Cancer Risk from Orthopedic Prostheses*f

Transcription

1 ANNALS OF CLINICAL AND LABORATORY SCIENCE, Vol. 26, No. 2 Copyright 1996, Institute for Clinical Science, Inc. Cancer Risk from Orthopedic Prostheses*f MICHEL P. COLEMAN, B.M., B.Ch. London School of Hygiene and Tropical Medicine, London, WC1E 7HT, United Kingdom ABSTRACT Permanent replacement of joints damaged by fracture or arthritis has become common over the last 50 years. Vigilance over possible long-term adverse effects of metal prostheses is required. Some of the metal components are potentially carcinogenic. Prolonged contact of metal alloys with body fluids results in gradual corrosion of even the most inert metals. Three cohost studies of persons with a hip prosthesis have been reported; they provide direct, quantitative observations of cancer risk in a human population with hip prosthesis. The design and the results of these studies are similar. Combining the results sharpens the precision of risk estimates. Collectively, the studies examined cancer risk in 40,945 patients followed up for a mean 8.5 years after hip replacement. Overall, the relative risk of cancer was 1.02 (95 percent Cl 1.00 to 1.05). There was an 8 percent excess of haemopoietic malignancy (leukaemia and lymphoma), with a total of 347 cases observed (RR 1.08, 95 percent Cl 0.97 to 1.20). Significant deficits of cancers of the breast and large bowel were seen in the two smaller studies, but combined results from all three studies suggest the relative risk is close to unity. Cancer risk in the first 10 years after hip replacement was not different from that expected, but there was an excess of borderline statistical significance 10 or more years after surgery, with a relative risk of 1.08 (95 percent Cl 1.00 to 1.13) based on 1,005 cases. All three studies were well-designed and executed. Their results are not alarming, but give no cause for complacency, since the number of patients with a prosthesis and the length of time they live with the prosthesis will increase. A register of malignancy complicating joint prosthesis would not help quantify any risk. Instead, a large cohort study of patients with joint prostheses is needed, including information on the type and composition of the prosthesis and on potential confounding exposures for each patient. Measures of corrosion in cancer cases and of tissue levels of relevant metal ions in cases and controls (prosthesis but no cancer) matched for age, sex, and time since insertion would be valuable. Such a study could be done internationally, using orthopedic units with good clinical records for 10 to 15 years in areas with long-term cancer registration. * Send reprint requests to Dr. Michel P. Coleman, Professor of Epidemiology and Vital Statistics, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, U.K. t This paper was presented at the Vth COMTOX Symposium on Toxicology and Clinical Chemistry of Metals, Vancouver, British Columbia, Canada, July 10 to 13, /96/ $01.20 Institute for Clinical Science, Inc.

2 1 40 COLEMAN Introduction Permanent replacement of joints which have become irreparably damaged or dysfunctional as a result of fracture or arthritis has become progressively more common over the last 50 years. This is particularly true for large joints such as the hip and knee in older persons, but more recently for smaller joints as well. Man-made materials are often implanted temporarily to provide physical support for the repair of complex fractures, usually in younger persons. Metal alloys containing cobalt, chromium, nickel, and molybdenum have been used most often, because of their strength, lightness and resistance to corrosion; however, plastics and organic cements have been used more recently. Two arguments suggest that a level of vigilance over the possible long-term adverse effects of metal prostheses is required.1 First, some of the metals used in current prostheses have carcinogenic potential, and can be corroded in situ and disseminated in the body.2,3 Second, both the number of patients who have a prosthesis and the length of time they live with the prosthesis in place are likely to increase. Prolonged contact of metal alloys with body fluids results in gradual erosion of even the most inert metals. Increased serum, urine, and tissue concentrations of cobalt, chromium, and nickel have been reported in patients with alloy prostheses containing these metals.1,4 Corrosion of the metal components of a typical total hip replacement has been estimated to release up to 20 mg of metal a year, which is about twice the total body burden of such metals in a typical 70 kg (154 lb) man.5 Cobalt metal powder and cobaltous oxide are carcinogenic in experimental animals, and there is evidence that metal alloys containing cobalt, chromium, and m olybdenum may also be carcinogenic. Technological advances in the design of prostheses suggest that the pool of patients eligible to receive them will w iden further, and that individual patients will live longer with their prosthesis than before. For hip and knee replacement in the elderly, demographic shift toward even older populations in the developed countries will also greatly increase the number of persons likely to be living with a permanent prosthesis. There have been a number of case reports of malignant neoplasms, mostly sarcomas, arising at the site of orthopaedic prostheses in both animals and humans.4,6,7 On the basis of experiments with tumour induction in rats, Lumb and Sunderman8 have proposed that the latency of tumours induced by nickel compounds might be expected to be years or more, although many of the case reports relate to tumours appearing at much shorter intervals since insertion of the prosthesis. Human Studies Three cohort studies of persons who have had a hip prosthesis have now been reported, and it is worth examining their results with care. This is because they provide direct and quantitative observations of the endpoint of interest in an essentially unselected human population exposed to the risk factor of interest, namely insertion of a hip prosthesis. If there is a carcinogenic risk attached to this procedure, it is reasonably expected that such studies may detect it, even if it may not be possible to say what mechanism underlies any carcinogenic risk. The first study was reported by Gillespie et al9 in which they followed up 1,358 persons who underwent a hip replacement in New Zealand during the period 1966 to 1973 for betw een 6 months and 17 years (mean follow-up,

