Increased Incidence and Mortality Associated With Skin Cancers After Cardiac Transplant

Authors


Murad Alam, m-alam@northwestern.edu

Abstract

Skin cancer incidence has been shown to be increased in the context of transplant-associated immunosuppression. There is, however, limited information specifically about the incidence of skin cancer after cardiac transplantation in the United States. A 10-year retrospective cohort study of 6271 heart transplants at 32 US transplant centers revealed increased postprocedure incidence of nonmelanoma and melanoma skin cancers, especially cutaneous squamous cell carcinoma, for which the incidence increased from 4- to 30-fold compared to the age and gender equivalent general population. Incidence of skin cancer in this study was consistent with prior single-center data regarding cardiac transplant patients. Comparison of all-cause mortality statistics for patients with basal cell carcinoma, squamous cell carcinoma and melanoma, respectively, demonstrated increased mortality associated with melanoma. Skin cancer screening and prophylaxis may be of some utility in reducing morbidity and mortality in cardiac transplant patients.

Abbreviations: 
CTRD

Cardiac Transplant Research Database

CDC

Centers for Disease Control

BCC

basal cell carcinoma

SCC

squamous cell carcinoma

SEER

Surveillance Epidemiology and End Results

Introduction

Transplant patients may develop many concurrent primary skin cancers that are unusually aggressive (1–4). Among the most common transplant-related skin cancers are cutaneous squamous cell carcinomas, with serious illness or death resulting if one of these lesions metastasizes to distant sites. Overall, cardiac transplant recipients are believed to be at greatest risk among solid organ transplant recipients for transplant-associated skin cancers due to their higher level of immunosuppression (5–8). There may be other causes for this elevated risk in cardiac transplant patients, but these are not well understood. Regardless of the etiology, correct estimation of this increased risk is important because it can help guide preventive and treatment strategies that can reduce the incidence and the impact of skin cancer in cardiac transplant patients.

Our understanding of the risk of cutaneous malignancy in transplant patients in general, and heart transplant patients in particular, has gradually evolved. Initial retrospective case series from individual centers suggested a dramatically increased incidence of skin cancers in cardiac transplant recipients (9,10). This elevated risk was confirmed by recent long-term retrospective single-center cohorts from the United States and Europe (11–13), as well as a two-center cohort from Europe (14). However, more recent anecdotal reports suggest that the incidence of transplant-associated skin cancer may be lower than expected, and the impact on cumulative morbidity and mortality may not be as dire as previously believed.

The purpose of this study was to assess the incidence, risk factors and likelihood of mortality associated with skin cancer after cardiac transplant via a comprehensive analysis of 10 years of prospectively acquired patient information on patients from 32 centers associated with the Cardiac Transplant Research Database (CTRD). We believe this methodologic approach was novel because it did not rely on data from a single center, which may in some cases represent an unusually high risk transplant-associated malignancy, but rather assessed the risk across many centers. Additionally, the CTRD database, unlike other available heart transplant databases, specifically elicited information on skin cancer risk for the duration of the study. As such, the risk of ascertainment bias, or inadvertent failure to report apparently ‘minor’ malignancies that were not specifically queried, would likely have been lower and the resulting data more robust. One clinically relevant purpose of this study was to determine whether cumulative risk of skin cancer after heart transplant was as high as originally believed or somewhat lower, as recently suspected. Another clinically relevant purpose of this study was to clarify the epidemiology of skin cancer in cardiac transplant patients; specifically, by determining the age and gender groups most at risk, and the specific types of skin cancer they were likely to develop, prevention and treatment strategies may be more effectively implemented. Finally, a third clinically relevant purpose of this study was to provide epidemiologic data to guide translational studies aimed at better understanding the reasons for high skin cancer risk in heart transplant patients.

Historically, large, multicenter studies of the type we report have not been performed. Virtually all prior studies are from one or two centers, and examine a very small subset of cardiac transplant patients in the United States. Beyond the limitation of small numbers, this is problematic for a variety of reasons: potential differences in local transplant protocols; local ethnic variation in median Fitzpatrick skin type (and hence skin cancer risk) among transplant subjects; and the known association of skin cancer risk with geographic latitude. As a result, there is a need for a study such as the one we report to confirm or refute early single-center data regarding both the cumulative incidence and demographic distribution of skin cancer in cardiac transplant patients.

