Role of Dietary Factors in the Development of Basal Cell Cancer and Squamous Cell Cancer of the Skin

Transcription

1 1596 Cancer Epidemiology, Biomarkers & Prevention Role of Dietary Factors in the Development of Basal Cell Cancer and Squamous Cell Cancer of the Skin Sarah A. McNaughton, 1 Geoffrey C. Marks, 1 and Adele C. Green 2 1 School of Population Health, University of Queensland and 2 Queensland Institute of Medical Research, Herston, Queensland Abstract The role of dietary factors in the development of skin has been investigated for many years; however, the results of epidemiologic studies have not been systematically reviewed. This article reviews human studies of basal cell (BCC) and squamous cell (SCC) and includes all studies identified in the published scientific literature investigating dietary exposure to fats, retinol, carotenoids, vitamin E, vitamin C, and selenium. A total of 26 studies were critically reviewed according to study design and quality of the epidemiologic evidence. Overall, the evidence suggests a positive relationship between fat intake and BCC and SCC, an inconsistent association for retinol, and little relation between B-carotene and BCC or SCC development. There is insufficient evidence on which to make a judgment about an association of other carotenoids with skin. The evidence for associations between vitamin E, vitamin C, and selenium and both BCC and SCC is weak. Many of the existing studies contain limitations, however, and further well-designed and implemented studies are required to clarify the role of diet in skin. Additionally, the role of other dietary factors, such as flavonoids and other polyphenols, which have been implicated in skin development in animal models, needs to be investigated. (Cancer Epidemiol Biomarkers Prev 2005;14(7): ) Introduction Diet may play a substantial role in the development of many s and it has been estimated that f35% of all s are due to dietary factors (1). Although the influence of diet on the development of skin is of considerable interest (2), the results of all salient epidemiologic studies have not been systematically reviewed. Keratinocytic is the most commonly occurring among light-skinned populations and includes two types of with distinct clinical, pathologic, and genetic features: basal cell (BCC) and squamous cell (SCC; refs. 3, 4). The key factor responsible for development of keratinocytic is UV radiation (5). Other factors, including diet, may play an important role (2, 6) and dietary factors are hypothesized to act at many points in the multistage process of carcinogenesis (7, 8). UV radiation induces skin through the formation of DNA mutations or lesions induced by the absorption of UV photons and damage to various immune mechanisms (9-12). UV-induced free radicals can also damage cellular proteins and cell membrane carbohydrates and fatty acids, thus influencing the process of carcinogenesis through altered cellular communication (11), changes to cell receptor functioning (7), and alterations in DNA repair systems and cell proliferation pathways. This review focuses on the role of fats, retinol, carotenoids, vitamin E, vitamin C, and selenium in the development of BCC and SCC and aims to evaluate the available epidemiologic evidence for these dietary factors. These particular dietary factors have been hypothesized to play a role in skin development based on the results of animal and in vitro studies and have been the focus of Received 1/12/05; revised 4/12/05; accepted 4/26/05. Grant support: National Health and Medical Research Council Public Health Postgraduate Research Scholarship (S.A. McNaughton). 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. Requests for reprints: Sarah McNaughton, Medical Research Council Human Nutrition Research, Elsie Widdowson Laboratory, Fulbourn Road, Cambridge CB1 9NL, United Kingdom. Phone: ; Fax: sarah.mcnaughton@mrc-hnr.cam.ac.uk Copyright D 2005 American Association for Cancer Research. epidemiologic research in humans (7, 8, 11, 13-22). Table 1 summarizes the potential mechanisms by which these dietary factors may act in the development and prevention of skin. Although there are other nutrients that may potentially influence skin development, such as folate and polyphenols, to date, few studies have examined their role. Materials and Methods All published human studies investigating dietary exposure to fats, retinol, carotenoids, vitamin E, vitamin C, and selenium in relation to BCC or SCC of the skin were reviewed. Literature searches were conducted via Medline (23) and the reference lists of relevant articles were crosschecked and any additional studies were identified. Studies were included if they investigated dietary intake or if they used biomarkers of dietary exposure, such as plasma/serum biomarkers. The studies were reviewed according to their study design, including details of the case ascertainment and whether the diagnosis of BCC or SCC was based on selfreport, clinical or histopathologic diagnosis, and aspects of the exposure measurement, such as the dietary assessment method and the timing of the biomarker measurement. Other considerations included selection of the study population, the number of skin cases, and the management of potential confounding factors. A few studies did not distinguish between the two types of keratinocytic ; that is, they combined BCC and SCC in the analysis and this limitation has been highlighted as necessary. The results of each study were considered in the context of a hierarchy of epidemiologic evidence. Case-control studies provide the weakest evidence of a relationship between an exposure and disease compared with other study designs and, in particular, hospital-based studies because of inherent selection bias compared with population-based designs. Methodologically sound cohort and nested case-control studies provide good evidence and sound intervention studies are considered to provide the best evidence (24).

