Comparison of the Incidence of Malignancy in Recipients of Different Types of Organ: A UK Registry Audit

Authors

  • D. Collett,

    1. Statistics and Clinical Audit, NHS Blood and Transplant, Fox Den Road, Stoke Gifford, Bristol BS34 8RR, UK
    Search for more papers by this author
  • L. Mumford,

    1. Statistics and Clinical Audit, NHS Blood and Transplant, Fox Den Road, Stoke Gifford, Bristol BS34 8RR, UK
    Search for more papers by this author
  • N. R. Banner,

    1. Departments of Cardiology and Transplant Medicine, Harefield Hospital, Middlesex, UK
    2. Faculty of Medicine, Imperial College, London, South Kensington Campus, London, UK
    Search for more papers by this author
    • On behalf of the UK Liver and Cardiothoracic Audit Steering Groups.

  • J. Neuberger,

    1. Organ Donation and Transplantation, NHS Blood and Transplant, Fox Den Road, Stoke Gifford, Bristol, UK
    Search for more papers by this author
    • On behalf of the UK Liver and Cardiothoracic Audit Steering Groups.

  • C. Watson

    1. University Department of Surgery and the NIHR Cambridge Biomedical Research Centre, Addenbrooke's Hospital, Hills Road, Cambridge, UK
    Search for more papers by this author

Corresponding author: Dave Collett, dave.collett@nhsbt.nhs.uk

Abstract

An increased incidence of malignancy is an established complication of organ transplantation and the associated immunosuppression. In this study on cancer incidence in solid organ transplant recipients in Britain, we describe the incidence of de novo cancers in the allograft recipient, and compare these incidences following the transplantation of different organs. Data in the UK Transplant Registry held by NHS Blood and Transplant (NHSBT) were linked with data made available by the cancer registries in England, Scotland and Wales. Incidence rates in the transplanted population were then compared with the general population, using standardized incidence ratios matched for age, gender and time period. The 10-year incidence of de novo cancer in transplant recipients is twice that of the general population, with the incidence of nonmelanoma skin cancer being 13 times greater. Nonmelanoma skin cancer, cancer of the lip, posttransplant lymphoproliferative disease and anal cancer have the largest standardized incidence ratios, but the incidence of different types of malignancy differs according to the organ transplanted. Patterns in standardized incidence ratios over time since transplantation are different for different types of transplant recipient, as well as for different malignancies. These results have implications for a national screening program.

Introduction

An increased risk of developing de novo cancer is an established complication of organ transplantation and the associated immunosuppression. Understanding the increased risk and defining those cancers that are increased in these patients will not only inform the patient prior to transplantation but also allow a rational approach to screening and surveillance in the postoperative period.

Numerous studies have investigated cancer incidence in recipients of a kidney, liver or cardiothoracic organ, but these often focus on a single transplant type, a specific form of malignancy, experience at a single center, or a limited duration of follow-up. National studies, based on registry data, have been reported for kidney transplant recipients in Canada (1), Sweden (2), Japan (3), Finland (4), Australia (5) and the USA (6). Other studies use data from large multicenter registries, such as the Israel Penn International Transplant Tumor Registry (IPITTR) and the Collaborative Transplant Study Registry. These registries rely on voluntary reporting, leading to the underestimation of cancer incidence following transplantation. Other registries, such as that of the International Society of Heart and Lung Transplantation, cover a number of countries, so it is not clear what the reference population should be. This is the first reported national study of de novo malignancy in British solid organ transplant recipients that makes comparisons with incidence rates in the general population and compares these incidences following the transplantation of different organs.

Materials and Methods

Data on transplant recipients in the UK Transplant Registry, held by NHS Blood and Transplant (NHSBT), has been linked to information on cancer registrations following solid organ transplantation recorded by the eight cancer registries in England and the separate registries for Wales and Scotland, using the NHS number. This number was not available for 17% of all transplant recipients since 1980, mainly those in the 1980s.

