ESKD indicates end-stage kidney disease; RRT, renal replacement therapy. The standardized incidence ratios and 95% confidence intervals for all cancer sites can be found at http://web.med.unsw.edu.au/nchecr.
Vajdic CM, McDonald SP, McCredie MRE, van Leeuwen MT, Stewart JH, Law M, Chapman JR, Webster AC, Kaldor JM, Grulich AE. Cancer Incidence Before and After Kidney Transplantation. JAMA. 2006;296(23):2823-2831. doi:10.1001/jama.296.23.2823
Author Affiliations: National Centre in HIV Epidemiology and Clinical Research, University of New South Wales, Sydney, Australia (Drs Vajdic, Law, Kaldor, and Grulich, and Ms van Leeuwen); Australia and New Zealand Dialysis and Transplant Registry, Queen Elizabeth Hospital, Adelaide, Australia (Drs McDonald, Chapman, and Webster); Disciplines of Medicine and Public Health, University of Adelaide, Adelaide, Australia (Dr McDonald); Department of Preventive and Social Medicine, University of Otago, Dunedin, New Zealand (Drs McCredie and Stewart); and Centre for Transplant and Renal Research, Millennium Institute, Westmead Hospital, University of Sydney, Sydney, Australia (Drs Chapman and Webster).
Context Immune suppression after organ transplantation is associated with a markedly increased risk of nonmelanoma skin cancer and a few virus-associated cancers. Although it is generally accepted that other cancers do not occur at increased rates, there have been few long-term population-based cohort studies performed.
Objective To compare the incidence of cancer in patients receiving immune suppression after kidney transplantation with incidence in the same population in 2 periods before receipt of immune suppression: during dialysis and during end-stage kidney disease before renal replacement therapy (RRT).
Design, Setting, and Participants A population-based cohort study of 28 855 patients with end-stage kidney disease who received RRT, with 273 407 person-years of follow-up. Incident cancers (1982-2003) were ascertained by record linkage between the Australia and New Zealand Dialysis and Transplant Registry and the Australian National Cancer Statistics Clearing House.
Main Outcome Measure Standardized incidence ratios (SIRs) of cancer, using age-specific, sex-specific, calendar year–specific, and state/territory–specific population cancer incidence rates.
Results The overall incidence of cancer, excluding nonmelanoma skin cancer and those cancers known to frequently cause end-stage kidney disease, was markedly increased after transplantation (n = 1236; SIR, 3.27; 95% confidence interval [CI], 3.09-3.46). In contrast, cancer incidence was only slightly increased during dialysis (n = 870; SIR, 1.35; 95% CI, 1.27-1.45) and before RRT (n = 689; SIR, 1.16; 95% CI, 1.08-1.25). After transplantation, cancer occurred at significantly increased incidence at 25 sites, and risk exceeded 3-fold at 18 of these sites. Most of these cancers were of known or suspected viral etiology.
Conclusions Kidney transplantation is associated with a marked increase in cancer risk at a wide variety of sites. Because SIRs for most types of cancer were not increased before transplantation, immune suppression may be responsible for the increased risk. These data suggest a broader than previously appreciated role of the interaction between the immune system and common viral infections in the etiology of cancer.
It is widely accepted that immune suppression after organ transplantation is associated with a markedly increased risk of nonmelanoma skin cancer, non-Hodgkin lymphoma, and Kaposi sarcoma.1 Whether other cancers occur at increased rates is controversial, because there have been few long-term population-based cohort studies.
The largest previous population-based study of cancer in kidney transplant recipients reported increased incidence of a wide range of cancer types,2 challenging the hypothesis that only a few cancers occur at increased rates. In an analysis of a subset of these recipients, there was greater excess cancer risk after transplantation than during dialysis.3 However, uncertainty remains about which cancers occur at increased rates after transplantation. In addition, there is also uncertainty as to whether the increased risk is due to immune suppression or related to preexisting cancer risk factors, and factors related to end-stage kidney disease (ESKD) or dialysis. Comparing cancer incidence rates in individuals before and after transplantation allows an analysis of the effect of immune suppression while controlling for these factors.