2 Cancer Epidemiology, Biomarkers & Prevention 1597 Results A total of 26 studies investigating BCC and SCC risk in relation to the specific dietary factors of interest were identified in the literature, published between 1983 and This included six case-control studies (five hospitalbased and one population-based), five cohort studies, seven nested case-control studies, and eight intervention studies as summarized in Tables 2, 3, 4, and 5 respectively. The findings for each dietary factor have been considered separately. Fats. There have been three cases-control studies (two hospital-based and one population-based), two cohort studies, and two intervention studies investigating the relationship between dietary fat and keratinocytic in humans (15, 25-30). Overall, the studies suggest there may be a possible relationship between fat intake and keratinocytic risk. One cohort study showed a significant positive association between long chain n 3 fatty acids and BCC and an inverse relationship with total fat and monounsaturated fat (30), whereas the remaining two hospital-based case-control studies (26, 29) and one cohort study (28) showed no significant effects for BCC. There was no association found between dietary fat intake and SCC in the hospital-based case-control study by Hakim et al. (27). Table 1. Dietary factors of interest with respect to keratinocytic and their potential mechanisms of action Dietary factor Carotenoids Fats Retinol Selenium Vitamin C Vitamin E NOTE: Refs. 7, 8, 11, Potential anti mechanism Ability to quench singlet oxygen h-carotene enhances immune system functioning Provitamin A carotenoids may influence prevention through conversion to retinoids Promoter of carcinogenesis Alterations in polyunsaturated fatty acid intake can influence the composition of the cell membrane lipids, which may affect intercellular communication and the responsiveness of tissues to growth factors Polyunsaturated fatty acids have a role in immune system functioning as precursors in eicosanoid synthesis Decreases cellular proliferation and enhances differentiation of cells along normal cell lines and prohibits the formation of a tumor mass of undifferentiated cells Enhances humoral and cell-mediated immunity Increases epidermal thickness and decreases the amount of UV light reaching the epidermal cells Cofactor for glutathione peroxidase, an enzyme that protects against oxidative tissue damage Enhances the mixed function oxidase system that can alter the metabolism of carcinogenic compounds Enhances the process of apoptosis or programmed cell death and result in the removal of mutated or damaged cells Functions as a free radical scavenger protecting against lipid peroxidation and restores activity of other antioxidants (e.g., vitamin E) Enhances and stimulates the immune response Hydroxylation of lysine and proline in the synthesis of connective tissue proteins, which affect the integrity of intracellular matrices and may prevent tumor growth Intracellular antioxidant that protects polyunsaturated fatty acids in cell membranes from oxidation Acts as a chain-breaking antioxidant during lipid peroxidation Able to maintain selenium in the reduced state There have been two intervention studies showing a reduction of keratinocytic risk (BCC and SCC combined) with a reduction of fat intake (15, 25). These intervention studies provide stronger evidence than observational studies; however, as BCC and SCC were not investigated separately in these studies, it is unclear whether these effects would apply equally to each type of. Of note, in the intervention study by Jaax et al. (15), the low fat intervention also resulted in increased intakes of h-carotene, vitamin C, and fiber, which suggests an increase in fruit and vegetable consumption. This is important considering the evidence supporting a protective role for fruit and vegetable consumption and these nutrients in many types of (8). Although only one of the existing observational studies found a relationship between keratinocytic and fat intakes, it is also important to consider that the level of fat intake applied in the intervention studies (20% of energy) would be considered very low in the general population (31). It is possible that the range of fat intakes among participants of the observational studies thus did not allow a corresponding effect to be identified and this may explain the contrasting results. Retinol. There have been three case-control studies (all hospital-based), two cohort studies, four nested case-control studies, and two intervention studies that have investigated the relationship between retinol (or vitamin A) and BCC or SCC (26, 28, 30, 32-39). The results of studies investigating the relationship between retinol and keratinocytic have been inconsistent. For BCC, an inverse association was shown with use of vitamin A supplements in the hospital-based case-control study by Wei et al. (39). No relationship was shown between either plasma retinol concentrations or intake in two case-control studies, two nested case-control studies (33), two cohort studies (28, 30), and two intervention studies (36, 37). In contrast, Breslow et al. (32) found that serum retinol levels were higher among patients with BCC compared with controls. With respect to SCC, two intervention studies provided contradictory results with the study by Moon et al. (37) showing a protective effect of retinol, whereas the study by Levine et al. (36) found no protective effect. Both studies used a similar dose of retinol and included subjects with a previous history of skin. The remaining observational studies (one hospital-based case-control study and two nested casecontrol studies) found no relationship with SCC (26, 32, 34). Additionally, in one case-control study and one nested casecontrol study in which BCC and SCC were not investigated separately, there was an inverse association between serum retinol and keratinocytic risk (35, 38). Carotenoids. There have been 15 studies investigating carotenoids and keratinocytic, including 2 case-control studies (both hospital-based), 5 cohort studies, 5 nested casecontrol studies, and 3 intervention trials (28-30, 32-34, 40-45). The majority of these studies have investigated either serum/plasma levels or dietary intake of h-carotene only. Intakes of carotenoids other than h-carotene have been investigated in analyses of the Nurses Health Study (46, 47) and in the cohort study by Dorgan et al. (45), and serum lycopene has been investigated in two nested case-control studies (32, 40). There is little evidence for a protective effect of h-carotene in BCC or SCC. None of the studies investigating intake or plasma h-carotene concentrations found an association with BCC risk (one hospital-based case-control, three nested casecontrol, four cohort studies, and three intervention studies; refs , 32, 34, 40-43, 45, 46). There have been fewer studies investigating h-carotene and SCC risk; however, no relationship was found in the two nested case-control studies (32, 34), two cohort studies (45, 47), and one intervention study (42)