The study cohort consists of all patients transplanted in England, Wales and Scotland, who received a first kidney, liver, heart or lung transplant between 1 January 1980 and 31 December 2007, and who had an NHS number. Those who received a combined heart and double lung transplant (n = 553) were classed as lung transplant recipients, and those who received a combined pancreas and kidney transplant (n = 764) were taken to be kidney recipients. The small number of recipients of a pancreas alone (n = 33) were excluded, together with all other multi-organ transplant recipients (n = 177).

The Cancer Registries code any new malignancy on the basis of available histology and other relevant information, such as cross-sectional imaging, according to the 10th World Health Organization International Classification of Disease (WHO ICD-10) (7,8). The date and ICD-10 code of the first diagnosis of any malignancy following transplantation was obtained from cancer registry information. In the case of patients whose first reported cancer following the date of transplantation is an in situ tumor (D00–D09), a benign tumor (D10–D36) or a tumor of uncertain or unstated behavior (D37–D48), the date and code of any subsequent malignant tumor (C00–C96) was used.

The time to first occurrence of a nonmelanoma skin cancer (NMSC) (ICD10 code: C44) was obtained for all individuals whose first registration was this cancer. However, for patients whose initial cancer was NMSC, the time to first occurrence of any subsequent cancer was used in our analysis. Where the indication for transplant was a cancer in the kidney or lung, any occurrence of tumor in the donor organ was ignored, and all occurrences of liver cancer in liver recipients were assumed not to be de novo.

To exclude cancers that may have been prevalent at the time of transplantation, all patients with cancer diagnosis within one month of transplant were excluded, except for the 620 patients diagnosed with posttransplant lymphoproliferative disease (PTLD), defined as the occurrence of Hodgkin's or non-Hodgkin's lymphoma (ICD10: C81–C85).

The cohort consists of 37 617 transplanted patients with known age and gender at transplantation. Patients were followed up from the date of first transplant to the earliest of the date of first reported de novo malignancy, or date of death within the study period.

Statistical methods

Cancer incidence in the transplant population was compared with the general UK population using the Standardized Incidence Ratio (SIR) (9). This is the ratio of the observed number of new cases of malignancy in the study cohort to the expected number in the general population.

The expected number of incident cases was obtained by calculating the person years at risk of malignancy in the transplant population in five year age groups [0–4, 5–9,…, 80–84, 85+] for each gender in each year between 1980 and 2007. Age and gender specific rates in the general population in each year, for different types of malignancy, are then applied to these numbers of person years.

Registrations of newly diagnosed cancers in England were obtained from the Office for National Statistics (10–12). The same rates were taken to apply to transplant recipients in Wales and Scotland. For some years, national incidence rates were not available for some cancer types and age groups. In these cases, estimates were obtained on the assumption of a log linear relationship over the time period for which data were available, a model that yields nonnegative estimates of incidence for rare cancers. These models were validated using an analysis of residuals (13).

Significance tests for deviations of an SIR from unity were based on the Poisson approximation to the distribution of the observed number of cases of malignancy, and exact confidence limits for the SIR calculated from the Poisson distribution (9).

The cumulative incidence of different cancers was summarized using the cumulative incidence function, the probability of cancer occurring in a given time period following transplantation, which closely approximates the cumulative incidence rate (9).

Patterns in cancer incidence over the time period following transplantation have been evaluated by further stratifying the SIRs by year since transplantation. Confidence intervals for these estimates and p-values for testing the hypothesis that the SIR is 1.0 were obtained. This analysis extended to 15 years following transplantation, and was carried out for each organ and all the different cancers.

For each organ type, the dependence of cancer incidence on recipient age group and gender, primary disease, smoking history, obesity (whether or not body mass index exceeds 30), whether or not a patient was on dialysis at the time of registration and transplant year [1980–1984, 1985–1989, 1990–1994, 1995–1999, 2000–2004, 2005–2007] was modeled. However, limitations of the data set meant that smoking history was not available for kidney recipients, data on body mass index was missing for 70% of the patients in the study, and whether or not a kidney transplant recipient was on dialysis at the time of registration was not known for patients transplanted before October, 1998. A competing risks Cox regression analysis was used to allow for the fact that the outcome variable is the time to first malignancy diagnosis; patients who develop one type of cancer are no longer at risk of a different cancer being their first. For a given type of cancer, the time from transplant to diagnosis was taken to be the outcome variable; this time was censored at the date of last follow-up or death when the cancer had not occurred (14,15).

