirty hips had a
healed fracture and a well-fixed, unrevised stem.
Forty-two hips received a cemented stem at the time of the index procedure
(Figs. 4-A and 4-B). After a
mean duration of follow-up of sixty-eight months (range, twenty-four to 185
months), six stems had been revised: three had been revised because of femoral
loosening, one had been revised because of femoral loosening and nonunion, one
had been revised because of recurrent dislocation, and one had been revised
because of a second fracture around the revision stem
(Table II). Two other stems had
been removed because of deep infection. Radiographically, six (18%) of the
thirty-four surviving implants were loose, two (6%) were associated with a
fracture nonunion, and one (3%) was loose and was associated with a fracture
nonunion. Overall, twenty-five (60%) of the forty-two hips had a healed
fracture and a well-fixed, unrevised stem.
Twenty-eight hips received an uncemented proximally porous-coated stem at
the time of the index procedure. After a mean duration of follow-up of
eighty-five months (range, thirty to 175 months), three stems had been
revised: two had been revised because of component loosening, and one had been
revised because of loosening in association with a nonunion of the fracture
(Table II). Radiographically,
twelve (48%) of the twenty-five hips with a surviving stem showed evidence of
loosening, two (8%) had a fracture nonunion with a well-fixed stem, and one
(4%) had a fracture nonunion with a loose stem. Overall, only ten (36%) of the
twenty-eight hips had a healed fracture and a well-fixed, unrevised stem.
Eighteen hips received either an allograft-prosthesis composite (fourteen)
or a tumor prosthesis (four) at the time of the index procedure. Fifteen of
the eighteen hips had a type-B3 fracture with poor proximal bone stock. In all
of these hips, the proximal fracture fragments were excised or wrapped around
an allograft. After a mean duration of follow-up of sixty-eight months (range,
twenty-four to 126 months), five prostheses had been revised because of
aseptic loosening, one had been resected because of a deep infection, and one
had been resected because of loosening
(Table II). Radiographically,
none of the surviving implants were loose. All hips had union at the
allograft-host junction. Overall, eleven (61%) of the eighteen femoral
components were well fixed.
Revision total hip arthroplasty is usually recommended for the treatment of
periprosthetic fractures associated with a failed femoral
stem22,23.
Therefore, an understanding of the results of this form of treatment is
valuable.
At our institution, almost all periprosthetic femoral fractures are treated
operatively. Although there was a roughly chronological progression of
preferred implants in the present series, some cemented stems were used
throughout the series and allograft-prosthesis composites and tumor prostheses
were used selectively for Vancouver type-B3 fractures throughout the series.
Vancouver type-B1 fractures were usually treated with retention of the
prosthesis and internal fixation. In the cases of the seven type-B1 fractures
in this series, the surgeon believed that the advantages of intramedullary
fracture fixation that were gained by revision to a long-stemmed implant
outweighed the disadvantages of removal of the prosthesis.
Several authors have reported the results of revision procedures performed
with use of a cemented stem for the treatment of a periprosthetic femoral
fracture24,25.
In the present study, twenty-five (60%) of the forty-two hips that received a
cemented stem for the treatment of a periprosthetic femoral fracture had a
well-fixed implant with fracture union at an average of about seven years
after the index revision. Proximally coated uncemented stems were associated
with poor results; only ten (36%) of the twenty-eight hips that received such
a stem had a stable implant and a healed fracture at the time of the most
recent follow-up. All of the proximally porous-coated stems in the present
study had a monoblock body, which has been associated with a high rate of
loosening in the general population of patients managed with femoral
revision26.
Allograft-prosthesis composites or proximal femoral replacement prostheses
(i.e., tumor prostheses) frequently have been used for fractures associated
with severe proximal bone loss (Vancouver type-B3
fractures)27,28.
In the present series, fifteen of the eighteen hips that were treated with an
allograft-prosthesis composite or a tumor prosthesis had a type-B3 fracture.
Overall, eleven of these eighteen were well fixed at the time of the latest
follow-up.
The uncemented, extensively porous-coated implants had the highest
likelihood of stable fixation and were not associated with any nonunions.
Several smaller studies have shown that uncemented, long-stemmed, extensively
porous-coated implants can be used successfully to treat periprosthetic
fractures of the
femur29-31.
These stems appear to be successful because the implant can gain initial and
long-term fixation in the well-preserved diaphysis distal to the fracture and
simultaneously can provide intramedullary fixation of most diaphyseal
fractures. Other designs of uncemented long-stemmed implants that can reliably
gain biological fixation in the revision setting and that also can provide
stable intramedullary fixation of the fracture site also may be
successful32,33.
Although the extensively porous-coated stems appeared to be associated with
the best results in the present study, the fact that this group had the
shortest duration of follow-up limits the strength of this conclusion.