3 1598 Dietary Factors and Skin Cancer Table 2. A summary of the case-control studies investigating the relationship between dietary exposure and keratinocytic Source Cases Controls Findings/results Limitations Graham (26) Clark et al. (49) Kune et al. (35) Wei et al. (39) Sahl et al. (29) Patients admitted to hospital with skin SCC (n = 96) Dermatology clinic patients with histologically diagnosed BCC/or SCC BCC (n = 142) n = 103 SCC (n = 48) BCC and SCC (n = 50) Consecutive hospital patients with histologically confirmed keratinocytic (BCC/SCC) All males Patients from same hospital free from neoplasms of the body systems at which the cases had and gastrointestinal diseases Past and current clinic patients without keratinocytic Randomly chosen elective surgery patients with no previous history of skin All males n =88 n =88 Histopathologically confirmed primary BCC Ages y Free of with non-premalignant skin disorders Recruited from a dermatology Food frequency questionnaire Vitamin A Vitamin C No relationship observed with any dietary variable and SCC Hospital-based selection of controls Limited dietary assessment method (nonquantitative) Fats Plasma selenium OR (95% CI) Blood samples BCC 3.91 ( ) collected after SCC 3.03 ( ) diagnosis of Keratinocytic 2.11 ( ) keratinocytic Comparing lowest to highest decile. Adjusted for age, sun damage, childhood farming, plasma retinol, and plasma total carotenoids Serum h-carotene (Ag/100 ml) Serum vitamin A (Ag/100 ml) Food frequency questionnaire (interview) Interview: vitamin supplement use Cases had lower mean serum h-carotene than controls (P < 0.001) Controls Cases 89 F F 23 Cases had lower mean serum vitamin A than controls (P = 0.02) Controls Cases 72 F F 25 A high intake of h-carotene/ vitamin C rich vegetables was protective with respect to keratinocytic (P < 0.05) OR (95% CI) clinic n = 131 Age matched Vitamin A b 0.20 ( ) n = 200 Vitamin C b 0.46 ( ) Vitamin E b 0.38 ( ) Compared with nonvitamin users. Adjusted for age, sex, smoking, number of severe sunburns, and skin elastosis. b Compared with nonvitamin users. Adjusted for age only. BCC patients referred for Mohs surgery Hakim et al. (27) Histopathologically confirmed nonmetastatic SCC within 4 mo of contact No prior history of skin Dermatology clinic patients without any type of Food frequency questionnaire: mean intakes compared between cases and controls Blood samples collected after diagnosis of keratinocytic Limited dietary assessment method Selection of hospital patients as controls BCC and SCC combined for analysis Low response rate (30% in cases and 21% in controls) Cases represent patients with large, recurrent, or difficult to manage BCCs Recruited from a Matched for Cases Controls P Low response university-based age, sex, and 3,714 F 458 4,292 F rate (43%) dermatology skin type 31.6 F F practice 178 F F n = 46 n = 46 h-carotene (Ag) 99 F F Vitamin E (mg) 4,288 F 472 4,729 F Vitamin C (mg) 77.6 F F Selenium (Ag) 32.6 F F Vitamin A (Ag) Fat (g) Fat (% of kj) Population-based 4 24-h recall OR (95% CI) controls Frequency matched to cases by age category Energy from fat a Total fat Saturated fat Monounsaturated Polyunsaturated (Continued on the following page) 1.34 ( ) 1.03 ( ) 1.18 ( ) 0.86 ( ) 0.93 ( )

4 Cancer Epidemiology, Biomarkers & Prevention 1599 Table 2. A summary of the case-control studies investigating the relationship between dietary exposure and keratinocytic (Cont d) Source Cases Controls Findings/results Limitations Ages z30 y Recruited from the Southeastern Arizona Skin Cancer Registry n = 301 No prior history of skin n = 267 Linolenic acid 1.01 ( ) Eicosapentaenoic acid 0.78 ( ) Docosahexaenoic acid 0.80 ( ) Arachidonic acid 0.70 ( ) a Comparison of >35% of energy intake from fat to V20% of energy intake from fat. All other ORs are a comparison of z90th percentile of intake to <50th percentile of intake. Adjusted for age, gender, total energy intake, history of actinic keratoses, ability to tan, number of freckles on arm, and sunscreen use. Further adjustment for alcohol intake and smoking status did not affect the results. Abbreviations: OR, odds ratio; 95% CI, 95% confidence interval. when either plasma h-carotene or intake was assessed. Only the hospital-based case-control study, which investigated BCC and SCC combined, found higher plasma h-carotene and a higher intake of h-carotene containing vegetables among controls compared with cases (35). There is insufficient evidence to draw solid conclusions on the relationship between other carotenoids and BCC or SCC risk, as the evidence is limited to three cohort studies of a- carotene, h-cryptoxanthin, and lutein and/or zeaxanthin (46, 47) and two nested case-control studies of serum lycopene (32, 40). With respect to a-carotene, no relationship between was found with either BCC (two cohort studies) or SCC (two cohort studies) using either serum or dietary intake measurements (45-47). For h-cryptoxanthin and lutein and/or zeaxanthin, no relationship was identified in two cohort studies of BCC (45, 46), but for SCC the two cohort studies were inconsistent with one showing no effect (47), whereas a recent study using serum levels showed a positive relationship with SCC (45). Similarly, lycopene was found not to be associated with either BCC (one nested case-control study and one cohort study) or SCC (two nested case-control studies and one cohort study; refs. 32, 40, 45). Vitamin E. There have been 12 studies investigating vitamin E and the risk of keratinocytic (28-30, 32-34, 39, 40, 45-48). Five of the studies investigated serum levels of vitamin E, whereas six of the studies investigated dietary intake of vitamin E and one study investigated intake of vitamin E supplements. The evidence for a protective effect of vitamin E in BCC or SCC is weak. For BCC risk, no relationship was found between plasma and intake in six studies (one