Data on immunosuppression was obtained at 3 months following transplantation, but as these data were available for less than 50% of the patients, we were only able to compare (i) use of azathioprine (AZA), Mycophenolate Mofetil (MMF), neither or both and (ii) use of tacrolimus or ciclosporin compared to nonuse, in kidney recipients. These comparisons were based on Cox regression models adjusted for recipient age group, gender, primary disease and year of transplant.

All calculations were carried out using Statistical Analysis Software (SAS Institute, Cary). Computation of the incidence functions and the person years at risk was based on SAS macros (16,17).

Results

The characteristics of the recipient cohort are shown in Table 1. Numbers in the period 1980–1984 are lower than expected because of incomplete recording at this stage.

Table 1.  Age, gender and year of transplant for recipients of different organs
CharacteristicTransplant type
KidneyLiverHeartLung
Number(%)Number(%)Number(%)Number(%)
Age at transplant
 0–9 637 (3) 667(10) 243 (7)  27 (1)
 10–191585 (6) 381 (6) 222 (6) 187 (9)
 20–345419(22) 693(10) 361(10) 464(23)
 35–498181(33)1904(28)1040(29) 542(26)
 50–595596(22)1974(29)1442(40) 676(33)
 60–3686(15)1227(18) 301 (8) 162 (8)
Year of transplant
 1980–19841420 (6)  15 (0)  35 (1)   0 (0)
 1985–19894034(16) 144 (2) 366(10)  77 (4)
 1990–19945560(22)1147(17)1023(28) 434(21)
 1995–19995490(22)1975(29)1069(30) 656(32)
 2000–20045296(21)2296(34) 737(20) 568(28)
 2005–20073304(13)1269(19) 379(11) 323(16)
Gender
 Male15511 (62)3592 (52)2875(80)1100(53)
 Female9593(38)3254 (48) 734(20) 958(47)
Total25104 (100) 6846(100)3609(100) 2058(100) 

Of the 37 617 transplanted patients, 25 104 (67%) received a kidney (or simultaneous kidney and pancreas), 18% a liver, 10% a heart and 5% a lung (or heart/lung). The median follow up period for this cohort of patients was 16 years (interquartile range: 8–26 years).

A total of 5706 patients (15%) developed a cancer in this period. In 3276 patients, the first cancer registered following transplantation was a NMSC (C44), and of these 426 were subsequently registered as having developed a different de novo cancer. Thus 2856 patients were registered with a de novo cancer other than a nonmelanoma skin cancer.

The cumulative incidence for all de novo cancers, excluding nonmelanoma skin cancer, in the transplanted population in England, Wales and Scotland, and for the general population of England, adjusted for age group, gender and time period, are shown in Figure 1. Ten years after transplantation, recipients have an overall incidence rate of 90 per thousand patients, which is more than double the 10-year incidence rate for the population of England which was 36 in the same period. This overall relative risk is more or less constant from two years after transplantation. NMSC is much more common, and the corresponding standardized incidence was 14.4 (95% confidence interval: 13.8–15.0), meaning that the risk of this type of cancer occurring in a transplant recipient is over 14 times that of the general population.

Figure 1.

Overall cumulative incidence of any de novo cancer (excluding nonmelanoma skin cancer) in the transplanted and general populations.

For kidney transplant recipients, the numbers of patients who developed a malignancy are shown in Table 2. Also given is the expected number of occurrences in the general population and standardized incidence ratios, obtained by applying age, gender and time period specific rates in England to the person years of follow-up.

