SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The goal of this study was to determine the risk factors for de novo cancer after liver transplantation (LTx). Retrospective analyses were performed in 385 LTx patients who underwent transplantation between 1986 and 2007. In total, 50 (13.0%) recipients developed de novo malignancy. The cumulative incidence of de novo cancer at 1, 5, 10, and 15 years after LTx was 2.9% ± 0.9%, 10.5% ± 1.8%, 19.4% ± 3.0%, and 33.6% ± 6.8%, respectively. The standardized incidence ratio of malignancy in LTx patients compared to the general population was 2.2 (95% confidence interval: 1.6-2.8). After excluding posttransplant lymphoproliferative disorder and skin cancer, patients with de novo cancer had a significantly lower survival rate compared to recipients who remained cancer-free. The identified univariate risk factors for de novo cancer were cyclosporine A (CsA) treatment, time period of LTx, and recipient age. In multivariate analysis, only CsA treatment emerged as an independent risk factor for de novo cancer, which was attributed to more aggressive cancer types. A surprising finding was that CsA treatment specifically enhanced cancer risk in patients who underwent transplantation after 2004, when C2 monitoring (blood concentration at 2 hours postdose) was introduced. In addition, these patients showed a significantly lower acute rejection rate, which might reflect a more robust immunosuppressive status caused by the CsA-C2 regimen. When age was considered, only patients ≤50 years had a higher cancer rate when treated with CsA compared to treatment with tacrolimus. Our data suggest that, compared to tacrolimus treatment, CsA treatment with C2 monitoring or in younger patients of ≤50 years is associated with a higher early de novo cancer risk after LTx. Liver Transpl 16:837–846, 2010. © 2010 AASLD.

The 1-year survival after liver transplantation (LTx) has dramatically increased in the past 3 decades. Currently, more than 80% of LTx patients are alive 1 year after LTx.1, 2 In contrast, the long-term outcome has improved less impressively and its improvement is one of the main focuses in modern transplantation medicine.1 Importantly, malignancy is one of the major leading causes of late death after LTx.2-5 The reported risk for de novo cancer in liver transplant recipients is up to 2.1-4.3 times higher compared to the matched general population, with an incidence varying between 3% and 26%.3-10 The higher cancer risk after organ transplantation has been reported to be directly related to the intensity as well as the cumulative dose of immunosuppression.11 A dose reduction of cyclosporine A (CsA) to maintain a trough blood level from 200 ng/mL to 100 ng/mL in kidney transplant recipients resulted in a significant reduction in the de novo cancer incidence.12 Furthermore, a history of usage of immunosuppressive drugs prior to liver transplantation was a significant risk factor for development of de novo malignancy.13 The cancer pathogenesis induced by immunosuppressants includes direct damage to the host DNA and impairment of the recipients' immunosurveillance, which reduces their antitumor and antiviral immunity.14, 15 Considering other risk factors, a higher age had been reported to play a significant role in the development of cancer after LTx.8, 13, 16

Since their introduction in the 1980s, calcineurin inhibitors (CNIs) have been the cornerstone of immunosuppressive treatment after transplantation. The relationship between cancer and the use of different CNIs, such as CsA or tacrolimus (TAC), is still to be elucidated.1 Several studies found no difference in the incidence of de novo cancer between CsA-based and TAC-based regimens,17-19 whereas other studies reported a higher de novo cancer risk for CsA-based20, 21 or TAC-based8 immunosuppressive protocols. Thus, the role of different CNIs in the occurrence of de novo cancer after LTx is still a matter of controversy.

Because the therapeutic window between efficacy and toxicity of CsA is small, blood level monitoring to guide dosing is an essential tool.22 Traditionally, CsA dose adjustment was based on trough blood level (C0). In more recent years, numerous studies have reported that dose level monitoring 2 hours after dosing (C2) is a more effective monitoring strategy compared to C0 level, because the absorption phase of CsA occurs within 2 hours and the concentration peak level is a better predictor of freedom from graft rejection.23, 24 Using CsA-C2 monitoring, the overall incidence of biopsy-proven acute rejection (22%-28%) appeared to be lower compared to CsA-C0 monitoring (36%-59%).25-28 Therefore, in 2005, the CsA-C2 monitoring strategy was introduced in our center.

