Amoderately obese, forty-seven-year-old woman fell and sustained a
transcervical fracture of the right femur, which was treated initially with
internal fixation and a spica cast. A subsequent nonunion was treated with a
Judet prosthesis, which was eventually removed, leaving the patient with a
Girdlestone resection of the right hip. In 1970, at the age of sixty-eight
years, the patient underwent total hip arthroplasty of the left hip for the
treatment of extensive osteoarthritis, and a metal-on-metal McKee-Farrar total
hip prosthesis was implanted. In 1995, the patient was contacted by our
research group, and she agreed to be evaluated as part of a long-term study of
McKee-Farrar total hip
replacements8,9.
She also consented to join the "Willed Joint" program of the Joint
Replacement Institute, thereby agreeing to donate her hip joints and to permit
postmortem biopsies of her organs. The patient was highly satisfied with the
results of the arthroplasty because the left hip was free of pain and did not
limit her function. Radiographs demonstrated good positioning and fixation of
the implant, with no evidence of loosening or osteolysis. Blood and urine
samples were obtained for the measurement of cobalt and chromium ions.
Following the patient's death in 1999, the left hip joint with the
surrounding acetabular and femoral bone, pseudocapsular tissue, and
McKee-Farrar prosthesis was removed en bloc. Tissue samples were obtained from
the liver, spleen, left kidney, left inguinal and abdominal lymph nodes, and
muscle of the left thigh. Whole blood was obtained from the femoral vein and
urine was aspirated from the bladder for the analysis of ion
content7.
Radiographs of the specimen were made in several planes. The components
were then removed with use of revision surgical techniques. The pseudocapsule
had a thin band of metal staining near the attachment to the femur. The
femoral component was not bonded at the metal-cement interface and was
extracted with minimal force. The cement mantle was well fixed to bone. The
acetabular component was well fixed to bone, without loosening at the
metal-cement or cement-bone interface. The bone was generally of poor quality,
consistent with extensive osteopenia in this elderly patient
(Fig. 1).
Component Analysis
Component wear was measured with use of a coordinate measuring machine, and
the surface roughness of the femoral component was measured with use of laser
profilometry (Mahr/Feinpruf, Göttingen, Germany) as previously
described10. The
stem was cleaned and was examined with scanning electron microscopy (DSM 960;
Zeiss, Oberkochen, Germany).
Tissue Analysis
The tissue samples that had been obtained at the time of autopsy for
routine histological analysis were fixed in formalin and processed in
paraffin. A modified Mirra grading
scale11 was used to
assess the numbers of macrophages, giant cells, and wear particles in sections
of the periprosthetic tissues. Sections from multiple samples of the lymph
nodes, spleen, liver, and kidney were examined with transmitted and polarized
light for assessment of the general histological appearance and the presence
of wear particles. Additional samples of those organs were mounted on a
beryllium plate for particle microanalysis by means of backscattered electron
imaging and electron microprobe with energy-dispersive x-ray analysis (model
JXA-8900RL electron probe microanalyzer; JEOL, Peabody, Massachusetts) as
previously
described5. With use
of previously described
techniques12,
representative samples of formalin-fixed periprosthetic tissue were used for
particle isolation. Because a metal blade had been used to obtain the
liver-tissue samples at the time of autopsy, the external surfaces of these
samples were trimmed with a ceramic knife to provide cleaner samples for
trace-metal analysis. Approximately 0.8 g of the material was digested in
ultrapure nitric acid for examination by atomic-absorption spectroscopy. Four
separate samples were tested in triplicate. Samples from a patient without
implants who was undergoing liver biopsy were similarly tested to serve as a
control.
The bearing surfaces were grossly smooth and shiny
(Fig. 2), with small areas of
light burnishing. A thin organic film was present on the femoral stem,
consistent with debonding or loosening. Several areas of the distal part of
the femoral stem showed light wear polishing that was consistent with relative
motion between the stem and the cement mantle. The acetabular component had a
central dimple from manufacturing that was 0.23 mm deep, and there was a
corresponding area on the femoral component that was 120 µm deep. The total
linear wear of the ball and socket was approximately 70 µm, and this wear
was spread over a comparable area of both surfaces. The diametral clearance
was approximately 100 µm.