Table 2.  Standardized Incidence Ratios (SIR) and associated 95% confidence limits for specific cancers in kidney transplant recipients. Rates are given for the first incidence of nonmelanoma skin cancer, and for de novo cancers other than nonmelanoma skin cancer
Site of cancerICD10 codeObserved numberExpected numberSIRp-Value95%CI
Skin: nonmelanomaC442440147.116.6<0.001(15.9, 17.3)
All cancers excluding nonmelanoma skin cancerC00–C43, C45–C961982824.82.4<0.001(2.3, 2.5)
   LipC00  590.965.6<0.001(49.9, 84.6)
   Oral cavityC01–C06  348.04.2<0.001(2.9, 5.9)
   OesophagusC15  4022.11.80.001(1.3, 2.5)
   StomachC16  4824.52.0<0.001(1.4, 2.6)
   ColorectalC18–C20 181100.01.8<0.001(1.6, 2.1)
   AnusC21  272.710.0<0.001(6.6, 14.6)
   LiverC22  197.82.40.001(1.5, 3.8)
   PancreasC25  2819.21.50.07(1.0, 2.1)
   Lung and bronchusC34 176126.01.4<0.001(1.2, 1.6)
   Skin – malignant melanomaC43  6826.42.6<0.001(2.0, 3.3)
   Kaposi sarcomaC46  120.717.1<0.001(8.9, 30.0)
   BreastC50 117121.61.00.72(0.8, 1.2)
   CervixC53  208.92.30.002(1.4, 3.5)
   UterusC54  1615.51.00.97(0.6, 1.7)
   OvaryC56  2417.71.40.18(0.9, 2.0)
   ProstateC61 11298.61.10.20(0.9, 1.4)
   KidneyC64 15519.57.9<0.001(6.7, 9.3)
   BladderC67  7631.92.4<0.001(1.9, 3.0)
   ThyroidC73  324.67.0<0.001(4.8, 9.8)
   Hodgkin's lymphomaC81  385.17.4<0.001(5.3, 10.2)
   Non-Hodgkin's lymphomaC82–C85 35328.312.5<0.001(11.2, 13.8)
   Multiple myelomaC90  3410.33.3<0.001(2.3, 4.6)
   LeukaemiaC91–C95  1918.31.00.93(0.6, 1.6)
   Other sites  294106.22.8<0.001(2.5, 3.1)

The overall incidence rate of cancer in the transplanted population is just over twice that of the general population. The standardized incidence ratios are particularly high for nonmelanoma skin cancer, cancer of the lip and Kaposi sarcoma.

Table 3 gives the standardized incidence ratios and associated confidence limits for recipients of different types of transplant, for those cancers with the greatest incidence in the transplant population.

Table 3.  Standardized Incidence Ratios (95% confidence intervals) for malignancy in recipients of different organs
Site of cancerTransplant type
KidneyLiverHeartLung
  1. 1SIR cannot be calculated for two cancers in lung transplant recipients where the observed number of cases is 1 and the corresponding expected number is zero.

  2. 2All occurrences of liver cancer in liver recipients are assumed not to be de novo.

All cancers (excluding nonmelanoma skin cancer)2.4 (2.3, 2.5)2.2 (2.0, 2.4)2.5 (2.2, 2.7)3.6 (3.0, 4.4)
Skin: nonmelanoma16.6 (15.9, 17.3)6.6 (5.8, 7.5)18.5 (16.9, 20.3)16.1 (13.1, 19.6)
Lip65.6 (49.9, 84.6)20.0 (5.4, 51.2)60.0 (31.0, 104.8)1
Hodgkin's lymphoma7.4 (5.3, 10.2)8.9 (3.8, 17.5)11.4 (4.9, 22.5)5.0 (0.1, 27.9)
Non-Hodgkin's lymphoma12.5 (11.2, 13.8)13.3 (10.6, 16.6)19.8 (16.1, 24.1)30.0 (20.6, 42.1)
Breast1.0 (0.8, 1.2)0.8 (0.5, 1.1)0.8 (0.3, 1.7)0.3 (0.0, 1.2)
Oral cavity4.2 (2.9, 5.9)10.0 (5.9, 15.8)5.0 (2.2, 9.8)5.0 (0.6, 18.1)
Colorectal1.8 (1.6, 2.1)2.3 (1.7, 3.0)1.1 (0.7, 1.7)1.1 (0.3, 2.9)
Anus10.0 (6.6, 14.6)3.3 (0.4, 12.0)7.5 (1.6, 21.9)20.0 (2.4, 72.2)
Liver2.4 (1.5, 3.8)21.2 (0.2, 4.5)10.0 (2.1, 29.2)
Lung and bronchus1.4 (1.2, 1.6)1.6 (1.2, 2.2)2.1 (1.6, 2.8)5.9 (3.7, 8.8)
Kaposi sarcoma17.1 (8.9, 30.0) 0.0 10.0 (0.2, 55.7)1
Kidney7.9 (6.7, 9.3)1.8 (0.8, 3.6)4.4 (2.5, 7.0)2.5 (0.3, 9.0)
Multiple myeloma3.3 (2.3, 4.6)0.8 (0.1, 3.0)3.2 (1.2, 6.9)2.5 (0.1, 13.9)

