Renal allograft recipients have a markedly increased incidence of nonmelanoma skin cancer (NMSC) compared with age-matched controls (Liddington et al, 1989;Glover et al, 1994). Tumors present at a younger age are more numerous, grow more rapidly, and metastasize earlier in the transplant group (Shiel et al, 1985;Liddington et al, 1989). Squamous cell carcinoma (SCC) behaves more aggressively than basal cell carcinoma (BCC) and is associated with increased mortality in the transplant population (Euvrard et al, 1995). Increased access to renal transplantation coupled with improved management has resulted in a progressive increase in the number of patients with functioning renal allografts worldwide. UK figures show that the number of kidney transplant patients with functioning allografts has increased from 11 700 in 1994 to 18 400 in 1999.1 Corresponding increases across Europe are estimated at 94 000 in 1994 to 147 000 in 1999.1 Consequently, there is a growing population of recipients who will develop NMSC. It is poorly understood why some patients develop early and/or multiple NMSC post-transplantation whilst other recipients appear protected. It is important to identify those individuals at greatest risk in order to develop more effective strategies for prevention and surveillance.
As in the general population, individual susceptibility to skin cancer following transplantation is likely to be determined by multiple and interacting factors including immunosuppression, ultraviolet light (UV) exposure, and genetic susceptibility. We and colleagues have previously shown that several clinical factors predispose to NMSC post-transplantation (Kelly et al, 1987;Roeger et al, 1992;Ramsay et al, 2000). These include duration of immunosuppression, age at transplantation, gender, occupation, smoking, and skin tanning in response to UV. The role of UV exposure is well established in cutaneous carcinogenesis, although the relationship between exposure and risk remains poorly understood. UV comprises an oxidative stress with the generation of reactive oxygen species (ROS) and consequent increased risk of DNA damage. This view is supported by data showing that UV-specific mutations in the tumor suppressor gene p53 have been identified in NMSC from both the general population and the renal transplant recipients (Ziegler et al, 1993;McGregor et al, 1997). We postulate that interindividual variation in ability to repair or modulate such damage is likely to influence susceptibility to NMSC. Indeed, in vitro studies have shown that reduced DNA repair capacity is associated with increased risk of BCC in the general nontransplant population (Wei et al, 1994).
The human glutathione S-transferase (GST) supergene family comprise a group of genes encoding enzymes involved in the detoxification of a variety of reactive and mutagenic compounds, including the products of UV-induced oxidative damage (Strange and Fryer, 1999). Polymorphisms have been identified in five major GST families – alpha, mu, theta, zeta, and pi. Increasing evidence suggests that enzymes of the mu (GSTM1, GSTM3), theta (GSTT1), and pi (GSTP1) classes are important in the protection of cells from the toxic products of oxidative stress-mediated reactions (Strange and Fryer, 1999). Individuals who are homozygous for null alleles (GSTM1, GSTT1) or those encoding low activity enzymes (GSTM3, GSTP1) may be more susceptible to deleterious effects of such compounds and be at greater risk of malignancy. Studies on GST polymorphisms in nontransplant BCC patients examining susceptibility and outcome (in terms of tumor numbers and rate of accrual) support this view (Lear et al, 1996;Ramachandran et al, 1999,2000a, b). Other studies from our laboratory have also shown that increased frequencies of the GSTM1 null genotype have been identified in nontransplant patients with both BCC and SCC compared with controls (Heagerty et al, 1994).
We hypothesize that polymorphism in genes encoding such antioxidant enzymes is important in determining individual risk of skin cancer following renal transplantation. We report a preliminary study examining the influence of allelism at GSTM1, GSTM3, GSTP1, and GSTT1 on NMSC type, number, and latency following renal transplantation. Since UV exposure is a critical factor in skin cancer carcinogenesis, we have also examined the interaction between markers of UV exposure and GST polymorphisms in determining NMSC risk in these patients.