In our study, we present cancer incidence based on cancer registration for a population-based nationwide register of 28 855 patients with ESKD who received renal replacement therapy (RRT) in 3 separate periods: the 5 years before RRT commencement, during dialysis, and after transplantation.
The Australia and New Zealand Dialysis and Transplant Registry (ANZDATA) is a population-based registry of all patients who start maintenance dialysis or receive kidney transplantation in Australia or New Zealand.4 Our study population was all Australian residents registered on ANZDATA before January 1, 1982, and still alive at January 1, 1982, or registered after this date up to September 30, 2003. The principal inclusion criterion was that the patient was eligible for registration by an Australian cancer registry, if he/she had developed cancer. On this basis, we excluded New Zealand patients, and those patients who only spent a portion of their follow-up time in Australia (n = 378), as well as second cancers of the same histological type and organ or organ system.5
Patients were not excluded on the basis of their cancer history before ESKD, dialysis, or transplantation. However, after a cancer diagnosis, patients did not contribute follow-up time to the determination of person-years at risk for that cancer type. These patients did continue to contribute follow-up time for all other cancer types. As myeloma, kidney, and urinary tract cancers are known to frequently cause ESKD, they were examined separately and were not included in the all-cancer rates. In addition, we excluded 32 cases of other cancer types that were diagnosed before transplantation and for which the cancer, or its treatment, was recorded by ANZDATA as the primary cause of ESKD. Human immunodeficiency virus (HIV) serology is not recorded by ANZDATA but anecdotal reports indicate that HIV infection is exceedingly rare in the Australian ESKD population. Racial origin was self-defined by patients using the ANZDATA classification scheme, which included Caucasoid, Australian Aborigine, Torres Strait Islander, and other options together with a free-text option (Chinese; Maori; Arab; Cook Islander; Samoan; Tongan; Pacific People, other [specify]; Indian; Indonesian; Malay; Filipino; Vietnamese; other [specify]).
Ethical approval was obtained from 9 institutional review boards. The requirement for informed consent from patients was waived because the researchers received only anonymized data.
The National Cancer Statistics Clearing House (NCSCH) is a compilation of data from all 8 Australian population-based state and territory cancer registries to which all cases of primary invasive cancer, except nonmelanoma skin cancer, must be reported by statute.6 The NCSCH data were available from January 1, 1982, to December 31, 2001, December 31, 2002, or December 31, 2003, depending on the jurisdiction of cancer registration.
Record linkage was performed by using an established probabilistic matching technique,7 with a customized algorithm and manual clerical review of potential matches using explicit rules. The ANZDATA and NCSCH records were matched on all elements of each registrant's name, sex, date of birth, date of death (if dead), and state or territory of residence to obtain date of diagnosis, topography, and morphology for each primary invasive neoplasm.
Population cancer incidence rates by 5-year age group, sex, year, and state/territory for each year from 1982 to 2001 were obtained for all cancer sites; rates for 2001 were used for later years. The exception was Kaposi sarcoma, in which the population rates for 1982 were applied to all calendar years because of the impact of AIDS-related Kaposi sarcoma in later years.
The ratio of observed to expected numbers of cancers, the standardized incidence ratio (SIR), was calculated with exact (Poisson) 95% confidence intervals (CIs) by using Stata version 8 (STATACorp LP, College Station, Tex). The expected numbers of incident cases were calculated by multiplying the person-years at risk by the corresponding population cancer incidence rates. Tests for linear trend across the exposure periods were performed by using Poisson maximum likelihood regression. P<.05 was considered statistically significant.
Person-years were accrued from January 1, 1982, in 3 periods of ESKD: (1) up to 5 years before commencing RRT; (2) time receiving dialysis treatment up to the date of first transplantation, and (3) time from the date of first transplantation. The last period therefore included all follow-up time after a transplant, including time on dialysis after transplant failure, and time after second- and higher-order transplantations. In each period, person-years were terminated at death, loss to follow-up (n = 120 [0.4%]), or December 31 of the year for which NCSCH records were available, and for each cancer type, at the date of diagnosis of the first such cancer.
