Evidence is sparse on the relative mortality risk posed by de novo cancers in liver and cardiothoracic transplant recipients. A retrospective cohort study was conducted in Australia using population-based liver (n = 1926) and cardiothoracic (n = 2718) registries (1984–2006). Standardized mortality ratios (SMRs) were computed by cancer type, transplanted organ, recipient age and sex. During a median 5-year follow-up, de novo cancer-related mortality risk in liver and cardiothoracic recipients was significantly elevated compared to the matched general population (n = 171; SMR = 2.83; 95% confidence interval [95%CI], 2.43–3.27). Excess risk was observed regardless of transplanted organ, recipient age group or sex. Non-Hodgkin lymphoma was the most common cancer-related death (n = 38; SMR = 16.6; 95%CI, 11.87–22.8). The highest relative risk was for nonmelanocytic skin cancer (n = 23; SMR = 49.6, 95%CI, 31.5–74.5), predominantly in males and in recipients of heart and lung transplants. Risk of death from de novo cancer was high in pediatric recipients (n = 5; SMR = 41.3; 95%CI, 13.4–96.5), four of the five deaths were non-Hodgkin lymphoma. De novo cancer was a leading cause of late death, particularly in heart and liver transplantation. These findings support tailored cancer prevention strategies, surveillance to promote early detection, and guidelines for managing immunosuppression once cancer occurs.
Australian and New Zealand Liver Transplant Registry
Australian and New Zealand Cardiothoracic Organ Transplant Registry
Australian Cancer Database; BCC, basal cell carcinoma; EBV, Epstein Barr virus
human papilloma virus; IQR, interquartile range; NDI, National Death Index; NHL, non-Hodgkin lymphoma; SCC, squamous cell carcinoma; SMR, standardized mortality ratio
Immunosuppressive therapy after solid organ transplantation is widely reported to be associated with a two- to threefold increased risk of cancer [1-3], but comparatively few studies have quantified the risk of death from cancer in this population [4-8]. Furthermore, some of these studies did not exclude deaths from recurrent cancers [6-8], and were limited by a large proportion of deaths of unknown cause [6, 7]. Documenting the extent and pattern of risk for de novo cancer-related death in transplant recipients will guide the identification of high-risk patient subgroups, potentially leading to further improvements in long-term survival.
We aimed to quantify the overall and site-specific risk of death from de novo cancer in Australian liver and cardiothoracic transplant recipients over a 23-year period using a population-based record linkage study design.
Materials and Methods
Our study included all Australian recipients of cardiothoracic and liver transplants between 1984 and 2006. Transplant recipients were registered on the Australian and New Zealand Liver Transplant Registry (ANZLTR) or the Australian and New Zealand Cardiothoracic Organ Transplant Registry (ANZCOTR). These national population-based registries recorded all liver and cardiothoracic transplantations since 1985 and 1984, respectively, and systematically record demographic and clinical data about each recipient.
Recipients with single (n = 4482) and second or and higher-order transplants (n = 162) were included in the analyses. Non-Australian recipients and patients who died within 30 days of transplantation (n = 287) were excluded from the study population. Patients with a history of cancer before transplantation (n = 367), including those whose indication for transplantation was a hepatobiliary tumor (n = 93), were not excluded because these patients were at risk of death from a different de novo cancer. However, these patients did not contribute person-years at risk for death from these cancers. A sensitivity analysis was performed to assess the impact of excluding patients with a history of cancer (before 30-day posttransplantation). Patients who received a combined liver and kidney transplant (n = 23), and those who received a combined liver, heart and lung transplant (n = 3), were classified as liver transplant recipients; those who received a combined heart and lung transplant (n = 137) were classified as lung transplant recipients.
Death and cancer ascertainment
Deaths were identified by record linkage with the National Death Index, a registry of all deaths in Australia since 1980, or from the transplant registry. The underlying cause of death was available for deaths ascertained from the National Death Index, coded to the 9th International Classification of Diseases, 9th revision (ICD-9), between 1984 and 1997, and ICD-10 from 1998 onwards. Deaths identified from the transplant registries alone had an unknown underlying cause and thus could not be included in cause-specific analyses. Cancers were ascertained by record linkage with the Australian Cancer Database (ACD), a register of incident primary invasive neoplasms in Australian residents. The date of diagnosis, topography and morphology was ascertained for each cancer diagnosed between 1984 and 2006. Record linkage was performed by the Australian Institute of Health and Welfare utilizing an established probabilistic record linkage algorithm .
Australian population mortality rates for any cancer and site-specific cancers were obtained from the Australian Institute of Health and Welfare by five-year age group, sex, calendar year and State or Territory, for 1984–2006.
Ethical approval was obtained and the requirement for informed participant consent was waived because the researchers received only de-identified data.
Classification of underlying cause of death
The underlying cause of death recorded on Australian death certificates is defined as “the disease or injury, which initiated the train of morbid events leading directly to death, or the circumstances of the accident or violence which produced the fatal injury”. For each death the cause was classified according to homogeneous groups defined by the Australian Bureau of Statistics and in accordance with WHO-ICD rules . Deaths were further classified as cancer-related, noncancer related, or unknown cause. Cancer-related deaths were considered recurrent if the cancer was diagnosed before transplantation or during the 30-day posttransplantation period. Cancers diagnosed during this period were almost certainly prevalent at transplantation and some were not registered until they were histopathologically diagnosed in the explanted organ. If the cancer attributed to the death was diagnosed more than 30 days after transplantation the cause of death was considered a de novo cancer. Some patients (n = 48) died from cancer but did not have a linked registered cancer, or they were registered with a cancer that was not attributed to their death. In these cases, if the cause of death was a liver cancer and there was a history of gallbladder cancer before transplantation or within 30 days of transplantation, the death was classified as a recurrent cancer (n = 2). All other cancer-related deaths, including those due to nonmelanocytic skin cancer (hereafter called skin cancer; C44, n = 14), which is not recorded by the Australian cancer registries, were classified as de novo cancer deaths.
All deaths were included in the calculation of overall mortality rates, whereas deaths due to an unknown cause (n = 69) were excluded from the cause-specific mortality rates. Person-years follow-up accrued from 30 days posttransplantation until the date of death, age 80 years, or December 31, 2006, whichever occurred first. Crude and age- and sex-standardized overall and cause-specific mortality rates (ASMR), standardized to the 1996 Australian population, and 95% confidence intervals (CIs; based on the normal approximation to the binomial distribution) were calculated using annual Australian population estimates from the Australian Bureau of Statistics. Cause-specific mortality rates were computed according to transplanted organ for any cancer, cardiovascular disease, respiratory disease, endocrine disease, nutritional and metabolic diseases, digestive disease and infectious disease.
Unadjusted overall survival probabilities by transplanted organ were estimated by the Kaplan–Meier method and compared using the log-rank test. Survival probabilities were computed from 30-days after transplantation to the date of death or December 31, 2006. The cumulative incidence of de novo cancer deaths was calculated, treating other deaths as a competing risk.
De novo cancer mortality risk relative to general population
De novo cancer mortality rates in transplant recipients were compared with rates in the Australian general population using the standardized mortality ratio (SMR), defined as the ratio of the observed and the expected numbers of cancer-related deaths. The likelihood ratio method was used to calculate 95%CIs for deaths with more than 10 expected cases, whereas exact CIs were used for deaths with less than 10 expected cases . The expected numbers of cancer deaths were calculated by multiplying cohort person-years at risk by the corresponding five-year age-, sex-, state- and calendar year-specific cancer mortality rates for the Australian population. SMRs were computed for the entire cohort and by transplanted organ, sex and age at transplantation (tertiles). SMRs were not compared statistically because of the heterogeneity in subgroup age and sex distributions .
Analyses were performed using SAS® software v9.2 (SAS Institute Inc., Cary, NC, USA). Person-years and SMRs were calculated using the %stratify macro written in SAS® . The Kaplan–Meier and cumulative incidence curves were generated using STATA, Version 11.2 (StataCorp, College Station, TX, USA).