Patients and methods
Patients
A total of 183 unrelated Caucasian renal transplant recipients (mean age 
SD = 38.8 15.6 y, 66.7% male) with functioning allografts were recruited in the North Staffordshire Hospital between May 1997 and June 1999; representing 92% of Caucasian renal transplants under follow-up at this center. Three patients withheld consent, seven were lost to follow-up, DNA samples were not available on a further five, and seven of skin types V and VI were excluded. The study was conducted with Local Hospital Ethics Committee approval and written informed consent was obtained from each participant.
A structured questionnaire and skin examination was completed by a dermatologist (HMR) blinded to previous dermatologic records at the time of examination and interview (Ramsay et al, 2000). Demographic information was gathered on age, gender, smoking history, skin type I–VI (Fitzpatrick, 1988), hair and eye color at age 21 y, years worked in an outdoor occupation (outdoor being defined as > 50% time usually spent outdoors), residence (> 3 mo) and holidays abroad in tropical/subtropical climates, sunbed use, recalled childhood sunburn episodes (painful erythema 
48 h), sunscreen use, arsenic exposure, and history of malignancy in a first degree relative. In addition to occupational exposure, sun exposure was determined from data on average number of hours spent outdoors (weekends and weekdays were separately assessed) during the periods 0–40, 40–60, and > 60 y of age. Cumulative sun exposure was then calculated as years spent outdoors. Sunbathing habits were examined by determining frequency of sunbathing (never, rarely, occasionally, frequently) during the same time periods. These were then scored (0, never; 1, rarely; 2, occasionally; 3, frequently) and the cumulative score calculated.
In order to assess interaction between markers of UV exposure and polymorphism in detoxifying enzyme genes, patients were grouped into high and low exposure categories for each variable; occupational exposure (yes versus no), residence abroad (> 3 versus 
3 mo), holidays abroad (> 2 versus 2 per y), sunbed use (any versus none), recalled childhood sunburn (any versus none), cumulative sun exposure (based on mean exposure; 4.5 versus < 4.5 y), sunbathing score (based on median score; > 3 versus 3). As there are no universally accepted markers of high exposure, these cut-offs were selected on the basis that patients in high exposure categories received higher levels than that normally encountered in a temperate climate population.
All suspicious lesions were biopsied, case-sheets and pathology reports were carefully reviewed for previous skin cancers. Only histologically proven SCC or BCC were considered for the purposes of this study. Recurrent tumors were excluded from the total tumor count. A nephrologist (PNH) collated information regarding time on dialysis, number and date(s) of transplantation(s), immunosuppressant therapy including use of antithymocyte globulin, and comorbid conditions.
Identification of genotypes
DNA was extracted from peripheral blood leukocytes by standard phenol/chloroform extraction. Determination of genotypes was as follows: GSTM1 A, B, A/B, and null genotypes were identified using a PCR-based amplification refractory mutation system approach with allele-specific primers (Elexperu-Camiruaga et al, 1996). GSTM3 AA, AB, and BB genotypes were identified using primers to exon 6/7 (Inskip et al, 1995). GSTP1 Ile/Ile, Ile/Val, and Val/Val genotypes were identified by PCR to identify the A-G transition at position 1578 (Ramachandran et al, 2000b). GSTT1 null and expressing subjects were also identified using PCR (Elexperu-Camiruaga et al, 1996). Because some DNA samples were refractory to amplification or were exhausted, it was not possible to obtain complete genotype data on all patients.