The overall SIR is greatest for lung transplant recipients. The incidence of nonmelanoma skin cancer is much less in liver recipients and the SIR for cancer of the lip in these patients is less than half that for other transplant recipients. Lymphoma rates are similar, except that non-Hodgkin's lymphoma has a greater incidence in cardiothoracic transplant recipients. The incidence of oral cancer in liver recipients is also high, but 15 of the 18 observed incidences of oral cancer occurred in those whose primary disease was alcoholic liver disease.

Trends over time following transplantation

The relationship between the SIR for all cancers and the time since transplantation for recipients of a kidney, liver, heart or lung is shown in Figure 2.

Figure 2.

Trends in the SIRs for any de novo cancer (excluding nonmelanoma skin cancer) in kidney, liver heart and lung transplant recipients. SIRs are relative to the general population of England; an SIR of 1.0 corresponds to equal incidence.

This figure shows that the incidence rate, compared to the general population, quickly increases to twice that of the general population following kidney or heart transplantation. The SIR following lung transplantation increases dramatically at first and then declines over time. The pattern is similar for liver recipients, although the SIR is generally lower.

For all organs, the SIR for non-Hodgkin's lymphoma increases to a maximum at one year after transplantation and then declines, but patterns in the incidence rates of many other malignancies vary between organs. Following kidney transplantation, the SIR increases for some cancers (NMSC, anus, cervix, Hodgkin's lymphoma, kidney, oesophagus and stomach) and decreases for others (Kaposi sarcoma, leukaemia). The SIRs for liver and ovarian cancer increase in the years immediately following transplantation and then decline, while the SIRs for breast, pancreas and uterine cancer remain constant. In Figure 3, time trends in the SIRs for a number of cancers are shown for kidney transplant recipients.

Figure 3.

Trends in the SIRs for particular cancers in kidney transplant recipients.

Following liver transplantation, the SIR for several cancers (cervix, colorectal, Hodgkin's lymphoma, liver, multiple myeloma and pancreas) peaks within 2 years of transplantation, and then declines steadily. In heart transplant recipients, the SIR for bladder, lung and oral cancer, as well as Hodgkin's lymphoma, increase over time, whereas those for most other cancers remain stable or decrease. The pattern is similar following lung transplantation, except there is less evidence of an increase over time in the SIR for leukaemia and lung cancer.

For kidney transplant recipients, SIRs according to the patient's age at transplant and gender are shown in Table 4, which gives the SIRs for NMSC (C44), PTLD (C81–C85) and all other cancers.