Materials and Methods

Description of the CTRD

The CTRD is a 32-center registry of US heart transplant patients from January 1, 1990 to December 31, 2008. Since 1990, the intake forms for this database have included baseline information on pretreatment skin cancer status; additionally, newly diagnosed posttransplant tumors, including skin cancers, are reported prospectively using a dedicated form.

Data extraction technique

Skin cancer incidence data were extracted from the CTRD for the period from July 1, 1993 to December 31, 2003. Data extraction was begun from 1993, not 1990, because postmalignancy transplant data were not collected until 1993. A total of 6271 heart transplants were included. Pretransplant skin cancer status was obtained from initial database submission forms, and posttreatment skin cancer incidence from malignancy report forms submitted at the time of diagnosis. Demographic information was extracted from initial data forms, with immunosuppression status estimated from periodic annual update forms sent in during the posttreatment period.

Evaluation of missing data and underreporting

Prima facie examination of the skin cancer incidence data revealed wide divergence in reporting rate from different transplant centers. Underreporting was presumptively ascribed to the ubiquity of skin cancers, the fact that these skin cancers were often viewed as less serious malignancies, the treatment of skin cancers by local dermatologists unaffiliated with the transplant service, and the difficulty inherent in separately counting each of several discrete primary lesions that may have occurred per person. Underreporting appeared to be confirmed by the wide disparity in the incidence of skin cancer at various centers. Specifically, the proportion of transplant patients developing at least one posttransplant skin cancer was more than fivefold higher at the centers reporting the highest rates (top quartile) than the centers reporting the lowest rates (bottom quartile).

Statistical methods

Reporting of nonmelanoma skin cancer incidence:  Significantly, in the United States there is no national mandate for reporting of the incidence of nonmelanoma skin cancer, including basal cell carcinoma (BCC) and cutaneous squamous cell carcinoma (SCC). Incident cases are not recorded by the Centers for Disease Control and Prevention (CDC), and there is no national registry of such tumors. As a consequence, population-based incidence data are sparse. Population-based incidence estimates have been attempted in several cases at the state level, and these are the best estimates available. For comparability, such estimates (including those from New Mexico and New Hampshire, discussed later) are adjusted for latitude, given that latitude is an independent factor for the absolute risk of BCC and SCC.

Comparing CTRD skin cancer incidence and mortality to general population and prior studies with cardiac transplant patients, respectively:  Apart from obtaining incidence and mortality data from the CTRD, as described earlier in the Methods section, we compared these data to both general population incidence and incidence reported in prior single and two-center studies of cardiac transplants.

Comparison to the general population was not possible in a single analysis because population-based incidence data for the entire United States are not available (discussed earlier), but two separate comparisons were made to the population-based data from New Mexico and New Hampshire, respectively. These two comparison data sets were selected because they are relatively cotemporaneous to the cardiac transplant data set, they offer detailed results by age and gender that can be compared to the cardiac transplant results, and they represent data from two widely separated latitudes in the United States.

In comparing CTRD incidence to New Mexico and New Hampshire incidence, incidence rates were adjusted for latitude, and rates were compared by tumor type, gender and age.

Comparison of the CTRD incidence to data from the best earlier single-center studies was achieved by comparison to cardiac transplant incidence rates reported at the Mayo Clinic in Rochester, MN, USA. Rates were compared by tumor type, and cumulative incidence rates were provided for 5, 10 and 15 years after transplantation.

The following sections detail methodologies for specific comparisons that require explanation and clarification beyond those aforementioned. Note that with one exception (immunosuppression, as discussed later), all analyses in the paper analyzed the event ‘first occurrence of skin cancer’, and multiple cancers per person were not counted.

Survival analyses:  Survival after skin cancer diagnosis and freedom from skin cancer after transplant were estimated using the Kaplan–Meier actuarial method (SAS software with user written macros (SAS Institute Inc., Cary, North Carolina)). A parametric method was used to resolve the number of phases of instantaneous risk (hazard function) and to estimate shaping parameters (15). (For additional details, see http://www.clevelandclinic.org/heartcenter/hazard.) Parametric survival analyses in the hazard domain were performed to determine risk factors for freedom from skin cancer after transplant and for the repeated incidence of new skin cancer diagnoses.