Statistical analysis
The Stata software package (version 6, Stata Corporation, Texas) was used for all statistical analyzes. Skin cancer risk was examined using presence of any NMSC, presence of SCC alone, and presence of BCC alone as endpoints. Logistic regression analysis was used to examine factors associated with risk. Outcome was also assessed in terms of the number of tumors accrued. To determine the factors associated with the number of skin cancers, negative binomial regression analysis was used, normalized for follow-up time, with number of NMSC, number of SCC, and number of BCC as outcome measures. A rate ratio, defined as the multiplicative effect of a change of a covariate by 1 was calculated (for this data usually being a change from 0 to 1). Thus, the rate ratio for males [1] against females [0] approximates to the mean number BCC in males/mean number BCC in females, when gender alone (i.e., not in the presence of other covariates) is considered. In the negative binomial regression, this will change in the presence of other covariates and following normalization for duration of follow–up. Associations with time from transplantation to appearance of the first NMSC was assessed using Cox's proportional hazards regression. All regression analyzes were corrected for gender and age at transplantation and, in the case of negative binomial regression, normalized for follow up. Clearly, some significant or near significant associations could arise artificially from multiple testing (Rosenberger, 1996). Performing such corrections presents difficulties as obtaining significant results after such correction would require study of thousands of cases with the relatively uncommon multiple cluster phenotype. Importantly,Perneger (1998) has argued that correction for multiple testing is inappropriate as it increases the likelihood of type II errors. Accordingly, we have presented uncorrected p-values but recognize that these preliminary findings require confirmation in a separate cohort of patients.
Results
The clinical features of the group have been reported in detail previously (Ramsay et al, 2000). Overall, 105 NMSC occurred in 29 patients, comprising 52 SCC in 21 patients and 53 BCC in 18 patients. Ten of these patients developed both SCC and BCC and are therefore included in the analysis of both groups, but only represented once in the total NMSC group.
GSTM1
Genotype frequencies in the transplant population were not significantly different to controls (Mattey et al, 1999). Table I shows that GSTM1 null was significantly increased in patients with SCC. This association remained significant after correction for age, gender, skin type, eye color, hair color, smoking history, occupation, cumulative sun exposure, sunbathing habits, childhood sunburn episodes, number of allografts, and follow-up (p = 0.012, OR = 8.44, 95% CI = 1.61–44.21, n = 138). To assess the interaction between GSTM1 genotype and UV exposure, the association was examined after categorizing patients according to low and high exposure. This secondary analysis showed that the association of GSTM1 null with SCC risk was particularly strong in patients with high UV indices: cumulative sun exposure > 4.5 y (p = 0.046, OR = 4.6, n = 64), greater than two holidays abroad per y (p = 0.029, OR = 8.5, n = 42), sunbathing score > 3 (p = 0.003, OR = 11.5, n = 49), outdoor occupation (p = 0.047, OR = 5.1, n = 57), though numbers of patients in some categories were small. The nature of the interaction with sunbathing score was further investigated by examining whether the two factors were synergistic; individuals with both GSTM1 null and a sunbathing score of > 3 were at particularly increased risk of SCC (p < 0.001, OR = 9.1, 95% CI = 2.8–29.3, n = 140). This remained significant when the main effects (GSTM1 null and sunbathing score > 3, individually) were included in the model (p = 0.031, OR = 12.0, 95% CI = 1.3–115.3, n = 140). Though no significant associations between GSTM1 alone and tumor numbers or latency were identified, individuals with both GSTM1 null and a sunbathing score of > 3 demonstrated a significantly shorter time to first SCC than those with other genotype/sunbathing score combinations (p = 0.012, HR = 7.1, 95% CI = 1.5–32.8, n = 138; Figure 1). We also determined whether the effect of GSTM1 null on SCC risk was related to tobacco consumption. The odds ratio for GSTM1 null was increased in ever smokers (p = 0.021, OR = 4.8, 95% CI = 1.3–18.1, n = 78), but not in nonsmokers (p = 0.898, OR = 0.9, 95%CI = 0.1–7.3, n = 63).
Figure 1.
The association between the interaction term comprising patients with both GSTM1 null and a sunbathing score of > 3, and SCC tumor latency. Individuals with this combination demonstrated a significantly shorter time from transplantation to the development of the first SCC than those with other genotype/sunbathing score combinations
Full figure and legend (10K)Table I - Genotype frequencies in patients with and without nonmelanoma skin cancer lesions in Caucasian renal transplant recipients.