At the present time, we use uncemented, extensively porous-coated,
long-stemmed implants for the majority of revisions performed for the
treatment of periprosthetic femoral fractures. We continue to use cemented
long-stemmed implants for selected older patients with poor bone and a simple
fracture pattern that can be reduced anatomically to preclude cement
extrusion. For Vancouver type-B3 fractures that are associated with severe
damage to the proximal part of the femur, we continue to use
allograft-prosthesis composites or tumor prostheses in selected patients and
uncemented, fluted, tapered stems that gain axial and rotational stability
distal to the fracture in some cases.
The present study demonstrates a reasonable rate of clinical success when
femoral revision is performed to treat a periprosthetic fracture; however,
there are important problems associated with such treatment. The three most
common problems were aseptic loosening of the revision femoral component,
fracture nonunion, and deep infection. Of these, implant loosening was the
most common. Ongoing efforts to improve the quality of femoral component
fixation at the time of revision for a periprosthetic femoral fracture are
important. Newer implants and greater operative experience with these implants
may improve the success rate. Recognition that fracture nonunion is not rare
should encourage measures to reduce that problem because the treatment of
periprosthetic fracture nonunions has been shown to be associated with a high
rate of complications and a low rate of
success34. The
nonunion rate may be reduced by careful handling of the soft tissues to avoid
devascularization of bone near the fracture site, the avoidance of cement
extrusion into the fracture site, and the use of autogenous bone graft or
cortical strut allografts to augment
fixation34-36.
On the basis of our results, we believe that the treatment of properly
selected periprosthetic femoral fractures with a total hip arthroplasty is
associated with a reasonable rate of clinical success, pain relief, and
restoration of function. The present study also demonstrated that the
complications of prosthetic loosening, fracture nonunion, and infection are
not rare.
A table showing the types of stem and the specific femoral components that
were used is 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).
Lewallen DG, Berry DJ.
Periprosthetic fracture of the femur after total hip arthroplasty: treatment
and results to date. Instr Course Lect.1998;47:
243-9.47243
1998
[PubMed]
Blatter G, Fiechter T, Magerl F.
[Peri-prosthesis fractures in total hip endoprostheses].
Orthopade.1989;18: 545-51.
German.18545
1989
[PubMed]
Rorabeck CH. Periprosthetic
fractures: a problem on the rise. Orthopedics.2000;23:
989-90.23989
2000
[PubMed]
Younger AS, Dunwoody I, Duncan
CP. Periprosthetic hip and knee fractures: the scope of the problem.
Instr Course Lect.1998;47:
251-6.47251
1998
[PubMed]
Berry DJ. Epidemiology: hip and
knee. Orthop Clin North Am.1999;30:
183-90.30183
1999
[PubMed][CrossRef]
Garbuz DS, Masri BA, Duncan CP.
Periprosthetic fractures of the femur: principles of prevention and
management. Instr Course Lect.1998;47:
237-42.47237
1998
[PubMed]
Brady OH, Garbuz DS, Masri BA, Duncan
CP. Classification of the hip. Orthop Clin North
Am.1999;30:
215-20.30215
1999
[CrossRef]
Kelley SS. Periprosthetic femoral
fractures. J Am Acad Orthop Surg.1994;
2: 164-72.2164
1994
[PubMed]
Kyle RF, Crickard GE 3rd.
Periprosthetic fractures associated with total hip arthroplasty.
Orthopedics.1998;21:
982-4.21982
1998
[PubMed]
Booth RE Jr. Management of
periprosthetic fractures. Orthopedics.1994; 17:
845-7.17845
1994
[PubMed]
Dennis MG, Simon JA, Kummer FJ, Koval
KJ, Di Cesare PE. Fixation of periprosthetic femoral shaft fractures: a
biomechanical comparison of two techniques. J Orthop
Trauma.2001;15:
177-80.15177
2001
[CrossRef]
Dennis MG, Simon JA, Kummer FJ, Koval
KJ, DiCesare PE. Fixation of periprosthetic femoral shaft fractures
occurring at the tip of the stem: a biomechanical study of 5 techniques.
J Arthroplasty.2000;15:
523-8.15523
2000
[PubMed][CrossRef]
Jukkala-Partio K, Partio EK,
Solovieva S, Paavilainen T, Hirvensalo E, Alho A. Treatment of
periprosthetic fractures in association with total hip arthroplasty—a
retrospective comparison between revision stem and plate fixation.
Ann Chir Gynaecol.1998;87:
229-35.87229
1998
[PubMed]
Radcliffe SN, Smith DN. The
Mennen plate in periprosthetic hip fractures. Injury.1996;27:
27-30.2727
1996
[PubMed][CrossRef]
Venu KM, Koka R, Garikipati R,
Shenava Y, Madhu TS. Dall-Miles cable and plate fixation for the treatment
of peri-prosthetic femoral fractures—analysis of results in 13 cases.