3 10.6 years) until death or the end of New cancers were identified by the New Zealand national cancer registry, which records practically all cancers arising in the whole population. Cancer incidence in the cohort of patients with a prosthesis was compared with that in the population as a whole. Overall, 164 cancers were observed, not significantly different from the number of cancers (179.4) that would have been expected if patients with a hip prosthesis had the same cancer risk as the general population in each age group and for each sex during the time period covered by the study. The relative risk (RR: the ratio of observed to expected incidence, w here 1.0 w ould represent equality with the general population) was thus 0.91 (164/179.4), with a 95 percent confidence interval (95 percent Cl) of 0.78 to 1.07 (table I). This result is basically reassuring, but it remains compatible with a small excess risk. When the different types of cancer are TABLE I Malignancy After Hip Replacement New Zealand to 1983* CANCER RISK FROM ORTHOPEDIC PROSTHESES 141 Obs Exp RR 95% Cl All cancers Haemopoietic Breast Colorectal Up to 10 years years & over *1,358 patients, mean 10.6 person-years' follow-up. Obs = observed cases. Exp = expected cases (see text). RR = relative risk (the ratio of observed to expected incidence, where 1.0 would represent equality with the general population). 95% Cl = 95% confidence interval. Gillespie WJ, Frampton CMA, Henderson RJ, Ryan PM. J Bone Joint Surg Br 1988;70: examined separately, however, a different picture emerges. There was a significant excess of leukaemias and lymphomas (21 cases observed, RR 1.68, 95 percent Cl 1.06 to 2.60), but significant deficits of cancers of the breast and large bowel. In view of the question of latency, examination of risk by time since insertion of the prosthesis is of interest. The risk of all cancers combined was significantly low in the first 10 years after surgery (107 cases observed, RR 0.74, 95 percent Cl 0.61 to 0.90), although this deficit was no longer significant when the 298 persons whose vital status at the end of the study period could not be satisfactorily determined were removed from the analysis. At 10 or more years after insertion of the prosthesis, overall cancer risk was significantly high (57 cases observed, RR 1.60, 95 percent Cl 1.22 to 2.08). A sim ilar but sm aller study was reported from Finland by Visuri and Koskenvuo.10 They followed up 443 patients (164 men and 279 women) who received one or more McKee-Farrar total hip replacements in the period 1967 to 1973 for an average of some 10 years until development of cancer, death, or the end of 1981, whichever came first. Cancers were ascertained from the Finnish national cancer registry, which also registers virtually all cancers arising in the whole population. Average follow-up was about 10 years. Overall mortality was not significantly different from that expected (97 deaths observed, expected, RR 0.94, 95 percent Cl 0.76 to 1.15). Overall cancer incidence was also similar to that expected (38 observed, 37.4 expected, RR 1.02, 95 percent Cl 0.72 to 1.39; table II). Although numbers of malignant neoplasms at any one site were small, there was a significant three-fold excess of leukaemias and lymphomas (6 observed, 1.99 expected, RR 3.01, 95 per-