Full table GSTM3
Comparison of GSTM3 genotype with that in a local control population (Mattey et al, 1999) showed that GSTM3 AA was lower (68.3% vs 74.9% in controls) and GSTM3 AB higher (29.8% vs 20.0% in controls) (
22 = 7.6, p = 0.022). The frequency of GSTM3 BB was significantly lower, after correction for age and gender, in patients with BCC when compared with all other GSTM3 genotypes (exact p = 0.029; Table I). No significant associations were identified between GSTM3 genotype and tumor numbers Table II, though GSTM3 BB, compared with GSTM3 AA, was associated with a shorter time to first BCC (p = 0.009, HR = 13.60, 95% CI = 1.93–95.7, n = 157). Though we identified an association between GSTM3 BB genotype with both BCC risk and latency, these data should be treated with caution given the low frequency of this genotype.
Table II - Significant associations with numbers of nonmelanoma skin cancer lesions in Caucasian renal transplant recipients.
Full table GSTP1
GSTP1 genotype frequencies were not significantly different from controls (Mattey et al, 1999). No significant associations between GSTP1 genotype and NMSC risk or latency were identified, though GSTP1 Ile/Ile was associated with increased number of SCC Table II. When compared with GSTP1 Ile/Val and Val/Val genotypes combined (mean SCC numbers 0.14), subjects with the GSTP1 Ile/Ile genotype demonstrated significantly more SCC (mean SCC numbers 0.58; p = 0.002, RR = 7.6, 95% CI = 2.2–26.9, n = 157). These associations, however, should be viewed as preliminary due to low patient numbers. Indeed, the absence of individuals with the GSTP1 Val/Val genotype in the NMSC group supports this view.
GSTT1
The GSTT1 null frequency in the total cohort (19.4%) was similar to that in controls (18.9%) (Mattey et al, 1999). Though no significant associations between GSTT1 genotype and NMSC risk or latency were identified, GSTT1 null genotype was associated with fewer NMSC, reflecting a reduced number of both BCC and SCC Table II. It was not possible to examine associations between this genotype and UV exposure or smoking history.
Discussion
We have identified associations between polymorphisms in detoxifying enzymes (GSTM1, GSTM3, GSTP1, GSTT1) and risk of NMSC in a population of well-characterized Caucasian renal transplant patients. Our data has examined associations with total NMSC as well as BCC and SCC individually using three related endpoints: overall risk, number of lesions, and tumor latency. Because UV constitutes an oxidative stress, we propose that individuals with low activity GST genotypes are less able to metabolize the products of oxidative stress leading to increased DNA damage, further local immunosuppression, and subsequent development of NMSC. The relevance of GSTP1 to skin cancer risk is also shown in studies on mice in whom the pi class GST genes had been deleted. Null mice developed more skin papillomas (mean 9.94) than controls (mean 2.89) after exposure to polycyclic aromatic hydrocarbon (PAH) (Henderson et al, 1998).
We have previously shown that the frequency of GSTM1 null is increased (70%) in a cohort of nontransplant patients with BCC and SCC (Heagerty et al, 1994). The finding that the association between GSTM1 null and SCC was particularly significant in patients with high UV exposure suggests that GSTM1 is involved in individual response to UV. This view is supported by data showing that GSTM1 null is associated with increased cutaneous UV sensitivity (Kerb et al, 1997) and that the frequency of this genotype is increased in photosensitive Ro positive patients with systemic lupus erythematosus (Ollier et al, 1996). GST have traditionally been viewed as carcinogen detoxifying enzymes and many studies have focused on their role in tobacco-smoke related cancers (e.g., lung, bladder). Because we have previously shown that smoking is associated with SCC (Ramsay et al, 2000), we examined our data to determine whether the effect of GST genotypes was related to tobacco consumption. This showed that the association of GSTM1 genotype with SCC risk was influenced by smoking status.
Data on the asp294his, val92met, and asp84glu polymorphisms at the melanocortin-1 receptor (MC1R), a further candidate gene for skin cancer risk, were also obtained during the course of this study. Correction of the observed associations for the effect of MC1R allelic variation showed that the significant associations were independent of MC1R genotype (data not shown), though numbers of individuals with mutant alleles at the asp294his and asp84glu sites were small.