Subset analyses restricted by subject race:  Notably, non-Caucasian recipients had almost no incidence of skin cancer, and thus no discernible risk factors for such cancer. For this reason, all our subset analyses of risk for skin cancer were on Caucasian recipients only (total n: 6271; Caucasian n: 5320; non-Caucasian n: 948). The only exception to this was assessment of duration of survival after getting a skin cancer; this was for the entire population, both white and nonwhite.

Assessment of impact of level of immunosuppression on cancer risk:  Given the limitations of the data, this analysis was performed on a subset of the patient base. We analyzed the outcomes of patients transplanted after 1995, for whom time zero was 1-year follow-up, and modeled all of the skin cancers as a repeated event.

Methods for comparing CTRD incidence of BCC or SCC to literature:  The methods used for comparing the incidence of SCC from this study to incidence in nontransplant patients described in the literature (16,17) was as follows: From the literature reference articles, within a 1-year time frame, individual persons who were newly diagnosed with a BCC or SCC were counted and used as the numerator of the incidence calculation. No person was counted more than once. The age of each person was calculated, and the population of the geographic area under study was taken from the census records to determine a denominator. At this point a straightforward incidence ratio was calculated for 1 year of follow-up.

In the CTRD data set, each patient was followed from transplant to first occurrence of a BCC or SCC. Patients with a pretransplant history of ‘skin cancer’ were deleted from the analysis. Each patient's follow-up interval was segmented according to their age. Each patients ‘exposure’ time to risk of skin cancer at a certain age could then be calculated by summing the total number of events at a certain age divided by the total years of exposure for that same age. This also allowed for a predictive model to be created for the probability of skin cancer with covariables age, gender, interaction of age and gender (since the known risk for skin cancer increases more quickly for males than for females as aging occurs), and the latitude of the transplant center (we did not have data for all individuals place of residence during the follow-up period, so the location of the transplant center was used). This model allowed for a predictive adjustment of the incidence rate after transplant for comparison to literature data that may have been geographically and demographically dissimilar from the transplant population described herein.

For comparison of CTRD BCC and SCC incidence to the prior literature on cardiac transplant-associated skin cancer, specifically the Brewer et al. paper (13), a similar methodology was used. The CTRD estimates are for the ‘white’ patient group; while we find no mention of ethnicity in the Brewer paper, at the Mayo clinic, approximately 90% of patients are expected to be white based on geographic location.

Methods for estimating increase in mortality risk associated with incidence of melanoma in CTRD:  By analyzing our data with posttransplant skin cancer as a time-varying covariable, we asked the question ‘Does a late skin cancer increase the hazard for death after its occurrence?’ Creating the variable in an ordered manner (i.e. a subsequent SCC makes a previous BCC of no interest, and a subsequent melanoma makes a previous BCC or SCC of no interest from that time forward), without adjustment for all of the many well-known risk factors for death, the increase in relative risk for death (hazard ratio) associated with a melanoma be computed. A similar statistic be computed if all of the other known risk factors for mortality are included multivariably in the model.

Methods for comparing actual number of CTRD melanomas to expected number in general population:  CTRD transplants between July 1, 1993 and December 31, 2003 who were ‘white’ and had sex and age data (N = 5316). Patients were censored at death, retransplant, transfer from care to a non-CTRD institution (this is standard censoring practice for all CTRD analyses) or at time of first melanoma diagnosis. Using data from the SEER (Surveillance Epidemiology and End Results) database for melanoma incidence specified by sex, age groups and year of observation, the cumulative incidence of melanoma in this transplant population was predicted.

Results

Among the 6271 cardiac transplant cases analyzed, there were reportedly 228 (3.6%) instances of BCC, 289 (4.6%) of SCC, 22 (0.003%) of melanoma and 6 (0.001%) of other skin cancers (Table 1). Freedom from any skin cancer was 98.2% (95% CI: 97.8–98.6%). One year after transplant, falling to 85% (95% CI: 84–87%) 10 years after transplant (Figure 1). White patients were much more susceptible to skin cancer than nonwhite patients (p < 0.0001), with 10-year freedom from skin cancer at 83% (95% CI: 81–85%) in Caucasian patients, and 99.2% (95% CI: 97.8–99.7%) in non-Caucasian patients. By type of cutaneous malignancy, freedom from skin cancer 10 years after transplant was 99.1% (95% CI: 98.5–99.5%) for melanoma, 92% (95% CI: 91–94%) for BCC and 89.1% (95% CI: 87.5–90.5%) for SCC.

