Background: Antioxidant nutrients can help prevent skin damage caused by ultraviolet radiation from sunlight, but it is not clear whether serum concentrations of such nutrients influence skin cancer risk.

Methods: We carried out a prospective study of the associations between serum concentrations of antioxidant nutrients and incidence (person-based and tumor-based) of basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) of the skin among a random subsample of 485 adults from an Australian community. Participants were divided into thirds, ranked according to their serum concentrations of carotenoids, α-tocopherol, and selenium measured in 1996 and were monitored for incident, histologically confirmed BCC and SCC tumors until 2004.

Results: Although there were no associations between baseline serum carotenoids or α-tocopherol concentrations and incidence of BCC or SCC, baseline serum selenium concentrations showed strong inverse associations with both BCC and SCC tumor incidence. Compared with participants with lowest selenium concentrations at baseline (0.4-1.0 μmol/L), those with the highest serum selenium concentrations (1.3-2.8 μmol/L) had a decreased incidence of BCC tumors (multivariate relative risk, 0.43; 95% confidence interval, 0.21-0.86; Ptrend = 0.02) and SCC tumors (multivariate relative risk, 0.36; 95% confidence interval, 0.15-0.82; Ptrend = 0.02).

Conclusion: Relatively high serum selenium concentrations are associated with an ∼60% decrease in subsequent tumor incidence of both BCC and SCC, whereas serum concentrations of carotenoids or α-tocopherol are not associated with later skin cancer incidence. A possible U-shaped association between serum selenium concentrations and SCC of the skin needs confirmation. (Cancer Epidemiol Biomarkers Prev 2009;18(4):1167–73)

Basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) are the most commonly occurring skin cancers in Caucasian populations. The costs of screening and treatment of these cancers continue to burden health systems of many countries around the world; thus, prevention could cause substantial public health and economic benefits (1-3).

The epidermis contains an array of antioxidants that help protect this outer layer of the skin against damage caused by exposure to ultraviolet radiation from sunlight, the main environmental cause of skin cancer. Ultraviolet radiation causes direct damage to DNA and the immune system (4, 5) and indirect damage through formation of free radicals such as reactive oxygen species (6). Carotenoids, α-tocopherol (vitamin E), and selenium (an important component of the antioxidant enzyme glutathione peroxidase) are found in the epidermis where they have been shown to protect against oxidative damage by neutralizing reactive oxygen species and other free radicals (7-11).

There is little consistent evidence about the relation between serum biomarkers of antioxidant nutrients and skin cancer risk in the general population. In a prospective study of antioxidant status and BCC and SCC risks, Dorgan et al. reported a positive association between serum lutein, zeaxanthin, and β-carotene and SCC risk in a 5-year follow-up study of skin cancer patients who participated in a drug trial in the United States (12). No such associations were found in another prospective study of persons who also all had a previous history of skin cancer (13). Results from a placebo-controlled trial, however, showed that selenium supplementation in persons with a history of skin cancer was associated with an increased incidence of BCC and SCC (14).

We have investigated associations between serum concentrations of antioxidant nutrients in a longitudinal study of a population sample of Australian adults whose sun exposure and skin cancer histories have been fully characterized.

Study Population

As part of the ongoing Nambour Skin Cancer Study, we conducted an 8-year prospective cohort study among adults who had been selected at random from the residents of the township of Nambour, a subtropical community in Queensland, Australia, for a baseline study of skin cancer (15). Between 1992 and 1996, 1,621 of these adults participated in a field trial to evaluate the role of a daily 30 mg β-carotene supplement and daily sunscreen use in skin cancer prevention (16, 17). In a two-by-two factorial design, participants were randomly allocated to taking daily either the β-carotene supplement or a placebo tablet (blinded) and daily sunscreen use or continuation of their usual sunscreen use habits (discretionary sunscreen use). At the end of the trial in 1996 (baseline of the current investigation), participants provided a blood sample, and during the course of the trial, they completed questionnaires on smoking habits, education, occupation, presence of selected medical conditions, and skin cancer risk factors such as skin color, tanning ability of the skin, and occupational and leisure-time sun exposure (17). During a physical examination in 1996, elastosis of the neck was recorded as a measure of long-term sun exposure history. Detailed descriptions of the community sample, the field trial, and its outcomes have been reported previously (16, 17).

Study participants were followed-up from 1996 until the end of 2004 to ascertain all occurrences of BCC and SCC. In the analyses, we considered all participants who provided a blood sample in 1996 and those who were randomized to the placebo tablets during the trial.

Ascertainment of skin cancers took place through an intensive surveillance system that had been set up during the Nambour trial and was continued during the complete post-trial follow-up period. Questionnaires were mailed twice yearly to participants and any reported skin cancers were confirmed through histologic reports. Independent pathology laboratories throughout Queensland provided pathology reports for all skin cancers diagnosed among study participants. These methods ensured virtually 100% ascertainment of histologically confirmed skin cancers in the study population (18).

There were no differences between the participants in the present study and the original 1,621 trial participants in terms of randomized sunscreen allocation during the trial, age, sex, education, occupation, smoking, use of dietary supplements, and skin cancer risk factors such as skin color, lifetime number of sunburns and other measures of sun exposure, and skin cancer history before 1996. This study was approved by the Ethics Committee of the Queensland Institute of Medical Research and all participants provided written informed consent.

Outcomes

We considered two outcomes in the analysis: (a) incidence of persons affected by a new basal or squamous cell cancer calculated as the number of persons affected by these cancers after the 1996 skin examination survey through December 31, 2004, divided by the person-years of follow-up accumulated between these dates and expressed per 100,000 person-years, and (b) incidence of basal or squamous cell tumors during the same person-years follow-up time as calculated for the person-based analysis. Tumors diagnosed during the 1996 skin survey were not included in the analyses to exclude disease that already existed at baseline. Tumors and person-years of follow-up were counted until date of withdrawal from the study, date of death, or December 31, 2004, whichever came first.

