Tel: +46-8-58587847, Fax: +46-8-58581020
Risk of skin cancer and other malignancies in kidney, liver, heart and lung transplant recipients 1970 to 2008—A Swedish population-based study
Article first published online: 28 AUG 2012
Copyright © 2012 UICC
International Journal of Cancer
Volume 132, Issue 6, pages 1429–1438, 15 March 2013
How to Cite
Krynitz, B., Edgren, G., Lindelöf, B., Baecklund, E., Brattström, C., Wilczek, H. and Smedby, K. E. (2013), Risk of skin cancer and other malignancies in kidney, liver, heart and lung transplant recipients 1970 to 2008—A Swedish population-based study. Int. J. Cancer, 132: 1429–1438. doi: 10.1002/ijc.27765
- Issue published online: 24 JAN 2013
- Article first published online: 28 AUG 2012
- Accepted manuscript online: 7 AUG 2012 03:26AM EST
- Manuscript Accepted: 18 JUL 2012
- Manuscript Received: 3 APR 2012
- Karolinska Institutet
- Strategic Research Program in Epidemiology
- Svenska Sällskapet för Medicinsk Forskning (Gustaf Edgren)
- organ transplantation;
- cancer risk;
- skin cancer;
- population-based study
Organ transplant recipients are at increased risk of a wide range of malignancies, especially cutaneous squamous cell carcinomas (SCC). Few previous population-based studies have quantified and compared cancer risks according to graft type and with long-term follow-up. Using nationwide Swedish registers, we identified 10,476 recipients transplanted from 1970 to 2008 and followed them for cancer occurrence. Relative risks of cancer in comparison with the general population were expressed as standardized incidence ratios (SIR) and within the transplanted cohort as incidence rate ratios (IRR). During a total follow-up of 93,432 person-years, patients were diagnosed with 1,175 cancers excluding SCC, and with 2,231 SCC, SIRcancer excl SCC 2.4 (95% CI, 2.2–2.5); SIRSCC 121 (95% CI, 116–127). Cancer risks were most increased among heart and/or lung recipients SIRcancer excl SCC 3.3 (95% CI, 2.8–4.0); SIRSCC 198 (95% CI, 174–224), followed by kidney SIRcancer excl SCC 2.3 (95% CI, 2.1–2.4); SIRSCC 121 (95% CI, 116–127) and liver recipients SIRcancer excl SCC 2.3 (95% CI, 1.9–2.8); SIRSCC 32 (95% CI, 24–42). During follow-up, risk of cancer excluding SCC remained stable while risk of SCC tripled over 20 years irrespective of graft type, partly due to a subgroup of patients developing new SCCs at a rapidly increasing rate. In summary, post-transplant cancer risk varied by transplanted organ and by cancer site, with the bulk of the excess risk driven by an exceptionally high and accelerating risk of SCC. These findings underscore the importance of regular skin screening in organ transplant recipients.
Organ transplant recipients are at greatly increased risk of a range of malignant tumors.1–3 With improved care, resulting in better graft and patient survival, post-transplant malignancies have become an increasingly important cause of both morbidity and mortality. Previous research indicates that cancer risks are driven primarily by the state of immunodeficiency, often in combination with oncogenic viral infections, rather than lifestyle or underlying disease.4, 5 Virus-related malignancies consistently associated with an increased risk in transplant recipients include non-Hodgkin-lymphoma,6 hepatocellular carcinoma,7 squamous cell carcinoma of the cervix, the outer genital tract, Kaposi's sarcoma4, 8 and Merkel cell carcinoma.9, 10 Despite still inconsistent evidence of a viral etiology in squamous cell carcinoma (SCC) of the skin11, 12—by far the most frequent type of post-transplant nonmelanoma skin cancer (NMSC)—65- to 250-fold increased risks of SCC have been reported in transplant recipients.9, 13, 14
During the past decade, population-based studies have reported two- to four-fold increased risks of cancer overall among kidney,1, 5, 15–18 liver19, 20 and heart transplanted patients.21 However, most previous studies have not included SCC, the most important driver of transplant patient cancer risk.1, 2, 5, 18, 20, 21 Also, few studies have directly compared risk of cancer overall2, 21, 22 or of skin cancer22, 23 by type of transplanted organ, and few studies have assessed cancer risks relative to the general population beyond 15 years posttransplantation.24 Therefore, we aimed to investigate cancer risk overall and by site, with a focus on SCC, among primarily kidney, liver, heart and lung transplanted patients in Sweden between 1970 and 2008.