In this study, we describe the incidence of de novo cancer after LTx in our center and determine the risk factors associated with de novo cancer risk. Because immunosuppressive regimens have changed over time and a delicate balance exists between effective prevention of rejection and oversuppression of immune surveillance, we additionally studied the incidence of de novo cancer over time and determined its correlation with changes in immunosuppressive regimens.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Patients

Between October 1986 and December 2007, 565 liver transplantations were performed in 500 adults in the Erasmus MC University Medical Centre Rotterdam, The Netherlands. Because malignancy is a time-dependent event, patients with a follow-up shorter than 4 months (n = 99) at the time of analysis were excluded. In addition, patients who underwent auxiliary LTx (n = 13), those who received a liver with the presence of occult malignant cells (n = 2), and those who had a cholangiocarcinoma in the native liver (n = 1) were excluded. Consequently, 385 patients were included in our cohort, which represented 3121 person-years of follow-up. A retrospective analysis was performed using the LTx database and medical records of the recipients. Analyzed data included age of recipient at time of transplantation, sex of recipient, primary LTx indication, type of primary immunosuppressive therapy, de novo malignancy after transplantation, interval from LTx to diagnosis of malignancy, interval from LTx or diagnosis of cancer to death, and interval from LTx to diagnosis of the first acute rejection, which was defined as biopsy-proven rejection (Snover-grade ≥ 2, or rejection activity index score ≥ 6) with a rise in aminotransferases and/or bilirubin, which responded to treatment with methylprednisolone, and in some patients with muromonab-CD3 (trade name, OKT3) or rabbit anti-thymocyte globulin (ATG). All patients were followed until December 2008.

The primary endpoint was de novo malignancy, which was defined as the development of cancer other than recurrent primary liver cancer. Recurrence of cancer was not accepted as an endpoint, because this is not exclusively dependent on the immunosuppression state, but also on the aggressiveness of the primary liver cancer. Diagnosis of malignancy was confirmed histologically, and the date of histological diagnosis was taken as date of malignancy.

Data were collected from follow-up examinations in the outpatient clinic at month 6 and every 12 months after transplantation. Prior to LTx, all patients had undergone a routine tumor screening, which included chest radiography, mammography, abdominal ultrasound, gastroscopy, and colonoscopy. After LTx, all patients participated in a yearly repeated clinical, laboratory, and radiology (ultrasound and chest x-ray) study. Women participated in a routine cervical cytology study on a yearly basis. Other studies were done on an individual patient basis. Epstein-Barr virus (EBV) immunoglobulin G in serum before LTx were recorded for 346 of 385 patients.

All patients declared that they did not object to the use of their data in the study.

Primary Immunosuppression

The choice of certain immunosuppressive protocols in our center over the years was based on general development and new insights in immunosuppressive drugs (based on clinical trials) in the field of liver transplantation medicine. During the first 10 years of the study, dual and triple CsA-C0-based maintenance immunosuppressive regimens have been routinely used. Dual therapy consisted of CsA and prednisolone, and azathioprine was added in the triple regimen. Between 1994 and 1999, the microemulsion formulation of CsA (Neoral; Novartis, East Hanover, NJ) had been altered with the oil-based CsA (Sandimmune; Novartis). The microemulsion formulation of CsA was routinely used since 1999. After CsA was initiated within 24 hours postreperfusion in a dose of 10-15 mg/kg body weight/day, the dosage was adjusted to predose levels according to a range from 200-400 ng/mL during the first 3 months and thereafter from 100-200 ng/mL. Moreover, CsA dosage was adjusted in case of rejection or CsA-related toxicity, as described.29 Since the introduction of TAC in 1990, a dual TAC-based regimen with prednisolone has been alternated with dual or triple CsA-based regimens. The monitoring of TAC was based on trough levels with a target range of 5-15 ng/mL. Starting in 2000, dual therapy with TAC and prednisolone was used as a standard regimen.

Since the introduction of the CsA-C2 level monitoring strategy in 2005, a CsA-C2-based immunosuppressive regimen was used as the standard regimen, interchanged with a TAC-based immunosuppressive regimen. The patients treated with CsA-C2 received an initial dose of 10-15 mg/kg/day, which was adjusted based on the C2 target level of 800-1200 ng/mL during the first 3 months. Thereafter, target levels were maintained at 700-900 ng/mL from 4 until 6 months and 480-720 ng/mL thereafter. The dosing strategy of TAC remained unchanged.

In the induction immunosuppressive regimen, ATG (n = 16) or OKT3 (n = 3) was used in the first decade of the study. Since 1998, these induction therapies have been replaced by interleukin-2 (IL2) receptor antagonists (IL2RA).

Prior to LTx, all patients with autoimmune cirrhosis were treated with prednisone and azathioprine.