The period before commencing RRT was retrospectively defined as 5 years before entry into the ANZDATA register. To account for the proportion of patients who would have developed and died of cancer before receiving RRT, the number of expected cases for this period was adjusted for survival, assuming kidney disease had no effect on survival and that site-specific, all-age relative survival applied. This approach has been used in studies of cancer incidence before AIDS registration.8- 10 Based on availability, Australian national11 or South Australian state12 population-based survival rates were used.
Standardized incidence ratios were calculated for all cancers, excluding nonmelanoma skin cancer and those cancers known to frequently cause ESKD, and each cancer type for each of the 3 periods of ESKD. Standardized incidence ratios were also computed for all cancers by time since dialysis, time since transplantation, and primary renal disease classified as due to diabetic nephropathy, glomerulonephritis, hypertensive and arteriopathic conditions, and other and unknown causes.
The total cohort consisted of 28 855 patients with ESKD, with 273 407 person-years of follow-up (Table 1). Of these patients, 25 685 (14 616 men and 11 069 women; mean age, 50 years) contributed to the period before RRT, 24 926 (14 144 men and 10 782 women; mean age, 54 years) to the dialysis-only period, and 10 180 (5966 men and 4214 women; mean age, 41 years) received 1 or more transplants. The mean (SD) length of follow-up was 4.6 (1.0) years before RRT, 2.7 (2.5) years during dialysis, and 8.5 (6.3) years after transplantation. A single transplant was received by 8607 patients, while 1327 received 2 and 246 received 3 or more. Immunosuppression before RRT, most commonly prednisolone with or without cyclophosphamide, was received by 8% of the cohort. Twenty-seven percent of recipients who underwent transplantation (n = 2712) received polyclonal or monoclonal anti–T-cell antibody treatment, of which 1137 (42%) were prophylactic and the remaining 1575 (58%) were for the treatment of rejection.
The incidence of all cancers, excluding nonmelanoma skin cancer and those cancers known to frequently cause ESKD (myeloma, kidney, and urinary tract), was significantly increased above the population rate for each of the ESKD periods (Table 2, Figure 1). The SIR for the dialysis period was significantly higher than that for the before RRT period (P = .002). After transplantation, the risk was markedly higher than in both preceding periods (P for trend<.001).
Up to 5 years before RRT, 689 incident cancers were observed compared with 594 expected (SIR, 1.16; 95% CI, 1.08-1.25). The corresponding values on dialysis and after transplantation were 870 vs 643 (SIR, 1.35; 95% CI, 1.27-1.45) and 1236 vs 378 (SIR, 3.27; 95% CI, 3.09-3.46), respectively. The overall risk was similar for men (SIR, 1.15; 95% CI, 1.04-1.26) and women (SIR, 1.19; 95% CI, 1.05-1.34) before RRT (P = .64), but higher in women (SIR, 1.53; 95% CI, 1.37-1.70) than in men (SIR, 1.26; 95% CI, 1.16-1.37) during dialysis (P = .006), and higher in men (SIR, 3.45; 95% CI, 3.20-3.70) than in women (SIR, 3.03; 95% CI, 2.77-3.31) after transplantation (P = .03).
Before RRT, the incidence of non-Hodgkin lymphoma and Kaposi sarcoma, and cancer of the lip, colon, and thyroid was significantly increased (but more than 2-fold only for Kaposi sarcoma and thyroid cancer), while anal cancer was significantly decreased (Figure 1).
During dialysis, the incidence of Kaposi sarcoma and cancer of the lip, tongue, stomach, small intestine, liver, lung, cervix, thyroid, and unspecified site was significantly increased, and anal and prostate cancer were significantly decreased (Figure 1). The increases were greater than 2-fold for all but stomach and lung cancer. The average time to cancer during dialysis was 2.8 years, which was very similar to the average length of follow-up during dialysis, and did not vary greatly by cancer site (Table 2). There was no trend in the incidence of all cancers with increasing time since dialysis (Table 3). The all-cancer SIR was not increased for those patients with diabetic nephropathy but was increased to a similar extent for those with all other causes of ESKD.