Material and Methods
The study cohort was established through the Swedish National Patient Register recording individual hospital discharges and surgical procedures since 1964, with coverage of all transplant centers from1970.25 The recorded information includes a unique 10-digit national registration number (NRN) for each person, dates of hospital admissions and discharges, surgical procedures and discharge diagnoses (ICD-8-10). Transplant recipients of any solid organ (kidney, liver, heart, lung, pancreas and small intestine) were identified using surgical procedure codes.26 Discharge diagnoses were used to assess the indication of transplantation. The overall register incompleteness has been estimated to be less than 2% and a prior evaluation of the Patient Register surgical procedure coding revealed high accuracy of the data.27
The occurrence of cancer was ascertained through linkage with the Swedish Cancer Register using NRNs.28, 29 Reporting by both diagnosing physicians and pathologists to the Cancer Register is mandatory by law, resulting in near complete registration.30 Invasive cancers were identified by basic codes for anatomical site according to ICD-7. Cancer in situ and basal cell carcinoma were not included. Patients with a history of cancer before transplantation and patients diagnosed with cancer within 30 days of first transplantation were excluded. Thus, follow-up started 30 days posttransplantation and extended until date of death, emigration or end of the study (December 31, 2008), whichever came first. Patients were allowed to have multiple primary cancer diagnoses, and therefore there was no censoring upon first cancer diagnosis.
For NMSC (ICD-7: 191), we further used histopathological classifications based on WHO/HS/CANC/24.1 coding, to define the following subgroups: cutaneous SCC (146), Merkel cell carcinoma (446), Kaposi's sarcoma (505, 566) and appendageal carcinomas (046, 096). Risks of malignant and premalignant tumors among pediatric organ recipients (5% of the current study population) and risk of non-Hodgkin lymphoma in this cohort have been reported separately.31, 32
As the indication for transplantation was not recorded in a standardized way in the registers, we imputed these using a hierarchical algorithm based on hospital discharge diagnoses at the time of and during two years prior to transplantation. For example, a kidney transplant patient with a diagnosis of polycystic kidney disease was assumed to have been transplanted for this disease, irrespective of any other kidney-related diagnoses. We also used combinations of indications where necessary, e.g., a patient with a diagnosis of unspecified chronic renal failure and diabetes was assumed to have been transplanted due to diabetic nephropathy.
Relative risks of cancer in transplanted patients compared to the general population were expressed as standardized incidence ratios (SIR), i.e., the ratio of the observed to the expected numbers. The expected number was calculated by multiplying the calendar period-, age- and sex-specific follow-up time in the cohort with the corresponding incidence in the general population. Ninety-five percent confidence intervals (CI) were computed under the assumption that the observed number of cases followed a Poisson distribution.33 The SIRs were calculated for cancer overall and for first cancer, by anatomical site, and for selected sites also by histopathological sub-classification. The analyses were stratified by transplanted organ (kidney, liver, heart and/or lung), time since transplantation and sex.
Within the cohort, we also performed analyses of risk factors for all cancer excluding SCC, of a first SCC and of additional SCC in individuals with a prior SCC. Relative risks (incidence rate ratios, IRR) were estimated using Poisson regression with the logarithm of time at risk as the offset. Risk factors considered in this comparison included age at transplantation, sex and type of transplanted organ, indication and decade of transplantation. The model also included time since transplantation fitted as a restricted cubic spline with five knots. For the analysis of risk of additional SCC, we additionally included variables for time since most recent cancer, also as a restricted cubic spine, and number of previous SCC (1, 2–3, 4–5, 6–9, ≥10 cancers). All variables were treated as categorical and were time-dependent, allowing patients to move between categories with time. To test for linear trends, we fitted models with year of transplantation as a continuous variable. Presence of nonproportional hazards were assessed in the Poisson models by including interaction terms between time, expressed as a linear term and the respective other variables.