The eligible cohort comprised 4644 transplant recipients; 1926 (41%) liver, 1518 (33%) heart and 1200 (26%) lung. The median duration of follow-up was 5.2 years (interquartile range, IQR 2.0–9.9), and 1558 deaths were observed over a total of 29 713 person-years. The median age at death was 53 years (IQR 41–60). The highest number of deaths from any cause was observed in lung transplant recipients (n = 590), followed by heart (n = 564) and liver (n = 404) recipients. There were 77 deaths in 415 pediatric liver and heart transplant recipients; the median age at death was 14 years (IQR 4–18), and most occurred either during the first year posttransplant (n = 28, 36.4%) or more than 5 years posttransplant (n = 28, 36.4%).
Underlying cause of death
There were a total of 1265 noncancer related deaths, led by cardiovascular disease (n = 410, 32.4%), respiratory disease (n = 235, 18.6%), and endocrine, nutritional and metabolic disease (n = 168, 13.3%). The leading causes of death corresponded to the underlying cause of end-stage organ disease, that is, digestive disease for liver transplantation, cardiovascular disease for heart transplantation, and respiratory disease for lung transplantation.
A total of 224 transplant recipients died of cancer; 171 (11%) deaths were de novo cancer deaths and 53 (3.4%) were recurrent cancer deaths. The median time between the first registered cancer and de novo cancer-related death was 7.1 months (IQR 1.2–24). Non-Hodgkin lymphoma (NHL; n = 38, 22.2%) was the most common cause of de novo cancer death, followed by cancer of unknown primary site (n = 25, 14.6%). For those who survived more than 5 years after transplantation (n = 2395, 51.6%), de novo cancer was the 2nd leading cause of death in heart (n = 72) transplant recipients, the 3rd leading cause of death in liver (n = 26) and the 4th leading cause of death in lung (n = 16) transplant recipients (Table S2). Five de novo cancer deaths occurred in pediatric transplant recipients, and the median time between transplantation and death was 6.6 years (IQR 6.4–7.6). For the 248 pediatric patients who survived more than five years after transplantation, de novo cancer was the leading (liver recipients) or the 2nd leading cause of death (heart recipients).
Of the 53 recurrent cancers (23% of all cancer-related deaths), 41 were recurrent liver cancers in liver transplant recipients. The median time between the first liver transplantation and recurrent liver cancer related death was 1.8 years (IQR 0.8–3.7).
The overall probability of survival for liver, heart and lung transplant recipients is shown in Figure 1(A). Compared to all other causes of death combined, the cumulative incidence of deaths due to de novo cancer increased steadily over time (Figures 1B–D). More than 5 years after transplantation, the rate of increase in cancer-related deaths seemed steeper for heart compared to liver and lung transplant recipients.
Crude and age–sex standardized mortality rates
Overall, the crude mortality rate was 5243 per 100 000 and the age–sex standardized mortality rate (ASMR) was 4948 (95%CI, 4052–5843) per 100 000. The crude mortality rate for de novo cancer related death was 575 per 100 000 and the ASMR was 359 (95%CI, 283–435) per 100 000. Liver transplant recipients had the lowest overall mortality rate (ASMR 2647; 95%CI, 2291–3002) and lowest rate of de novo cancer mortality (ASMR 260; 95%CI, 149–370). The crude mortality rate for recurrent liver cancer in liver transplant recipients was 322 per 100 000 and the ASMR was 268 (95%CI, 162–375) per 100 000.
The ASMR for de novo cancer deaths was 121 (95%CI, 38.6–204) per 100 000 up to 2 years after transplantation, 258 (95%CI, 143–373) between 2 and 5 years, and 554 (95%CI, 402–707) more than 5 years after transplantation. The ASMR for NHL was 77.5 (95%CI, 9.05–146) per 100 000 during the first 2 years, 53.3 (95%CI, 2.94–104) between 2 and 5 years, and 172 (95%CI, 70.9–273) beyond 5 years posttransplantation.
Mortality risk compared to the general population: all transplant recipients
Transplant recipients were at 10-fold higher risk of death from any cause compared to the general population (SMR = 10.8; 95%CI, 10.3–11.4). The 171 de novo cancer deaths corresponded to a 2.8-fold higher risk compared to the general population. Figure 2 shows the risk of death by cancer site for sites with at least three deaths. The risk of death was significantly elevated for skin cancer, NHL, melanoma, cancer of unknown primary site, liver cancer, connective and soft tissue cancer and lung cancer (Figure 2). There was no excess risk of death due to the most common epithelial cancers in the general population, specifically colon (n = 6; SMR = 1.19; 95%CI, 0.44–2.59), prostate (n = 3; SMR = 1.00; 95%CI, 0.21–2.94), and breast cancer (n = 1; SMR = 0.33; 95%CI, 0.01–1.82). The risk estimates did not change when those with a history of cancer before transplantation were excluded (data not shown).
Mortality risk compared to the general population: by age and sex
Of the 77 deaths in pediatric transplant recipients, five were attributed to de novo cancer (SMR = 41.3; 95%CI, 13.4–96.5). The median age at de novo cancer death was 17 (IQR 8–20) years. Of the 10 pediatric recipients diagnosed with NHL after transplantation, 7 died and in four cases (3 heart and 1 liver transplant recipient) the deaths were attributed to NHL.
The risk of death from any de novo cancer was increased at least twofold regardless of age at transplantation (Table 1). A significant excess risk was also observed for skin cancer, NHL and melanoma for the three age groups examined. The risk of death from cancer of unknown primary site was only increased in those more than 40 years of age at transplantation.
Table 1. Site-specific cancer mortality risk relative to the general population by age at transplantation
Age at transplantation (years)
Obs = observed number of cancer deaths; SMR = standardized mortality ratio.
An excess risk of death from any de novo cancer was observed for both males and females (Table 2). Both sexes also exhibited an increased risk of death due to NHL and cancer of unknown primary site. All skin cancer deaths were observed in males, and a significantly elevated risk of death due to melanoma and lung cancer was also only observed in males.
Table 2. Site-specific cancer mortality risk relative to the general population by sex
Obs = observed number of cancer deaths; SMR = standardized mortality ratio.
Mortality risk compared to the general population: by transplanted organ
The risk of death due to any de novo cancer was twofold for liver (SMR = 1.96; 95%CI, 1.42–2.61), threefold for heart (SMR = 3.05; 95%CI, 2.49–3.7) and more than fourfold for lung transplant recipients (SMR = 4.41; 95%CI, 3.02–6.23; Figure 3). The risk of death due to NHL and cancer of unknown primary site was significantly increased for all three transplanted organs. An excess risk of mortality due to skin cancer was observed for both heart (SMR = 66.1; 95%CI, 39.2–104) and lung (SMR = 86.4; 95%CI, 23.6–221) but not liver (SMR = 6.91; 95%CI, 0.17–38.5) recipients. Risk of melanoma-related death was significantly increased after liver (SMR = 5.26; 95%CI, 1.43–13.5) and heart (SMR = 8.81; 95%CI, 4.22–16.2) but not lung (SMR = 3.67; 95%CI, 0.09–20.4) transplantation. An increased risk of death due to lung cancer was only observed in lung transplant recipients (SMR = 3.98; 95%CI, 1.46–8.66). There was no excess risk of de novo liver cancer death for liver, heart or lung transplant recipients. Lung transplant recipients exhibited a significantly increased risk of death from de novo cancer at all time periods posttransplantation (Table 3). A significantly increased risk of death from de novo cancer was only observed beyond 2 years after transplantation for liver and heart transplant recipients.
Table 3. Risk of mortality due to de novo cancer and NHL by time since transplant and transplanted organ
In this national population-based cohort study we found a 2.8-fold excess risk of death from de novo cancer in liver and cardiothoracic transplant recipients compared to the matched general population. This risk estimate is similar in magnitude to the excess risk of incident cancer in this cohort (SIR 2.62), a finding that is predominantly attributed to immunosuppression . The excess risk of death from de novo cancer was observed regardless of recipient sex and age at transplantation, and was greatest for lung transplant recipients. NHL was the most common cancer-related death. De novo cancer was a leading cause of late death, particularly in heart and liver transplantation, and in pediatric recipients. The excess mortality from cancer in this population reinforces the need for tailored cancer prevention strategies, evidence-based surveillance to promote early detection and guidelines for managing immunosuppression and cancer treatment after diagnosis.
There are relatively few prior estimates of cancer-related mortality after solid organ transplantation. Our estimate of a 1.96-fold risk of death from de novo cancer after liver transplantation is similar to that observed in a single-center Spanish study (1990–2001) that also excluded recurrent cancers (2.93; 95%CI, 1.56–5.02; Ref. ). A 2.3-fold risk of death from any cancer (de novo and recurrent) was reported in a population-based study of kidney transplant recipients in Hong Kong (1972–2011; Ref. ). On the other hand, no excess risk of cancer death was observed in a population-based study of 164 078 kidney transplant recipients registered on the United States Renal Data System (1990–2004; Ref. ). However, this finding is likely to have underestimated cancer deaths because cause of death was unknown for 41% of recipients and all of these deaths were classified as noncancer related.
The site-specific pattern of cancer-related death by transplanted organ very closely mirrored that observed for incident cancers in this cohort . Specifically, risk of death from NHL and cancer of unknown primary site was elevated for all three transplanted organs; risk of death from skin cancer was increased in heart and lung but not liver recipients, risk of death from melanoma was increased in liver and heart but not in lung transplant recipients, and risk of death from lung cancer was confined to heart and lung transplant recipients. Furthermore, there was no excess risk of death due to prostate or breast cancer in any transplanted group. In contrast to the cancer incidence profile, we observed no excess risk of death attributed to colon cancer (n = 6) or to lip/oral cancer (n = 3).
There are no prior estimates of the risk of death from NHL in transplant recipients relative to the general population. However, in keeping with previous studies [15, 16], NHL was the most common cause of cancer-related death in this cohort. This may reflect the high incidence of this neoplasm in this patient group rather than a poor treatment response, as a single study found no difference in NHL survival for transplant recipients and the Surveillance, Epidemiology and End Results (SEER) general population (1964–2007; Ref. ). NHL in transplant recipients is managed by reducing immunosuppression, chemotherapy with or without rituximab, antiviral therapy for Epstein-Barr viral (EBV)-positive patients, surgery and radiation [18, 19].
Skin cancer carried the highest relative risk of cancer-related death in our cohort (SMR = 50). This estimate is very similar to the risk of death from cutaneous squamous cell carcinoma (SCC) in Swedish solid organ transplant recipients (1970–1997; SMR = 52.2; 95%CI, 21.0–107.6), predominantly kidney . Although the skin cancer histology was not recorded in our data, a previous Australian study of cardiothoracic transplant recipients found that 95% of nonmelanocytic skin cancers were SCC . We did not find an excess risk of mortality from skin cancer in liver transplant recipients, which is consistent with prior evidence that most deaths in this patient group occur from noncutaneous neoplasms . This finding, like the different patterns in cancer incidence by transplanted organ, is likely to be largely related to the less intensive immunosuppressive therapy received by liver compared to cardiothoracic transplant recipients [3, 21, 22].
For pediatric transplant recipients, most deaths of any cause occurred during the first year and more than 5 years after transplantation. De novo cancer was the leading cause of late death, the risk relative to the general population was around 40-fold and most cancer-related deaths were attributed to NHL. In the only prior study to assess mortality in pediatric transplant recipients with NHL, heart transplant recipients had significantly lower overall survival compared to kidney or liver transplant recipients . Given the relatively high mortality of pediatric recipients with NHL, serial monitoring of EBV load has been recommended as a screening strategy for high-risk patients, and may improve outcomes [23-26].
The increased incidence and mortality from de novo cancer in solid organ transplant recipients justifies patient education about the need for sun protection and monthly self skin examinations, and early reporting of new signs and symptoms. Our findings indicate that males are at greatest risk of dying from nonmelanoma skin cancer and melanoma. Clinically, cancer risk may be reduced by minimizing immunosuppression, viral prophylaxis, EBV viral load monitoring, and targeted cancer screening in high-risk patients [18, 19, 27]. Significantly greater non-skin cancer–related survival and overall survival was reported for a period of intensive cancer surveillance (2002–2007) compared to historical surveillance (1982–2001) in Austrian liver transplant recipients . However, there was no adjustment for either cancer treatment or improvements in cancer outcomes over time. The intensive surveillance was an annual chest and abdominal CT scan, urological evaluation (including prostate specific antigen [PSA] test), gynaecological evaluation (including Papanicolau smear and mammography), dermatological screening and 3-yearly colonoscopy except in patients with a history of adenoma or inflammatory bowel disease (1-yearly). Nevertheless, as incidence rates of prostate and breast cancer are not increased in organ transplant recipients, there is no evidence to suggest the need for screening practices for these malignancies that would differ from the general population.
The strengths of this study include the population-based design and the relatively large size, allowing stratification by organ and cancer type. A limitation is that some of the cancer-related deaths did not have a matching linked cancer registry record; however, most of these were skin cancers, the majority of which are not notifiable neoplasms in Australia. Also, there is the potential for differential misclassification of the underlying cause of death in individuals with multiple morbidities (such as those with end-stage organ disease and cancer), and this may have led to an under-estimation of the true risk of cancer-related death in this patient group . Furthermore, 4% of deaths were due to an unknown cause, and these cases were not included in the cause-specific analyses. Despite these limitations, the rank order of cause-specific mortality was in accordance with previous reports [15, 30, 31].
Our study quantifies the risk of de novo cancer-related mortality in liver and cardiothoracic transplant recipients, overall and in relation to organ type, time since transplantation and recipient age and sex. This population-based data provides additional momentum to calls for a review of guidelines for cancer prevention and screening for solid organ transplant recipients, and the generation of robust evidence in these areas .
This study was funded by the National Health and Medical Research Council (ID510254). Andrew Grulich is supported by an NHMRC principal research fellowship (ID568819). Claire Vajdic is supported by a National Health and Medical Research Council Career Development Fellowship (ID1023159) and a Cancer Institute New South Wales Career Development Fellowship (ID10/CDF/2–42). Renhua Na is supported by a Translational Cancer Research Network (TCRN) PhD Scholarship Top-up Award. The TCRN is a translational cancer research centre program funded by the Cancer Institute NSW.
We thank the Australian and New Zealand Liver Transplant Registry, the Australian and New Zealand Cardiothoracic Organ Transplant Registry, and also the Australian Institute of Health and Welfare for conducting the data linkage.
We thank: Phyllis Larkins (Royal Prince Alfred Hospital, Sydney); Geraldine Lipka (Princess Alexandra Hospital, Brisbane); Cassandra Kastaneas (St Vincent's Hospital, Sydney); Vicki Jermyn and Brooke Andersen (The Children's Hospital, Westmead); Jo Maddicks-Law, Nicole Ostenfeld, Sara Gray and Muhtashimuddin Ahmed (Prince Charles Hospital, Brisbane); Kerrie Beale (Royal Childrens Hospital, Brisbane); Libby John, Nicole Williams (Flinders Medical Centre, Adelaide); Kathryn Marshall, Ailsa Cowie, Connie Kambanaros, Jasmin Board, and Colleen Farrell (Alfred Hospital, Melbourne); Lyn Crellin, Kathe Beyerle, Kate Schurmann, Anne Shipp, Janette McEwan, Danielle Kamolins, Angie Wood and Hollie Gilmore (Royal Childrens Hospital, Melbourne); Julie Pavlovic and Betheia Lele (Austin Hospital, Melbourne); Sharon Lawrence, Clare Wood and Sharlene Beinke (Royal Perth Hospital); Barb Chester, Judith Bull, Joanne Plummer, Nikki Copland and Megan O-Dea (Sir Charles Gairdner Hospital, Perth).
The following lead investigators participated in this study: George Alex, Glenda Balderson, Peter Bergin, John Chen, Weng Chin, Lawrence Dembo, Pamela Dilworth, Looi Ee, Allan Glanville, Winita Hardikar, Peter Hopkins, George Javorsky, Gary P Jeffrey, Robert Jones, Bronwyn Levvey, Steven Lynch, Michael Musk, Ross Pettersson, Greg Snell, Michael Stormon, and Robert Weintraub.
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.