After transplantation, a significant excess was observed for melanoma, Kaposi sarcoma, non-Hodgkin lymphoma, Hodgkin disease, leukemia, and cancer of the lip, tongue, mouth, salivary gland, esophagus, stomach, colon, anus, liver, gallbladder, lung, connective and other soft tissue, vulva, cervix, penis, eye, thyroid, and unspecified site (Figure 1). A significant excess, based on less than 5 cases, was also observed for nasal cavity (SIR, 5.54; 95% CI, 1.51-14.19) and vaginal cancer (SIR, 15.57; 95% CI, 4.24-39.85). Except for thyroid cancer, the SIRs were greater than those before transplantation. The magnitude of the increase was more than 3-fold for 18 of these 25 sites. No cancer occurred at significantly decreased incidence after transplantation. The average time to cancer after transplantation was 9.4 years, and exceeded 10 years for 15 sites (Table 2). The all-cancer SIR increased with increasing time since transplantation (P for trend<.001) (Table 3). The all-cancer SIR was increased to a similar extent for each category of primary renal disease (Table 3).
The incidence of cancers known to frequently cause ESKD (myeloma, kidney, and urinary tract) was significantly increased before RRT, during dialysis, and after transplantation (Table 4 and Figure 2). During dialysis, some of these cancers were diagnosed soon after dialysis commenced, but after transplantation the average time to cancer was very similar to that for the non-ESKD–related sites (Table 4). The SIR for all cancers when the ESKD-related sites were included was 1.98 (95% CI, 1.88-2.09) before RRT, 1.63 (95% CI, 1.53-1.72) during dialysis, and 3.40 (95% CI, 3.22-3.59) after transplantation.
After kidney transplantation, there was a marked increase in incidence of a wide range of cancer types. Although the incidence of some cancers was increased during dialysis, and the incidence of a few was increased before RRT, the magnitude and breadth of the increased risk after transplantation suggests that immune suppression causes a substantial and broad-ranging increase in cancer risk. Our comparison between the 3 periods demonstrates that preexisting personal cancer risk factors, and factors related to primary renal disease, ESKD, or dialysis can be excluded as major contributors to the posttransplantation excess risk.
There are no previous population-based estimates of cancer incidence for patients with ESKD before RRT. We found a slight increase in incidence of all cancers, and a significant increase for only 5 cancer types, including Kaposi sarcoma and non-Hodgkin lymphoma. Given that these are virus-associated cancers (Table 5), the likely cause of their increased incidence is uremic immune dysfunction.19,20 This explanation is supported by evidence of reactivation of latent Epstein-Barr virus infection in uremic immunodeficiency.21 A contributing factor may be the use of immunosuppressive agents to treat diseases that cause ESKD.
Prior population-based studies of cancer during dialysis have shown no22 or only a slight increase in overall risk.3 A viral etiology has been established or suggested (Table 5) for 6 of the 9 cancers that we observed at significantly increased incidence during dialysis (tongue, stomach, lung, cervix, Kaposi sarcoma, and liver). Other characteristics of uremia and dialysis that may contribute to the excess cancer include nutritional deficiencies and metabolic changes, the retention of carcinogenic compounds, the underlying renal disease or its treatment, or an interaction of uremic and dialysis-induced immune dysfunction with established risk factors,23 such as UV radiation (lip cancer), tobacco (lip, lung, and cervical cancer), or alcohol (liver cancer).
Previous population-based estimates of cancer risk after kidney transplantation, which have included nonmelanoma skin cancer and cancers that frequently cause ESKD, have found a 3- to 4-fold increase in risk.2,3 Our study confirms and extends this finding, being the first to our knowledge with sufficient power to examine site-specific cancer risk for a wide spectrum of cancers. Cancers known or suspected to be associated with viral agents appear to comprise much, but not all, of the excess risk we found after transplantation.
Of the 18 specific cancers with greater than 3-fold increase in risk, 5 were at sites that are known to be caused by human papillomavirus (tongue, mouth, vulva, vagina, penis) and an additional 4 were at sites for which the evidence for human papillomavirus is limited or inconclusive (eye, salivary gland, esophagus, nasal cavity). Two cancers were known to be causally related to Epstein-Barr virus (Hodgkin disease, non-Hodgkin lymphoma), one was of a type known to be related to hepatitis B or hepatitis C virus (liver cancer), and the other was caused by human herpesvirus 8 (Kaposi sarcoma).