case-control, three cohort, and two nested case-control; refs , 32, 40, 45). An inverse relationship was found between intake and BCC risk in two studies (one hospital-based case-control and one nested case-control; refs. 33, 39); in contrast, one cohort study showed a significant positive relationship between vitamin E intake and BCC risk (46). The four studies that investigated SCC showed no relationship with vitamin E (two nested casecontrol studies and one cohort study; refs. 32, 34, 45, 47). Similarly, the nested case-control study by Wald et al. (48), which did not distinguish between BCC and SCC, also found no significant association between serum vitamin E and keratinocytic risk. Vitamin C. Three case-control studies (all hospital-based), four cohort studies, and one nested case-control study have investigated the dietary intake of vitamin C and one casecontrol study investigated the use of vitamin C supplements and risk of keratinocytic (26, 28-30, 33, 35, 39). With respect to vitamin C and BCC, the available evidence for a protective effect is weak. In analysis of the Nurses Health Study cohort, Fung et al. (46) found a significant, but small, positive relationship between vitamin C intake and BCC, whereas the hospital-based case-control study by Wei et al. (39) showed an inverse association between BCC risk and use of vitamin C supplements. The four remaining studies (one hospital-based case-control study, two cohort studies, and one nested case-control study) found no relationship between vitamin C and BCC (28-30, 33). There are fewer studies investigating vitamin C and SCC risk, with only one hospital-based case-control (26) and one Table 3. A summary of cohort studies investigating the relationship between dietary exposure and keratinocytic Source Sample characteristics Outcome variables Dietary variables and assessment method Findings/results Limitations Hunter et al. (28) Nurses Health Study cohort Self-report of BCC occurrence Semiquantitative food frequency questionnaire RR (95% CI)* 73,366 females 4-y follow-up Diet Diet + supplements Ages y n = 771 BCCs Retinol 1.00 ( ) 1.00 ( ) in 1976 h-carotene equivalents 1.15 ( ) Vitamin C 0.96 ( ) 1.07 ( ) Vitamin E 1.04 ( ) 1.01 ( ) Saturated fat 0.90 ( ) Monounsaturated fat 0.85 ( ) Polyunsaturated fat 0.94 ( ) Adjusted for age and energy intake. No effect in adjusting for sun exposure and smoking. (Continued on the following page) Diagnosis of BCC based on self-reports

5 1600 Dietary Factors and Skin Cancer Table 3. A summary of cohort studies investigating the relationship between dietary exposure and keratinocytic (Cont d) Source Sample characteristics Outcome variables Dietary variables and assessment method Findings/results Limitations Van Dam et al. (30) Fung et al. (46) Fung et al. (47) Dorgan et al. (45) Health Professionals Follow-up Study Self-report of BCC occurrence Semiquantitative food frequency questionnaire RR (95% CI)* 43,217 males 8-y follow-up Total fat 0.81 ( ) Ages y n = 3,190 BCCs Saturated fat 1.03 ( ) Monounsaturated fat 0.79 ( ) Polyunsaturated fat 1.07 ( ) Long chain n ( ) Retinol 0.99 ( ) Carotene 0.94 ( ) Vitamin C 1.12 ( ) Vitamin E 0.94 ( ) Adjusted for age, BMI, 2-y follow-up periods, hair color, energy intake, ancestry, smoking, frequency of routine physical examinations, and mean annual solar radiation in residence region. Dietary fat intake variables adjusted other dietary fat variables. Micronutrients adjusted for other micronutrients. Nurses Health Study cohort Self-report of BCC Semiquantitative food frequency questionnaire RR (95% CI)* 85,836 females 12-y follow-up Vitamin A 1.16 ( ) y in 1976 n = 5,392 BCCs Vitamin C 1.13 ( ) Vitamin E 1.15 ( ) a-carotene 1.00 ( ) h-carotene 1.10 ( ) h-cryptoxanthin 0.94 ( ) Lutein/zeaxanthin 1.10 ( ) Adjusted for health, sun exposure, and sun sensitivity factors Nurses Health Study cohort Self-report of SCC Semiquantitative food frequency questionnaire RR (95% CI)* 85,944 females 10- to 14-y follow-up Retinol 0.85 ( ) Ages y Nurses Health Vitamin C 1.09 ( ) in 1976 Study cohort Vitamin E 1.10 ( ) Health Professionals n = 369 SCCs a-carotene 0.92 ( ) Follow-up Study h-carotene 1.21 ( ) 43,867 males Health Professionals h-cryptoxanthin 1.13 ( ) Ages y Follow-up Study n = 305 SCCs Lutein/zeaxanthin 0.98 ( ) Adjusted for age, current and childhood state of residence, energy, BMI, smoking, and alcohol n = 302 Histologically BCC SCC confirmed BCC or SCC Ages >40 y 5-y follow-up RR (95% CI) c RR (95% CI) c n = 221 BCC Serum a-carotene 1.06 ( ) 1.45 ( ) n = 85 SCC Serum h-carotene 1.01 ( ) 1.47 ( ) Serum lycopene 1.01 ( ) 0.99 ( ) Serum lutein 1.04 ( ) 1.63 ( ) Serum zeaxanthin 0.97 ( ) 2.40 ( ) Serum h-cryptoxanthin 1.06 ( ) 2.15 ( ) Serum a-tocopherol 1.15 ( ) 1.60 ( ) BCC analysis adjusted for clinic, gender, age, solar damage, skin type, number of prior BCCs, BMI, treatment group, high density lipoprotein cholesterol, and low density lipoprotein cholesterol SCC analysis adjusted for clinic, gender, age, solar damage, skin type, number of prior BCCs, number of prior SCCs, BMI, treatment group, high density lipoprotein cholesterol, and low density lipoprotein cholesterol. Abbreviations: RR, relative risk; BMI, body mass index. *Comparing highest quintile with lowest quintile. ccomparing lowest tertile with highest tertile. Diagnosis of BCC based on self-reports Diagnosis of BCC based on self-reports Diagnosis of BCC based on self-reports

6 Cancer Epidemiology, Biomarkers & Prevention 1601 cohort study (47); both found no relationship between vitamin C intake and SCC risk. In addition, the hospital-based case-control study by Kune et al. (35) investigated BCC and SCC combined and found that a high intake of h-carotene and vitamin C containing foods was significantly related to reduced risk of keratinocytic. Selenium. There have been two case-control studies (both hospital-based), three nested case-control studies, and one intervention trial investigating the relationship between selenium and keratinocytic (29, 32-34, 49-51). Three of these studies investigated selenium exposure using serum/ plasma measurements, whereas three studies investigated dietary intakes of selenium. The evidence suggests that there is no protective effect of selenium in BCC or SCC. With respect to selenium and BCC, only one case-control study found an inverse relationship between plasma selenium and BCC (49), whereas no relationship with selenium intake or plasma concentrations was shown in the remaining studies (two case-control studies, two nested case-control studies, and one intervention study; ref. 50). Of the four studies investigating the relationship between selenium and SCC (one case-control study, two nested case-control studies, and one intervention study), none identified a significant relationship (32, 34, 49, 50). Discussion This review has focused on evaluating the possible role of fats, retinol, carotenoids, vitamin E, vitamin C, and selenium in the development of cutaneous BCC and SCC. A total of 26 studies were critically reviewed according to study design and quality of the epidemiologic evidence. Overall, there is a possible positive relationship between fat intake and BCC and SCC, but for the remaining dietary factors the evidence is, at best, weak. The potential mechanisms through which diet may influence the development of skin, as summarized in Table 1, are well supported by the results of animal studies. Overall, the existing epidemiologic studies reviewed here have not provided strong evidence to support the role of dietary factors in skin development. Dietary factors may not be of sufficient importance in keratinocytic risk at a population level for an effect to be detected, and the effect of UV radiation may overwhelm any effects of diet. However, an ability to detect as association may also be hampered by limitations in the existing epidemiologic studies in key areas, including dietary exposure assessment, inclusion of subjects with previous skin s, and case ascertainment. A potential explanation for the lack of effect of many of the dietary factors is that the relevant period of dietary exposure was not measured (28, 50, 52). It is acknowledged that BCCs have a long induction period and that the origins of disease may occur early in life (53). Serum/plasma biomarkers used in many of the studies are hypothesized to represent dietary exposure over the short term, and although the dietary intake measurements used have tended to represent a longer-term measurement of habitual intake (54), the appropriate exposure period may not have been investigated. Therefore, it is still possible that dietary factors may act at earlier stages of keratinocytic development. However, it has been shown that some dietary factors may act in the late stages of carcinogenesis. Synthetic retinoids have rapid effects on BCC risk and probably act in the late stages of carcinogenesis (28). This suggests that different dietary factors may act at different stages of the carcinogenic process, and the relevant exposure period may be different according the dietary factor under consideration. Similarly, with respect to many of the interventions trials, it has been suggested that the treatment period may have been too short (50, 52). In reference to the potential effects of selenium, Clark et al. suggested that although an effect of selenium supplementation on total mortality and lung, colorectal, and prostate incidence could be detected the treatment period for BCC and SCC may have been to short (50). It was suggested that UV radiation increases the risk of BCC and SCC through a genetic mutation preventing apoptosis, which occurs early in the process of skin carcinogenesis, whereas in other types of this mutation occurs late in the process. They proposed that if the primary protective action of selenium is the stimulation of cell death, prevention of SCC and BCC may require a longer treatment period than other s; otherwise, a significant number of cells already contain the mutation and selenium treatment cannot reverse the effects (50). With respect to trials of h- carotene supplementation, Manson et al. (52) also suggested that the treatment and follow-up periods of 5 years may have been too short. These authors also suggested that many of the skin s detected during the study period may have been present when the treatment began; therefore, h-carotene would not have had an effect unless it acted late in carcinogenesis. However, later studies with longer treatment periods also showed no effect, suggesting that this was not the reason for a lack of effect of h-carotene (41). An alternate explanation is that the -protective action seen in observational dietary studies, which has been attributed to h-carotene, may not actually be due to h-carotene but due to some other dietary component that is closely associated with h-carotene (40, 55, 56). This has been suggested in the prevention of lung where a large number of observational studies have supported the role of h-carotene, but large intervention trials have shown either no effect or an adverse effect (57). This review included six case-control studies; however, case-control studies (and in particular, hospital-based casecontrol studies) provide the weakest evidence of a relationship between an exposure and disease compared with other study designs (24). Prospective study designs, such as nested case-control and cohort studies, avoid concerns over recall bias and clearly define the temporal relationship between exposure and onset of disease. This is particularly important when considering serum/plasma biomarkers as measures of dietary exposure (58). Contrasting results have been shown in studies using plasma biomarkers depending on the timing of the plasma biomarkers. In a case-control study of plasma selenium, there were significant inverse relations with BCC (49); however, in a study based on prospectively collected data, there was no association between keratinocytic and selenium (32). Similar discrepancies were identified in the study by Wald et al. (38) in which mean serum retinol levels were significantly lower among cases compared with controls but only when patients with <1 year between collection of the blood sample and diagnosis of were investigated. These results highlight the importance of using prospectively collected blood samples. Many of the studies of BCC and SCC included participants that had had a previous BCC. These subjects are at a higher risk of subsequent skin s (59). It is unclear from the existing literature whether dietary factors act differently in the development of a first skin or in subsequent skin s. The inclusion of participants with a previous history of skin may only affect the findings if dietary factors act in the early stage of keratinocytic skin. As described above, some dietary factors, such as retinol, have been shown to act in the later stages of carcinogenesis; however, early stage effects have not been completely dismissed.

