The aim of our study was to examine whether an extensive surveillance protocol will promote early diagnosis and improved survival in patients with de novo cancer following liver transplantation (LT). Of 779 consecutive LT recipients, 96 (12.3%) developed 105 malignancies. The cumulative risk for the development of de novo cancer was 10%, 24%, 32% and 42% at 5, 10, 15 and 20 years after LT, respectively. The most frequent tumor types were skin (17%), lung (16%), oropharyngeal (11%) and prostate cancer (11%). The overall standard incidence ratio as compared to that of the general population was 1.9 (95% CI: 1.5–2.3). The median survival of patients with de novo non-skin cancers was 3.1 years after diagnosis. Only patients with skin cancers and solid tumors, diagnosed at early stages, showed an excellent outcome. After introducing an intensified surveillance protocol, the detection rate of de novo cancers increased from 4.9% to 13% and more de novo malignancies were diagnosed in earlier stages. For non-skin cancers, the median tumor-related survival significantly improved from 1.2 to 3.3 years as well as the median overall survival post-LT. This study indicates that an extensive tumor surveillance program is highly recommendable in LT recipients.
Initially, infectious complications were the main causes of death after liver transplantation (LT). Improvements in immunosuppressive and antiinfective drug therapies have led to a significant improvement in graft and patient survival. With the growing number of long-term survivors, the development of de novo malignancies has emerged as one of the major complications following LT, and de novo cancer has become the second-leading cause of late mortality following cardiovascular disease (1).
Several studies have reported an incidence of de novo malignancies ranging from 3% to 26.2%, mainly depending on the length of follow-up (2–8). The major cause of the development of de novo malignancies is related to the loss of immunovigilance induced by immunosuppressive drugs. In addition, several other risk factors, such as age, alcohol, hepatitis C, Epstein-Barr and JC virus infections, have been identified (2,7,9). So far, there have been only few reports characterizing incidence, therapeutic options and outcome of de novo cancers after LT as compared to malignancies in the general population (2,7,8,10). The outcome of patients who develop de novo malignancies is very poor, mainly due to the fact that tumors are often diagnosed in advanced stages, thus limiting therapeutic options. No optimal screening protocol for these patients has been defined and the question whether extended surveillance protocols will improve patient survival remains controversial. Therefore, the aim of our study was to examine the impact of an extensive tumor surveillance program on early diagnosis of de novo malignancies and survival of LT recipients. Furthermore, we analyzed the incidence of de novo cancer in LT recipients as compared to the general population of our region as well as survival of the most common tumor entities and possible risk factors associated with de novo malignancies.
Patients and Methods
The medical records of all adult patients who underwent LT at our center between September 1982 and December 2007 were reviewed. The study was approved by the local ethics committee.
The study group consisted of 779 patients (235 female and 544 male) with a mean age at LT of 53 (range: 15–76) years. The Median follow-up was 4.1 (range: 0–24) years. The numbers of patients alive at 5, 10, 15 and 20 years after LT were 349, 125, 30 and 17, respectively. Patients who received a second or third graft were counted only once.
The baseline characteristics of the 779 patients are listed in Table 1. The most common indication for LT was end-stage liver disease due to viral hepatitis or alcoholic liver disease (ALD). A pretransplant history of malignancy was diagnosed in 276 patients. In two-thirds (65%) of these patients, a hepatocelluar carcinoma (HCC) was diagnosed pre-LT. In 29 patients a cholangiocellular carcinoma (CCC) or other rare liver tumor (hepatic angiosarcoma, hemangioendothelioma) was seen. Eleven patients had a pretransplant history of an extrahepatic tumor, among them five with urogenital cancer, two with breast cancer, two with lymphomas and one with a colorectal carcinoma and nonmelanoma skin cancer. All patients with nonliver-related tumors had to be disease-free for at least 3 years prior to LT.
Table 1. Clinical characteristics of the 779 liver transplant recipients
Number of patients receiving two or more allografts
Median follow-up after LT in years (range)
Immunosuppressive therapy was based on a triple-drug regimen of cyclosporine A or tacrolimus in combination with corticosteroids and azathioprine. From 2000 on, mycophenolate mofetil was substituted for azathioprine. Corticosteroids were gradually tapered and discontinued within 3 months, except for patients with hepatitis C who received a low-dose maintenance steroid therapy for at least 1 year. Azathioprine (daily dose of 1.0–1.5 mg/kg body weight) or mycophenolate mofetil (targeted daily dose of 2000 mg) was continued for 1 year following LT, unless contraindicated.