4 1 42 COLEMAN TABLE II Malignancy After Hip Replacement Finland to 1981* Obs Exp RR 95% Cl All cancers Haemopoietic Breast Colorectal Up to 10 years years & over patients, mean 13.2 person-years' follow-up. Obs = observed cases. Exp = expected cases (see text). RR = relative risk (the ratio of observed to expected incidence, where 1.0 would represent equality with the general population). 95% Cl = 95% confidence interval. Visuri T, Koskenvuo M. Orthopedics 1991 ;14: cent Cl 1.11 to 6.56). There were fewer cancers of the breast than expected. As in the New Zealand study, no overall excess of cancer was seen in the first 10 years of follow-up. There was a substantial deficit in the first two years after hip replacement, presumably owing to selection of relatively healthy patients for surgery. This is analogous to the healthy worker effect in occupational studies, in which newly recruited workers have extremely low cancer risk and general mortality for the first few years after recruitment. At 10 or more years after hip replacement, cancer risk was higher than expected, though not significantly so (9 cases observed, RR 1.42, 95 percent Cl 0.65 to 2.69). A recent study from Sweden included over 39,000 patients who had one or more hip replacements between 1965 and 1983, with follow-up for a mean 8.4 person-years to the end of Hospital discharge registers covering between 60 to 85 percent of the population during this period were linked to the National Cancer Registry to ascertain all cancers diagnosed at least one year after insertion of the hip prosthesis. Overall cancer risk was some 3 percent higher than in the general population, a small but significant increase (4,572 cases observed, RR 1.03, 95 percent Cl 1.00 to 1.06) (table III). There were only 6 bone tumours (4.3 expected, RR 1.40, 95 percent Cl 0.5 to 3.1), a non-significant excess, and there was no clear trend in this risk with time since surgery. The overall risk of leukaemia and lymphoma was 1.04, much lower than in the previous studies, and, again, there was no increase with time since operation. In further contrast to the results of the previous studies, the risk of breast cancer in women with a hip prosthesis was not different from that of women in the general population. There were significant excess risks of cancer of the kidney (RR 1.41, 95 percent Cl 1.15 to 1.71, based on 105 cases) and melanoma TABLE III Malignancy After Hip Replacement Sweden to 1989* Obs Exp RR 95% Cl All cancers Haemopoietic Breast (F) Colorectal Up to 10 years years & over *39,154 patients, mean 8.4 person-years follow-up. Obs = observed cases. Exp = expected cases (see text). RR = relative risk (the ratio of observed to expected incidence, where 1.0 would represent equality with the general population). 95% Cl = 95% confidence Interval. Nyren O, McLaughlin JK, Gridley G, et al. JNatl Cancer Inst 1995;87:28-33.

5 in women, and of prostate cancer in men (RR 1.13, 95 percent Cl 1.04 to 1.22, based on 638 cases). Overall cancer risk 10 or more years after surgery was again higher than in the first decade, but the difference was small, and much less marked than in the earlier studies. Combined Results The design of these studies and their main results bear some striking similarities. It is worth combining the results, in order to sharpen the precision of the risk estimates by increasing the number of patients under observation. This has the potential to make the studies more informative together than they can be individually. Collectively, the studies examined cancer risk in over 40,000 patients followed up for a mean 8.5 years after hip replacement. Overall, 4,774 cancers were diagnosed in these patients, some 115 more than expected. This represents a small but statistically significant overall increase of 2 percent in cancer risk (RR 1.02, 95 percent Cl 1.00 to 1.05; table IV). CANCER RISK FROM ORTHOPEDIC PROSTHESES 143 In the first two studies, there was an almost two-fold and significant excess of haemopoietic malignancy (lymphoma and leukaemia; 27 cases, RR 1.86), but inclusion of the much larger Swedish data set transforms the picture, and the overall excess is more like 8 percent (table IV). Again, in the first two studies, there were deficits of cancer of the breast, but inclusion of the Swedish data suggests no overall change in breast cancer risk. In combined analysis of the three studies, the overall risk of cancer in the first 10 years after hip replacement was not different from expected (table V). There was a significant 6 percent excess of cancer 10 or more years after surgery, however, with a relative risk of 1.06 (95 percent Cl 1.00 to 1.13, based on 1,005 cases). Comment All three studies were well designed and executed. Cancer ascertainment is likely to have been reliable and fairly complete, and statistical analyses were standard. The results of the small studies TABLE IV Malignancy After Hip Replacement - Type of Cancer* New Zealand Finland Sweden Combined Obs RR Obs RR Obs RR RR 95% Cl All cancers Haemopoietic Breast (F) Colorectal Combined total of 40,945 patiants, mean 8.5 person-years' follow-up. Obs = observed cases. RR = relative risk (the ratio of observed to expected incidence, where 1.0 would represent equality with the general population. 95% Cl = 95% confidence interval. Gillespie, WJ, Frampton CMA, Henderson RJ, Ryan PM. J Bone Joint Surg Br 1988;70: Visuri T, Koskenvuo M. Orthopedics 1991 ;14: Nyren O, McLaughlin JK, Gridley G, et al. J Natl Cancer Inst 1995;87:28-33.