Table 4.  Standardised Incidence Ratios (95% confidence intervals) for nonmelanoma skin cancer, PTLD and other cancers in kidney transplant recipients by age and gender
GenderAge groupType of malignancy
Skin cancerPTLDOther cancer
Male 0–1983.3 (53.9, 123.0)32.9 (20.8, 49.3)6.7 (3.8, 10.8)
20–3430.2 (25.6, 35.3)23.7 (18.5, 29.8)4.8 (3.9, 5.8)
35–4922.9 (21.1, 25.0)14.1 (11.4, 17.4)2.7 (2.4, 3.1)
50–5918.3 (16.9, 19.8)8.2 (6.3, 10.6)1.7 (1.6, 1.9)
60–  13.3 (12.1, 14.6)6.8 (4.8, 9.2)1.6 (1.4, 1.7)
Female 0–1940.0 (20.7, 69.9)40.0 (20.7, 69.9)7.7 (4.7, 11.9)
20–3420.0 (15.8, 25.0)20.0 (12.8, 29.8)3.6 (2.9, 4.4)
35–4914.6 (12.7, 16.8)9.6 (6.3, 14.1)2.1 (1.8, 2.4)
50–5912.2 (10.6, 13.9)7.3 (4.7, 10.8)1.7 (1.4, 1.9)
60–  9.2 (7.7, 10.9)7.3 (4.4, 11.4)1.5 (1.2, 1.8)

The general pattern is that the SIRs for each cancer decrease with increasing age, and many of the rates are higher for males than females. The incidence of cancers other than NMSC and PTLD in recipients aged over 50 is not much greater than that of the general population.

Factors affecting time to cancer diagnosis

The impact of a number of factors on the time from transplantation to cancer diagnosis is shown in Table 5 for kidney transplant recipients. The table gives estimated hazard ratios for the occurrence of skin, kidney, lung, and colorectal cancer, and PTLD.

Table 5.  Relative hazard (95% confidence interval) of the occurrence of malignancy across age groups, gender and primary disease in kidney transplant recipients. Baseline categories are the 20–34 age group, male patient, those with a primary disease in the ‘other’ category and a transplant year of 1980–1984.
FactorType of malignancy
SkinPTLDKidneyLungColorectal
Recipient age group
 0–90.0 (0.0, 0.2)0.7 (0.3, 1.4)0.3 (0.0, 2.0)0.00.0
 10–190.5 (0.3, 0.7)1.0 (0.6, 1.5)0.7 (0.3, 1.7)0.00.0
 20–341.01.01.01.01.0
 35–493.4 (2.9, 3.9)1.1 (0.8, 1.4)1.9 (1.2, 2.9)19.9 (4.8, 82.2)3.6 (1.9, 6.8)
 50–597.8 (6.7, 9.0)1.4 (1.0, 1.8)2.2 (1.4, 3.7)  55.8 (13.6, 228.4)11.6 (6.2, 21.7)
 60– 13.7 (11.7, 16.1)2.1 (1.5, 3.0)3.1 (1.8, 5.6)  104.8 (25.4, 432.6)   23.3 (12.2, 44.5) 
Recipient gender
 Male2.0 (1.8, 2.2)1.7 (1.4, 2.1)1.7 (1.2, 2.5)1.4 (1.0, 1.8)1.4 (1.0, 1.9)
 Female1.01.01.01.01.0
Primary disease
 Renal cancer0.4 (0.1, 2.6) 1.9 (0.3, 13.9)0.00.0 6.7 (0.9, 49.0)
 GN1.3 (1.1, 1.5)1.1 (0.8, 1.6)1.1 (0.7, 2.0)1.4 (0.8, 2.3)1.1 (0.6, 1.9)
 Diabetes0.7 (0.6, 0.9)0.9 (0.5, 1.5)0.6 (0.2, 1.6)0.7 (0.3, 1.7)1.4 (0.7, 2.9)
 Other1.01.01.01.01.0
Transplant year
 1980–19841.01.01.01.01.0
 1985–19891.0 (0.9, 1.2)2.9 (1.7, 4.9)2.5 (1.2, 5.2)2.2 (0.9, 5.4)2.0 (1.0, 4.1)
 1990–19941.3 (1.1, 1.6)4.3 (2.7, 7.4)2.8 (1.3, 6.0)2.3 (0.9, 5.9)1.9 (0.9, 4.1)
 1995–19991.3 (1.0, 1.5)4.8 (2.7, 8.4)3.5 (1.6, 7.9)2.5 (1.0, 6.6)1.5 (0.7, 3.3)
 2000–20041.3 (1.0, 1.5)4.7 (2.5, 8.8) 3.9 (1.5, 10.1)2.3 (0.8, 6.3)2.6 (1.1, 6.3)
 2005–20071.0 (0.6, 1.6) 7.5 (3.4, 16.8) 3.9 (0.8, 20.4) 2.1 (0.4, 11.4) 2.6 (0.5, 13.4)

For each of the types of cancer shown, the hazard of occurrence is greater in males than females, and increases with increasing age. The risks of developing PTLD and kidney cancer have increased over the study period.

Body mass index, and whether or not a patient was on dialysis at time of listing for transplantation, where available, did not significantly affect the hazard of cancer after adjusting for the factors in Table 5 (p = 0.75 and 0.54, respectively).

For recipients of a liver, recipient age group and primary disease affected the risk of skin cancer (p < 0.001 and 0.01, respectively) and colorectal cancer (p = 0.06 and 0.002, respectively). The risk of PTLD only depended on recipient gender (p = 0.02), and the risk of lung cancer depended only on recipient age group (p = 0.02). For recipients of cardiothoracic organs, we only identified a significant dependence of the risk of PTLD on recipient age group for heart recipients (p = 0.004).

Impact of initial immunosuppression

From the available data, the incidence of NMSC in kidney transplant recipients was unaffected by whether azathioprine or MMF was used at three months, nor was it affected by the use of tacrolimus or ciclosporin. For other types of cancer, there was insufficient data for reliable estimates.

Discussion

This first detailed registry study of cancer occurrence in British transplant recipients has shown evidence of different patterns of cancer occurrence in recipients of different organ types. The overall incidence of cancer (excluding NMSC), in recipients of a heart, kidney or liver, is over twice that of the general population (SIRs: 2.5, 2.4, 2.2, respectively), and over three times higher in recipients of lung or combined heart and lung transplants (SIR 3.6). NMSC is the most common malignancy, with an SIR of over 16 for kidney and cardiothoracic transplant recipients, which is three times higher than that for liver recipients. This increased incidence may be an underestimate since reporting of skin cancers may not be complete because some lesions will be removed or ablated without histological confirmation and some of the cancer registries did not collect data on basal cell carcinomas. Similarly, incident skin cancers in the general population may also be underreported. Liver recipients also appear to be relatively protected against cancer of the lip and anal canal, diseases which, like NMSC, have an underlying viral association and the lower rise in incidence for which may reflect lower doses of maintenance immunosuppression, in particular avoidance of azathioprine and the absence of biological induction agents in UK liver transplant immunosuppressive protocols.

The incidence rates of many cancers that we are reporting for the UK are similar to those reported by Canada, Australia and Sweden, but there are also some important differences. We have found higher rates of non-Hodgkin's lymphoma compared to Sweden, Finland and Canada, and the high rate of kidney cancer seen in Canada and Australia, but not to such an extent in Sweden, is similar to the UK. The incidence rates for a number of cancers reported in the US study (6), are considerably higher than those found in other countries. In particular, the high rates of hepatobiliary, pancreatic, prostate, kidney, uterine and cervical cancer have not been seen in the UK.

Some diseases for which transplantation is undertaken may be associated with malignancy elsewhere, or may be a marker of exposure to potential carcinogens. Cigarette smoking, for example, may contribute to the requirement for lung transplantation (emphysema) or heart transplantation (ischaemic cardiomyopathy), while also being associated with an increase in risk of many cancer types. Primary sclerosing cholangitis may require liver transplantation, but is also associated with colitis, which might account for the increase in incidence of cancer of the colon and rectum; likewise alcohol (and the exposure to cigarette smoking which accompanies it) may account for the higher risk of oral cancer in liver recipients (18). Analgesic nephropathy or aristolochia use may necessitate renal transplantation, while being associated with transitional cell carcinoma of the native urothelium (19,20).