7 1602 Dietary Factors and Skin Cancer Table 4. A summary of the nested case-control studies investigating the relationship between dietary exposure and keratinocytic Source Cases Controls Wald et al. (38) Cases of skin identified through the National Health Service records No distinction between SCC and BCC n =43 Initial cohort: 22,000 men ages y Controls (two per case) matched for age, duration of serum sample storage, and smoking status Wald et al. (48) See Wald et al. (1986) See Wald et al. (1986) n =56 n = 107 Wald et al. (44) See Wald et al. (1986) See Wald et al. (1986) n =56 n = 107 Comstock et al. (40) Breslow et al. (32) Patients with BCC identified through the local county registry n = 21 Cases of BCC and SCC in participants of a blood collection survey. Identified using a local registry BCC (n = 32) SCC (n = 37) Initial cohort: 25,802 volunteers Controls (two per case) were matched for race and sex Initial cohort Controls were matched (two per case) for age and gender Karagas et al. (34) Patients who developed a new, nonrecurrent SCC during the follow-up period of the study n = 132 Patients who did not develop SCC during the study period Controls were matched (two per case) for age, sex, and study center Davies et al. (33) Developed a new BCC Controls (two per cases) were matched for age, sex, and date of dietary assessment 52 females, 57 males y NOTE: Comparing lowest and highest quantile group. *SD not presented. cno 95% CI reported; P for trend across quantile groups was >0.05 in all cases. bdietary intake is modeled as a continuous variable. A further limitation of several studies relates to the methods used for ascertaining cases of keratinocytic. Firstly, several studies relied on health records and registries, which increase the potential for misclassification with respect to skin outcome (32, 33, 38, 40, 44, 48). Secondly, several studies only used self-report of keratinocytic (28, 30, 46, 47). The validity of the self-report measures was assessed in analyses of the Nurses Health Study and the Health

8 Cancer Epidemiology, Biomarkers & Prevention 1603 Table 4. A summary of the nested case-control studies investigating the relationship between dietary exposure and keratinocytic (Cont d) Dietary variables Findings/results Limitations Serum retinol Only mean serum concentrations at <1 y were significantly different (P = 0.01) * BCC and SCC combined for analysis Controls Cases Reliance on local health records for case ascertainment <1 y 790 Ag/L 654 Ag/L 1-2 y 696 Ag/L 666 Ag/L z3 y 676 Ag/L 722 Ag/L Serum vitamin E No significant differences between mean serum concentrations* <1 y 12.0 mg/l 10.4 mg/l 1-2 y 11.5 mg/l 11.1 mg/l z3 y 9.5 mg/l 10.4 mg/l BCC and SCC combined for analysis Controls Cases Reliance on local health records for case Serum h-carotene No significant differences between mean serum concentrations* BCC and SCC combined for analysis Controls Cases Reliance on local health records for case ascertainment OR c 234 Ag/L 226 Ag/L Reliance on registry Serum h-carotene 1.1 records for case ascertainment Serum vitamin E 0.4 Small number of cases Serum lycopene 1.5 BCC SCC OR (95% CI) OR (95% CI) Serum retinol 3.3 ( ) 1.8 ( ) Serum h-carotene 1.3 ( ) 1.4 ( ) Serum lycopene 1.4 ( ) 1.0 ( ) Serum a-tocopherol 2.6 ( ) 1.5 ( ) Serum selenium 0.8 ( ) 0.6 ( ) Adjustment for education, smoking, and hours between last meal and blood collection did not alter the results No significant differences between BCC/SCC cases and controls for any nutrients, except retinol (higher in BCC cases; P = 0.02) OR 95% CI All subjects (cases and controls) had a previous history of at least 1 SCC or BCC Plasma selenium 0.86 ( ) Unknown proportion of controls Plasma a-tocopherol 0.89 ( ) developed BCC during follow-up Plasma h-carotene 0.73 ( ) Plasma retinol 1.43 ( ) No significant differences in mean plasma levels between cases and controls. Adjustment for skin risk factors did not effect risk estimates 7-d food diary OR (95% CI) b Reliance on registry records h-carotene equivalents 1.06 ( ) for case ascertainment Total fat 0.86 ( ) Vitamin A 0.82 ( ) Selenium 1.07 ( ) Vitamin C 1.02 ( ) Professionals Follow-up Study (28, 30, 46), although in the former study it was conducted among a small number of subjects. A few studies did not distinguish between the two types of keratinocytic ; that is, they combined BCC and SCC in the analysis (15, 25, 35, 38, 44, 48). Despite this, these studies were retained in the review as they included two intervention studies, which are considered to provide strong epidemiologic evidence. However, it is

9 1604 Dietary Factors and Skin Cancer Table 5. A summary of the intervention trials investigating the relationship between dietary exposure and keratinocytic Source Sample characteristics Treatment Greenberg et al. (43) 1,805 patients with previous history of skin Oral administration of 50 mg h-carotene or placebo Black et al. (25) 101 patients with previous history of skin Contribution of fat to energy intake in baseline diet: Control group: 40 F 4% Intervention group: 39 F 4% Recommended: Low-fat diet with 20% of energy provided by fat for a 2-y period Compliance: Intervention group: mean fat intake, 21 F 7% of energy intake Control group: mean fat intake, 38 F 6% of energy intake Clark et al. (50) 1,312 patients with a previous BCC or SCC Further analysis in (51) Oral administration of 200 Ag/d selenium or placebo Mean length of treatment = 4.5 y Jaax et al. (15) 115 patients with previous history of skin Contribution of fat to energy intake in baseline diet: Control group: 40 F 4% Intervention group: 39 F 3% Recommended: Low-fat diet with 20% of energy provided by fat for a 2-y period Compliance: Intervention group: mean fat intake, 21 F 6% of energy intake Control group: mean fat intake, 38 F 4% of energy intake Moon et al. (37) Levine et al. (36) z10 actinic keratoses (most recent occurrence within 1 y of recruitment) and V2 pathologically confirmed BCCs or SCCs Ages y Retinol (n = 1,157) Placebo (n = 1,157) Males and females, y with z4 previous BCC/SCC, most recent within the previous year Retinol (n = 173) Isotretinoin (n = 178) Placebo (n = 174) Oral administration of retinol (25,000 IU) or placebo daily for up to 5 y Oral administration for 3 y of either 25,000 IU retinol, 5-10 mg isotretinoin, or placebo Green et al. (42) Random sample of adult community, y Oral administration of 30 mg/d n = 1,621 h-carotene for 4.5 y Frieling et al. (41) Healthy male physicians, y Oral administration of 50 mg every n = 22,071 second day for 12 y unclear whether the effects of diet would apply equally to each type of skin and it is possible that combining the outcomes in the analysis may attenuate any effects of diet. Another important limitation of the existing published literature is that in many of the earlier studies there is inadequate investigation and control for potential confounders (26). Many of the case-control studies match cases and controls