Table 1.  Demographic composition of cohort, and incidence of posttransplant skin cancer (CTRD, 1993–2003)
 n%
Total number of transplants6271100
Ethnicity of transplant recipients
 White532085
 Nonwhite 95115
  African-American 77012
  Hispanic  570.9
  Middle-eastern/Arabian  390.6
  Indian subcontinent  250.4
  Asian/Pacific Islander  390.6
  Native American/Alaskan native   70.1
  Unknown   30.05
Sex of transplant recipients (white only)
 Male420279
 Female111321
 Unknown   30.06
Etiology of cardiac disease in recipients (white only)
 Ischemic294155
 Idiopathic171632
 Congenital 1352.5
 Acquired valvular 1322.5
 Hypertrophic  841.6
 Restrictive  751.4
 Myocarditis  511.0
 Postpartum  350.7
 Alcoholic  150.3
 Other 1352.5
 Unknown   10.2
Patients with ≥1 posttreatment BCC(s) 2283.6
Patients with ≥1 posttreatment SCC(s) 2894.6
Patients with ≥1 posttreatment melanoma(s)  220.003
Patients with ≥1 posttreatment ‘other’ skin cancer(s)   60.001
Patients with ≥1 posttreatment any skin cancer 4407
Figure 1.

Cumulative incidence of first occurrence of BCC, SCC and melanoma (The three curves are not mutually exclusive and cannot be compared statistically. For each event the number of patients still at risk is given below the x-axis at yearly intervals. Error bars are 95% confidence intervals).

The shape of the most parsimonious parametric model for any skin cancer after transplant was a constant hazard. Parametric multiple regression analysis in the hazard domain revealed several associations with elevated risk of cardiac transplant-associated skin cancer (Table 2). Older recipients (age 60 vs. 30) had a relative risk of skin cancer of 7.6 (p < 0.0001); 10 years after transplant, 74% of those aged 60 and 86% of those aged 30 were free of skin cancer. Lower latitude (32nd parallel vs. 42nd parallel) had a relative risk of skin cancer of 1.2 (p = 0.03); those living in areas such as Texas and Florida had a 10-year skin cancer free likelihood of 77% (95% CI: 73–81%) compared to a likelihood of 90% (95% CI: 87–93%) in patients living in northern states. Pretransplantation history of skin cancer was associated with a 2.0 relative risk of skin cancer (p = 0.001); two-thirds of patients with skin cancers before their transplants developed a new skin cancer within 10 years of their transplant. Number of posttransplant skin cancers prior to 1-year follow-up was also associated with subsequent risk of posttransplant skin cancers (p < 0.0001), with a relative risk of 2.4 (for 1 vs. 0 lesions within the 1-year window). History of smoking within 6 months of transplantation had a 1.5 relative risk of posttransplant skin cancer (p = 0.01). Male recipients had a relative risk of posttransplant skin cancer of 2.1; 10 years after transplantation, 81% (95% CI: 79–83%) of men and 90% (95% CI: 87–93%) of women were free of skin cancer. The predicted probability of first posttransplant skin cancer within 5 years of transplantation varied by age and sex: for patients age 35 at transplantation, the predicted probabilities were 3% and 2%, respectively, for men and women; for transplants at age 65, the predicted probabilities were 22% and 13%. Age-adjusted BCC and SCC incidence for men and women of different age groups is displayed in Tables 3A and B. Many patients developed multiple skin cancers, as seen in Table 4.

Table 2.  Risk factors for occurrence of any skin cancer by multivariate analysis (white recipients)
VariableRelative risk (unadjusted)1p-Value (unadjusted)1Relative risk (adjusted)2p-Value (adjusted)2
  1. 1Univariable.

  2. 2Multivariable.

  3. 3Pretreatment.