Statistical Analyses

Statistical analyses were performed using SPSS software package, version 15.0 (SPSS, Chicago, IL). The cumulative incidences of de novo malignancy and survival were assessed by the Kaplan-Meier method. For the analysis of malignancy, the starting point was the first LTx and the endpoint was the first malignancy. When reviewing the interval from LTx to specific cancer types, we initially categorized cancer into 3 groups based on the highest incidence in our center, which were nonmelanoma skin cancer, posttransplant lymphoproliferative disorder (PTLD), and other cancers. The endpoints were then defined as: (1) nonmelanoma skin cancer, (2) PTLD, and (3) other cancers. In all analyses, patients were censored in the case of death and loss to follow-up. Underlying this approach was the assumption that patients were alive, and therefore at risk, in the follow-up.

For survival analysis when comparing patients who developed de novo cancer with those who remained cancer-free, de novo cancer was considered as a time-dependent factor, because cancer developed at various follow-up times. This means that all patients entered at time 0 as cancer-free. At the time of the development of de novo cancer, the patient was censored in the cancer-free group and entered the de novo cancer group. In this way, the period that a patient lived as cancer-free was calculated as the “event-free survival period” in the Kaplan-Meier analysis. The same method was used to estimate the association of different types of cancer with patient survival. Patients could have more than one malignancy, but only the first malignancy was included in the survival analysis.

Univariate Cox regression and log-rank analyses were used to identify risk factors associated with de novo cancer development. To investigate whether these factors were independently associated with cancer development, all risk factors with P ≤ 0.05 were used as a covariate in a multivariate Cox regression model. P ≤ 0.05 was considered significant. The distribution of categorical variables was compared using the chi-squared test and differences between means were compared by Student t test.

In the analyses of the transplantation period as a risk factor for de novo cancer, malignancy incidence curves were made of the different LTx time periods containing 4-year intervals. The recent LTx time period consisted of 3 years in order to provide more equal numbers in the groups, because a large number of LTx were performed in the latest years. The first LTx period starts at 1989, because all LTx performed in our center before 1989 were auxiliary LTx.

For analyses of the mean daily dose of CsA (given in milligrams per kilogram body weight) at different time intervals within the first 12 months post-LTx, cumulative daily doses were divided by the mean body weight and the number of days in the same time interval. Differences between means were compared by Student t test.

The standardized incidence ratio (SIR) of malignancy was calculated by comparing the observed number of cancers in the study cohort with the expected number of cancers based on age-specific and sex-specific rates for the Netherlands. The Dutch incidence rates of cancer were obtained from the Comprehensive Cancer Centre in The Netherlands30 for the calendar year 1997, which is the median of our study period. We assume that the SIR estimation will not differ considerably when we matched our population with calendar-year-specific rates in The Netherlands, because the Dutch cancer rates have not changed significantly over time.30 Then, the incidence rates were stratified into 5-year age groups per 100,000 person-years for both sexes. The expected numbers of cases were calculated by the multiplication of the Dutch cancer incidence rate in both sex and age group by the number of person-years in the corresponding period of observation. The standard error for hazard ratio was calculated for estimation of the 95% confidence interval (CI).

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Incidence of De Novo Malignancy

Patient characteristics are summarized in Table 1. From the 385 liver graft recipients, who had at least 4 months of follow-up after the first LTx, 50 (13.0%) patients developed at least one de novo cancer. The cumulative incidence of de novo malignancy at 1, 5, 10, and 15 years after LTx was 2.9% ± 0.9%, 10.5% ± 1.8%, 19.4% ± 3.0%, and 33.6% ± 6.8%, respectively (Fig. 1).

Table 1. Patient Characteristics
Characteristic NPercent
  • *

    NOTE: Viral cirrhosis: sum of hepatitis B cirrhosis (n = 24), hepatitis B/D cirrhosis (n = 7), and hepatitis C cirrhosis (n = 34).

  • Cholestastic disease: sum of primary biliary cirrhosis (PBC; n = 25), primary sclerosing cholangitis (PSC; n = 64) and secondary biliary cirrhosis (n = 7).

  • Cancer: sum of hepatocellular carcinoma (n = 43) and hepatic epithelioid hemangioendothelioma (n = 1).

  • §

    Sixteen patients did not have either cyclosporine or tacrolimus.