Mammalian target of rapamycin (mTOR) inhibitors (sirolimus, everolimus) were used in some patients with calcineurin inhibitor (CNI)-induced side effects.
The dosage of CNIs or mTOR inhibitors was based on general recommendations. Trough or 2 h postdose levels (starting in 2002) for cyclosporine A were monitored at every outpatient visit.
Induction immunosuppressive protocols using antithymocyte globulin or OKT 3 were used in the early years of the LT program. From 1999, patients with poor renal function were given an interleukin-2 receptor antagonist instead of CNIs in the early postoperative period.
Acute rejection episodes were treated with bolus corticosteroid therapy and OKT 3 was used only in refractory cases.
Pre- and posttransplant cancer surveillance
Pre-LT evaluation guidelines at our center did not change significantly over time and include chest and abdominal CT scans, upper gastrointestinal (UGI) endoscopy and colonoscopy, dermatological, urological, ear–nose–throat (ENT) and gynaecological examination (including mammography and Papanicolau smear). In 2002, a urological examination was also added in women. Bone scinitigraphy was only performed in patients with HCC until 2001 but was restricted to patients only with suspicious lesions on CT scans thereafter.
Our ‘historical’ screening program for de novo neoplasia consisted of annual chest X-rays and abdominal ultrasound. Chest and abdominal CT scans were only performed in patients with a preLT history of liver or nonliver malignancies. Dermatological screening was not performed at a regular basis and mammographies as well as urological screenings were only based on the standard of care. The compliance was not strictly supervised during the ‘historical’ era.
In 2002, we started an ‘intensified’ surveillance protocol. The following examinations were performed annually in all patients regardless of age: chest and abdominal CT, urological evaluation (including measurement of PSA), gynaecological (including Papanicolau smear and mammography) and dermatological screening. Colonoscopy was performed 3 years after LT and every 5 years thereafter, except in patients with an adenoma prior to LT or a history of inflammatory bowel disease. In these patients, the first colonoscopy was performed 1 year after LT.
Post-LT HCC surveillance has not changed over the study period. In HCC patients, an annual chest and abdominal CT scan was performed pre- and post-2002.
Concerning the standard routine health maintenance for the general population, annual gynecologic examinations (PAP smear) starting at the age of 20 years, mammographies every 2 years starting at the age of 40 years, annual urologic examinations including PSA measuring starting at the age of 45 years and colonoscopies starting at the age of 50, except patients with a positive family history for colorectal malignancies (starting at the age of 40), are recommended.
Patient survival and cumulative probability of de novo cancers were calculated using the Kaplan–Meier method. Subgroups were compared by means of the log-rank test. A univariate Cox proportional hazards model was used to analyze the variables considered to be possible risk factors for the development of de novo malignancies. The same variables were used in a multivariate Cox proportional hazards model to study the independent effects of these variables. A value of p < 0.05 was considered statistically significant.
Recurrent cancers of patients with a pre-LT history of malignancy were not considered as de novo malignancies. For the purpose of statistical analysis, tumors were divided into the following subgroups: gastrointestinal tumors, lymphomas, lung cancers, skin cancers, oropharyngeal tumors, prostate cancers and miscellaneous tumors.
Standard incidence ratios (SIR) were calculated by applying the method of indirect age-standardized rates. Standardized rates were provided by the cancer registries of Tyrol and South Tyrol, whose data have been included in cancer incidence in five continents (11). Observed cancers in our study cohort were compared with anticipated cases based on age-, sex- and calendar year-specific rates. The 95% confidence limits for the observed to expected ratios were obtained by applying the STATA procedure, which computes exact confidence intervals. Nonmelanoma skin cancer was excluded from the person-year analysis, because no reliable incidence data are available for this cancer in our regional tumor registry. In patients with more than one malignancy, only the first tumor was included in the analysis of the risk for all cancers combined, but the second malignancies were included in the analysis of the specific cancers.
SIR analysis was performed using STATA software version 9.2 (StataCorp, College Station, TX) by applying a self-written code. The remaining statistical analysis was performed with SPSS software version 15.0.1 (SPSS Inc., Chicago, IL).
Prevalence of de novo malignancies
In total, 96 (12.3%) of 779 allograft recipients developed 105 malignancies. The various tumor types are listed in Table 2.
Table 2. Frequencies of various tumor types in LT recipients and mean survival of the main tumor types.
Panel A Tumor types
1Values expressed as number (percent).
2Hodgkin's and non-Hodgkin's lymphomas.