 (RR [95% CI]) (RR) 
Male recipient2.1 (1.5–2.8)<0.00011.8 (age = 56)0.01
Older recipient age7.6 (5.2–11) <0.0001 0.07
 (age 30 vs. 60) 9.2 (male) 
Interaction: male recipient with age__2.1 (female) (age 30 vs. 60)0.002
Lower Latitude (transplant center)1.65(1.3–2.2)0.00021.2 (latitude 32 vs. 42)0.03
Pre-Tx3 history of skin cancer3.6 (1.8–7.0)0.00022.90.001
History of smoking within 6 months of list1.2 (0.9–1.7)0.21.50.01
Table 3.  Raw incidence of (A) BCC and (B) SCC by age and gender
(A)
Basal cell carcinoma
 
AgeStandard US population proportionIncidence/100 000
ActualIncidence adjusted for US population
Females
 <200.2868600
 20–290.1310100
 30–390.15181482.09 73.186
 40–490.15397783.96120.706
 50–590.11117696.10 77.385
 60–690.07306707.57 51.695
 70–790.058771426.34 83.826
 80–890.02760No data
 >900.00575No data
Overall 677406.798
Males
 <200.2868600
 20–290.13101221.52 29.021
 30–390.15181392.47 59.581
 40–490.15397376.36 57.947
 50–590.11117896.44 99.658
 60–690.073061847.56134.983
 70–790.058771802.81105.951
 80–890.0276000
 >900.00575No data
Overall 1194487.141
(B)
Squamous cell carcinoma
 
AgeStandard US population proportionIncidence/100 000
ActualIncidence adjusted for US population
Females
 <200.2868600
 20–290.1310100
 30–390.15181238.74 36.243
 40–490.15397622.38 95.827
 50–590.11117504.43 56.078
 60–690.07306615.49 44.968
 70–790.058771415.15 83.168
 80–890.02760No data
 >900.00575No data
Overall 532316.284
Males
 <200.2868600
 20–290.1310100
 30–390.15181258.39 39.227
 40–490.15397564.7 86.947
 50–590.11117950.02105.613
 60–690.073062796.44204.308
 70–790.058773093.15181.784
 80–890.0276000
 >900.00575No data
Overall 1634617.879
Table 4.  Number of reported skin cancers for an individual patient (by number and by type of cancer, white patients only)
Number of reported skin cancers #Number of patients
AllBasal cellSquamous cellOthersMelanoma
N%N%N%N%N%
 0488492509596503295531499.9539899.6
 1 2574.8 1793.4 1863.5   6 0.1  17 0.3
 2  901.7  300.6  531.0     4 0.08
 3  460.9  140.3  220.4     1 0.02
 4  140.3   10.02  100.2    
 5  110.2   10.02   80.1    
 6   60.1     40.08    
 7   50.09     50.09    
 8   30.06        
 9   20.04        
10   10.02        
12   10.02        
Total53201005320100532010053201005320100

Level of immunosuppression was also related to the likelihood of posttransplant skin cancer. Relative risk of skin cancer was 1.5 for higher dosages of cyclosporine measured at the 1-year follow-up (6 vs. 2 mg/kg/d, p = 0.01), 1.4 for higher dosages of azathioprine (2.5 vs. 1 mg/kg/d, measured at 1-year follow-up, p = 0.007) and 1.4 for higher dosages of mycophenolate mofetil (40 vs. 10 mg/kg/d, at one-year follow-up, p = 0.04).

All-cause mortality after posttransplant skin cancer varied by type of skin cancer. For BCC, the 83% of patients were alive 5 years (95% CI: 73–90%) after diagnosis of skin cancer, and 78% were alive at 8 years (95% CI: 54–86%) (Figure 2). For SCC, 5-year survival was 80% (95% CI: 74–86%), and 10-year was 55% (95% CI: 35–73%). For melanoma, 3-year survival was 50% (95% CI: 27–73%), and given the small numbers of patients with this malignancy, further survival data were not computed.

Figure 2.

Survival (%) after diagnosis of BCC, SCC and melanoma (The three curves are ‘not’ mutually exclusive and therefore cannot be compared statistically. Death after first BCC diagnosis (15 deaths): If SCC or melanoma is diagnosed before BCC, the patient is not included. Death after first SCC diagnosis (49 deaths): If melanoma is diagnosed before SCC, the patient is not included. Death after first melanoma diagnosis (9 deaths). The mean ages at first diagnosis are 60.0 ± 8.12 years (BCC), 62.2 ± 6.88 years (BCC) and 57.4 ± 11.7 years (melanoma). The error bars are 95% confidence limits. Patients who had a pretransplant skin cancer are not excluded.