Serum Biomarkers

In 1996, nonfasting venous blood samples of 30 mL were collected using standard venipuncture techniques done by experienced phlebotomists. Blood was not taken from subjects who had consumed more than a light breakfast (e.g., toast or cereal, no cooked breakfast). Blood samples were processed at the time of collection and serum samples were stored in ∼1 mL aliquots at -70°C until analysis. Measurements of carotenoids, α-tocopherol, selenium, and cholesterol were conducted by Queensland Health Pathology Services at the Royal Brisbane Hospital. Measurements of carotenoids and α-tocopherol were conducted simultaneously by high-performance liquid chromatography using the method of Sowell et al. (19). Total concentrations of lutein and zeaxanthin have been presented as a combined estimate because these carotenoids were not separated in the high-performance liquid chromatography analysis. Total cholesterol was measured using an enzymatic colorimetric test (20). Serum selenium was analyzed by atomic absorption spectrometry using a graphite furnace and Zeeman background correction (21). Coefficients of variations of pooled serum measured in duplicate in every run (n = 21) were α-carotene 10.7%, β-carotene 9.7%, β-cryptoxanthin 12.1%, lutein and zeaxanthin 17.3%, α-tocopherol 4.7%, selenium 9.8%, and cholesterol 2.6%. Due to an error in the automated reading of chromatograms, results for lycopene serum concentrations could not be included.

Stability of these analytes over time was investigated in 45 participants of this study population in 1992 to 1993. This was done in participants who received the placebo and thus not the β-carotene supplement. Spearman correlation coefficients (r) of serum concentrations measured in two blood samples that were taken 16 months apart were α-carotene 0.72, β-carotene 0.70, β-cryptoxanthin 0.60, lutein and zeaxanthin 0.85, and α-tocopherol 0.63.

Statistical Models

Serum carotenoids and α-tocopherol were adjusted for serum cholesterol using the residual method to account for variations in serum carotenoids and α-tocopherol that are due to variations in serum lipids (22). Distributions of the biomarkers were identified as skewed and variables were log transformed to improve normality before calculation of the residuals. Tertiles were calculated for each biomarker and used as cut points for grouping. For person-based analysis, relative risks (RR) with 95% confidence intervals (95% CI) for increasing concentrations of the biomarker compared with the lowest group were derived from generalized linear models specifying Poisson distribution with a robust error variance (23) and person-years of follow-up as offset. For tumor-based analyses, RR (95% CI) were derived using generalized linear models with negative binomial distribution. The negative binomial distribution has been recommended for analyzing nonnegative integer data with variance greater than the mean (24) and provided the best fit to our tumor-count data.

We first applied models controlling for age and sex. The expanded multivariate models also controlled for usual time spent outdoors on weekdays, history of skin cancer before 1996, pack-years of smoking until 1996, and alcohol intake (continuous variable). These confounders were selected based on their association with the exposure and outcome variables and on previous studies of diet and skin cancer. Covariates were retained in the model if they changed the risk estimate by >10%, whereas age and sex were retained in all multivariate models. There was no additional confounding by sunscreen treatment allocation during the trial, skin color, tanning ability of skin, elastosis of the neck, number of painful sunburns during life, other indicators of past sunlight exposure, or use of dietary supplements (yes/no).

To test for linear trends, we assigned the median to each tertile group and modeled these values as a continuous variable for each biomarker. All analyses were done with SAS statistical software (version 9.1; SAS Institute). All reported P values are two-sided.

Serum antioxidants were measured in 485 participants. In the 8-year follow-up of these participants, a total of 173 histologically confirmed new BCC tumors were diagnosed in 77 participants during 3608 person-years of follow-up (tumor-based incidence: 4,795/100,000; person-based incidence: 2,134/100,000). For SCC, a total of 124 histologically confirmed new tumors were diagnosed in 59 participants during the same person-years of follow-up (tumor-based incidence: 3,437/100,000; person-based incidence: 1,635/100,000). In general, participants who had either an incident SCC or BCC in the follow-up period were more likely (P < 0.05) to be male, to be older, to have fair skin, and to have a history of skin cancer than those who did not develop a skin cancer (Table 1). Those who developed SCC were also more likely (P < 0.05) to have a tendency to sun-burn and to have smoked, but those who developed BCC did not differ by these characteristics compared with participants who did not develop BCC. The number of painful sunburns during life was not associated with developing BCC or SCC in this study population (although borderline significant for SCC; P = 0.06).

Table 1.

Characteristics by skin cancer status of 485 participants of the Nambour Skin Cancer Study, 1996-2004