Nonbasal skin cancer
Basal cell skin cancer
Miscellaneous tumors types
Renal cell cancer
De novo HCC
In situ conjunctival cancer
The cumulative risk for development of de novo neoplasia increased with the time after LT (Figure 1). Cumulative risk for all malignancies was 10.0%, 23.7%, 31.8% and 42.3% at 5, 10, 15 and 20 years after LT, respectively.
Time to diagnosis of de novo cancer and survival
The median time from LT to (first) de novo neoplasia was 4.4 years (range: 0.21–19.6 years). In 14 patients (14.6%), the malignancy was diagnosed within the first year after LT. Two had oropharyngeal and one pancreatic cancer. All three tumors were diagnosed at advanced stages. PTLD was seen in three cases and Kaposi's sarcoma in one patient. Three skin cancers were detected in early stages as well as one breast, renal, esophageal and prostatic cancer, respectively.
No significant difference in time to diagnosis was seen among the various tumor types. Nine patients developed a second malignancy. The median time interval between first and second neoplasia was 2.0 (range: 0.8–4.6) years.
Median survival following diagnosis of the first tumor was 4.3 years. The 1-, 5- and 10-year survival rates for all tumors were 76.0%, 49.6% and 46.3%, respectively.
Survival was highly dependent on tumor entity. The best survival was seen in patients with skin cancer, followed by patients with miscellaneous tumors and prostate cancer. Patients with gastrointestinal tumors, especially esophageal and pancreatic cancer, lung cancer, oropharyngeal tumors or lymphomas had significantly poorer survival (Table 2 Panel A, Figure 2).
Thirty-two (29.2%) patients with de novo malignancy died during the study period due to cancer. Seven patients (11.5%) died due to HCV recurrence or other nontumor-related causes (sepsis or bleeding). The actual risk of a tumor-related death was 4%, 11% and 14% at 5, 10 and 15 years, respectively, after LT.
No significant difference in survival after LT was found between patients with or without de novo neoplasia (p = 0.158). The actuarial 1-, 5-, 10- and 15-year survival rates following LT were 82.9%, 67.7%, 54.7% and 45.1% in patients without and 99%, 80.8%, 59.4% and 34.8% in patients with de novo cancer, respectively. However, excluding patients with a follow-up of less than 1 year, a significant difference in survival between patients with and without de novo noncutaneous cancer (median survival 10.4 vs. 17.0 years; p = 0.015) was found.
Survival as a factor of surveillance and therapeutic options
As de novo malignancies became increasingly important in the care of LT recipients, we started an extensive surveillance program in 2002. Interestingly, the detection rate of de novo malignancies increased from 4.9% (24 out of 490 LT patients) during the early period from 1982 to 2001 to 13.0% thereafter (81 out of 625; p < 0.001). In fact, nearly 75% of all 105 malignancies were diagnosed after the year 2001. All 21 non-skin carcinomas in the preintensified era were diagnosed due to symptomatic disease, whereas 19 of the 66 (29%) in the era of the intensified protocol were diagnosed during surveillance visits. In addition, 71% of non-skin tumor patients were diagnosed at stages III and IV before 2002, in contrast to only 46% after 2002. The median non-skin tumor-related survival significantly improved from 1.2 years in the historical to 3.3 years in the intensified surveillance era (p = 0.021; Figure 3) as well as the median overall survival (post-LT) from 3.1 to 11.3 years (p = 0.001). No significant difference in the incidence of the major cancer entities was seen between the two periods.
Comparison with the general population and risk factors
Incidence of de novo malignancies was significantly increased in LT recipients as compared to the general population of our region, with a SIR of 1.9 (95% CI 1.5–2.4). Esophageal cancer, lymphomas (Hodgkin's and non-Hodgkin’s) and oropharyngeal cancer had the highest risks with an 8.3-fold, 8-fold and 4.8-fold increase, respectively (Table 3).
Table 3. Standard incidence ratio for developing a neoplasia following LT
De novo neoplasias
95% confidence interval
1Nonmelanoma skin cancer excluded.
2Esophageal, gastric, colorectal and pancreatic cancers.
*p < 0.05.
The age at tumor diagnosis differed between the general population and LT recipients. Regarding all cancers, a median difference of 4 years was found which was statistically significant (p < 0.001). In the subgroup analysis, only gastrointestinal tumors developed at a significantly younger age (Table 4).
Table 4. Median age at diagnosis in years
The following possible risk factors for the development of de novo malignancies were analyzed: gender, age at LT, alcoholic liver disease, history of nicotine abuse, preoperative history of malignancy, rejection episodes, reLT and the use of monoclonal or polyclonal antibodies.
In univariate analysis, age (HR 1.05; 95% CI 1.02–1.07; p < 0.001) and history of smoking (HR 1.7; 95% CI 1.1–2.7; p = 0.016) were significantly associated with the risk of de novo malignancy. In multivariate analysis, only age at LT proved to be an independent predictor for de novo cancer. The hazard ratio of 1.04 (p = 0.037) implies that the risk of de novo malignancy increases by 4% with every year of age.
Smoking proved to be an independent predictor for two tumor entities: lung (HR = 8.3; 95% CI 2.2–27.3; p = 0.001) and oropharyngeal cancer (HR = 9.2; 95% CI 2.0–42.0; p = 0.004).
With increasing numbers of long-term survivors after LT, de novo malignancies have become one of the leading causes of late mortality (7,12). We were interested in determining whether an extensive surveillance program promotes early diagnosis of de novo cancers and consequently improves patient survival. Furthermore, we analyzed the tumor-specific characteristics of de novo malignancies in LT recipients as compared to the general population.
The results of this single-center study highlight the clinical implication of de novo malignancies after LT. A total of 96 patients developed 105 malignancies. The overall incidence of 12.3% is consistent with previous studies (2–8).
Compared to the general population of our region, the overall risk of tumor development was doubled in LT recipients and increased up to eight times for specific tumor entities (Table 3). To date, only few data are available on tumor incidence as a factor of age-, sex- and calendar year-specific rates. Two studies show the relative risk for de novo neoplasia to be 2.7 and 4.3 (2,8). Another study reported a relative risk of 3.23 for noncutaneous and of 16.9 for cutaneous malignancies (7). There is strong evidence that the increased cancer risk after solid organ transplantation is mediated through several pathogenetic factors, in particular impaired immunosurveillance and immunoediting caused by immunosuppressive drugs (13,14).
Apart from the increased risk of cancer, tumor development occurs at a younger age in LT recipients. This is particularly true for gastrointestinal cancers as well as for oropharyngeal and lung neoplasia (Table 4). The median age at diagnosis of patients with gastrointestinal cancer was 12 years lower in LT recipients as compared to the general population. This finding confirms a previous report, which found that the mean age of patients with de novo colon cancer in solid organ transplant patients was 58 years as compared to 70 years in the general population (15).
Almost one-third of the patients with de novo malignancies died during the study period. The median survival after diagnosis was 4.3 years. In contrast to previous studies suggesting that cancer-related death is not clinically relevant after LT (16,17), more recent reports have shown that de novo malignancies are one of the leading causes of late mortality following LT (2,18,19). We found a cancer-related mortality risk of 14% at 15 years after LT. This is in accordance with the 15% risk after 15 years published by Haagsma et al. (2).
Increased cancer risk, tumor development at younger age and the significant impact of de novo cancer on long-term survival of LT recipients endorse the importance of surveillance protocols for this cohort of patients. An intensified surveillance protocol was introduced at our center in 2002. As a result, the detection rate of de novo cancers rose from 4.9% to 13% and, more importantly, survival of tumor patients post-LT significantly improved. The improved cancer-related survival of de novo cancer patients is most probably related to a ‘stage shift’ from advanced to earlier stages. Before 2002, 71% of non-skin tumor patients were diagnosed at stages III and IV, in contrast to only 46% after 2002.
De novo malignancies were diagnosed in 14 patients during the first year after LT. This raises the question if these patients are a matter of preoperative screening failures or cases of donor-transmitted cancer. The pre-LT radiographic studies were re-evaluated in all cases of solid organ malignancies and the negative results were confirmed. However, especially in the cases of advanced oropharyngeal and pancreatic cancers, screening failures cannot be excluded. The donor histories for pre-existing cancers were negative.
Lung cancer was the most common noncutaneous malignancy in our study population, with a frequency of 16.2% and a 3.1-fold increased incidence as compared to the general population. Our results confirm data from England and Wales, reporting a 2-fold increased risk of lung cancer in LT patients (8).
Early detection is essential for a favorable outcome of patients with lung cancer, as potentially curative therapies are restricted for patients diagnosed at early tumor stages. It has been shown that CT screening is highly effective in the early detection of lung cancer in a high-risk population (20). Having screened 31 567 patients with a ≥10 pack-year smoking history with a low-dose spiral chest CT, the international lung cancer action project (I-ELCAP) detected 405 lung cancers, the majority at early stage I (20). In our cohort of LT recipients, the number of patients needed to screen (NNS) to detect one lung cancer is 46 and, therefore, lower than the reported number of 78 by the I-ELCAP. Given the low NNS in our cohort, the costs and the radiation burden associated with annual chest CT scans may be justified at least in these high-risk patients.