The Ra (average roughness) at the dome of the femoral ball
averaged 0.08 ± 0.01 µm, and the Rmax (maximum roughness
depth) averaged 1.58 ± 0.48 µm. In an area of light burnishing, the
Ra averaged 0.13 ± 0.05 µm and the Rmax
averaged 3.7 ± 1.4 µm. These values averaged 0.12 ± 0.02
µm and 2.68 ± 0.7 µm, respectively, in an unworn region.
Scanning electron microscopy of the femoral stem showed an irregularly
textured surface (Fig. 3-A).
The grossly polished region had a slightly less textured appearance that was
consistent with wear and flattening of the peaks in that area
(Fig. 3-B).
Tissue Analysis
A gray membrane ranging from <0.5 to 4 mm in thickness was present
within the femoral canal. Histological examination of this membrane revealed
fibrous tissue containing histiocytes and giant cells with spaces consistent
with dissolved bone cement. Electron microprobe analysis identified
cobalt-chromiumalloy particles that ranged from <1 to approximately 5 µm
in size. The results of the semiquantitative histological analyses of the
periprosthetic tissues are summarized in the Appendix. Generally, the capsular
tissues were fibrous, with areas of adipose or necrotic tissue. A distinct
inflammatory zone on the edge facing the implant contained macrophages, giant
cells, blood vessels, and wear particles of cobalt-chromium and
polymethylmethacrylate. Lymphocytes and plasma cells were rare, and
polymorphonuclear leukocytes were absent. No histological abnormalities were
noted in any of the organ samples.
Particle Analysis
Three types of particles were identified in the periprosthetic tissues.
Most particles were oval (approximately 40%) or needle-shaped (approximately
40%), and the remainders were round. An EDAX system (Princeton Gamma Tech,
Princeton, New Jersey), used with a JEOL 2000 fx scanning electron microscope
(Akishima, Japan), revealed that approximately half of the particles were
consistent with cobalt-chromium alloy and the remainders were consistent with
chromium oxides. All particles ranged in size from 18 to 472 nm (average and
standard deviation, 77 ± 50 nm).
Small numbers of dark, dense particles were noted in the liver, spleen, and
lymph nodes, but none were associated with necrosis or other pathological
changes. Most of the particles were <1 µm in size, and some were
birefringent. The particles were contained within macrophages that occurred in
clusters of six to fifteen cells in the portal tracts of the liver, near the
splenic trabeculae, and within the medullary sinuses of the lymph nodes.
Energy-dispersive x-ray analysis identified most of the particles as silicates
of environmental origin. The greatest numbers of particles were in the
abdominal lymph nodes, where small numbers of macrophages with abundant
particles were located perivascularly in the sinus tissue. These particles
ranged from barely visible to several micrometers in size and were mostly
round. The spectra of these particles had major peaks for silicon and oxygen
and minor peaks for aluminum, titanium, potassium, or magnesium. None of the
particles in any of the organ tissues were identified as cobalt-chromium
alloy, except for one such particle among a group of similar-appearing
particles in phagocytic cells within the liver.
Ion Analysis
Because postmortem deterioration of red blood cells had contaminated the
sample, it was not possible to obtain the serum levels of cobalt or chromium
ions. The value for chromium in the urine, 1.59 ng/mL, was three times the
value that was measured in 1995 (0.515 ng/mL). The role of postmortem
concentration or contamination in this value is not known.
The liver samples contained measurable levels of cobalt and chromium that
were higher than those found in the control samples (see Appendix). Metals
with nickel and titanium alloy constituents were also present.