Among the remaining 5 cancer types, viral infection is not generally accepted as the primary cause. Infection with unknown agents is a possible explanation of increased risk. Infection with hepatitis C virus, hepatitis B virus, and liver flukes has been suspected as causing biliary tract cancer.24,25 For lip cancer, exposure to UV radiation and tobacco products are thought to be important causal factors,26 but the role of human papillomavirus is under study.27 It is not clear why rates of thyroid cancer should occur at increased risk in all 3 periods examined. The increased rates of connective tissue cancer may represent misclassified Kaposi sarcoma, and the only other cancer type occurring at more than 3-fold risk was the unclassified category.
There were an additional 7 cancer types that were significantly increased, but increased less than 3-fold, after transplantation. Among these, 2 are known to be human papillomavirus–related (cervical and anal cancer), and there is inconclusive evidence of a role for human papillomavirus in 3 others (stomach, colon, and lung cancer). There is no evidence for an infective agent causing melanoma or the majority of cases of leukemia.
Our findings are supportive of a broad-ranging role of viral infection in the development of cancer in immune deficiency. Many of the cancers occurring at increased risk were human papillomavirus–associated, but some cancers that have been suggested as being caused by human papillomavirus were not increased posttransplantation. There was no excess posttransplantation risk for the 2 most common cancers in the Australian population, breast and prostate cancer.6
In HIV-related immune deficiency, it has been generally accepted that only a limited range of cancers, including Kaposi sarcoma, non-Hodgkin lymphoma, and cervical cancer, occur with increased incidence.28 The recent advent of effective antiretroviral therapies for the treatment of HIV infection has led to much longer survival from HIV, and our data raise concern that long-term HIV-related immune deficiency may also be associated with increased risk of a wider range of cancers.
Our study had several strengths, including the large size of the cohort encompassing all patients with RRT in Australia, the long period of follow-up, and the ascertainment of cancers using a national registry. Therefore, patients with RRT were under follow-up for cancer unless they emigrated. Other strengths are the use of regional population cancer rates to take into account the marked variation in incidence for some cancers,6 and the exclusion of cancers known to frequently cause ESKD from the all-cancer incidence rates.
Our study also had some limitations. Because the period before RRT was retrospectively defined, it is possible that the incidence of cancer may be underestimated if some patients were not considered for RRT because of their cancer history, or because they died before RRT. Our adjustment for cancer survival will have lessened this effect,8 but if patients with ESKD with cancer have shorter than average survival, some underestimation will remain. It is also possible that the estimate of cancer risk due to ESKD may be reduced by those patients with rapidly progressive renal disease. On the other hand, if increased medical surveillance due to cancer diagnosis led to the diagnosis of ESKD and commencement of RRT, overestimation of cancer incidence would result. Despite these limitations, the similarity in pattern of increased cancer risk before RRT and during dialysis supports a lack of major bias in our estimates of risk.
Another possible source of bias was heightened surveillance for cancer. The lack of excess breast and prostate cancer, 2 cancers commonly diagnosed by opportunistic screening, as well as the distinct pattern of cancers in the different periods despite close medical supervision in all 3 periods, argues against this bias playing an important role. The exception may be cervical cancer for which the risk was similar during dialysis and after transplantation, in sharp contrast with the pattern for the other human papillomavirus–related cancers. This apparent paradox may reflect the regular screening for cervical but not the other human papillomavirus–related cancers.
The extent of any misclassification error will depend on the validity of our linkage techniques, and the accuracy and completeness of the ESKD and cancer registries. Our linkage methodology proved to be 99% sensitive and 100% specific in identifying cases of AIDS-related non-Hodgkin lymphoma on the New South Wales cancer registry.29 The accuracy and completeness of ESKD registration is routinely verified by active follow-up, and Australian cancer registration is of high quality.30
We believe that the restriction of markedly increased cancer risks to the posttransplantation period argues against other common causes of cancer acting on their own as an explanation of the increased risks. In support of this, for adults in our cohort registered on ANZDATA since 2000, there was no evidence that rates of smoking were higher than for the general population.31 In addition, the all-cancer SIRs in the posttransplantation period were similar among all causes of ESKD examined. We do not have data on the alcohol intake for the cohort.