SCC
BCC
Yes (n = 59), n (%)No (n = 426), n (%)P*Yes (n = 77), n (%)No (n = 408), n (%)P*
Sex       
    Male 38 (64) 185 (43) 0.002 46 (60) 177 (43) 0.008 
    Female 21 (36) 241 (57)  31 (40) 231 (57)  
Age       
    Age (mean) in 1996 (y) 63 y 54 y <0.0001 61 y 54 y <0.0001 
Skin color       
    Fair 41 (69) 226 (53) 0.04 53 (69) 214 (53) 0.04 
    Medium 15 (25) 172 (41)  19 (25) 168 (41)  
    Olive 3 (5) 27 (6)  5 (6) 25 (6)  
Propensity to burn/tan after acute sun exposure       
    Always burn 20 (34) 77 (18) 0.004 21 (27) 76 (19) 0.12 
    Burn then tan 36 (61) 299 (70)  49 (64) 286 (70)  
    Tan only 3 (5) 49 (12)  7 (9) 45 (11)  
Painful sunburns during life       
    None 14 (25) 74 (18) 0.06 19 (25) 69 (17) 0.33 
    1 Burn 12 (21) 58 (14)  7 (9) 63 (16)  
    ≥2 Burns 31 (54) 285 (68)  49 (65) 267 (67)  
Smoking status in 1996       
    Never 22 (37) 236 (55) 0.005 36 (47) 222 (55) 0.12 
    Current 5 (8) 38 (9)  5 (6) 38 (9)  
    Ex 32 (54) 152 (36)  36 (47) 148 (36)  
History of skin cancer before 1996       
    No 17 (29) 309 (73) <0.0001 19 (25) 307 (75) <0.0001 
    Yes 42 (71) 117 (27)  58 (75) 101 (25)  
SCC
BCC
Yes (n = 59), n (%)No (n = 426), n (%)P*Yes (n = 77), n (%)No (n = 408), n (%)P*
Sex       
    Male 38 (64) 185 (43) 0.002 46 (60) 177 (43) 0.008 
    Female 21 (36) 241 (57)  31 (40) 231 (57)  
Age       
    Age (mean) in 1996 (y) 63 y 54 y <0.0001 61 y 54 y <0.0001 
Skin color       
    Fair 41 (69) 226 (53) 0.04 53 (69) 214 (53) 0.04 
    Medium 15 (25) 172 (41)  19 (25) 168 (41)  
    Olive 3 (5) 27 (6)  5 (6) 25 (6)  
Propensity to burn/tan after acute sun exposure       
    Always burn 20 (34) 77 (18) 0.004 21 (27) 76 (19) 0.12 
    Burn then tan 36 (61) 299 (70)  49 (64) 286 (70)  
    Tan only 3 (5) 49 (12)  7 (9) 45 (11)  
Painful sunburns during life       
    None 14 (25) 74 (18) 0.06 19 (25) 69 (17) 0.33 
    1 Burn 12 (21) 58 (14)  7 (9) 63 (16)  
    ≥2 Burns 31 (54) 285 (68)  49 (65) 267 (67)  
Smoking status in 1996       
    Never 22 (37) 236 (55) 0.005 36 (47) 222 (55) 0.12 
    Current 5 (8) 38 (9)  5 (6) 38 (9)  
    Ex 32 (54) 152 (36)  36 (47) 148 (36)  
History of skin cancer before 1996       
    No 17 (29) 309 (73) <0.0001 19 (25) 307 (75) <0.0001 
    Yes 42 (71) 117 (27)  58 (75) 101 (25)  
*

P from Mantel-Haenzel χ2 test (categorical data) or ANOVA (continuous data).

There were no associations between serum carotenoids or α-tocopherol concentrations and person-based incidence of BCC or SCC (Tables 2 and 3). However, participants with the highest compared with the lowest selenium concentrations appeared less likely to be affected by BCC (highest versus lowest group multivariate-adjusted RR, 0.58; 95% CI, 0.32-1.07; Ptrend = 0.08) and SCC (highest versus lowest group multivariate-adjusted RR, 0.49; 95% CI, 0.24-0.99; Ptrend = 0.05; Table 4). The tumor-based analyses confirmed and strengthened these inverse associations, such that participants with the highest compared with the lowest selenium concentrations had a substantially decreased incidence of BCC (highest versus lowest group multivariate-adjusted RR, 0.43; 95% CI, 0.21-0.86; Ptrend = 0.02) and SCC (highest versus lowest group multivariate-adjusted RR, 0.36; 95% CI, 0.15-0.82; Ptrend = 0.02; Table 4). None of the other serum concentrations showed associations with skin cancer incidence in the tumor-based analyses (results not shown).

Table 2.

RR (95% CI) of BCC by group of serum concentration of antioxidants at baseline, person-based analyses in 485 participants of the Nambour Skin Cancer Study, 1996-2004

Serum biomarkerBiomarker concentration
Ptrend
Group 1Group 2Group 3
α-Carotene     
    Median (μmol/L; min-max) 0.05 (0.01-0.07) 0.11 (0.08-0.15) 0.22 (0.16-1.04)  
    No. tumors/participants 18/153 35/171 24/161  
    Basic RR (95% CI)* 1.00 1.87 (1.05-3.30) 1.33 (0.72-2.46) 0.91 
    Multivariate RR (95% CI) 1.00 1.70 (0.95-3.03) 1.28 (0.68-2.41) 0.69 
β-Carotene     
    Median (μmol/L; min-max) 0.30 (0.02-0.44) 0.59 (0.45-0.79) 1.10 (0.80-5.10)  
    No. tumors/participants 23/163 30/159 24/163  
    Basic RR (95% CI)* 1.00 1.33 (0.77-2.30) 1.06 (0.59-1.90) 0.99 
    Multivariate RR (95% CI) 1.00 1.30 (0.74-2.30) 1.07 (0.59-1.96) 0.95 
β-Cryptoxanthin     
    Median (μmol/L; min-max) 0.10 (0.02-0.16) 0.29 (0.18-0.42) 0.73 (0.43- 3.20)  
    No. tumors/participants 23/159 30/168 24/158  
    Basic RR (95% CI)* 1.00 1.10 (0.64-1.90) 0.90 (0.50-1.63) 0.63 
    Multivariate RR (95% CI) 1.00 1.07 (0.61-1.87) 0.91 (0.50-1.67) 0.68 
Lutein and zeaxanthin     
    Median (μmol/L; min-max) 0.20 (0.04-0.28) 0.37 (0.29-0.47) 0.64 (0.48-1.60)  
    No. tumors/participants 26/162 21/158 30/165  
    Basic RR (95% CI)* 1.00 0.85 (0.48-1.51) 1.05 (0.62-1.77) 0.78 
    Multivariate RR (95% CI) 1.00 0.81 (0.45-1.46) 1.06 (0.62-1.83) 0.71 
α-Tocopherol     
    Median (μmol/L; min-max) 24.0 (9.0-27.0) 30.0 (28.0-33.0) 38.5 (34.0-130.0)  
    No. tumors/participants 22/163 29/158 26/164  
    Basic RR (95% CI)* 1.00 1.12 (0.63 1.98) 0.90 (0.50-1.65) 0.65 
    Multivariate RR (95% CI) 1.00 0.88 (0.49-1.56) 0.95 (0.51-1.75) 0.92 
Serum biomarkerBiomarker concentration
Ptrend
Group 1Group 2Group 3
α-Carotene     
    Median (μmol/L; min-max) 0.05 (0.01-0.07) 0.11 (0.08-0.15) 0.22 (0.16-1.04)  
    No. tumors/participants 18/153 35/171 24/161  
    Basic RR (95% CI)* 1.00 1.87 (1.05-3.30) 1.33 (0.72-2.46) 0.91 
    Multivariate RR (95% CI) 1.00 1.70 (0.95-3.03) 1.28 (0.68-2.41) 0.69 
β-Carotene     
    Median (μmol/L; min-max) 0.30 (0.02-0.44) 0.59 (0.45-0.79) 1.10 (0.80-5.10)  
    No. tumors/participants 23/163 30/159 24/163  
    Basic RR (95% CI)* 1.00 1.33 (0.77-2.30) 1.06 (0.59-1.90) 0.99 
    Multivariate RR (95% CI) 1.00 1.30 (0.74-2.30) 1.07 (0.59-1.96) 0.95 
β-Cryptoxanthin     
    Median (μmol/L; min-max) 0.10 (0.02-0.16) 0.29 (0.18-0.42) 0.73 (0.43- 3.20)  
    No. tumors/participants 23/159 30/168 24/158  
    Basic RR (95% CI)* 1.00 1.10 (0.64-1.90) 0.90 (0.50-1.63) 0.63 
    Multivariate RR (95% CI) 1.00 1.07 (0.61-1.87) 0.91 (0.50-1.67) 0.68 
Lutein and zeaxanthin     
    Median (μmol/L; min-max) 0.20 (0.04-0.28) 0.37 (0.29-0.47) 0.64 (0.48-1.60)  
    No. tumors/participants 26/162 21/158 30/165  
    Basic RR (95% CI)* 1.00 0.85 (0.48-1.51) 1.05 (0.62-1.77) 0.78 
    Multivariate RR (95% CI) 1.00 0.81 (0.45-1.46) 1.06 (0.62-1.83) 0.71 
α-Tocopherol     
    Median (μmol/L; min-max) 24.0 (9.0-27.0) 30.0 (28.0-33.0) 38.5 (34.0-130.0)  
    No. tumors/participants 22/163 29/158 26/164  
    Basic RR (95% CI)* 1.00 1.12 (0.63 1.98) 0.90 (0.50-1.65) 0.65 
    Multivariate RR (95% CI) 1.00 0.88 (0.49-1.56) 0.95 (0.51-1.75) 0.92 
*

Adjusted for age, sex, and allocation of a β-carotene supplement during the Nambour trial. All RRs from multivariate Poisson regression.

Adjusted for age, sex, pack-years of smoking, β-carotene supplement allocation during the trial, alcohol intake (continuous), time spent outdoors on weekdays, and history of skin cancer before 1996.

Table 3.

RR (95% CI) of SCC by group of serum concentration of antioxidants at baseline, person-based analyses in 485 participants of the Nambour Skin Cancer Study, 1996-2004