6 144 COLEMAN TABLE V Malignancy After Hip Replacement - Time Since Prosthesis* New Zealand Finland Sweden Combined Obs RR Obs RR Obs RR RR 95% Cl Up to 10 years years & over Total 'Combined total of 40,945 patients, mean 8.5 person-years' follow-up. Obs = observed cases RR = relative risk (the ratio of observed to expected incidence, where 1.0 would represent equality with the general population). 95% Cl = 95% confidence interval. Gillespie, WJ, Frampton CMA, Henderson RJ, Ryan PM. J Bone Joint Surg Br 1988;70: VisuriT, Koskenvuo M. Orthopedics 1991; 14: Nyren O, McLaughlin JK, Gridley G, et al. J Natl Cancer Inst 1995;87: in New Zealand and Finland are strikingly similar, but the large recent study from Sweden does not confirm them all. The Swedish study has substantially greater statistical power, however, and it shows a small but significant overall increase in risk. Although the Swedish study does not show a clear-cut trend in risk with time since insertion of the prosthesis, there is a higher risk after five years than in the first five years, as in the first two studies. In public health terms, the available studies suggest that there is no cause for alarm about general (all causes) mortality or overall cancer risk in patients who have undergone total hip replacement with metal alloy prostheses, especially in view of the obvious benefits of successful joint replacement in restoring mobility and independence. There are several good reasons for avoiding complacency, however. First, some of the metals (and their compounds) used in metal alloy prostheses are known to have carcinogenic potential. Second, corrosion of the prosthesis and deposition of metal compounds in distant tissue causes a range of chronic inflammatory responses.3 Third, increasing numbers of people are now living for extended periods with a joint prosthesis. Fourth, there is a small but significant excess risk of cancer 10 or more years after hip replacement in combined analysis of data from the three available human studies. The first two studies included patients operated up to 1973, and even the recent study from Sweden only included patients first operated upon up to These points provide sufficient reason to mount a large, prosp ective ep id em io lo g ica l study of patients who have had joint prostheses inserted since the 1980s. It has been suggested that further case reports of malignancy arising as a late complication of arthroplasty should be published so that any possible risk can eventually be quantified. 5 The selective nature of the case reports is clear, however. In almost every case, the patient had a tumour at or near the site of the prosthesis, and would in all probability have been referred to the orthopaedic surgeon who inserted it. None of the case reports identified lymphoma or leu kaemia, or a solid epithelial tumour. This

7 is presum ably because in any one patient, coexistence of a prosthesis and the malignancy would be seen as a coincidence, if indeed the treating physician noticed the existence of a prosthesis at all. An international registry of implantrelated tumours has also been suggested by several authors,1,12 but it seems unlikely that this would be successful. How would the tumours be detected? Unless the malignancy actually compromised the prosthesis, referral of the patient back to the orthopaedic surgeon would be unlikely. It seems implausible that sufficiently complete reporting of any malignancy arising in a patient with a previous joint implant could be achieved in this way. The international register of tumours related to breast implants-an initiative of comparable goals and scopeprovided no useful information on cancer risk.13 Case reports are valuable in identifying the possibility of a hazard, but they are of no use at all in quantifying such an infrequent risk. This is for two important reasons. The first is that case reports provide a biased and incomplete view of malignancies arising in patients who have had a prosthesis, as discussed previously. Second, cancer can occur in anyone, and what clinicians want to know is whether or not cancer risk is different in patients who have a prosthesis than in those who did not have one. Case reports cannot tell us this, because they do not provide a quantitative estimate of the risk of malignancy in patients with a prosthesis relative to the risk in the absence of a prosthesis. Cohort studies such as those discussed here address this problem directly. Sunderman1 also makes the point that ep id em io logical studies are needed. It can be argued that there is now a need for a large prospective cohort study of patients with joint prostheses. It should include patients with various CANCER RISK FROM ORTHOPEDIC PROSTHESES 145 types of prosthesis, including if possible information on the type and composition of the alloys used, as well as on potential confounding exposures for the major cancers. Interpretation of the results would also be enhanced by measurements of the degree of corrosion of any implants removed at tumour diagnosis, and of the concentration of the relevant metal ions in body tissues and fluids, both for subjects who do develop a cancer and for at least a sample of patients who do not (controls). Controls could be identified from within the cohort, matched for age, sex, and perhaps duration of implant, at the time of diagnosis of each patient with cancer (case). Conventional case-control techniques would then be directly applicable to the analysis of the data. Visuri and Koskenvuo10 gave a figure of 350,000 total hip replacements in the world each year, and others have estimated that 10 million such operations have been done, while Nyren et al.11 cite a 1983 estimate of 300,000 to 400,000 each year. The U.S. Food and Drug Administration does not collect figures,* but prosthetic joint replacement is now the second most common surgical procedure in the United Kingdom, and a third of the patients are aged under 65 years.3 More than 50,000 hip prostheses were inserted in NHS hospitals in the UK alone in the financial year (the last year for which such figures were available14) when they represented 45 percent of all open orthopaedic procedures (table VI). The figures given by previous authors seem likely to be underestimates. It would only be necessary to study a small proportion of this number of patients in order to reach a definitive and quantitative answer to the question of carcinogenicity of metal alloy joint prostheses. * Personal communication with J. Auerbach, FDA, April, 1995.