Comparison with population-based cutaneous SCC incidence data from New Mexico and New Hampshire, respectively, revealed that incidence was much higher in the CTRD population, when this was adjusted for age, gender and geographic latitude (Tables 5A and B, 6A and B). Specifically, the relative risk ranged from 30.5 for women under the age of 35 in New Hampshire to 4.3 for men aged 45–54 in New Mexico. For men and women older than 65, the relative risk of SCC was not markedly greater in the comparison populations than the CTRD. Unlike with SCC, for BCC, the relative risk in the CTRD cohort was not consistently greater than in the two population-based samples. For the youngest patients under age 35, the risk of BCC was higher in the CTRD group, but this difference disappeared in older patients.

Table 5.  Comparison of CTRD incidence model to population-based data from New Mexico, Athas et al. for (A) BCC and (B) SCC [the latitude of 35 was used for Albuquerque, NM; Prediction was made at each age (whole year) within the interval (10 points) and then averaged]
(A)
Basal cell skin cancer incidence (New Mexico; Athas et al.)
 
AgeIncidence/100 000
Athas et al. 1998–1999Mean model prediction1Prediction at min and max age within group95% CI of prediction
Females
 25–3484.8463.4423.0–505.9177.0–1222.9
 35–44362.1565.2516.0–617.1291.5–1103.0
 45–54572.5689.5629.5–752.7447.1–1065.5
 55–64926.4841.0767.8–918.2560.2–1263.9
 65–741886.41025.9936.6–1120.1559.6–1891.8
 75+2199.81226.11142.6–1313.1515.7–2930.1
Males
 25–3463.5242.0183.6–310.2134.5–436.6
 35–44339.8433.5328.9–555.7285.9–658.9
 45–541008.5776.5589.0–995.3599.1–1008.0
 55–641756.01390.81055.1–1782.81179.2–1640.5
 65–743530.52491.11889.8–3193.31995.6–3113.4
 75+5651.74188.03384.9–5090.32950.5–5952.6
(B)
Squamous cell skin cancer incidence (New Mexico; Athas et al.)
 
AgeIncidence/100 000
Athas et al. 1998–1999Mean model predictionPrediction at min and max age within group95% CI of prediction
  1. 1Includes SCC, BCC, melanoma and others.

Females
 25–3417.0217.2181.8–256.662.6–763.6
 35–4447.1318.6266.6–376.2135.2–759.5
 45–54143.1467.2390.9–551.7273.4–803.0
 55–64249.6685.2573.2–809.1440.7–1066.0
 65–74603.91004.8840.7–1186.5517.0–1969.6
 75+1007.31415.01232.8–1611.7528.9–3815.25
Males
 25–3412.7174.5119.6–242.3100.3–304.4
 35–4448.5382.6262.1–531.1257.8–568.9
 45–54193.6838.7574.5–1164.2655.8–1074.4
 55–64590.51838.61259.3–2552.11586.7–2130.9
 65–741486.24030.42760.4–5594.43335.2–4875.0
 75+2765.78093.56051.1–10481.25961.1–11001.3
Table 6.  Comparison of CTRD incidence model to population-based data from New Hampshire, Karagas et al. for (A) BCC and (B) SCC [the latitude of 44 was used for New Hampshire; Prediction was made at each age (whole year) within the interval (10 points) and then averaged]
(A)
Basal cell skin cancer incidence (New Hampshire; Karagas et al.)
 
AgeIncidence / 100 000
Karagas et al. 1993–1994Mean model prediction1Prediction at min and max age within group95% CI of prediction
Females
 <3510.4316.85269.00–369.69109.46–939.44
 35–44113.7413.06377.11–450.97210.40–815.97
 45–54244.6503.87460.02–550.10319.69–795.71
 55–64388.5614.64561.15–671.04398.81–948.23
 65–74729.0749.76684.51–818.56400.71–1410.72
 75+1090.8896.06834.99–959.60370.94–2175.11
Males
 <357.1148.1089.22–226.7278.77–280.27
 35–44113.1316.8240.33–406.10206.29–487.43
 45–54308.0567.44430.47–727.38422.94–762.12
 55–64682.31016.38771.04–1302.86805.66–1282.26
 65–741513.21820.51381.06–2333.641361.67–2436.24
 75+2762.13060.632473.7–3719.992038.84–4599.79
(B)
Squamous cell skin cancer incidence (New Hampshire; Karagas et al.)
 