10 Cancer Epidemiology, Biomarkers & Prevention 1605 Table 5. A summary of the intervention trials investigating the relationship between dietary exposure and keratinocytic (Cont d) Outcome variable Results/findings Limitations Occurrence of first new histologically confirmed BCC or SCC Number of histologically confirmed new skin s Comparison of the last 8-mo period with first 8-mo period of study No difference between the control and intervention groups in the rate of occurrence the first new keratinocytic RR (95% CI) = 1.04 ( ) Adjusted for age, sex, skin type, previous skin, smoking, study center, baseline plasma h-carotene and retinol Skin occurrence did not change significantly between the first 8-mo period and the last 8-mo period in the control group and significantly decreased between the first 8-mo period and the last 8-mo period in the intervention group Significantly fewer skin s in the intervention group than the control group in the last 8-mo period (0.02 versus 0.22 s/patient) Subjects had a previous history of skin SCC and BCC combined for analysis Incidence of histologically confirmed BCC or SCC Time to first BCC or SCC Number of histologically confirmed new confirmed skin s Comparison of the last 8-mo period with first 8-mo period of study Time to first new occurrence of BCC or SCC (pathologically confirmed) RR (95% CI) SCC 1.14 ( ) BCC 1.10 ( ) Hazard ratio (95% CI) BCC 1.09 ( ) SCC 1.25 ( ) Keratinocytic 1.17 ( ) Skin occurrence: In the control group: did not change between the first 8-mo period and the second or third 8-mo periods In the intervention group: significantly lower in the last 8-mo period compared with the first 8-mo period In control and intervention groups in the last 8-mo period: significantly different (0.02 vs 0.26 s/patient) Comparing placebo and retinol supplemented subjects: Hazard ratio (95% CI) BCC 1.06 ( ) SCC 0.74 ( ) Percentage of calories from fat was not the only component of diet that was altered. Vitamin C, h-carotene and fiber increased in the intervention group Subjects had a previous history of skin SCC and BCC combined for analysis Previous history of skin Time to first new occurrence of BCC or SCC (pathologically confirmed) No differences between those who received retinol, isotretinoin, or placebo with respect to time to first new occurrence of either BCC or SCC or total number of tumors Incidence of histologically confirmed BCC or SCC First keratinocytic, BCC, or SCC RR (95% CI) BCC 1.04 ( ) SCC 1.35 ( ) RR (95% CI) Keratinocytic 0.98 ( ) BCC 0.99 ( ) SCC 0.97 ( ) Adjusted for age for major factors, such as age and sex (29), but the potential confounding effects of other skin risk factors are overlooked. This is important as there is clustering of healthy lifestyle behaviors (60, 61); for example, intakes of fruits and vegetables are higher in nonsmokers compared with smokers (62). These relationships may also exist between healthy eating behaviors and a variety of sun protection behaviors and other skin risk factors.

11 1606 Dietary Factors and Skin Cancer Conclusions Overall, the studies suggest that there may be a possible positive relationship between fat intake and BCC and SCC, whereas the results for retinol are inconsistent with both positive and negative relationships with both BCC and SCC observed. There is little evidence for a role for h-carotene in BCC or SCC development, whereas there is insufficient evidence on which to make a judgment for other carotenoids for either BCC or SCC. The evidence for associations between vitamin E, vitamin C, and selenium and BCC or SCC is weak. Many of the studies of diet and keratinocytic to date have limitations. The most common problems relate to limitations in the dietary exposure assessment, reliance on health records and registries for cases ascertainment, combined analysis of BCC and SCC, small numbers of BCC or SCC cases, a lack of adjustment for potential confounding factors, and a lack of population-based studies. High-quality epidemiologic studies with adequate diagnosis of skin, sufficient power, and adjustment for important confounding factors are required to clarify the role of many dietary factors in the development of skin. In addition, there remain several dietary factors that may have anti potential that have not been substantially investigated with respect to skin, including flavonoids and other polyphenols, folate, vitamin D, Allium compounds, coumarins, riboflavin, and zinc (8). Further investigation into these dietary factors that may be involved in BCC and SCC development is required. References 1. Working Group on Diet and Cancer. Nutritional aspects of the development of : report of the Working Group on Diet and Cancer of the Committee on Medical Aspects of Food and Nutrition Policy. London: Department of Health (Great Britain); Sies H, Stahl W. Nutritional protection against skin damage from sunlight. Annu Rev Nutr 2004;24: Leigh IM, Newton-Bishop JA, Kripke ML. Skin. An introduction. Cancer Surv 1996;26: 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: IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Solar and ultraviolet radiation. IARC monographs on the evaluation of carcinogenic risks to humans. Vol. 55. Lyon: IARC; Green A, Beardmore G, Hart V, et al. Skin in a Queensland population. J