Median age at LTx (range)49 (16-69) 
Median follow-up in years (range)5.0 (0.3-19.1) 
Sex   
 Male 22057
 Female 16543
Primary LTx indication   
 Viral cirrhosis* 6517
 Cholestatic disease 9625
 Autoimmune cirrhosis 144
 Alcoholic cirrhosis 3910
 Acute liver failure 6216
 Cancer 4411
 Other 6517
Time periods of LTx   
 1989-1992 256
 1993-1996 6717
 1997-2000 9424
 2001-2004 10126
 2005-2007 9825
Primary immunosuppressive drug§   
 Cyclosporine 16752
 Tacrolimus 20243
Induction therapy   
 ATG/OKT3 195
 IL2RA 23060
EBV IgG prior to LTx   
 Positive 33487
 Negative 123
 Unknown 3910
thumbnail image

Figure 1. Cumulative incidence of de novo malignancy after LTx in liver graft recipients who underwent transplantation in the Rotterdam LTx cohort (n = 385). The gray lines represent the standardized error. The Kaplan-Meier method was used under the assumption that patients were at risk during follow-up.

Download figure to PowerPoint

In the Dutch population, the expected number of age-matched and sex-matched individuals with de novo malignancy in the same follow-up period as our cohort was calculated to be 22.8. Because our observed number of patients with de novo cancer was 50, the SIR of de novo malignancy in liver transplant patients in comparison with the general population was 2.20 (95% CI = 1.59-2.80).

Types of Cancer

In total, 66 de novo malignancies appeared in 50 recipients. Table 2 shows the distribution of the types of cancer in our study. The most common cancer types were nonmelanoma skin cancer (55%) and PTLD (21%). Skin cancer included basal cell carcinoma (41%, 27 of 66 patients) and squamous cell carcinoma (14%, 9 of 66 patients). The cumulative incidences of nonmelanoma skin cancer, PTLD, and other cancers are depicted in Table 3.

Table 2. Type of De Novo Cancer in Patients After Liver Transplantation
TypeNumber of TumorsTotal (%)
Nonmelanoma skin3655
PTLD1421
Gastrointestinal46
Lung35
Melanoma23
Gynecological23
Kaposi sarcoma12
Kidney12
Liver12
Other23
Total66100
Table 3. Cumulative Incidence of De Novo Cancer Types in Patients After Liver Transplantation
 Time After LTx
Type of Cancer1 year5 year10 year15 year
  • *

    NOTE: Other cancer includes the remaining cancer types.

Nonmelanoma skin0.8%5.1%10.2%19.7%
PTLD1.3%2.8%6.4%8.3%
Other cancer*0.5%2.6%4.4%10.1%

Influence of Cancer on Survival

Patients who developed de novo malignancy after LTx had a significantly shorter survival than those who remained cancer-free. At 10 years after cancer diagnosis, survival was 50.8% ± 8.6% in the cancer group versus 79.0% ± 2.9% in the cancer-free group at 10 years post-LTx (P < 0.001).

With regards to the type of cancer, the survival was not impaired in patients who developed nonmelanoma skin cancer and PTLD compared to patients who remained cancer-free (P = 0.122 and P = 0.330, respectively; Fig. 2). However, a highly reduced survival was found in patients who developed the other cancer types summed in Table 2, including the more aggressive gastrointestinal, lung, and gynecological tumors (P < 0.001). The 1-year survival in these patients after diagnosis of cancer was 28.6% ± 1.2%. None of the patients who developed nonmelanoma skin cancer and PTLD died from their disease, whereas 9 patients (64%, 9 of 14) who developed the remaining cancer types died due to cancer.

thumbnail image

Figure 2. Estimated survival in LTx patients who developed different types of cancer (after diagnosis of de novo cancer) and those who remained cancer-free (after LTx). All survival rates were compared with the survival rate of cancer-free recipients (*P = 0.122; **P = 0.330, ***P < 0.001).

Download figure to PowerPoint

Risk Factors for Malignancy After Transplantation

When analyzing potential factors influencing the development of de novo cancer after LTx by univariate analyses, age > 50 year (P = 0.025), immunosuppressive treatment with CsA (P = 0.035), and the time period of LTx (P = 0.049) emerged to be significant risk factors (Table 4).