AgeIncidence/100 000
Karagas et al. 1993–1994Mean model prediction1Prediction at min and max age within group95% CI of prediction
  1. 1Includes SCC, BCC, melanoma and others.

Females
 <350.4149.27107.98–199.2438.64–598.95
 35–448.2247.43207.02–292.19104.37–593.11
 45–5425.1362.85303.59–428.48209.78–630.96
 55–6455.5532.12445.21–628.36336.38–842.36
 65–74170.5780.34652.88–921.48395.92–1551.02
 75+361.01098.95957.44–1251.71406.29–2995.11
Males
 <350.3108.0153.60–188.1660.15–195.04
 35–443.1297.15203.51–412.47199.19–444.06
 45–5435.0651.39446.14–904.17497.53–853.71
 55–64142.41427.91977.99–1982.031166.37–1748.19
 65–74540.63130.122143.85–4344.792437.07–4023.30
 75+1239.76285.724699.53–8140.604401.63–8985.08

The predicted incidence of melanoma in a similar general population (see Statistical Methods) was 10.2. The actual number of melanomas in this patient population was 22. The p-value of observed versus expected by chi-square test is p = 0.0002.

Having a melanoma in the CTRD increased the relative risk for death (hazard ratio) by a factor of 5. If all of the other known risk factors for mortality were included multivariably in the model, the hazard for death increased by a factor of 6 for a melanoma.

Comparison of the cumulative incidence of skin cancer among the CTRD cohort with single-center data on cardiac transplant recipients from the Mayo Clinic as measured by Brewer et al. is presented in Table 7. At 5 years after transplant, Mayo Clinic patients were approximately twice as likely as CTRD patients to have developed BCC and SCC (notably, the Mayo Clinic is a CTRD member). However, the cumulative incidence rates reported by Brewer were markedly lower, approximately half as high at 5 years, than the highest institutional incidence rates within the CTRD.

Table 7.  Percentage of patients diagnosed with skin cancer in CTRD compared to Mayo Clinic (Brewer et al. data) at 5, 10 and 15 years after transplant, and by type of cancer (numbers in parentheses are extrapolations from a single phase hazard model for comparison with the Brewer paper)
Years after transplantPercentage of patients who have been diagnosed with a skin cancer of the specified type
CTRDHighest institutional incidence within CTRDBrewer paper (Mayo Clinic)
  1. 1Includes SCC, BCC, melanoma and others.

SCC
 5 years 7.4%26%15.4%
 10 years12.7%37%32.3%
 15 years(23%)(59%)38.2%
  estimate estimate 
BCC
 5 years 5.8%20%10.3%
 10 years 8.6%34%19.2%
 15 years(14%)(47%)31.6%
  estimate estimate 
Any skin cancer1
 5 years11.2%31%20.4%
 10 years17.2%48%37.5%
 15 years(29%)(70%)46.4%
  estimate estimate 
Age range18.0–76.6 0.1–73.3

Discussion

This US database-derived investigation of the epidemiology of skin cancer in cardiac transplant recipients corroborates, clarifies and corrects trends detected by smaller, retrospective, and non-US trials. Overall, the risk of skin cancers, especially cutaneous SCC was found to be much higher for transplant recipients than nontransplant recipients of similar age and gender living at comparable latitudes; however, the risk was lower than suggested by prior single-center studies of heart transplant patients.

From 1993 through 2003, there were 24 785 heart transplants done in the United States. Although our results do not encompass all cardiac transplant cases in the United States, our 6271 patients constitute approximately 25% of such cases over a 10-year period. As such, this study contributes to the literature by providing the only large-scale epidemiologic review of skin cancer risk in US transplant patients.