Oropharyngeal tumors were the second-leading tumor entity in our study cohort. Compared to the nontransplant population, tumor incidence was 4.8-fold increased, which is similar to the 7.6-fold increased risk found in the Pittsburgh study (10). These findings contrast other reports in which oropharyngeal tumors occurred less frequently or were even absent (2,4,6). Alcohol has been reported to be a major risk factor for the development of oropharyngeal cancer in the general population, and two studies found a significantly higher incidence of overall and oropharyngeal cancer in patients transplanted for ALD as compared to non-ALD patients (5,21,22). In our study, smoking seemed to be more important for the development of oropharyngeal cancer than alcohol, as 10 of our 12 patients were smokers and smoking was the only parameter that significantly correlated with the development of oropharyngeal cancer. Of patients with oropharyngeal cancer 64% were diagnosed in UICC stages III or IV. Therefore, mortality was high and the 5-year survival rate was lower than the survival rates reported for oropharyngeal cancer in the general population of our region (37% vs. 50%) (23). Benlloch et al. recommend annual screening for oropharyngeal tumors in patients with a history of alcohol abuse (21). Based on our findings, we suggest that this recommendation should also be expanded to LT recipients with a history of nicotine abuse. However, there is no proven evidence for the efficacy of an annual ENT surveillance examination, which needs to be tested in a prospective manner.
Although skin cancer is very common in liver transplant recipients, it does not negatively impact patient survival. There were no skin cancer-related deaths in our study, thus confirming the low mortality observed by other authors (2,3,7,10).
In contrast to previous reports, de novo colorectal cancers were rare in our cohort (5.7%). This might be explained by the very small number of LT recipients transplanted for primary sclerosing cholangitis (PSC) and concomitant ulcerative colitis (UC) at our center, as it has been reported that PSC patients with UC have a particular high risk for colorectal cancer post-LT (8,24,25). Only one patient with PSC and UC developed colon cancer, whereas the remaining five patients with colon cancer had no history of UC.
Identification of predisposing factors for the development of de novo neoplasia is important for the identification of patients with a high risk for cancer. Univariate analysis showed a history of smoking and age at LT to be significantly associated with de novo malignancies. Only age at LT proved to be an independent predictor of overall de novo malignancies in multivariate analysis. This may in part be due to the fact that no data on smoking habits were available for patients transplanted in the early years of our program. However, for lung and oropharyngeal cancer, a positive nicotine history was a statistically significant risk factor for both cancer entities, which confirms several previous reports (26,27).
Our data suggest different screening protocols for patients with high and low risk of de novo malignancies following LT. Especially in older patients, smokers and in patients with a positive pre-LT history of alcoholism, an intensified screening protocol including annual chest and abdominal CT scans and ENT examination seems to be justified. In contrast, screening for breast and prostate cancers can be performed based on the routine health maintenance recommendations, but compliance should be monitored. Certain cancers, such as pancreatic, esophageal and gastric cancers, are not detectable at early stages by radiographic surveillance protocols. UGI endoscopy and endoscopic ultrasound seem to be the methods of choice; however, their efficacy in the setting of cancer surveillance is unknown and needs to be proven in controlled trials.
A further limitation of our study is the fact that we compared tumor specific survival of patients from two different time periods. Other confounding variables, such as the enormous progress made in the treatment of malignancies over the last years, might have influenced the improved survival in patients diagnosed in the later surveillance era. Only an analysis of patients followed and treated during the same time period can definitely prove the efficacy of an extensive surveillance protocol on the survival of LT recipients with de novo malignancies.
Our proposed intensified screening protocol is certainly associated with high expenses. The cost effectiveness of such an extensive surveillance protocol must be addressed in further studies. In addition, the radiation burden of annual chest and abdominal CT scans has to be outweighed by the possibly positive effect of these imaging studies on patient survival. At least in high-risk patients, radiation burden and costs of annual CT scans might also be justified by the relatively low NNS in this patient cohort.
In conclusion, our study shows that the risk for LT recipients to develop a malignancy is markedly increased as compared to that of the general population. De novo cancers are an important cause of late death after LT. In addition, our data suggest that an extensive surveillance program promotes early diagnosis of de novo malignancies and, therefore, might improve long-term outcome in LT recipients. However, our findings need to be confirmed by a prospective, controlled study.
This work was supported by ‘Verein zur Förderung der Forschung in Gastroenterologie und Hepatologie an der Medizinischen Universität Innsbruck’.