Am Acad Dermatol 1988;19: Flagg EW, Coates RJ, Greenberg RS. Epidemiologic studies of antioxidants and in humans. J Am Coll Nutr 1995;14: World Cancer Research Fund. Food, nutrition and the prevention of : a global perspective. Washington: American Institute for Cancer Research; Ullrich SE. Modulation of immunity by ultraviolet radiation key effects on antigen presentation. J Invest Dermatol 1995;105:30 6S. 10. Camplejohn RS. DNA damage and repair in melanoma and non-melanoma skin. Cancer Surv 1996;26: Axelrod M, Serafin D, Klitzman B. Ultraviolet light and free radicals: an immunologic theory of epidermal carcinogenesis. Plast Reconstr Surg 1990; 86: Grossman D, Leffell DJ. The molecular basis of nonmelanoma skin : new understanding. Arch Dermatol 1997;133: Carroll KK. Dietary fat and : specific action or caloric effect? J Nutr 1986;116: Greenwald P, Clifford CK, Milner JA. Diet and prevention. Eur J Cancer 2001;37: Jaax S, Scott LW, Wolf JE, Thornby JI, Black HS. General guidelines for a low-fat diet effective in the management and prevention of nonmelanoma skin. Nutr Cancer 1997;27: Craven NM, Griffiths CE. Retinoids in the management of non-melanoma skin and melanoma. Cancer Surv 1996;26: De Luca LM, Darwiche N, Celli G, et al. Vitamin A in epithelial differentiation and skin carcinogenesis. Nutr Rev 1994;52:S Fryer MJ. Evidence for the photoprotective effects of vitamin E. Photochem Photobiol 1993;58: Steinmetz KA, Potter JD. Vegetables, fruit and. II. Mechanisms. Cancer Causes Control 1991;2: Black HS, Lenger WA, Gerguis J, Thornby JI. Relation of antioxidants and level of dietary lipid to epidermal lipid peroxidation and ultraviolet carcinogenesis. Cancer Res 1985;45: Black HS, Mathews-Roth MM. Protective role of butylated hydroxytoluene and certain carotenoids in photocarcinogenesis. Photochem Photobiol 1991; 53: Pence BC, Delver E, Dunn DM. Effects of dietary selenium on UVB-induced skin carcinogenesis and epidermal antioxidant status. J Invest Dermatol 1994;102: Medline. Bethesda: U.S. National Library of Medicine; Beaglehole R, Bonita R, Kjellstrom T. Basic epidemiology. Geneva: WHO; Black HS, Thornby JI, Wolf JE, et al. Evidence that a low-fat diet reduces the occurrence of non-melanoma skin. Int J Cancer 1995; 62: Graham S. Results of case-control studies of diet and in Buffalo, New York. Cancer Res 1983;43: S. 27. Hakim IA, Harris RB, Ritenbaugh C. Fat intake and risk of squamous cell carcinoma of the skin. Nutr Cancer 2000;36: Hunter DJ, Colditz GA, Stampfer MJ, et al. Diet and risk of basal cell carcinoma of the skin in a prospective cohort of women. Ann Epidemiol 1992;2: Sahl WJ, Glore S, Garrison P, Oakleaf K, Johnson SD. Basal cell carcinoma and lifestyle characteristics. Int J Dermatol 1995;34: Van Dam RM, Huang Z, Giovannucci E, et al. Diet and basal cell carcinoma of the skin in a prospective cohort of men. Am J Clin Nutr 2000;71: Mahan LK, Arlin S. Krause s food, nutrition, & diet therapy. Philadelphia: Saunders; Breslow RA, Alberg AJ, Helzlsouer KJ, et al. Serological precursors of : malignant melanoma, basal and squamous cell skin, and prediagnostic levels of retinol, h-carotene, lycopene, a-tocopherol, and selenium. Cancer Epidemiol Biomarkers Prev 1995;4: Davies TW, Treasure FP, Welch AA, Day NE. Diet and basal cell skin : results from the EPIC-Norfolk cohort. Br J Dermatol 2002;146: Karagas MR, Greenberg ER, Nierenberg D, et al. Risk of squamous cell carcinoma of the skin in relation to plasma selenium, a-tocopherol, h-carotene, and retinol: a nested case-control study. Cancer Epidemiol Biomarkers Prev 1997;6: Kune GA, Bannerman S, Field B, et al. Diet, alcohol, smoking, serum h-carotene, and vitamin A in male nonmelanocytic skin patients and controls. Nutr Cancer 1992;18: Levine N, Moon TE, Cartmel B, et al. Trial of retinol and isotretinoin in skin prevention: a randomized, double-blind, controlled trial. Cancer Epidemiol Biomarkers Prev 1997;6: Moon TE, Levine N, Cartmel B, et al. Effect of retinol in preventing squamous cell skin in moderate-risk subjects: a randomized, double-blind, controlled trial. Cancer Epidemiol Biomarkers Prev 1997;6: Wald N, Boreham J, Bailey A. Serum retinol and subsequent risk of. Br J Cancer 1986;54: Wei Q, Matanoski GM, Farmer ER, Strickland P, Grossman L. Vitamin supplementation and reduced risk of basal cell carcinoma. J Clin Epidemiol 1994;47: Comstock GW, Helzlsouer KJ, Bush TL. Prediagnostic serum levels of carotenoids and vitamin E as related to subsequent in Washington County, Maryland. Am J Clin Nutr 1991;53:260 4S. 41. Frieling UM, Schaumberg DA, Kupper TS, Muntwyler J, Hennekens CH. A randomized, 12-year primary-prevention trial of h carotene supplementation for nonmelanoma skin in the Physician s Health Study. Arch Dermatol 2000;136: Green A, Wiiliams G, Neale R, et al. Daily sunscreen application and h carotene supplementation in prevention of basal-cell and squamous-cell carcinomas of the skin: a randomised trial. Lancet 1999;354: Greenberg ER, Baron JA, Stukel TA, et al. A clinical trial of h carotene to prevent basal-cell and squamous-cell s of the skin. N Engl J Med 1990;323: Wald NJ, Thompson SG, Densem JW, Boreham J, Bailey A. Serum h-carotene and subsequent risk of : results from the BUPA Study. Br J Cancer 1988;57: Dorgan JF, Boakye NA, Fears TR, et al. Serum carotenoids and a-tocopherol and risk of nonmelanoma skin. Cancer Epidemiol Biomarkers Prev 2004;13: Fung TT, Hunter DJ, Spiegelman D, et al. Vitamins and carotenoids intake and the risk of basal cell carcinoma of the skin in women (United States). Cancer Causes Control 2002;13: Fung TT, Spiegelman D, Egan KM, et al. Vitamin and carotenoid intake and risk of squamous cell carcinoma of the skin. Int J Cancer 2003;103: Wald NJ, Thompson SG, Densem JW, Boreham J, Bailey A. Serum vitamin E and subsequent risk of. Br J Cancer 1987;56: Clark LC, Graham GF, Crounse RG, et al. Plasma selenium and skin neoplasms: a case-control study. Nutr Cancer 1984;6: Clark LC, Combs GF, Turnbull BW, et al. Effects of selenium supplementation for prevention in patients with carcinoma of the skin. A randomized controlled trial. JAMA 1996;276: