The first well-documented case of cancer associated
with total joint replacement was in a patient in whom a malignant fibrous
histiocytoma developed three and one-half years after a McKee-Farrar
total hip replacement that was performed in December 19691. At least twenty-four additional cases
of malignant disease occurring in association with a total hip or
knee prosthesis have been reported in the English-language literature2-20. On the basis of this
small number of case reports, there is no apparent relationship
between any specific implanted material and cancer.
The carcinogenic properties of the metals used in joint prostheses
have been studied. Cobalt and chromium wear particles have been
shown to induce carcinoma in animal models21,
giving rise to the concern that such alloys could have the same effect
in human tissue if present in sufficient amounts for a sufficient
length of time. Elevated levels of chromium and cobalt have been
found in human tissues surrounding orthopaedic implants and in tissues
at remote sites22. There also
has been concern that metal-on-metal bearings used
in total hip arthroplasty may pose a higher risk of malignant degeneration
because of an increased exposure to metal particles and ions. Although
metal particles and ions have been the prominent concern, there
also have been reports of cancer induction in association with polymethylmethacrylate
(bone cement)23 and polyethylene24 in animal models.
The speculation that total hip and total knee replacements could cause malignancy
has inspired epidemiological studies aimed at evaluating this concern.
Of the twenty-five reported cases of cancer following a
total hip or knee replacement, twenty-one1-9,11-16,18-20 involved
sarcomas. Because the prevalence of such cancers is low in the general
population, combining the available data enhances the ability to
examine the relative risk of cancer in association with total hip
or knee arthroplasty. The available epidemiological studies regarding
this question encompass more than 140,000 total hip and knee replacements25-33.
While there has been much research on the local effects of total
joint arthroplasty, such as osteolysis, there has been comparatively
little investigation of systemic effects, such as cancer. If a goal
of research is to develop prostheses that will function for thirty
years or more, it is increasingly important to understand the systemic
consequences of joint arthroplasty. This review presents the available
data on cancer associated with total joint arthroplasty in order
to (1) define the current state of knowledge, (2) identify limitations
of the available data, and (3) direct future studies.
With use of MEDLINE via PubMed, the Internet database of the
National Library of Medicine, all articles published between January
1966 and October 1999 in the English-language literature that include
the key words joint, hip, knee, replacement, prosthesis,
arthroplasty, implant, cancer, sarcoma, lymphoma, leukemia, histiocytoma,
hematopoietic, tumor, polyethylene, cobalt, chromium, ions, toxicity,
stainless steel, titanium, and UHMWPE (ultra-high molecular weight
polyethylene) were reviewed.
The search revealed nine studies25-33 that
included the following data: (1) the relative risk of cancer in patients
undergoing elective total hip or knee arthroplasty compared with
that in the general population of the study or a control group and
(2) a standard incidence ratio (SIR) or relative risk and the data
needed to calculate these values. The pooled data from the nine
studies encompassed 110,792 total hip replacements and 29,800 total knee
replacements. Seven of the studies were from Scandinavia27-33; one, from New Zealand25; and one, from Seattle, Washington26.
Relative risk was calculated by dividing the sum of the observed number
of cancer cases associated with hip or knee arthroplasty by the
sum of the expected number of cases in the general population from
each of the nine relevant studies. The expected number of cases
in the general population was derived from reported data from that
geographical region as cited in each study. A 95% confidence
interval was calculated with use of these data through a Poisson
model.
Total Hip Replacement
Risk of All Cancers
A combined total of 12,052 observed and 12,435.4 expected cases
of cancer were identified following total hip replacement (relative
risk, 0.97; 95% confidence interval, 0.95 to 0.99) (Fig. 1-A). Mathiesen
et al.28 separately evaluated
the data for bilateral and revision total hip arthroplasty to obtain
the relative risk for each procedure. Among the 2005 patients who
underwent bilateral total hip replacement, there were 117 observed
and 138.7 expected cases of cancer (relative risk, 0.84; 95% confidence
interval, 0.70 to 1.01). Among the 1258 patients who underwent revision
total hip replacement, there were ninety observed and 86.8 expected cases
of cancer (relative risk, 1.04; 95% confidence interval, 0.83
to 1.27).
In only three studies was the development of cancer categorized
on the basis of the time to diagnosis following total joint replacement25,29,32. The relative risk tended
to stay relatively constant with increasing durations of follow-up, except
in one study consisting of only 1358 patients25,
in which the relative risk increased to 1.60 after ten years of
follow-up. However, in two larger studies29,32, encompassing a total of 70,805 total
hip replacements, no substantial increase was reported over time.
In one of these studies, the relative risk of cancer was <1.0,
regardless of the time to diagnosis following total joint replacement32.
Risk of Hematopoietic Cancer
Hematopoietic cancers are identified as ICD-7 (International Classification
of Disease) numbers 200 through 209 and include non-Hodgkin
lymphoma, Hodgkin disease, myeloma, and leukemia. There was a combined total
of 732 observed and 718.75 expected cases of hematopoietic cancer
following total hip arthroplasty (relative risk, 1.02; 95% confidence
interval, 0.94 to 1.08) (Fig. 1-B). Among the 2005 patients who underwent
bilateral total hip replacement28,
there were five observed and 10.5 expected cases of hematopoietic
cancer (relative risk, 0.48; 95% confidence interval, 0.16
to 1.12). Among the 1258 patients who underwent revision hip replacement,
there were five observed and 6.7 expected cases of hematopoietic
cancer (relative risk, 0.75; 95% confidence interval, 0.24
to 1.75).
In three studies25,29,32, the
development of hematopoietic cancer was categorized on the basis
of the time to diagnosis following total joint replacement. These
studies had variable results. One study32 demonstrated
an increased risk in association with more than ten years of follow-up,
and two25,29 demonstrated a decreased
risk. In one of the latter two studies, the highest relative risk
occurred in the one to four-year follow-up period29.
Risk of Sarcoma
Although the studies did not specifically differentiate sarcomas of
bone and connective tissue from other malignant lesions involving
these tissues, an elevated risk of sarcoma would present in this
group of cancers. The occurrence of bone, soft-tissue, and connective-tissue
malignancy (ICD-7 numbers 196 and 197) following total
hip replacement was specifically reported in five studies involving
a total of 105,166 patients28-30,32,33.
Among these patients, there were sixty-six observed and 63.86 expected
cases of sarcoma (relative risk of 1.03; 95% confidence
interval, 0.80 to 1.31) (Fig. 1-C).
Total Knee Replacement
Risk of All Cancers
In the four studies that included data on knee replacement26,27,30,31, there was a combined total
of 1978 observed and 2142.3 expected cases of cancer (relative risk,
0.92; 95% confidence interval, 0.88 to 0.96). Data on the
development of cancer after bilateral or revision total knee replacement
were not available, nor were data on latency.
Risk of Hematopoietic Cancer
In the four studies that included data on total knee replacement26,27,30,31, there were 154 observed
and 139.6 expected cases of hematopoietic cancer (relative risk, 1.10;
95% confidence interval, 0.94 to 1.29). Data on the development
of hematopoietic cancer after bilateral or revision total knee replacement
were not available, nor were data on the time to diagnosis of cancer
following total knee replacement.
Risk of Sarcoma
In the three studies that included data on bone, connective-tissue, and
soft-tissue malignancy after total knee replacement27,30,31, there were twenty-three observed
and 20.6 expected cases of sarcoma (relative risk, 1.12; 95% confidence
interval, 0.71 to 1.68).
Metal-on-Metal Compared with Metal-on-Polyethylene
Total Hip Arthroplasty
Only Visuri et al.33 and Paavolainen
et al.32 segregated metal-on-metal
prostheses from metal-on-polyethylene hip prostheses.
Visuri et al. evaluated the occurrence of cancer following total hip replacement
performed with the McKee-Farrar prosthesis, a metal-on-metal
device made of a cast cobalt-chromium alloy. In the group of 579
patients who received this prosthesis, there were 113 observed and
118.36 expected cases of cancer (relative risk, 0.95; 95% confidence interval,
0.79 to 1.13). In the study by Paavolainen et al., which involved
only metal-on-polyethylene total hip replacements,
there were 2367 observed and 2626 expected cases of cancer (relative
risk, 0.90; 95% confidence interval, 0.87 to 0.94).
Visuri et al. reported twelve observed and 7.56 expected cases
of hematopoietic cancer (relative risk, 1.59; 95% confidence
interval, 0.82 to 2.77) and zero observed and 0.56 expected cases
of sarcoma (relative risk, 0.00; 95% confidence interval, 0.00
to 6.59) following metal-on-metal total hip replacement33. When Visuri et al. compared the
occurrence of leukemia in patients who had had a metal-on-metal
total hip replacement with that in patients who had had a metal-on-polyethylene
replacement, they found a relative risk of 3.77 (95% confidence
interval, 0.96 to 17.6), suggesting a higher risk among patients
receiving a metal-on-metal implant. Paavolainen
et al. reported 173 observed and 187 expected cases of
hematopoietic cancer (relative risk, 0.93; 95% confidence interval,
0.69 to 1.22) and thirteen observed and seventeen expected cases
of sarcoma (relative risk, 0.76; 95% confidence interval,
0.41 to 1.31) following metal-on-polyethylene
total hip replacement32.
Data on the development of cancer after bilateral or revision metal-on-metal
or metal-on-polyethylene total hip replacement were not available
in either study. Data on the time to diagnosis of cancer following
total joint replacement were available only for metal-on-polyethylene implants
in the study by Paavolainen et al.32.
According to these data, the overall relative risk of all cancers remained <1.0
regardless of latency, although the relative risk of hematopoietic
cancer increased after ten years.
To our knowledge, there have been no studies on the risk of cancer
associate with ceramic bearings and no studies in which data were
stratified on the basis of the type of fixation (use of bone cement
compared with no use of bone cement). Similarly, there are insufficient
data to allow comparisons of different metallic alloys
(cobalt-chromium, titanium, or stainless steel) used in the implants.
Duration of Follow-up
Six of the nine epidemiological studies included an analysis
of the data with regard to the mean duration and range of follow-up25,29-33. The mean duration of follow-up for
the six studies was 7.5 years (range, six months to seventeen years).
In only three of these studies were data on person-years
at risk stratified according to the duration of follow-up25,28,33. In the study by Visuri et
al.33, 36% of the person-years
at risk were associated with patients who had undergone surgery
zero to four years prior to examination; 53%, with those
who had had surgery five to fourteen years prior to examination;
and only 11%, with those who had had surgery at least fifteen
years prior to examination. Only four studies included a follow-up period
of at least ten years25,29,32,33.
A substantial proportion of the cancers were observed during the
first few years after joint replacement. The combined data of Gillespie
et al.25, Mathiesen et al.28, and Paavolainen et al.32 indicate that 34% of all
cancers and 36% of hematopoietic cancers were observed
during the first two postoperative years. The combined data of Mathiesen
et al. and Paavolainen et al. indicate that 39% of sarcomas
were observed during the first two postoperative years. The combined
data of Gillespie et al.25 and
Nyren et al.29 indicate that 32% of
all cancers and 36% of hematopoietic cancers were observed
at zero to four years; 47% and 43%, at five to
nine years; and 21% and 20%, at ten years or more.
Only Nyren et al. evaluated sarcoma in this manner; the rate was
26% at zero to four years, 62% at five to nine years,
and 12% at ten years or more (Fig. 2).
This analysis reveals that the available data have limitations, including
an insufficient length of follow-up for patients who have
undergone total joint replacement, a lack of information regarding
dose-response, and the presence of confounding comorbidities. In
addition, there are limited data on patients from countries outside
of Scandinavia. Therefore, it is unclear if the available data are
relevant to other races, cultures, or geographic regions.
The relative risks of 0.97 for total hip replacement and 0.92 for
total knee replacement suggest that there is a decrease in the risk
of cancer following total hip and total knee replacement compared
with that in the general population. The 95% confidence intervals
are also less than unity, indicating a statistically lower risk.
These data could be interpreted as indicating that total joint replacement
has a protective effect against cancer. It is also possible, however,
that the anesthetic and surgical concerns associated with total
joint arthroplasty result in the selection of a comparatively healthy
patient population, in which case the general health status, and
not the total joint replacement per se, may confer
a reduced risk of cancer.
There has been concern that metal-on-metal
total joint replacements may be associated with an increased risk
of cancer because of an increased exposure to metal particles or
ions. The risk of cancer after metal-on-metal
total hip replacement has been assessed specifically in only one epidemiological
study33. In that study, the relative
risk of cancer was reported to be 0.95 (95% confidence
interval, 0.79 to 1.13), suggesting that there is no apparent increased
risk of cancer after metal-on-metal total hip arthroplasty. In addition,
the risk of sarcoma after metal-on-metal total hip replacement was
found to be 0.00 (95% confidence interval, 0.00 to 6.59)33. However, those same authors found
the relative risk of hematopoietic cancer to be 1.59 (95% confidence
interval, 0.82 to 2.77) following metal-on-metal
total hip replacement and 3.77 (95% confidence interval,
0.96 to 17.6) for leukemia when metal-on-metal implants
were compared with metal-on-polyethylene implants.
Again, the confidence intervals for these data are very broad and
encompass unity, indicating that the risk is statistically neither increased
nor decreased. From an epidemiological perspective, these data are
limited because of the small number of patients (579) who underwent metal-on-metal
total hip replacement. Because this number is small and the numbers
of both observed and expected cases are also small, the strength
of the probability analysis is quite limited. Taken literally, the
relative risk for sarcoma in this cohort (0.00) would suggest that patients
with a metal-on-metal total hip replacement have
a conferred immunity to sarcoma due to the surgery or the implant. This
point highlights the limitations of the data and the need for cautious
interpretation.
Latency is an important consideration in the determination of the
cause of malignant transformation. For a carcinoma to result from
exposure to an exogenous stimulus, cellular changes must occur
and many cycles of division must follow. A malignant cell
would then require numerous divisions before becoming a clinically
apparent cancer. Case et al., in a biochemical study of premalignant changes
in bone marrow adjacent to total hip replacements, found no evidence
of cellular transformation less than ten years after surgery34, suggesting that epidemiological studies
should focus on patients who have had surgery more than ten years
prior to evaluation.
The time from the initial exposure to the diagnosis of cancer has
been reported for carcinogens such as asbestos. Asbestos fibers
are associated primarily with mesotheliomas and bronchogenic carcinomas35. In theory, the carcinogenic properties
of asbestos have some similarity to those of prosthetic particles.
Asbestos fibers are microscopic and phagocytized, inducing chromosomal
mutations (aneuploidy and aberrations) and transformations in mesothelial
cells35 and causing a chronic
foreign-body inflammatory response similar to that induced by
particles from total joint replacement. The malignant degeneration
following asbestos exposure has a much longer latency period than
that following total joint replacement as suggested by these epidemiological
studies. In two studies, mean latency periods of twenty-two years36 and thirty-seven years37 were reported for development of
cancer following asbestos exposure.
With regard to latency, the length of follow-up is another
limitation of these epidemiological studies. Gillespie et al.25 reported that the relative risk
of cancer associated with total hip replacement increased to 1.60 (95% confidence
interval, 1.22 to 2.09) when the duration of follow-up
exceeded ten years. In that small study, there were fifty-seven
cases of cancer among the 1358 patients who had undergone surgery
at least ten years earlier. However, Nyren et al.29 and
Paavolainen et al.32, in studies
based on larger sample populations, did not report such an increase
after ten years. Nyren et al. reported 939 cases of cancer in 39,154
patients who had had a total hip replacement at least ten years
earlier (relative risk, 1.04; 95% confidence interval,
0.98 to 1.11). Paavolainen et al. reported 236 cases of cancer in
31,651 patients who had had a total hip replacement at least ten years
earlier (relative risk, 0.95; 95% confidence interval,
0.83 to 1.07).
Another argument against a causal relationship between total hip
and knee arthroplasty and cancer is that bilateral arthroplasties
have not been associated with a higher risk of cancer. Known carcinogens,
such as asbestos and tobacco, have been shown to have a dose-dependent role
in carcinogenesis38. Thus, a greater
exposure to the materials used for total joint arthroplasty, as
in patients managed with bilateral replacement, would be expected
to result in a greater risk of cancer. However, Mathiesen et al.28 found a relative risk of 0.84 (95% confidence
interval, 0.70 to 1.01) for bilateral total hip arthroplasty, which
contrasts with the dose-dependent properties of established
carcinogens.
An additional limitation of the available data is that comorbidities
that may influence the occurrence of cancer were not accounted for
in most of these epidemiological studies. Some conditions that necessitate
or that are associated with hip or knee replacement have themselves been
associated with an increased risk of cancer. Rheumatoid arthritis
is a common condition that leads to total joint arthroplasty, and
such autoimmune disorders have been associated with a predisposition
to lymphoma39 and leukemia40. Lewold et al.27 stratified
their data on the development of cancer following total knee replacement
according to whether the patients had rheumatoid arthritis or osteoarthritis. Figure 3 demonstrates
that the risk of cancer in patients with rheumatoid arthritis remained
high compared with that in patients with osteoarthritis, regardless
of latency time. This suggests that the elevated risk is associated with
the systemic disease (rheumatoid arthritis) and that it may not
be associated with the local treatment (total knee replacement).
Future studies should stratify data according to diagnosis or should
exclude patients with rheumatoid arthritis in order to avoid a confounding
effect on the association between cancer and total joint replacement.
Other potential comorbidities that have been independently associated
with cancer include Paget disease41,
osteomyelitis42, hereditary bone
dysplasia43, and bone infarcts44. These conditions have been linked
to malignant fibrous histiocytoma, a soft-tissue sarcoma that
has been reported frequently in studies of cancer following total
hip and knee arthroplasty.
Thermal necrosis can result from the use of polymethylmethacrylate
bone cement during orthopaedic procedures45.
Areas of bone immediately adjacent to cement become necrotic and
undergo repair in a manner similar to bone infarcts. The association
between bone infarcts and malignant fibrous histiocytoma is the
basis for the theory that this type of tumor may arise in bone adjacent
to cement as a result of the repair process becoming malignant3. However, there have been no studies,
to our knowledge, in which the risk of cancer in patients who have
had total joint replacement with bone cement is compared with that
in patients who have had total joint replacement without cement.
In summary, the available data do not support a causal link between
total hip or knee arthroplasty and the development of cancer. Although
it is biologically plausible for the materials used in total joint
replacement to induce malignant degeneration, this relationship
has not been demonstrated. There is ongoing concern about an association between
hematopoietic cancer and total joint replacement. The most important
finding of this review, however, is the identification of the limitations
of the available data. In order to better define the relationship
between total hip or total knee replacement and cancer, future studies must
include larger patient populations with more racial and cultural
diversity, adjust for comorbidities, focus on patients who had the
surgery more than ten years previously, and stratify the data on
the basis of the specific materials implanted and whether or not
bone cement was used.