Serum biomarkerBiomarker concentration
Ptrend
Group 1Group 2Group 3
α-Carotene     
    Median (μmol/L; min-max) 0.05 (0.01-0.07) 0.11 (0.08-0.15) 0.22 (0.16-1.04)  
    No. tumors/participants 19/153 26/171 14/161  
    Basic RR (95% CI)* 1.00 1.28 (0.71-2.32) 0.74 (0.37-1.49) 0.59 
    Multivariate RR (95% CI) 1.00 1.30 (0.70-2.39) 0.71 (0.35-1.45) 0.49 
β-Carotene     
    Median (μmol/L; min-max) 0.30 (0.02-0.44) 0.59 (0.45-0.79) 1.10 (0.80-5.10)  
    No. tumors/participants 20/163 22/159 17/163  
    Basic RR (95% CI)* 1.00 1.09 (0.59-2.01) 0.88 (0.45-1.69) 0.64 
    Multivariate RR (95% CI) 1.00 1.08 (0.57-2.03) 0.92 (0.47-1.81) 0.78 
β-Cryptoxanthin     
    Median (μmol/L; min-max) 0.10 (0.01-0.17) 0.29 (0.18-0.42) 0.73 (0.43- 3.20)  
    No. tumors/participants 16/159 22/168 21/158  
    Basic RR (95% CI)* 1.00 1.18 (0.62-2.25) 1.11 (0.56-2.16) 0.87 
    Multivariate RR (95% CI) 1.00 1.16 (0.60-2.24 1.17 (0.59-2.33) 0.70 
Lutein and zeaxanthin     
    Median (μmol/L; min-max) 0.20 (0.04-0.28) 0.37 (0.29-0.47) 0.64 (0.48-1.60)  
    No. tumors/participants 17/162 21/158 21/165  
    Basic RR (95% CI)* 1.00 1.34 (0.71-2.55) 1.04 (0.55-1.98) 0.96 
    Multivariate RR (95% CI) 1.00 1.26 (0.66-2.42) 1.01 (0.53-1.93) 0.91 
α-Tocopherol     
    Median (μmol/L; min-max) 24.0 (9.0-27.0) 30.0 (28.0-33.0) 38.5 (34.0-130.0)  
    No. tumors/participants 19/163 20/158 20/164  
    Basic RR (95% CI)* 1.00 0.84 (0.44-1.60) 0.76 (0.39-1.47) 0.44 
    Multivariate RR (95% CI) 1.00 0.78 (0.41-1.50) 0.85 (0.43-1.65) 0.68 
Serum biomarkerBiomarker concentration
Ptrend
Group 1Group 2Group 3
α-Carotene     
    Median (μmol/L; min-max) 0.05 (0.01-0.07) 0.11 (0.08-0.15) 0.22 (0.16-1.04)  
    No. tumors/participants 19/153 26/171 14/161  
    Basic RR (95% CI)* 1.00 1.28 (0.71-2.32) 0.74 (0.37-1.49) 0.59 
    Multivariate RR (95% CI) 1.00 1.30 (0.70-2.39) 0.71 (0.35-1.45) 0.49 
β-Carotene     
    Median (μmol/L; min-max) 0.30 (0.02-0.44) 0.59 (0.45-0.79) 1.10 (0.80-5.10)  
    No. tumors/participants 20/163 22/159 17/163  
    Basic RR (95% CI)* 1.00 1.09 (0.59-2.01) 0.88 (0.45-1.69) 0.64 
    Multivariate RR (95% CI) 1.00 1.08 (0.57-2.03) 0.92 (0.47-1.81) 0.78 
β-Cryptoxanthin     
    Median (μmol/L; min-max) 0.10 (0.01-0.17) 0.29 (0.18-0.42) 0.73 (0.43- 3.20)  
    No. tumors/participants 16/159 22/168 21/158  
    Basic RR (95% CI)* 1.00 1.18 (0.62-2.25) 1.11 (0.56-2.16) 0.87 
    Multivariate RR (95% CI) 1.00 1.16 (0.60-2.24 1.17 (0.59-2.33) 0.70 
Lutein and zeaxanthin     
    Median (μmol/L; min-max) 0.20 (0.04-0.28) 0.37 (0.29-0.47) 0.64 (0.48-1.60)  
    No. tumors/participants 17/162 21/158 21/165  
    Basic RR (95% CI)* 1.00 1.34 (0.71-2.55) 1.04 (0.55-1.98) 0.96 
    Multivariate RR (95% CI) 1.00 1.26 (0.66-2.42) 1.01 (0.53-1.93) 0.91 
α-Tocopherol     
    Median (μmol/L; min-max) 24.0 (9.0-27.0) 30.0 (28.0-33.0) 38.5 (34.0-130.0)  
    No. tumors/participants 19/163 20/158 20/164  
    Basic RR (95% CI)* 1.00 0.84 (0.44-1.60) 0.76 (0.39-1.47) 0.44 
    Multivariate RR (95% CI) 1.00 0.78 (0.41-1.50) 0.85 (0.43-1.65) 0.68 
*

Adjusted for age and sex. All RRs from multivariate Poisson regression.

Adjusted for age, sex, pack-years of smoking, alcohol intake (continuous), time spent outdoors on weekdays, and history of skin cancer before 1996.

Table 4.

RR (95% CI) of BCC and SCC by group of serum selenium concentration at baseline in 485 participants of the Nambour Skin Cancer Study, 1996-2004

Selenium concentration
Ptrend
Group 1Group 2Group 3
Median (μmol/L*; min-max) 0.9 (0.4-1.0) 1.1 (1.1-1.2) 1.4 (1.3-2.8)  
BCC: person-based     
    No. persons affected/total participants 33/194 28/163 16/128  
    Basic RR (95% CI) 1.00 1.02 (0.61-1.69) 0.67 (0.37-1.22) 0.19 
    Multivariate RR (95% CI) 1.00 0.97 (0.58-1.61) 0.58 (0.32-1.07) 0.08 
BCC: tumor-based     
    No. tumors/total participants 20/163 22/159 17/163  
    Basic RR (95% CI) 1.00 1.09 (0.56-2.10) 0.57 (0.28-1.17) 0.14 
    Multivariate RR (95% CI) 1.00 1.02 (0.56-1.87) 0.43 (0.21-0.86) 0.02 
SCC: person-based     
    No. persons affected/total participants 27/194 21/163 11/128  
    Basic RR (95% CI) 1.00 0.92 (0.52-1.63) 0.54 (0.27-1.08) 0.08 
    Multivariate RR (95% CI) 1.00 0.89 (0.49-1.60) 0.49 (0.24-0.99) 0.05 
SCC: tumor-based     
    No. tumors/total participants 17/162 21/158 21/165  
    Basic RR (95% CI) 1.00 0.97 (0.48-1.95) 0.44 (0.19-1.00) 0.06 
    Multivariate RR (95% CI) 1.00 0.86 (0.44-1.67) 0.36 (0.15-0.82) 0.02 
Selenium concentration
Ptrend
Group 1Group 2Group 3
Median (μmol/L*; min-max) 0.9 (0.4-1.0) 1.1 (1.1-1.2) 1.4 (1.3-2.8)  
BCC: person-based     
    No. persons affected/total participants 33/194 28/163 16/128  
    Basic RR (95% CI) 1.00 1.02 (0.61-1.69) 0.67 (0.37-1.22) 0.19 
    Multivariate RR (95% CI) 1.00 0.97 (0.58-1.61) 0.58 (0.32-1.07) 0.08 
BCC: tumor-based     
    No. tumors/total participants 20/163 22/159 17/163  
    Basic RR (95% CI) 1.00 1.09 (0.56-2.10) 0.57 (0.28-1.17) 0.14 
    Multivariate RR (95% CI) 1.00 1.02 (0.56-1.87) 0.43 (0.21-0.86) 0.02 
SCC: person-based     
    No. persons affected/total participants 27/194 21/163 11/128  
    Basic RR (95% CI) 1.00 0.92 (0.52-1.63) 0.54 (0.27-1.08) 0.08 
    Multivariate RR (95% CI) 1.00 0.89 (0.49-1.60) 0.49 (0.24-0.99) 0.05 
SCC: tumor-based     
    No. tumors/total participants 17/162 21/158 21/165  
    Basic RR (95% CI) 1.00 0.97 (0.48-1.95) 0.44 (0.19-1.00) 0.06 
    Multivariate RR (95% CI) 1.00 0.86 (0.44-1.67) 0.36 (0.15-0.82) 0.02 
*

1 μmol/L ≈ 79 μg/L.