Urban et al.5
reported on the dissemination of wear particles to the abdominal lymph nodes,
liver, and spleen in a study of postmortem specimens from twenty-nine patients
who had metal-on-polyethylene hip or knee replacements that had been in situ
for up to twenty-four years. Metallic particles were detected in the lymph
nodes in two-thirds of the specimens and in the liver or spleen in about
one-third of the specimens, but there was usually no histological or clinical
evidence of pathological changes associated with the presence of the
particles. The lymphatic system is thought to be the major route of particles
from the joint. Particles distributed beyond the pseudocapsule were usually
<1 µm in size and were found more commonly in patients who had a loose
component or who had undergone revision surgery. The principal source of the
metal particles was nonbearing surfaces. Other authors have reported that the
dissemination of wear particles from total joint replacements is more common
when the component is
loose13,14.
It is important to recognize that the production of metal particles and ions
can be greater from secondary sources, such as the fretting corrosion of
modular tapers or loose stems, than from well-functioning metal-on-metal
bearings15-17.
In the present report, the loose stem was a likely source of particles and
ions. The tissues about the joint showed only a moderate histiocytic response.
The presence of giant cells and large tissue spaces suggests that
polymethylmethacrylate, rather than metallic particles, provoked this
response. Despite debonding (if not frank loosening) of the femoral stem,
cobalt and/or chromium-based particles large enough to be visible at the light
microscopic level were essentially absent from reticuloendothelial tissues.
Presuming that the tissues sampled were representative of the particle burden
of the organ as a whole, the absence of visible particles suggests either that
migration away from the joint was limited or that the particles that had
migrated were too small to be detected using the methods employed in this
study (or had already dissolved). The detection of metal ions in the liver
samples could have been the result of very small particles and/or ions. Their
presence verifies that the products of implant wear are stored in the liver.
It is encouraging that there was no evidence of end-organ damage in any of the
tissues that were examined in this case after nearly thirty years of exposure
to metal particulates and ions.
Unfortunately, it was not possible to measure postmortem serum levels of
cobalt or chromium, but the urine level of cobalt was elevated compared with
that measured in 1995. At that time, both the serum and the urine values were
comparable with those measured in a group of eight patients who had a
well-functioning metal-on-metal (McKee-Farrar)
implant9. The
average values for that group of patients at an average of 295 months after
the operation were 1.28 ng/mL for serum chromium, 1.22 ng/mL for urine cobalt,
and 0.9 ng/mL for serum cobalt. The values at that time for our patient were
1.03 ng/mL for serum chromium, 0.51 ng/mL for urine cobalt, and 0.66 ng/mL for
serum cobalt.
The postmortem urine value for cobalt (1.59 ng/mL) was three times that of
the 1995 value. This value may reflect additional ions from the fretting of
the loose implant. Assuming that this was not the result of postmortem
concentration or contamination, and assuming that the blood values were
comparably increased, the resulting levels would be approximately 2.0 ng/mL
for serum cobalt and 3.1 ng/mL for serum chromium. These levels are higher
than the average levels for the group of patients with McKee-Farrar implants
as measured in
19959, but they are
comparable with the levels for the group of patients with McMinn surface
replacements who were described in the same study. A wide range of ion levels
(<1 to >7 ng/mL) was reported by Skipor et al. in a group of twenty-five
patients who had well-functioning metal-on-metal surface
implants15.
It is difficult to assess the possible clinical implications of ion levels
of cobalt and chromium in the urine, blood, and liver without validated
threshold levels for toxicity, carcinogenicity, and other possible systemic
effects. As of this writing, studies that have examined the rates of cancer in
patients who had total hip replacements, including patients who had
metal-on-metal prostheses, have demonstrated conflicting results, but there is
consensus that longer and broader studies need to be carried
out3,18.
Despite the years of experience with loose, failed implants and
epidemiological investigations into possible causal relationships between
metallic implants and cancer, such a relationship has yet to be proven. An
increasing number of the first-generation metal-on-metal components are
entering their third decade of use. Second-generation metal-on-metal bearings
are entering their second decade of use. Continued scrutiny of these patients
with expanded studies of ion levels and autopsy analyses will help to define
the risks and benefits associated with the use of metal-on-metal bearings.
Tables showing the histological results and the ion levels in the liver
samples 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).