Injury.2001;32:
395-400.32395
2001
[PubMed][CrossRef]
Incavo SJ, Beard DM, Pupparo F, Ries
M, Wiedel J. One-stage revision of periprosthetic fractures around loose
cemented total hip arthroplasty. Am J Orthop.1998;27:
35-41.2735
1998
[PubMed]
Siegmeth A, Menth-Chiari WA, Wozasek
GE, Vecsei V. Femur fractures in patients with hip arthroplasty:
indications for revision arthroplasty. J South Orthop
Assoc.1998;7:
251-8.7251
1998
Duncan CP, Masri BA. Fractures of
the femur after hip replacement. Instr Course Lect.1995;44:
293-304.44293
1995
[PubMed]
Brady OH, Garbuz DS, Masri BA, Duncan
CP. The reliability and validity of the Vancouver classification of
femoral fractures after hip replacement. J
Arthroplasty.2000;15:
59-62.1559
2000
[CrossRef]
Harris WH, McCarthy JC Jr, O'Neill
DA. Femoral component loosening using contemporary techniques of femoral
cement fixation. J Bone Joint Surg Am.1982;64:
1063-7.641063
1982
[PubMed]
Engh CA, Massin P. Cementless
total hip arthroplasty using the anatomic medullary locking stem. Results
using a survivorship analysis. Clin Orthop.1989;249:
141-58.249141
1989
[PubMed]
Haddad FS, Masri BA, Garbuz DS,
Duncan CP. Femoral bone loss in total hip arthroplasty: classification and
preoperative planning. Instr Course Lect.2000;49:
83-96.4983
2000
[PubMed]
Ries MD. Periprosthetic
fractures: early and late. Orthopedics.1997;20:
798-800.20798
1997
[PubMed]
Johansson JE, McBroom R, Barrington
TW, Hunter GA. Fracture of the ipsilateral femur in patients with total
hip replacement. J Bone Joint Surg Am.1981;63:
1435-42.631435
1981
[PubMed]
Beals RK, Tower SS.
Periprosthetic fractures of the femur. An analysis of 93 fractures.
Clin Orthop.1996;327:
238-46.327238
1996
[PubMed][CrossRef]
Berry DJ, Harmsen WS, Ilstrup D,
Lewallen DG, Cabanela ME. Survivorship of uncemented proximally
porous-coated femoral components. Clin Orthop.1995;319:
168-77.319168
1995
[PubMed]
Chandler HP, Tigges RG. The role
of allografts in the treatment of periprosthetic femoral fractures.
Instr Course Lect.1998;47:
257-64.47257
1998
[PubMed]
Wong P, Gross AE. The use of
structural allografts for treating periprosthetic fractures about the hip and
knee. Orthop Clin North Am.1999;
30: 259-64.30259
1999
[PubMed][CrossRef]
Moran MC. Treatment of
periprosthetic fractures around total hip arthroplasty with an extensively
coated femoral component. J Arthroplasty.1996;11:
981-8.11981
1996
[PubMed][CrossRef]
Mulliken BD, Rorabeck CH, Bourne
RB. Uncemented revision total hip arthroplasty: a 4-to-6-year review.
Clin Orthop.1996;325:
156-62.325156
1996
[PubMed][CrossRef]
Macdonald SJ, Paprosky WG, Jablonsky
WS, Magnus RG. Periprosthetic femoral fractures treated with a long-stem
cementless component. J Arthroplasty.2001;16:
379-83.16379
2001
[PubMed][CrossRef]
Berry DJ. Femoral revision:
distal fixation with fluted, tapered grit-blasted stems. J
Arthroplasty.2002;17 (4
Suppl 1): 142-6.17142
2002
[CrossRef]
Kolstad K. Revision THR after
periprosthetic femoral fractures. An analysis of 23 cases. Acta
Orthop Scand.1994;65:
505-8.65505
1994
[CrossRef]
Crockarell JR Jr, Berry DJ, Lewallen
DG. Nonunion after periprosthetic femoral fracture associated with total
hip arthroplasty. J Bone Joint Surg Am.1999;81:
1073-9.811073
1999
[PubMed]
Brady OH, Garbuz DS, Masri BA, Duncan
CP. The treatment of periprosthetic fractures of the femur using cortical
onlay allograft struts. Orthop Clin North Am.1999;30:
249-57.30249
1999
[PubMed][CrossRef]
Larson JE, Chao EY, Fitzgerald
RH. Bypassing femoral cortical defects with cemented intramedullary stems.
J Orthop Res.1991;9:
414-21.9414
1991
[PubMed][CrossRef]