Adjusted for age and sex. RRs from Poisson regression (person-based) or negative binomial regression (tumor-based).

Adjusted for age, sex, pack-years of smoking, alcohol intake (continuous), time spent outdoors on weekdays, and history of skin cancer before 1996.

In this prospective community-based study of Australian adults, those with relatively high serum selenium concentrations at baseline had decreased incidence rates of both BCC and SCC tumors in the 8-year follow-up period. Previously, few, if any, nutrients have been shown to influence BCC occurrence (in contrast to SCC). These results reflected the inverse associations observed between serum selenium concentrations and overall incidence of persons newly affected by BCC and SCC, although mostly these associations did not reach statistical significance. This apparently protective effect of selenium on skin cancer incidence is consistent with in vivo and in vitro studies that have shown that topical and oral selenium can provide protection against ultraviolet-induced sunburn, tanning, and skin cancer (7, 25-27). Previous human data are sparse and conflicting, however. A clinic-based study showed lower plasma selenium concentrations in patients with SCC or BCC compared with controls (28), whereas other studies found no associations (13, 29). Notably, our observations are in apparent contrast to the results of a selenium supplementation trial in which a daily 200 μg selenium supplement was associated with an increased risk of SCC after 13 years of follow-up (hazard ratio for selenium supplement versus placebo, 1.25; 95% CI, 1.03-1.51; P = 0.03; refs. 14, 30). The authors reported that the increased risk of SCC in their trial was greatest among participants with the highest baseline concentrations of plasma selenium (>105.6 ng/mL or 1.3 μmol/L). These concentrations were equivalent to those in the highest serum selenium group in our study population, in which we found a protective association. It is possible that selenium supplementation in people who already have adequate levels of this nutrient increases rather than decreases skin cancer risk. Such unintended adverse effects have also occurred in supplementation trials of β-carotene (31, 32). Furthermore, the investigators of the selenium supplementation trial have indicated that ∼60% of the trial participants had punctate keratoses of the palms, characteristic of arsenic exposure (14). Selenium can enhance the toxic effects of arsenic by increasing its retention in tissues and by suppressing its methylation (14, 33), although the risk of premalignant skin lesions in arsenic-exposed populations appears to be higher among persons with lower than average whole-blood selenium concentrations (34). Thus, the generalizability of the findings of the selenium supplementation trial is not clear. No arsenic exposure-related lesions have been seen in our study population who had no unusual background exposures.

The majority of our study participants had selenium concentrations in the normal range (0.5-2.5 μmol/L) including those in the lowest group. The selenium content of serum and plasma is generally comparable and both respond to short-term changes in dietary selenium intake (35, 36). A general cancer-preventive effect of selenium supplementation has been described in persons who were initially in the lowest two tertiles of serum selenium concentrations within their study population (<121.6 ng/mL or 1.6 μmol/L; ref. 37). Furthermore, experimental study has shown that the selenium plasma concentration needed to achieve optimal platelet glutathione peroxidase activity lies between 1.3 and 1.5 μmol/L (38). In our study, such concentrations were achieved only by the participants in the highest group of serum concentrations, for whom the observed protection against SCC was strongest.

In previous analysis of nutrient intake in this study population, selenium intake per se was not associated with BCC or SCC incidence (39), and correlations between intake and serum concentrations were very weak (r = 0.09; P < 0.01). None of our study participants have reported use of selenium-containing supplements. Due to large variations in food selenium content, serum selenium concentration is a better indicator of selenium status than estimates of dietary intake, although correlations with functional indicators such as glutathione peroxidase are progressively weaker for blood concentrations >1.0 μmol/L (40). Globally soil levels are highly variable and this is also the case within Australia. The selenium content of the foods is known to vary up to 5-fold between states but also can vary up to 5-fold within states (41). Furthermore, the food supply in Australia is very mobile, and produce grown in a region tends to get distributed to many different areas of the country. Thus, dietary intake is generally not strongly dependent on local selenium soil concentrations, although direct evidence for that is lacking.

Serum selenium was the only biomarker of selenium status that was measured in our study; toenail or erythrocyte selenium concentrations are better indicators of long-term selenium status (34). Although in a previous nested case-control study of BCC in this population we found no associations between serum concentrations of selenium (or carotenoids or α-tocopherol) and BCC (42), the previous study was based only on cases identified after a 4.5-year follow-up period, which would explain the different earlier findings.

A major strength of this study was its prospective nature and the extensive characterization of past sun exposure and of skin cancer occurrence in all participants. Our study was based on analysis of histologically confirmed BCC and SCC, which were ascertained through an extensive surveillance system. We consider any bias due to misclassification of participants due to misdiagnosis or missed diagnosis of skin cancer very unlikely. Any confounding effect of the β-carotene supplement taken during the Nambour field trial that preceded this study was ruled out by including only those who received placebo treatment in the present analyses. Although only one measurement of serum antioxidant concentrations was available, in a randomly selected subgroup of our study population, repeated measurements of serum antioxidant concentrations were very stable (r ≥ 0.60). Furthermore, serum concentrations of antioxidants have shown high correlations with concentrations in the skin, with both a single-blood analysis and the average of repeated blood analyses (43). One third of the participants had already had skin cancer before the start of the study. The results from all multivariable models were adjusted for and are thus independent of the presence or absence of a skin cancer history before baseline.

A drawback in our study was the high coefficient of variation in our serum measurements of lutein and zeaxanthin, which may have caused dilution of associations with skin cancer incidence. We measured serum concentrations of carotenoids, α-tocopherol, and selenium because of their possible antioxidant properties. However, the antioxidant capacity of human tissues is influenced by a wide variety of nutrients and other factors (44). Thus, the effect of the antioxidants measured in this study on the overall antioxidant defense capabilities of the skin may be limited. Although serum selenium concentrations were not included in our study of stability over time, food is the main source of selenium, so we expect that the stability of selenium concentrations was as high as that of the other five measured analytes, indicating the generally stable overall dietary intake of our study population. We tested a large number of associations in this study; thus, caution needs to be exercised in its interpretation due to the possibility of chance findings. However, our observation that selenium concentrations specifically showed inverse associations with both BCC and SCC tumor incidence, and moreover these reflected the results based on people newly affected in the study population at large, suggest that chance is an unlikely explanation for this finding.

From this population-based prospective study, we conclude that although concentrations of α-carotene, β-carotene, β-cryptoxanthin, lutein and zeaxanthin, and α-tocopherol appear not to be associated with skin cancer incidence, relatively high serum selenium concentrations at baseline are associated with ∼60% lower incidence of BCC and of SCC tumors in this population. The inverse association with BCC is especially salient given the dearth of existing evidence about the possible dietary links of BCC. Given the results of a previous intervention trial, our findings suggest that a U-shaped relationship between serum selenium concentrations and cutaneous SCC may exist. Evidence from other populations, particularly those that include persons with relatively low serum selenium concentrations, is needed to confirm this.

No potential conflicts of interest were disclosed.

Grant support: National Health and Medical Research Council of Australia (data collection) and World Cancer Research Fund International (data analysis).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank the Nambour Skin Cancer Study participants for continued support and Prof. Gail Williams (School of Population Health, University of Queensland) and Nirmala Pandeya (Queensland Institute of Medical Research) for statistical advice.