After kidney transplantation, a wide variety of cancers across a number of organ systems occur with substantially increased incidence. Most, but not all, of these cancers are those with known or suspected viral causes. In contrast, cancer incidence was only slightly increased before kidney transplantation. Our findings point to an important role of the interaction between common viral infections and the immune system in the etiology of cancers at a broad range of sites.
Corresponding Author: Claire M. Vajdic, PhD, National Centre in HIV Epidemiology and Clinical Research, University of New South Wales, 376 Victoria St, Level 2, Darlinghurst, New South Wales 2010, Australia (firstname.lastname@example.org).
Author Contributions: Dr Vajdic had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Vajdic, McDonald, McCredie, Stewart, Law, Chapman, Kaldor, Grulich.
Acquisition of data: Vajdic, McDonald, van Leeuwen.
Analysis and interpretation of data: Vajdic, McDonald, McCredie, van Leeuwen, Stewart, Law, Chapman, Webster, Kaldor, Grulich.
Drafting of the manuscript: Vajdic, McDonald, McCredie, van Leeuwen, Stewart, Grulich.
Critical revision of the manuscript for important intellectual content: Vajdic, McDonald, McCredie, van Leeuwen, Stewart, Law, Chapman, Webster, Kaldor, Grulich.
Statistical analysis: van Leeuwen, Law.
Obtained funding: Vajdic, McDonald, McCredie, Stewart, Law, Chapman, Kaldor, Grulich.
Administrative, technical, or material support: Vajdic, van Leeuwen, Law, Kaldor, Grulich.
Study supervision: Vajdic, Grulich.
Financial Disclosures: Dr Chapman reports being on advisory boards and speaker panels for Astellas, Novartis, Wyeth, and Hoffmann la Roche, and receiving research support from the National Health and Medical Research Council, Juvenile Diabetes Foundation International, Novartis, Wyeth, Jannsen-Cilag, and Hoffmann la Roche. Dr Webster reports receiving an International Postgraduate Research Scholarship from the Australian Government Department of Education, Science, and Training, and an International Postgraduate Research Award from the University of Sydney. Dr Grulich reports being on the advisory board for the Gardasil human papillomavirus vaccine for the Commonwealth Serum Laboratories. No other authors reported financial disclosures.
Funding/Support: The study was funded by the Cancer Council NSW (RG 47/03). Dr Vajdic has a postdoctoral research fellowship from the National Health and Medical Research Council (ID 209668). Ms van Leeuwen has a postgraduate scholarship from the National Health and Medical Research Council (ID 401131). The ANZDATA Registry administrative office is supported by funding from the Australian Government Department of Health and Aging, the New Zealand Ministry of Health and Kidney Health Australia; data collection costs are borne by contributing renal units. The National Centre in HIV Epidemiology and Clinical Research is funded by the Australian Government Department of Health and Aging.
Role of the Sponsors: The funding organizations played no role in the design and conduct of the study, in the collection, management, analysis, and interpretation of the data, or in the preparation, review, or approval of the manuscript.
Disclaimer: The interpretation and reporting of the ANZDATA Registry data are the responsibility of the authors and in no way should be seen as official policy or interpretation of the Australia and New Zealand Dialysis and Transplant Registry.
Acknowledgment: We gratefully thank the dedication and care with which dialysis and transplantation units throughout Australia have regularly submitted the information on which this analysis has been performed, and the ANZDATA staff who have created and maintained the database so accurately. We are also grateful to the staff of the state and territory cancer registries for the use of their data. We thank the Australian Institute of Health and Welfare, and the Cancer Council Victoria, for their commissioned work on this study. The Australian Institute of Health and Welfare was paid for the data linkage they conducted and the population cancer rates they provided to us. The Cancer Council Victoria was paid for the data linkage they conducted. We thank the gratis biostatistical advice and support we received from Janaki Amin, BSc(Hons), MPH(Hons), National Centre in HIV Epidemiology and Clinical Research.