Many studies on risk factors for skin cancer in transplant patients have considered NMSC as a group; however, our results identify factors associated with SCC but not BCC, indicating that the two lesions may have different, if overlapping, pathogenetic mechanisms. NMSC poses a significant clinical problem post-transplantation, and there is currently no standard practice with regard to surveillance. In the long term, it may prove possible to develop predictive indices utilizing known clinical and genetic risk factors. High risk patients could be entered into skin surveillance programs to promote earlier detection of NMSC. We acknowledge that, given the size of the population and the proportion of patients with NMSC, this study is exploratory and we are currently recruiting a confirmatory cohort in which to test the significant associations identified here.
Notes
1 Statistics prepared by the UK Transplant Support Service Authority from the National Transplant Database maintained on behalf of transplant services in the UK and Republic of Ireland.
References
| 1. | Elexperu-Camiruaga J, Buxton N & Kandula V et al. Susceptibility to astrocytoma and meningioma: Influence of allelism at glutathione S-transferase, GSTT1 and GSTM1 and cytochrome P450, CYP2D6 loci. Cancer Res (1996) 55: 4237–4239. |
| 2. | Euvrard S, Kanitakis J & Pouteil-Noble C et al. Aggressive squamous cell carcinomas in organ transplant recipients. Transplant Proc The (1995) 27: 1767–1768. | ChemPort | |
| 3. | Fitzpatrick TB. The validity and practicality of sun reaction types I through VI. Arch Dermatol (1988) 124: 869–871. | Article | PubMed | ISI | ChemPort | |
| 4. | Glover MT, Niranjan N, Kwan JTC & Leigh IM. Non-melanoma skin cancer in renal transplant recipients: the extent of the problem and a strategy for management. Br J Plast Surg (1994) 47: 86–89. | Article | PubMed | ISI | ChemPort | |
| 5. | Heagerty AH, Fitzgerald D & Smith A et al. Glutathione S-transferase GSTM1 phenotypes and protection against cutaneous tumours. Lancet (1994) 343: 266–268. | Article | PubMed | ISI | ChemPort | |
| 6. | Henderson CJ, Smith AG, Ure J, Brown K, Bacon EJ & Wolf CR. Increased skin tumorigenesis in mice lacking pi class glutathione S-transferases. Proc Natl Acad Sci USA (1998) 95: 5275–5280. | Article | PubMed | ChemPort | |
| 7. | Inskip A, Elexperu-Camiruaga J & Buxton N et al. Identification of polymorphism at the glutathione S-transferase, GSTM3 locus: Evidence for linkage with GSTM1*A. Biochem J (1995) 312: 713–716. | PubMed | ISI | ChemPort | |
| 8. | Kelly GE, Mahony JF, Sheil AGR, Meikle WD, Tiller DS & Horvarth J. Risk factors for skin carcinogenesis in immunosuppressed kidney transplant recipients. Clin Transplant (1987) 1: 271–277. |
| 9. | Kerb R, Brockmoller J, Reum T & Roots I. Deficiency of Glutathione S-transferases T1 and M1 as heritable factors of increased cutaneous UV sensitivity. J Invest Dermatol (1997) 108: 229–232. | Article | PubMed | ISI | ChemPort | |
| 10. | Lear JT, Heagerty AHM & Smith A et al. Multiple cutaneous basal cell carcinomas: Glutathione S-transferase (GSTM1, GSTT1) and cytochrome P450 (CYP2D6, CYP1A1) polymorphisms influence tumour numbers and accrual. Carcinogenesis (1996) 17: 1891–1896. | PubMed | ISI | ChemPort | |
| 11. | Liddington M, Richardson AJ, Higgins RM, Endre ZH, Venning VA, Murie JA & Morris PJ. Skin cancer in renal transplant recipients. Br J Surg (1989) 76: 1002–1005. | PubMed | ISI | ChemPort | |
| 12. | Mattey DL, Hassell AB & Plant M et al. Association of polymorphism in glutathione S-transferase loci with susceptibility and outcome in rheumatoid arthritis: Comparison with the shared epitope. Ann Rheum Dis (1999) 58: 164–168. | PubMed | ISI | ChemPort | |
| 13. | McGregor JM, Berkhout RJ, Rozycka M, ter Schegget J, Bouwes Bavinck JN, Brooks L & Crook T. p53 mutations implicate sunlight in post-transplant skin cancer irrespective of human papillomavirus status. Oncogene (1997) 15: 1737–1740. | Article | PubMed | ISI | ChemPort | |
| 14. | Ollier WER, Davies E, Snowden N, Alldersea J, Fryer AA & Jones P. Strange: Association of homozygosity for glutathione-S-transferase GSTM1 null alleles is associated with the Ro+/La-autoantibody profile in patients with systemic lupus erythematosus. Arth Rheum (1996) 39: 1763–1764. | ISI | ChemPort | |
| 15. | Perneger TV. What's wrong with Bonferroni adjustments. Br Med J (1998) 316: 1236–1238. | ISI | ChemPort | |
| 16. | Ramachandran S, Fryer AA & Smith AG et al. Basal cell carcinoma: Tumor clustering is associated with increased accrual in high-risk subgroups. Cancer (2000a) 89: 1012–1018 10.1002/1097-0142(20000901)89:5<1012::aid-cncr10>3.3.co;2-f. | Article | ISI | ChemPort | |
| 17. | Ramachandran S, Hoban PR & Ichii-Jones F et al. Glutathione S-transferase GSTP1 and cyclin D1 genotypes. Association with numbers of basal cell carcinomas in a patient subgroup at high-risk of multiple tumours. Pharmacogenetics (2000b) 10: 545–556. | Article | ISI | ChemPort | |
| 18. | Ramachandran S, Lear JT & Ramsay H et al. Presentation with multiple cutaneous basal cell carcinomas: Association of glutathione S-transferase and cytochrome P450 genotypes with clinical phenotype. Cancer Epidemiol Biomarkers Prev (1999) 8: 61–67. | PubMed | ISI | ChemPort | |
| 19. | Ramsay HM, Fryer AA, Smith AG & Harden Pn. Cinical risk factors associated with non-melanoma skin cancer in renal transplant recipients. Am J Kidney Dis (2000) 36: 167–176. | PubMed | ISI | ChemPort | |
| 20. | Roeger LS, Sheil AGR & Disnet APS et al. Risk factors associated with the development of squamous cell carcinomas in immunosupressed renal transplant recipients. Clin Transpl (1992) 6: 202–211. | ISI | |
| 21. | Rosenberger WF. Dealing with multiplicities in pharmacoepidemiologic studies. Pharmacoepidem Drug Safety (1996) 5: 95–100. | ChemPort | |
| 22. | Shiel AGR, Flavel S, Disney APS & Matthew TH. Cancer development in patients progressing to dialysis and renal transplantation. Transplant Proc The (1985) 17: 1685–1688. |
| 23. | Strange RC & Fryer AA. The glutathione S-transferases: influence of polymorphism on susceptibility to non familial cancers. In: Boffetta, P, Caporaso, N, Cuzick, J, Lang, M, Vineis, P, eds.Metabolic Polymorphisms and Cancer (1999) Lyon: IARC Scientific Publications pp 303–322. |
| 24. | Wei Q, Matanoski GM, Farmer ER, Hedayati MA & Grossman L. DNA repair related to multiple skin cancers and drug use. Cancer Res (1994) 54: 437–440. | PubMed | ISI | ChemPort | |
| 25. | Ziegler A, Leffell DJ & Kunala S et al. Mutation hotspots due to sunlight in the p53 gene of nonmelanoma skin cancers. Proc Natl Acad Sci USA (1993) 90: 4216–4220. | PubMed | ChemPort | |
Acknowledgments
We would like to thank the North Staffordshire Medical Institute for financial support, and Ms Joanne Bath and Mrs Julie Alldersea for technical support.