Table 4. Risk Factors Associated with De Novo Cancer Development (Univariate Cox Regression Model)
Risk FactorHazard Ratio (95% CI)P Value
Age  
 ≤50 year10.025
 >50 year1.92 (1.09-3.38) 
Sex  
 Male10.431
 Female0.79 (0.45-1.41) 
Indication for LTx  
 Viral cirrhosis10.753
 Cholestatic disease1.03 (0.45-2.35) 
 Autoimmune cirrhosis0 (0-∞) 
 Alcoholic cirrhosis0.99 (0.33-2.95) 
 Acute liver failure0.71 (0.27-1.86) 
 Cancer1.80 (0.63-5.11) 
 Other0.74 (0.26-2.10) 
Immunosuppressive regimen  
 Dual10.795
 Triple1.10 (0.62-1.89) 
Primary immunosuppressive drug  
 Tacrolimus10.035
 Cyclosporine2.02 (1.05-3.87) 
Induction immunosuppressive treatment  
 Non-IL2 receptor antagonist10.070
 IL2 receptor antagonist0.55 (0.29-1.05) 
Time period of LTx  
 1989-19920.83 (0.28-2.45) 
 1993-19960.51 (0.19-1.40) 
 1997-20000.37 (0.13-1.02) 
 2001-20040.22 (0.07-0.71) 
 2005-200710.049
EBV IgG prior to LTx  
 Positive10.514
 Negative2.51 (0.89-7.07) 
 Unknown1.12 (0.52-2.42) 

In order to see how the period of transplantation influenced the appearance of de novo cancer, malignancy incidence curves were made of the different LTx time periods containing 3-year and 4-year intervals (Fig. 3). A high incidence was found in patients who underwent transplantation in the earliest LTx period (1989-1992). Comparing this period with the period from 2001-2004, patients transplanted between 1989 and 1992 had a significantly higher risk for cancer (P = 0.004). After this earliest LTx period, a considerable decline of the incidence of de novo cancer was observed after each period of transplantation. In contrast, in patients who underwent transplantation in the most recent period of 2005-2007, an increase of the incidence of de novo malignancy was observed. Patients transplanted in this period had a significantly higher rate of de novo malignancy compared to those transplanted in the period of 2001-2004, during the first 4 years posttransplantation (P = 0.022). The rise in the de novo cancer incidence was particularly seen in the first year after LTx, which was 7.2% ± 2.6% in the LTx period from 2005-2007 versus 1% ± 1.0% in the LTx period from 2001-2004. After identifying which cancer types were increased in the recent period, we found that the incidence of all cancer types, rather than a specific cancer type, increased (data not shown).

thumbnail image

Figure 3. Cumulative incidence of de novo cancer in LTx patients who underwent transplantation in different time periods. A higher risk of de novo malignancy was observed in liver graft recipients who underwent transplantation in 1989-1992 and 2005-2007 compared to 2001-2004 (*P = 0.004 and **P = 0.022 compared to 2001-2004).

Download figure to PowerPoint

In order to see which variables were independently associated with a higher cancer risk after LTx, a multivariate analysis was performed including factors that in the univariate analyses were associated with de novo cancer. In this Cox-regression analysis, CsA treatment remained a significant risk factor (P = 0.029), whereas LTx period (P = 0.401) and age > 50 years (P = 0.219) failed to show an independent association with the development of cancer. This data indicates that CsA treatment was the only independent risk factor for development of de novo cancer after LTx and that the effects of both LTx period and age on cancer risk were related to CsA treatment. Remarkably, when the types of cancer appearing in patients who were treated with CsA compared to TAC were studied, we observed that CsA-treated patients had a 2.5-times higher risk of developing more aggressive cancer types that do not belong to the nonmelanoma skin cancer and PTLD summed in Table 2 (data not shown). These data indicate that CsA is not only associated with a higher early de novo cancer risk, but also with cancer types having a worse prognosis.

Increased Cancer Rate in CsA-Treated Patients Was Only Observed in Recent Years

To investigate how the influence of the LTx period on cancer risk was related to CsA treatment, we compared the incidence of de novo cancer in CsA-treated patients with TAC-treated patients in the different LTx periods. As depicted in Fig. 4, we could not find a difference in the cancer rate between CsA-treated and TAC-treated patients in the LTx period before 2005. Most striking was that CsA-treated patients who underwent transplantation since 2005 showed a significantly higher de novo cancer risk in the early phase after LTx compared to the TAC-treated patients, which was 9.9-fold (95% CI = 1.2-80.5, P = 0.032) higher compared to patients treated with TAC. These data indicate that only the specific CsA treatment used in recent years was associated with a higher risk for early development of de novo cancer.

thumbnail image

Figure 4. Cumulative incidence of de novo malignancy in CsA-treated and TAC-treated liver graft recipients who underwent transplantation before 2005 and those since 2005. Only CsA treatment in patients who underwent transplantation since 2005 showed a significantly higher early de novo cancer rate compared to TAC treatment. In this period, 7 de novo cancers occurred in the CsA-treated group versus 1 de novo cancer in the TAC-treated group. (*P = 0.009 compared to TAC from 2005, **P = 0.255 compared to TAC before 2005).