Cumulative incidence of a first cancer excluding SCC, and of a first SCC, stratified by type of organ, was estimated using the CUMINCID macro supplied with SAS Statistical Analysis Software, version 9.2. Similar plots were also constructed to visualize the cumulative incidence of additional SCCs in relation to number of previous SCCs. Both analyses accounted for the competing risk of death.
This study was approved by the Regional Ethical review board, Karolinska Institutet, Stockholm, Sweden (Application No. 2007/1485-31/4).
We identified 11,277 patients transplanted with a solid organ (38% females and 62% males) between the years 1970 and 2008. A total of 801 patients (7.1%) were excluded from further analyses due to a cancer diagnosis more than 30 days before (417, 3.7%) or within 30 days of transplantation (384, 3.4%), resulting in a final cohort of 10,476 patients (Table 1). The most frequent cancer types among excluded patients were liver and kidney cancers, and cancer in the endocrine glands. Kidney recipients constituted the majority of all patients (n = 7,952), followed by liver (n = 1,221), heart (n = 557), lung (n = 438), multiple organ (n = 238), heart and lung (n = 17), pancreas (n = 44) and small bowel (n = 9) recipients. The most common indications for kidney transplantation were glomerulo-pathological diseases (37%) and polycystic kidney disease and other congenital conditions (19%). For liver transplantation, the most common indications were primary sclerosing cholangitis (22%) and viral hepatitis (17%). Heart transplanted patients were most often diagnosed with dilated (52%) or ischemic cardiomyopathy (31%) and lung transplanted patients with chronic obstructive pulmonary disease (33%) or alfa-1-antitrypsin deficiency (21%). The median follow-up time was 7.9 years in kidney recipients, 5.1 years in liver and 4.7 years in heart and/or lung recipients. The longest individual follow-up time was 38 years.
Cancer risks by transplanted organ and overall
Altogether 1,610 transplanted patients developed 3,406 malignancies during follow-up. Of these, 1,036 patients were diagnosed with 1,175 malignancies excluding SCC and 668 patients with 2,231 SSC. Risk of all cancer excluding SCC was increased threefold among heart and/or lung recipients and twofold among kidney and liver recipients, compared with the general population, withholding more pronounced risks for a wide range of malignancy types (Table 2). Risk of all cancer excluding SCC was similar among men and women (Table 1, Supporting Information). The estimates for risk of SCC were 198-fold increased in heart and/or lung, 121-fold in kidney and 32-fold in liver recipients (Table 2). When only considering first SCC, risk estimates were up to 69-fold increased (SIRheart/lung 69 (95% CI, 52–88), SIRkidney 52 (95% CI, 48–57), SIRliver 22 (95% CI, 14–32)), (Table 2, Supporting Information).
Cumulative incidences of first cancer excluding SCC and of first SCC according to type of transplanted organ are shown in Figure 1. At 20 years posttransplantation, 12% (95% CI, 11–13%) of the kidney recipients, 24% (95% CI, 19–30%) of the heart and/or lung and 27% (95% CI, 18–39%) of the liver transplanted patients were diagnosed with a malignancy other than SCC, whereas SCC had been contracted by 9% (95% CI, 8–10%) of the kidney recipients, 19% (95% CI, 14–25%) of the heart and/or lung and 18% (95% CI, 12–29%) of the liver recipients.