The results of this study are clinically significant and may change clinical practice. The reasons for this include that the following: (1) This is the only multicenter study of its type, and as such provides detailed skin cancer risk information about cardiac transplant recipients that is not markedly skewed by the peculiarities of particular centers (e.g. limited to a single latitude, based on a cohort with restricted ethnic and Fitzpatrick skin type variability, subject to center-specific transplant procedures and protocols). Such risk factor data can help direct preventive and treatment strategies to the most at-risk individuals. (2) The stratification of risk by age, gender and latitude obtained in this study made possible precise comparison of skin cancer risk in our cohort to the general population risk and the risk reported by earlier single center heart transplant studies. Interestingly, these comparisons revealed that cumulative risk for skin cancer is much higher in heart transplant patients than in the general population but lower than as reported by the most pessimistic single center studies. The more moderate risk levels we found may help increase the effectiveness of interventions to treat skin cancer in heart transplant patients by reducing misallocation of resources to those heart transplants who, by virtue of age, gender or more northern latitudes, are at relatively lower risk. (3) The epidemiologic data obtained in this study may guide investigation into the etiology of heightened skin cancer risk in cardiac transplant patients. At present, the reasons why heart transplant patients get more skin cancers are not well understood, and studying the highest risk subsets, including those with fairer skin and hair and those residing in more equatorial latitudes, may be instructive in this regard.

Notably, precise comparison of our incidence data with population-based estimates from the literature is prevented by the fact that BCC and cutaneous SCC are not reportable tumors in the United States and the national population-based data are not available. However, adjusted comparisons were made to relatively recent population-based incidence data from New Mexico and New Hampshire, respectively (16,17). As noted in the Methods, latitude was among the variables used for adjustment. These comparisons (Tables 5A and B, 6A and B) show a cardiac transplant-associated marked increase in cutaneous SCC incidence at virtually all ages for men and women. We have also compared our data to nonpopulation-based incidence data from the recent single a recent single-center study of cardiac transplantation (Table 7).

Significantly, the cumulative skin cancer incidence in cardiac transplant recipients detected in the CTRD cohort was significantly lower than that in a recent single-center study from the Mayo Clinic. At 5 years after transplant, the incidence of BCC and SCC in the CTRD cohort was approximately half that of the comparison cohort, and this disparity widened at 10 years. Although the highest incidences reported at any CTRD institution were indeed markedly higher than in the Mayo–Brewer cohort, the mean incidence in CTRD was lower. Underreporting remains a risk, and the true underlying rates may be higher. This suggests that the association between immunosuppression and induction of skin cancers may also be greater than reported.

That being said, the burden of skin cancer in cardiac transplant recipients remains substantial. Even nonfatal skin cancers may result in substantial pain, disfigurement, loss of function and range of motion and time away from work. Moreover, advanced melanoma and metastatic cutaneous SCC are potentially fatal malignancies. In the CTRD cohort, the risk of melanoma was markedly increased compared to the SEER database, with almost twice as many melanomas as would be expected in a similar nontransplant population. Additionally, having a melanoma increased a given patient's risk of mortality by sixfold.

The limitations of this study are many. Data were obtained at 32 centers; by conservative estimate, these accounted for approximately one-third of cardiac transplants performed in the United States during the study period. Further, some centers were clearly likely to underreport skin cancers. Since patients with skin cancers are known to commonly have multiple serial primary skin tumors, and since such few patients with multiple lesions were reported, underreporting of second, third and subsequent skin cancers was likely particularly severe. While pretransplant education at most centers contributing data to the CTRD included recommendations to use sunscreen and practice sun avoidance after transplantation, the database did not elicit information regarding patient compliance with these instructions. Therefore, it was not possible to correlate sun protective behaviors with the incidence of posttransplant skin cancer.

While improved cancer risk assessment is generally useful prognostic information, for nonmelanoma and melanoma skin cancer, it additionally has clinical relevance because these malignancies are both highly preventable and treatable. Frequent skin examinations by a dermatologist can detect early cutaneous SCCs and melanomas. These can be cured by simple in-office destructive modalities, like electrodessication and curettage (SCC), or skin-only surgical excision (SCC and melanoma). Actinic keratoses, which are a precursor for cutaneous SCC, can be prevented with topical and systemic chemotherapeutic agents like retinoids (18), or destroyed with liquid nitrogen cryotherapy. Patients at high risk, or residing in southern latitudes, can also be directed to organ transplant skin cancer clinics, which are operated in many medical centers (19–23). Similarly, pretransplant counseling can be directed to patients most likely to eventually develop skin cancers (24). Modifying the frequency of routine skin examinations in cardiac transplant patients can thus help to match surveillance to risk, eliminating precancers and preventing serious malignancies.

Acknowledgments

Funding sources: Northwestern University and Cardiac Transplant Research Database research funds.

Disclosure

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.

Ancillary