Download figure to PowerPoint

In patients who underwent transplantation since 2005, 7 CsA-treated patients developed de novo cancer, of which 3 were nonmelanoma skin cancer, 3 were PTLD, and 1 was epithelioid sarcoma. In the TAC-treated group, only one bone marrow metastasis of unknown origin occurred.

To see whether the CsA-treated group was comparable with the TAC-treated group in the LTx period from 2005, chi-squared tests were performed, showing that there were no differences in age (≤50 years of age: 41.5% in the CsA group versus 58.5% in the TAC group; P = 0.10), sex (male: 46.4% in the CsA group versus 53.6% in the TAC group; P = 0.18), and LTx indication (P = 0.35) between both treatment groups. Next, to determine whether CsA was more frequently used in a triple immunosuppressive regimen in the recent LTx period, a chi-squared test was performed to calculate the distribution of CsA and TAC in triple versus dual immunosuppressive regimens. We found that CsA and TAC were equally distributed in triple as well as in dual immunosuppressive regimens (data not shown).

Since January 2005, CsA dosing based on the conventional C0 level monitoring was replaced by dosing based on C2 level monitoring in all de novo LTx patients. Because CsA blood level monitoring was the only major change in CsA treatment in the recent LTx period, our data suggest that the C2 monitoring strategy was the reason for the increased early de novo cancer risk. In order to see whether patients treated with CsA-C2 received a higher dose that could have caused the higher cancer risk, the mean daily CsA doses of patients treated with CsA-C2 and CsA-C0 were compared. Analysis was performed in different time intervals within the first 12 months after LTx, because all malignancies in CsA-C2-treated patients occurred within the first year post-LTx. During the induction phase (first 14 days after LTx), patients treated with CsA-C2 received a significantly higher dose than patients treated with CsA-C0, which averaged 11.76 mg/kg/day in CsA-C2-treated patients versus 9.79 mg/kg/day in CsA-C0-treated patients (P = 0.043). Thereafter, no differences in daily CsA dose were observed between patients treated with CsA-C2 and CsA-C0 up to 6 months post-LTx (5.26 mg/kg/day in CsA-C2 versus 6.05 mg/kg/day in CsA-C0, P = 0.157). However, at 12 months post-LTx, patients treated with CsA-C2 received a significantly lower dose than patients treated with CsA-C0 (3.69 mg/kg/day in CsA-C2 versus 5.24 mg/kg/day in CsA-C0, P < 0.001).

Nevertheless, CsA doses given to patients may not adequately represent drug exposure due to interpatient and intrapatient variability in CsA absorption. A better surrogate marker for drug exposure is the rejection rate, which is strongly correlated with CsA bioavailability.31-34 Therefore, analyses were performed to compare acute rejection rates during the first year post-LTx between CsA-treated and TAC-treated patients who underwent transplantation before and after 2005. At 1 year post-LTx, CsA-treated patients who underwent transplantation since 2005 showed a low incidence of acute rejection (9.6%), whereas this is significantly higher in CsA-treated patients who underwent transplantation before 2005 (35.2%, P = 0.002), and TAC-treated patients who underwent transplantation since 2005 (27.5%, P = 0.030) and before 2005 (23.9%, P = 0.047).

Increased Cancer Risk for CsA Treatment Was Associated with Younger Recipient Age

In order to see how recipient age was correlated with the difference in cancer rate between CsA and TAC treatment, we compared the incidence of de novo cancer between CsA-treated and TAC-treated patients in different age groups, of which 50 years was the best discriminating cut-off value (data not shown). We found that CsA-treated patients in both age groups (≤50 and >50 years) had a high de novo cancer risk (18.3% >50 years versus 18.8% in ≤50 years, P = 0.560), which was statistically not different from TAC-treated patients who were older than 50 years (9.2%, P = 0.459). In contrast, TAC treatment in patients who were 50 years or younger resulted in a significantly lower cancer risk as compared to CsA treatment in the same age group (5.2% in TAC versus 18.8% in CsA, P = 0.027). These data indicate that only in the younger patients a higher cancer risk was found for CsA treatment compared to TAC treatment.