As expected the study shows a very high incidence of posttransplant lymphoma and Kaposi's sarcoma, much in excess of the incidence of all other cancers with the exception of NMSC. The incidence of posttransplant lymphoma in kidney recipients has increased ninefold since 1980–1984, an observation which probably reflects the more aggressive immunosuppressive protocols available today, as well as the development of antimicrobial drugs which has enabled patients to tolerate more aggressive immunosuppression without succumbing to opportunistic infection. Unfortunately we do not have sufficient data on immunosuppression to enable further analysis. The other cancer whose incidence is clearly increasing in renal transplant recipients over the study period is renal cell carcinoma. The reason for this is not clear, but may in part reflect the increasing age of the recipient in whom renal malignancy is more common and in part the occurrence of cancer in acquired cystic disease of the native kidneys (21,22). Finally, the registries do not distinguish between cancers arising in the native kidneys or in the transplant, and it is possible that the increase in incidence also reflects the aging donor population who may have an unrecognized renal cancer at the time of transplantation which becomes manifest later in the posttransplant course.

One cancer whose incidence is not increased in our study, and indeed which may be less common, is breast cancer. The reason for this is not clear, but the finding is consistent with data from an earlier report from the Collaborative Transplant Study registry (recipients between 1983 and 1994 from Europe and North America) (23) but not the most recent US analysis (recipients 1995–2001) in which the incidence of breast cancer was twice that of the normal population, a similar rise to that noted with other cancers (6). Another tumor whose incidence might be expected to be higher is cervical carcinoma, since it is known to be a consequence of papilloma virus infection (24). However, the active cervical screening program in the UK probably explains this observation, with the cancer being detected and treated before it becomes invasive.

A weakness of the study is the lack of complete data on immunosuppression. Heart transplant recipients have generally received biological agents such as antithymocyte globulin for induction with higher levels of calcineurin inhibitor (CNI) based triple therapy than recipients of other organs. In contrast, liver recipients receive a lower immunosuppressive burden using protocols, which involve tapering to CNI monotherapy (usually tacrolimus) in all but autoimmune liver disease.

It is likely that the cancer incidence rates in this study are underestimates, since the study has censored a recipient at the first occurrence of any cancer (except NMSC), so if a recipient developed a second cancer this would not be recorded. This effect will, in part, balance the increase in cancer detection that one would expect in a population undergoing regular follow up by clinicians who are now aware that cancer occurrence is increased. Moreover, it appears that the increased risk of cancer in transplant recipients is a reflection of being immunosuppressed rather than being attributable to any particular drugs, since patients immunosuppressed as a complication of human immunodeficiency virus infection have a similar pattern of increased cancer incidence (25).

Studies such as ours identify patient groups and cancer types, which should form part of a cancer surveillance program, and provide data for counselling pretransplant patients on the risk of malignancy. Following transplantation, our study confirms that all transplant recipients should be screened regularly for NMSC, including lip and anal cancer, and that this should be combined with lifestyle changes to reduce sun exposure (26). However, there is no need for any additional breast or cervical cancer screening. Patients undergoing liver transplant for alcohol should have regular surveillance of the aerodigestive tract, and those with a diagnosis of primary sclerosing cholangitis who have their colon and rectum in situ should undergo regular colonoscopy. The presence of recurrent graft cirrhosis should also be a cue to regular ultrasound surveillance for hepatocellular carcinoma. Similarly a case can be made for subjecting renal transplant recipients to regular ultrasound scans of their transplant and native kidneys for evidence of tumor, particular those transplanted for analgesic nephropathy or aristolochia-related interstitial nephritis in whom a cystoscopic examination of the urothelium is also indicated. What is not so clear is how to manage the increased risk of other cancers, which, in spite of having a higher incidence than the normal population, are uncommon (27). Also uncertain is whether the benefits of early cancer detection in this immunosuppressed population translate into superior cure rates.

Acknowledgments

We thank the Office for National Statistics and also the cancer registries in England, Wales and Scotland for providing data on cancer occurrence in transplant recipients. We are grateful to the editor and reviewers for comments that have led to an improved version of this paper.

Ancillary