8 1 4 6 COLEMAN TABLE VI Hip Operations, National Health Service Hospitals, England, * Total Hip Head of Prosthesis Femur Total No. (%) No. (%) No. (%) With cement 23,602 (72) 7,528 (42) 31,130 (61) Without cement 3,567 (11) 7,428 (42) 10,995 (22) Other 5,729 2,871 8,600 Total 32,898 (100) 17,827 (100) 50,725 (100) 'Department of Health, London, England, Such a study could be carried out internationally, recruiting patients from large orthopaedic units which have retained good clinical records for at least 10 to 15 years and are located in countries or regions where excellent populationbased cancer registration has been in operation for the same period. References 1. Sunderman FW, Hopfer SM, Swift T, Rezuke WN, Ziebka L, Highman P, et al. Cobalt, chromium, and nickel concentrations in body fluids of patients with porous-coated knee or hip prostheses. J Orthop Res 1989;7: Jacobs JJ, Skipor AK, Black J, Urban RM, Galante JO. Release and excretion of metal in patients who have a total hip-replacement component made of titanium-base alloy [see comments]. J Bone Joint Surg Am 1991;73: Case CP, Langkamer VG, James C, Palmer MR, Kemp AJ, Heap PF, et al. Widespread dissemination of metal debris from implants. J Bone Joint Surg Br 1994;76: Sunderman FW. Carcinogenicity of metal alloys in orthopedic prostheses: clinical and experimental studies. Fundam Appl Toxicol 1989;13: Hamblen DL, Carter RL. Sarcoma and joint replacement. J Bone Joint Surg Br 1984;66: Jacobs JJ, Rosenbaum DH, Hay RM, Gitelis S, Black J. Early sarcomatous degeneration near a cementless hip replacement: a case report and review. J Bone Joint Surg Br 1992;74: Aboulafia AJ, L ittelton K, Shm ookler B, Malawer MM. Malignant fibrous histiocytoma at the site of hip replacement in association with chronic infection. Orthop Rev 1994;23: Lumb GD, Sunderman FW. The problem of latency in the development of tumors following exposure to nickel compounds. The Science of the Total Environment 1994;148: Gillespie WJ, Frampton CMA, Henderson RJ, Ryan PM. The incidence of cancer following total hip replacement. J Bone Joint Surg Br 1988;70: Visuri T, Koskenvuo M. Cancer risk after McKee-Farrar total hip replacement. Orthopedics 1991;14: Nyren O, McLaughlin JK, Gridley G, Ekbom A, Johnell O, Fraumeni JF, et al. Cancer risk after hip replacement with metal implants: a population-based cohort study in Sweden. J Natl Cancer Inst 1995;87: Apley AG. Malignancy and joint replacement: the tip of an iceberg? J Bone Joint Surg Br 1989; 71: Coleman MP. Surgery. In: Coleman MP, editor. Cancer risk after medical treatment. Oxford: Oxford University Press, 1991: Department of Health. Hospital episode statistics. Volume 1-Finished consultant episodes by diagnosis, operation and specialty. England: financial year London: Department of Health, 1994.