Factors Predisposing to the Development of Specific Cancer Types

We questioned whether the risks of the individual cancer types were associated with the same risk factors or might be influenced by confounding factors, such as EBV seronegativity for development of PTLD. Indeed, we found that patients who were EBV-seronegative prior to LTx had a 4.95-fold (95% CI = 1.06-23.15, P = 0.042) higher risk to develop PTLD. By performing chi-squared test and Kaplan-Meier analysis, we found that there were no differences in the EBV status before LTx (P = 0.337) nor in the incidence of PTLD (P = 0.823) among the older and younger CsA-treated patients. This suggests that the higher cancer risk in the younger CsA-treated patients was not due to a negative EBV status before LTx. In addition, the rise of PTLD in CsA-treated patients transplanted from 2005 could not be explained by the influence of EBV serology status prior to LTx, because the EBV serology did not differ between the CsA-treated and TAC-treated patients who underwent transplantation in this recent LTx period (P = 0.362). The risk for nonmelanoma skin cancer appeared to be 3.08-fold (95% CI = 1.31-7.24, P = 0.01) higher in patients who were older than 50 years. For the development of the other cancer types summed in Table 2, we found that the only independent risk factor appeared to be CsA treatment (hazard ratio = 4.79, 95% CI = 1.01-22.66, P = 0.048).

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

In this study, we showed that CsA in comparison to TAC treatment is the most important risk factor for de novo malignancy after LTx. This higher cancer risk was, however, not observed in all CsA-treated patients, but CsA specifically enhanced development of de novo cancer in patients who underwent LTx in more recent years (2005-2007) and in younger patients (≤50 years of age). In addition, we showed that CsA treatment particularly resulted in more aggressive types of cancer compared to TAC, with a 1-year survival rate <30%. Interestingly, the incidence of de novo malignancies after LTx in our center considerably decreased over time after the first LTx period (1989-1992), but increased again in liver graft recipients who underwent transplantation in the most recent years (2005-2007).

The high incidence of de novo cancer in patients who underwent transplantation in the earliest LTx period might be explained by the ATG induction treatment in 16 of 25 LTx patients, because ATG induction therapy has been described to be associated with posttransplantation cancer.17, 19, 35, 36 However, because the numbers of ATG-treated patients were small, this association could not be confirmed in our cohort. Thereafter, since 2005, the trend of the decreasing de novo cancer incidence was noticeably turned into an increasing incidence, which was significantly related to CsA treatment. From January 2005, CsA dosing based on the conventional C0 level monitoring was replaced by dosing based on C2 level monitoring in all de novo LTx patients. Because the CsA blood level monitoring was the only major change in the CsA treatment in the recent LTx period, our data suggest that the C2 monitoring strategy was the reason for the increased early de novo cancer risk.

We were not able to compare head to head the incidence of de novo cancer risk in patients treated with either CsA-C0 and CsA-C2, because these dosing strategies were used in different time periods in our center. Therefore, we compared patients treated with CsA-C0 and CsA-C2, respectively, with TAC-treated patients who underwent transplantation within the same time period, because TAC dosing strategy has not changed over time. This comparison corroborates that only in the period of C2-based CsA dosing, patients treated with CsA had a higher risk of cancer compared to TAC-treated patients (Fig. 4).

Our finding that all malignancies in the CsA-C2-treated patients occurred within the first year after LTx suggests that CsA-C2-treated patients were exposed to a high dose during the early post-LTx phase. We found that CsA-C2-treated patients received a significantly higher dose, but only during the induction phase after LTx. However, a far better correlate of drug exposure is the incidence of acute rejection, because CsA dose is dependent on drug absorption. Interestingly, we found a lower incidence of acute rejection during the first year after LTx in CsA-treated patients who underwent transplantation from 2005 with C2 based monitoring (9.6%), compared to CsA-treated patients who underwent transplantation before 2005 with C0-based monitoring (35.2%). Furthermore, to exclude a bias from the different LTx periods, the acute rejection rates were compared within the same LTx period (2005-2007) between CsA-treated and TAC-treated patients, showing that within this period, the CsA group had a lower acute rejection incidence (9.6% in the CsA group versus 27.5% in the TAC group). Corroborating our observations, several studies have shown that CsA-C2 monitoring is associated with a lower acute rejection rate compared to CsA-C0.25, 26, 37 Collectively, these observations indicate that dosing of CsA based on CsA-C2 monitoring results in more robust immunosuppression compared to treatment based on CsA-C0, which consequently might impair immunosurveillance of developing malignancies. Regarding to the types of cancer arising in the CsA-treated patients who underwent transplantation in recent years, we found a relative rise in nonmelanoma skin cancer and PTLD. The higher PTLD risk was not a consequence of negative EBV serology prior to LTx, because EBV status was not different in the CsA-treated compared to TAC-treated patients who underwent transplantation in this recent LTx period.