The skin cancer group included also 52 malignant melanomas diagnosed in 51 patients, SIR 2.2 (95% CI, 1.7–2.9). Only relatively few rare skin cancer types were noted: nine Merkel cell carcinomas, SIR 65 (95% CI, 30–124), seven Kaposi's sarcomas, SIR 39 (95% CI, 16-81) and 22 appendageal carcinomas, SIR 40 (95% CI, 25-61) (Table 1, Supporting Information). Lip cancer followed a pattern of increased risks by transplanted organ similar to that of SCC, with the most pronounced risk (84-fold increased) among heart and/or lung recipients, a 46-fold risk among kidney recipients, and 19-fold risk among liver recipients (Table 2). In skin-adjacent genital mucosal sites, i.e., anus, vulva/vagina and penis, cancer risks were sevenfold, 13-fold and sevenfold increased, respectively, among all transplant recipients (Table 1, Supporting Information). Risk of cervical cancer was doubled among kidney recipients while risk of cervical intraepithelial neoplasia, grade III (CIN III), was tripled among kidney and liver recipients, and 4.7-fold increased among heart and/or lung recipients (Table 2).
We observed a pronounced risk of cancer at the respective sites of transplantation, specifically a 14-fold increased risk of primary liver cancer in liver recipients, a 5.4-fold risk of lung cancer in heart and/or lung recipients, SIRheart 3.8 (95% CI, 1.5–7.9), SIRlung 8.4 (95% CI, 3.6–17) and a sixfold risk of kidney cancer in kidney recipients (Table 2). While the risk estimate of non-Hodgkin lymphoma was substantially increased among all recipients regardless of graft type, risks of other hematological malignancies were more unstable and inconsistent (Table 2). The relative risk of malignancy types occurring most frequently in the background population was only moderately elevated, e.g., colon cancer, SIR 2.2 (95% CI, 1.7–2.8), or not elevated, e.g., breast, prostate and rectal cancer, regardless of type of transplanted organ (Table 2 and Table 1, Supporting Information).
Cancer risks by follow-up time and other factors
With longer time since transplantation, the SIR of all malignancies excluding SCC remained twofold increased throughout the study period (Table 3). In contrast, the SIR of SCC increased substantially with follow-up time. Among heart and/or lung recipients, the SIR of SCC was 60-fold increased during the first five years posttransplantation, then rose to 200-fold at 5–9 years and to more than 300-fold after 10–19 years. For kidney recipients, the initial 50-fold increased risk of SCC rose to a 160-fold risk and for liver recipients, a 15-fold increased risk rose to a 50-fold risk beyond 10 years posttransplantation. Risk estimates for first SCC also increased over time: for all organs SIR 41 (95% CI, 35–48) <5 years posttransplantation, SIR 65 (95% CI, 57–75) at 5–9 years and SIR 69 (95% CI, 61–79) after 10–19 years.
In the internal comparison, we investigated determinants for all cancer excluding SCC, a first SCC and for additional SCC (Table 4). Patients transplanted at ages below 60 years were at significantly lower risk of cancer excluding SCC and of first SCC compared with individuals transplanted at 60 years or above. In contrast, risk of additional SCCs was significantly increased among younger recipients (20–49 years). The risk of a first SCC was 35% lower among women than men, while risk of additional SCCs and of cancer excluding SCC did not differ by sex. When comparing risks by calendar period, risk of a first SCC was significantly lower in the 1970s than during the most recent decade. Tests for linear trends over calendar time showed a nonsignificant decrease of 0.6% per year of cancer excluding SCC (p = 0.22) and a nonsignificant increase of 1.0% per year of first SCC (p = 0.13), while risk of additional SCCs decreased significantly by 2.1% per year (p = 0.0005). The number of previous SCC was strongly associated with risk of additional SCC. Close to 30% of the patients with a first SCC contracted at least one additional SCC within two years, whereas this was true for more than 50% of the patients with two or three former SCCs and for 80% of the patients with a minimum of six previous SCCs (Fig. 2). Additional adjustment for underlying disease diagnosis leading to transplantation did not have other than marginal effects on the risk estimates by age, sex and other factors (data not shown). There was some evidence of nonproportional hazards in the model of risk of all cancer excluding SCC, with the male excess risk and age gradient decreasing slightly over time, but no such effects for type of organ or calendar year of transplantation. The pattern was similar for risk of first cutaneous SCC, with the addition of a diminishing effect of calendar period with time. There was no evidence of nonproportional hazards for the analysis of second or later SCCs.
In this nation-wide, population-based cohort of over 10,000 organ transplant recipients followed for more than 20 years, we show distinct variations in risk of skin and non-skin cancer by type of transplanted organ and follow-up. Our four most important findings were (i) the risk of cutaneous squamous cell carcinoma (SCC) outranked all other cancer risks, with the highest estimates observed among heart and/or lung recipients followed by kidney and lowest in liver recipients; (ii) generally high relative risks were observed of rare skin malignancies and of cancer at mucosal skin-adjacent sites; (iii) the risk of SCC followed a striking pattern with a subgroup of patients accumulating large numbers of cancers at an accelerating rate; (iv) the risk of lung, kidney and liver cancer were substantially increased among lung, kidney and liver recipients, respectively.
Higher risks not only of SCC, but also of cancer overall in heart compared with kidney recipients is well recognized and has been attributed to the often considerably higher doses of immunosuppressive agents required for cardiac transplant patients.3, 13, 15, 22, 23, 34 In our study, liver recipients contracted SCC less frequently which, along the same lines, could be a consequence of generally less aggressive immunosuppression due to a presumed lower immunogenicity of liver allografts.35 In two Finnish population-based studies of kidney16 and liver transplanted patients,19 no difference in relative risk of SCC was seen by organ, perhaps reflecting insufficient power. In the present study, risk of lip cancer was highly increased and followed the same risk pattern by organ type as did SCC.
We also observed very high relative risks for a number of rare skin cancer types, e.g., Merkel cell carcinoma, Kaposi's sarcoma and appendageal carcinoma, but the absolute numbers were still small. In recent studies from the United Kingdom22 and the United States2 risk of Kaposi's sarcoma was increased 17-fold and 61-fold, respectively. Most other previous studies of skin cancer risk after transplantation have not estimated relative risks for these rare entities.13, 14, 16, 17, 23 Regarding cancer types arising in mucosal, skin-adjacent sites, we found an increased risk of cancer of the oral cavity among kidney as well as heart and/or lung recipients, where an association with HPV has been suggested.36 Risk of cervical cancer was doubled among kidney recipients and risk of its precursor CIN III was three- to four-fold increased among all organ groups. These estimates are in line with some previous studies5, 15, 18 but lower than in other reports.1, 17, 37 A thorough pre-transplant PAP smear screening of female transplant candidates in Sweden could possibly explain lower SIRs in our study. No such preventive measures are routinely instituted for other known or suspected HPV-related cancers, e.g., penile, anal, vaginal and oral cancers, for which we and others have described markedly higher risks posttransplantation.1, 2, 15, 19–22, 38
During follow-up, risk of SCC increased to over 300-fold among heart and/or lung recipients after 10–20 years and to almost 200-fold among kidney recipients at 20 years and beyond. This pattern was largely, but not only, explained by a large number of tumors and an accelerating rate of tumor occurrence seen in a comparatively small number of patients. For example, after a fifth SCC, the risk of developing yet another SCC within two years exceeded 80%. Cumulative incidences of subsequent SCC have previously mainly been described in single-center studies.39–43 Our population-based figures therefore add valuable information for tumor pace in patients with an already high tumor burden. Despite the fact that SCC are widely considered benign, in the context of immunosuppression, 50-fold elevated mortality rates have been reported compared with a nontransplanted population.9, 44 Based on our findings, a minimum of three to four clinical skin check-up visits per year should be recommended for transplanted patients with at least two to three previous SCCs. Curiously, once patients had contracted a first SCC, the risk of additional SCC did not vary noticeably with age nor with sex. Thus, we speculate that age may be a proxy for some other effect, presumably accumulation of somatic mutations due to exposure to UV light or some other carcinogens.
Although the majority of the elevated site-specific cancer risks appear to be driven mostly by immune suppression, linked with oncogenic viral or bacterial infections,12 tumor-promoting effects of the immunosuppressive drugs themselves may also contribute.45 The accelerated skin carcinogenesis is likely complex, involving both genetic and local immune microenvironment aberrations due to UV light exposure (age) and possibly HPV exposure.11, 46 Why risk of malignant melanoma was only modestly increased (2.2-fold), similar to previously published data,2, 13, 22, 47 is not easily understood, but presumably reflects that melanoma development is mainly linked to UV light through mechanisms not affected by immunosuppression and/or that viral carcinogenesis is unimportant.
The fact that the results with regard to risk of cancer excluding SCC and of SCC were unaffected by adjustment for indication of transplantation seems to support the notion that cancer risk in organ-transplanted patients is primarily driven by immune suppression and not by the underlying condition, although our classification was based on a retrospective review of discharge diagnoses which may not be entirely consistent with the actual indication, and effects could not be evaluated by specific cancer site. The strikingly elevated cancer risks at the respective site of transplantation are however clearly related to the indication of transplantation, e.g., chronic obstructive pulmonary disease and lung cancer due to smoking,48 glomerulo-pathologic disease3, 5 as well as polycystic kidney disease49 and kidney cancer and viral hepatitis or sclerosing cholangitis and primary liver cancer.7 The excess risk of thyroid cancer, which also has been demonstrated previously,50 may result from increased surveillance, although a true biologic effect cannot be excluded.
We consider the internal validity of this investigation to be high, in view of the well-defined national cohort based on population-based registers with high coverage, use of robust statistical methods and virtually no loss to follow-up. Some limitations should, however, also be emphasized. As we were unable to scrutinize over 10,000 patient records, we lacked information about possible explantation and retransplantation. Thus, minor misclassification of exposure, i.e., including follow-up time on and off immunosuppressive treatment, could have led to an underestimation of the excess risk actually entailed by these drugs, presumably mainly for kidney recipients. With regard to risk of skin cancer, basal cell carcinomas are usually included in the nonmelanoma skin cancer group, but this cancer type could not be included due to incomplete registration during the study period.
In conclusion, in this large population-based study of cancer risk in transplanted patients, we have shown elevated cancer risks at a range of sites, with the most pronounced risks for SCC. Over time, risks increased dramatically for SCC but remained stable for all other cancer types when considered together. Our results reinforce the importance of regular skin screening, especially among heart, lung and kidney recipients and following the diagnosis of a first SCC. In the future, we suggest that new strategies for post-transplant routine visits should be developed, including specific checklists for cancer screening, according to risk profiles of patient groups and of the transplanted organ.
- 25National Board of Health and Welfare CfE. In-patient diseases in Sweden 1987–2006, vol. 2012, 2007.
- 26National Board of Health and Welfare CfE. Classification of Surgical Procedures Version 1.9, vol. 2012, 2004.
- 28National Board of Health and Welfare CfE. In English: The Swedish Cancer Register, vol. 2012, 2011.
- 34Are heart transplant recipients more likely to develop skin cancer than kidney transplant recipients? Transpl Int 2000; 13: S380–1., , , et al.
- 35Liver transplant tolerance: mechanistic insights from animal models, with particular reference to the mouse. Transplant Rev 1997; 11: 151–64., , , et al.
- 46Basic scientific mechanisms of accelerated development of squamous cell carcinoma in organ transplant recipients. In: Otely CC, Stasko, T, eds. Skin Disease in Organ Transplantation. New York: Cambridge University Press, 2008. 53–59..
Additional Supporting Information may be found in the online version of this article.
|IJC_27765_sm_SuppTab1.xls||32K||Supplemental Table 1. Standardized incidence ratio of all cancer by anatomical site and/or histology in solid organ transplant recipients, presented overall and by gender.|
|IJC_27765_sm_SuppTab2.xls||32K||Supplemental Table 2. Standardized incidence ratios of first cancer and specific cancer types by anatomical site and/or histology among kidney, liver and heart and/or lung transplant recipients.|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.