1
Diepgen TL, Mahler V. The epidemiology of skin cancer.
Br J Dermatol
2002
;
146
Suppl 61:
1
–6.
2
Christenson LJ, Borrowman TA, Vachon CM, et al. Incidence of basal cell and squamous cell carcinomas in a population younger than 40 years.
JAMA
2005
;
294
:
681
–90.
3
de Vries E, Louwman M, Bastiaens M, de Gruijl F, Coebergh JW. Rapid and continuous increases in incidence rates of basal cell carcinoma in the southeast Netherlands since 1973.
J Invest Dermatol
2004
;
123
:
634
–8.
4
Ullrich SE. Modulation of immunity by ultraviolet radiation: key effects on antigen presentation.
J Invest Dermatol
1995
;
105
:
30
–6S.
5
Camplejohn RS. DNA damage and repair in melanoma and non-melanoma skin cancer.
Cancer Surv
1996
;
26
:
193
–206.
6
Takeshita K, Takajo T, Hirata H, Ono M, Utsumi H. In vivo oxygen radical generation in the skin of the protoporphyria model mouse with visible light exposure: an L-band ESR study.
J Invest Dermatol
2004
;
122
:
1463
–70.
7
Pence BC, Delver E, Dunn DM. Effects of dietary selenium on UVB-induced skin carcinogenesis and epidermal antioxidant status.
J Invest Dermatol
1994
;
102
:
759
–61.
8
Sies H, Stahl W. Nutritional protection against skin damage from sunlight.
Annu Rev Nutr
2004
;
24
:
173
–200.
9
Black HS, Gerguis J. Modulation of dietary vitamins E and C fails to ameliorate β-carotene exacerbation of UV carcinogenesis in mice.
Nutr Cancer
2003
;
45
:
36
–45.
10
Steenvoorden DP, van Henegouwen GM. The use of endogenous antioxidants to improve photoprotection.
J Photochem Photobiol B
1997
;
41
:
1
–10.
11
Bohm F, Edge R, Lange L, Truscott TG. Enhanced protection of human cells against ultraviolet light by antioxidant combinations involving dietary carotenoids.
J Photochem Photobiol B
1998
;
44
:
211
–5.
12
Dorgan JF, Boakye NA, Fears TR, et al. Serum carotenoids and α-tocopherol and risk of nonmelanoma skin cancer.
Cancer Epidemiol Biomarkers Prev
2004
;
13
:
1276
–82.
13
Karagas MR, Greenberg ER, Nierenberg D, et al. Risk of squamous cell carcinoma of the skin in relation to plasma selenium, α-tocopherol, β-carotene, and retinol: a nested case-control study.
Cancer Epidemiol Biomarkers Prev
1997
;
6
:
25
–9.
14
Duffield-Lillico AJ, Slate EH, Reid ME, et al. Selenium supplementation and secondary prevention of nonmelanoma skin cancer in a randomized trial.
J Natl Cancer Inst
2003
;
95
:
1477
–81.
15
Green A, Beardmore G, Hart V, Leslie D, Marks R, Staines D. Skin cancer in a Queensland population.
J Am Acad Dermatol
1988
;
19
:
1045
–52.
16
Green A, Battistutta D, Hart V, et al. The Nambour Skin Cancer and Actinic Eye Disease Prevention Trial: design and baseline characteristics of participants.
Control Clin Trials
1994
;
15
:
512
–22.
17
Green A, Williams G, Neale R, et al. Daily sunscreen application and β-carotene supplementation in prevention of basal-cell and squamous-cell carcinomas of the skin: a randomised controlled trial.
Lancet
1999
;
354
:
723
–9.
18
Valery PC, Neale R, Williams G, Pandeya N, Siller G, Green A. The effect of skin examination surveys on the incidence of basal cell carcinoma in a Queensland community sample: a 10-year longitudinal study.
J Investig Dermatol Symp Proc
2004
;
9
:
148
–51.
19
Sowell AL, Huff DL, Yeager PR, Caudill SP, Gunter EW. Retinol, α-tocopherol, lutein/zeaxanthin, β-cryptoxanthin, lycopene, α-carotene, trans-β-carotene, and four retinyl esters in serum determined simultaneously by reversed-phase HPLC with multiwavelength detection.
Clin Chem
1994
;
40
:
411
–6.
20
Kattermann R, Jaworek D, Moller G, et al. Multicentre study of a new enzymatic method of cholesterol determination.
J Clin Chem Clin Biochem
1984
;
22
:
245
–51.
21
Carnrick GR, Manning DC, Slavin W. Determination of selenium in biological materials with platform furnace atomic-absorption spectroscopy and Zeeman background correction.
Analyst
1983
;
108
:
1297
–312.
22
Willett WC. Nutritional epidemiology. Monographs in epidemiology and biostatistics. New York: Oxford University Press; 1998.
23
Zou G. A modified Poisson regression approach to prospective studies with binary data.
Am J Epidemiol
2004
;
159
:
702
–6.
24
Gardner W, Mulvey EP, Shaw EC. Regression analyses of counts and rates: Poisson, overdispersed Poisson, and negative binomial models.
Psychol Bull
1995
;
118
:
392
–404.
25
Burke KE, Combs GF, Jr., Gross EG, Bhuyan KC, Abu-Libdeh H. The effects of topical and oral l-selenomethionine on pigmentation and skin cancer induced by ultraviolet irradiation.
Nutr Cancer
1992
;
17
:
123
–37.
26
Burke KE, Clive J, Combs GF, Jr., Nakamura RM. Effects of topical l-selenomethionine with topical and oral vitamin E on pigmentation and skin cancer induced by ultraviolet irradiation in Skh:2 hairless mice.
J Am Acad Dermatol
2003
;
49
:
458
–72.
27
Rafferty TS, McKenzie RC, Hunter JA, et al. Differential expression of selenoproteins by human skin cells and protection by selenium from UVB-radiation-induced cell death.
Biochem J
1998
;
332
:
231
–6.
28
Clark LC, Graham GF, Crounse RG, Grimson R, Hulka B, Shy CM. Plasma selenium and skin neoplasms: a case-control study.
Nutr Cancer
1984
;
6
:
13
–21.
29
Breslow RA, Alberg AJ, Helzlsouer KJ, et al. Serological precursors of cancer: malignant melanoma, basal and squamous cell skin cancer, and prediagnostic levels of retinol, β-carotene, lycopene, α-tocopherol, and selenium.
Cancer Epidemiol Biomarkers Prev
1995
;
4
:
837
–42.
30
Clark LC, Combs GF, Jr., Turnbull BW, et al. Effects of selenium supplementation for cancer prevention in patients with carcinoma of the skin. A randomized controlled trial. Nutritional Prevention of Cancer Study Group.
JAMA
1996
;
276
:
1957
–63.
31
Albanes D, Heinonen OP, Taylor PR, et al. α-Tocopherol and β-carotene supplements and lung cancer incidence in the α-tocopherol, β-carotene cancer prevention study: effects of base-line characteristics and study compliance.
J Natl Cancer Inst
1996
;
88
:
1560
–70.
32
Omenn GS, Goodman GE, Thornquist MD, et al. Risk factors for lung cancer and for intervention effects in CARET, the β-Carotene and Retinol Efficacy Trial.
J Natl Cancer Inst
1996
;
88
:
1550
–9.
33
Styblo M, Thomas DJ. Selenium modifies the metabolism and toxicity of arsenic in primary rat hepatocytes.
Toxicol Appl Pharmacol
2001
;
172
:
52
–61.
34
Chen Y, Hall M, Graziano JH, et al. A prospective study of blood selenium levels and the risk of arsenic-related premalignant skin lesions.
Cancer Epidemiol Biomarkers Prev
2007
;
16
:
207
–13.
35
Gibson RS. Principles of nutritional assessment. Oxford: Oxford University Press; 2005.
36
Levander OA, Alfthan G, Arvilommi H, et al. Bioavailability of selenium to Finnish men as assessed by platelet glutathione peroxidase activity and other blood parameters.
Am J Clin Nutr
1983
;
37
:
887
–97.
37
Duffield-Lillico AJ, Reid ME, Turnbull BW, et al. Baseline characteristics and the effect of selenium supplementation on cancer incidence in a randomized clinical trial: a summary report of the Nutritional Prevention of Cancer Trial.
Cancer Epidemiol Biomarkers Prev
2002
;
11
:
630
–9.
38
Alfthan G, Aro A, Arvilommi H, Huttunen JK. Selenium metabolism and platelet glutathione peroxidase activity in healthy Finnish men: effects of selenium yeast, selenite, and selenate.
Am J Clin Nutr
1991
;
53
:
120
–5.
39
Heinen MM, Hughes MC, Ibiebele TI, Marks GC, Green AC, van der Pols JC. Intake of antioxidant nutrients and the risk of skin cancer. Eur J Cancer 2007;43:2707–16.
40
Diplock AT. Indexes of selenium status in human populations.
Am J Clin Nutr
1993
;
57
:
256
–8S.
41
McNaughton SA, Marks GC. Selenium content of Australian foods: a review of literature values.
J Food Comp Anal
2002
;
15
:
169
–82.
42
McNaughton SA, Marks GC, Gaffney P, Williams G, Green AC. Antioxidants and basal cell carcinoma of the skin: a nested case-control study.
Cancer Causes Control
2005
;
16
:
609
–18.
43
Peng YM, Peng YS, Lin Y, Moon T, Baier M. Micronutrient concentrations in paired skin and plasma of patients with actinic keratoses: effect of prolonged retinol supplementation.
Cancer Epidemiol Biomarkers Prev
1993
;
2
:
145
–50.
44
Collins AR. Assays for oxidative stress and antioxidant status: applications to research into the biological effectiveness of polyphenols.
Am J Clin Nutr
2005
;
81
:
261
–7S.