Compared to TAC, CsA treatment was only unfavorable in younger patients with regards to de novo cancer development after LTx. This higher cancer risk in the younger CsA-treated patients was not attributable to EBV seronegativity prior to LTx, because no difference in EBV status was present in the different age groups. Moreover, the younger CsA-treated patients did not develop more PTLD.

Until now, it has been stated that the basis for the higher cancer incidence after LTx is the long-term immunosuppressive state, and not the specific immunosuppressive protocol that is used.9 Our data showed that CsA treatment is a strong determinant for de novo cancer development after LTx. First, we found that comparison of CsA treatment to TAC treatment resulted in a higher de novo cancer risk in younger patients, of which is known that they have a low cancer risk.8, 13, 16 Second, we found that dosing strategy does influence cancer development, because the intense dosing in the CsA-C2 treatment protocol resulted in a more rapid development of de novo malignancy within the first year after LTx. Both observations suggest that not only the duration of immunosuppressive state, but also the immunosuppressive treatment protocol influences the development of cancer. In addition, we observed that CsA treatment resulted in more aggressive types of cancer.

There is evidence that CsA has procarcinogenic effects that are independent of its effect on the host immune response, including the inhibition of DNA repair, stimulation of tumor growth factor-β and/or vascular endothelial growth factor synthesis, which diminishes clearance of altered cells and transforms cancer cells into aggressive cancer cells.15, 38, 39 Although TAC has been reported to have similar prometastatic mechanism described for CsA,38, 40 there is also evidence that TAC has antimetastatic effects not seen with CsA.41 Our observation that CsA treatment resulted in more diverse and aggressive types of cancer supports the notion that CsA may promote a higher level of carcinogenesis than TAC.

In contrast to other studies,5, 42 sex was not significantly associated with a higher cancer risk. Furthermore, we did not find pretransplant use of immunosuppressive drugs to be correlated with a higher cancer risk, because all autoimmune patients with cirrhosis received immunosuppressive drugs prior to LTx, which did not result in a higher de novo cancer risk.13 In agreement with other reports, we found that IL2RA administered as an induction therapy did not result in a higher incidence of cancer.19, 42, 43

Overall, we found a 2.2-fold higher risk for de novo cancer development in the entire cohort of liver graft recipients compared to the age-matched and sex-matched Dutch population. Although this risk is high, it is within the lower part of the range of SIRs reported by other studies, which range from 2.1-4.3.5, 6 The highest incidence of cancer was found for nonmelanoma skin cancer and PTLD, which is consistent with other reports.5, 6, 9, 13, 14, 19, 20, 42, 44-47 It was not surprising that we found a relation between EBV seronegativity prior to LTx and PTLD development, because PTLD is mainly related to EBV infection,48 and transmission of EBV is frequent when the naïve transplant recipient receives an organ from an EBV-infected donor. Because a high frequency of the donor population (>85%) is EBV seropositive, mismatching for EBV is common.49 However, as discussed previously, EBV-negativity before LTx did not explain the higher cancer risk induced by CsA treatment. Considering the survival, it was remarkable that we did not find a higher mortality rate for recipients who developed PTLD compared to cancer-free recipients, whereas various studies showed a low survival for patients with PTLD (40.8% at 5 years post-LTx).44, 50, 51 Our observed high survival rate of PTLD patients may be explained by successful treatment of PTLD by reduction of immunosuppressive drugs, administration of rituximab, and systemic chemotherapy. Indeed, disease management with adequate immunosuppression reduction is associated with improved survival in patients with PTLD.14

In summary, we have shown that CsA treatment in comparison to TAC treatment is the most significant risk factor for development of de novo cancer in LTx patients. However, this effect was confined to patients treated with CsA-C2 and to patients of 50 years and younger. Furthermore, CsA treatment gave rise to more aggressive cancer types compared to TAC treatment. So far, to our knowledge, our study is the first study suggesting that CsA treatment based on C2 monitoring in de novo LTx patients and CsA treatment in younger patients are associated with a significantly higher early de novo cancer risk compared to TAC. The limitations of our study include the retrospective study design, the limited number of patients, and the lack of opportunity to compare C0 and C2 treatment within the same time period. So, larger randomized trials with long-term follow-up are needed to confirm this hypothesis. The ongoing international randomized trial comparing TAC with CsA-C2 in hepatitis C-related liver transplant patients might provide support to our finding. This study highlights the importance of re-evaluation and optimization of currently used immunosuppressive regimens.

Acknowledgements

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We thank S.N. de Visser and M.E. Azimpour Gilani for their great contribution to the Rotterdam LTx database.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES