Measures of Cumulative Exposure from a Standardized Sun Exposure History Questionnaire: A Comparison with Histologic Assessment of Solar Skin Damage

Transcription

1 American Journal of Epidemiology Copyright ª 2007 by the Johns Hopkins Bloomberg School of Public Health All rights reserved; printed in U.S.A. Vol. 165, 6 DOI: /aje/kwk055 Advance Access publication January 4, 2007 Practice of Epidemiology Measures of Cumulative Exposure from a Standardized Sun Exposure History Questionnaire: A Comparison with Histologic Assessment of Solar Skin Damage Margaret R. Karagas 1, Michael S. Zens 1, Heather H. Nelson 2, Kiyohiko Mabuchi 3, Ann E. Perry 4, Therese A. Stukel 5, Leila A. Mott 1, Angeline S. Andrew 1, Katie M. Applebaum 2, and Martha Linet 3 1 Section of Biostatistics and Epidemiology, Department of Community and Family Medicine, and the Norris Cotton Cancer Center, Dartmouth Medical School, Lebanon, NH. 2 Department of Environmental Health, Harvard School of Public Health, Boston, MA. 3 Radiation Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD. 4 Department of Pathology, Dartmouth Hitchcock Medical Center, Lebanon, NH. 5 Institute for Clinical Evaluative Sciences, Toronto, Canada. Received for publication April 14, 2006; accepted for publication August 22, Ultraviolet radiation exposure is the dominant environmental determinant of all major forms of skin cancer; however, the nature of the association is incompletely understood. Existing instruments to capture sun exposure history tend to yield reproducible results, but the validity of these responses is unknown. To address this question, the authors examined the relation between responses to a standardized sun exposure instrument and histologic evidence of actinic damage in a population-based study of keratinocyte cancers from New Hampshire diagnosed from July 1, 1997, through March 31, A single study dermatopathologist histologically reviewed the adjacent skin of 925 skin cancer biopsies for the presence of solar keratoses and the extent of solar elastosis. The authors compared these measures with responses to a personal interview on history of sunburns, sunbathing, and time spent outdoors. Focusing on site-specific exposure, they found variables that estimated cumulative exposure related to histologic evidence of actinic damage. In contrast, measures of acute/intermittent exposure were generally unrelated to solar damage histologically. Findings suggest that cumulative, but not intermittent, measures of sun exposure derived from a personal interview appear to reflect a person s exposure history based on histologic evidence. case-control studies; questionnaires; skin neoplasms; ultraviolet rays; validation studies Abbreviations: BCC, basal cell carcinoma; SCC, squamous cell carcinoma; UVR, ultraviolet radiation. Ultraviolet radiation (UVR) exposure plays a major role in the etiology of all major forms of skin cancer: basal cell carcinoma (BCC), squamous cell carcinoma (SCC), and melanoma; however, the nature of the association is incompletely understood (1). SCC relates primarily to measures of cumulative exposure, but some studies suggest that, similar to melanomas, BCC in particular may be caused by lower levels of intense, intermittent sun exposure (i.e., recreational) and by exposure during childhood (2 4). Additionally, risks of both SCC and BCC may be affected by recent sun exposure (2, 5). Standardized interview instruments to estimate sun exposure history attempt to capture time spent outdoors during daylight hours throughout a person s lifetime during both recreational and occupational activities. Furthermore, they gather information regarding specific sun exposure behaviors such as frequency of sunbathing. Although these instruments tend to yield reproducible results (4, 6), there are Correspondence to Dr. Margaret R. Karagas, Dartmouth Medical School, Section of Biostatistics and Epidemiology, 7927 Rubin Building, One Medical Center Drive, Lebanon, NH ( margaret.karagas@dartmouth.edu). 719

2 720 Karagas et al. TABLE 1. Association of anatomic site and age with the presence of solar keratoses* and severe solar elastosis among men and women residents of New Hampshire aged 25ÿ74 years diagnosed from July 1, 1997, through March 31, 2000 Characteristic a number of challenges to eliciting responses. Among them is the length of the interview, typically about 45 minutes per subject for the sun exposure history component alone. In addition, the responses result in more than 100 possible variables. Selection of the relevant analytic variables is especially problematic in light of the virtual absence of data on the validity of the instruments in capturing a person s sun exposure; that is, responses to the standardized questions on sun exposure history have not been compared with objective measures of actinic damage. Presently, differences in the accuracy of sun exposure responses among men and women remain unknown even though their patterns of exposure differ (i.e., men have more opportunities than women for occupational UVR exposure). Understanding the validity of sun exposure questionnaire responses would help to improve UVR exposure classification and in turn lead to more accurate and efficient characterization of UVR exposure assessment useful in future studies. For this reason, we sought to determine the sun exposure variables derived from a structured interview that best predict histologic evidence of actinic damage in men and women, using data collected as part of a populationbased case-control study of keratinocyte malignancies in New Hampshire. MATERIALS AND METHODS Study group Solar keratoses For the current study, we used the pathology materials from skin cancer cases selected to take part in our casecontrol study. To identify these cases, we enlisted the collaboration of dermatologists and pathology laboratories throughout New Hampshire and bordering regions (7). Briefly, we selected all invasive SCC cases, and, for efficiency, a roughly equal number of BCC cases (at random), diagnosed from July 1, 1997, through March 31, 2000, among New Hampshire residents aged years. We recruited 1,118 cases to take part and obtained the histologic materials from the original diagnostic tumor for 925 cases (83 percent) (443 BCC and 482 SCC). Interview data Severe solar elastosis Men (n ¼ 252) Women (n ¼ 149) Men (n ¼ 435) Women (n ¼ 313) ORy,z 95% CIy,z ORz 95% CIz ORz 95% CIz ORz 95% CIz Anatomic site Rest of body 47/ / / / Head and neck 94/ , / , / , / , 12.0 Reference age (years) 50 16/ / / / / , / , / , / , / , / , / , / , 6.8 >70 42/ , / , / , 5.5 7/ , 13.5 p for trend ¼ * Solar keratoses were limited to squamous cell carcinoma only. y OR, odds ratio; CI, confidence interval. z Adjusted for age and sex. Subject s age at diagnosis. p for trend ¼ p for trend ¼ p for trend < All participants provided informed consent in accordance with the Committee for the Protection of Human Subjects at Dartmouth College (Hanover, New Hampshire). Study participants completed a structured personal interview, usually at their homes. The interview included sociodemographic information (e.g., level of education), use of tobacco, and medical history (e.g., history of radiotherapy). To ascertain skin sensitivity to the sun, we asked subjects their usual skin reaction to sun exposure (i.e., tanning or burning), both acutely (first exposure of the summer for 1 hour) and after repeated and prolonged exposure to sunlight. Skin color was measured in both a sun-exposed (forearm) and an unexposed (underarm) area by using a Minolta CR200 colorimeter (Minolta Camera Co., Ltd., Osaka, Japan), for which 100 percent reflectance would represent pure white and 0 percent reflectance pure black for the parameter L. For hair color, subjects chose the color that corresponded with their natural hair color as a young adult (prior to graying or balding) from hair samples. Interviewers assessed eye color under natural light by using an eye photo chart and by self-report (as blue, green, gray, hazel, light brown, dark brown, or black). We used a modified version of the sun exposure questionnaire designed by Kricker et al. (8) for a previous populationbased, nonmelanoma skin cancer case-control study from

3 Measurement of Sun Exposure 721 TABLE 2. Association of pigmentary and other sun sensitivity characteristics with presence of solar keratoses and severe solar elastosis among New Hampshire residents aged 25ÿ74 years diagnosed from July 1, 1997, through March 31, 2000 Characteristic Solar keratoses (n ¼ 401) Severe solar elastosis (n ¼ 748) OR*,y 95% CI*,y ORy 95% CIy Skin color (underarm, colorimeter L)z Q*1 (29 <67) 69/ / Q2 (67 <69) 49/ , / , 2.4 Q3 (69 <71) 54/ , / , 1.3 Q4 (71 77) 44/ , / , 2.0 p for trend ¼ p for trend ¼ Natural hair color (matched to wig samples) Black 15/ , 1.4 8/ , 2.3 Dark brown 54/ (ref*) 48/ (ref) Light 66/ , / , 1.9 Blond 56/ , / , 1.5 Red 26/ , / , 2.6 p for trend ¼ p for trend ¼ Eye color (matched to eye photo chart) Brown or black 24/ / Green or hazel 46/ , / , 1.9 Blue or grey 148/ , / , 1.5 p for trend ¼ p for trend ¼ Sun sensitivity, first summer exposure Tan 19/ / Burn/tan 111/ , / , 1.2 Burn/peel 73/ , / , 0.9 Burn/blister 21/ , / , 1.0 p for trend ¼ p for trend ¼ Sun sensitivity, prolonged exposure Deep tan 41/ / Moderate tan 106/ , / , 1.3 Tend to peel 59/ , / , 1.8 No tan 18/ , / , 3.1 p for trend ¼ p for trend ¼ * OR, odds ratio; CI, confidence interval; Q, quartile; ref, reference. y Adjusted for age and sex (refer to the text). z Observed reflectance of underarm skin from a Minolta CR200 colorimeter (Minolta Camera Co., Ltd., Osaka, Japan), where, for the parameter L, 0 is black and 100 is white. Quartiles were established from the entire study population. Geraldton, Australia, and tested for its reproducibility (9). This instrument was derived from earlier skin cancer studies (10 13). First, subjects were sent a personal residence and work history calendar to complete prior to the interview. The calendar included the average number of days subjects worked and did not work in a given year. During the interview, the calendar was used to identify periods of consistent outdoor activities, then questions relating to time spent outdoors, sunbathing, and sunburns were asked for each of these periods. Questions included the amount of time spent outdoors: 1) from 9 a.m. to 5 p.m. and from 10 a.m. to 2 p.m., 2) on workdays and nonworkdays, and

4 722 Karagas et al. TABLE 3. Relation between measures of sun exposure and histologic evidence of solar keratoses and severe solar elastosis among New Hampshire residents aged 25ÿ74 years diagnosed from July 1, 1997, through March 31, ) in the summer and the rest of the year. For each period, subjects also were asked their frequency of sunbathing, number of painful sunburns (one that caused pain for 2 or more days), and number of blistering sunburns. Additionally, subjects were asked whether the anatomic site of the skin cancer was usually exposed and the number of sunburns at this site. For cases with multiple tumors at the time of diagnosis, we asked about the predominant site or a randomly selected site when multiple tumors occurred at the same site. Histopathologic assessment Exposure to specific anatomic site We requested the diagnostic pathology materials of skin cancers from the original pathology laboratory or dermatopathologist. The study pathologist verified the histologic Solar keratoses OR*,y 95% CI* Severe solar elastosis ORy 95% CI Measures of acute intense or intermittent exposure of lifetime painful sunburns 0 108/ / / , / , / , / , / , / , 1.3 p for trend ¼ p for trend ¼ Lifetime frequency of sunbathing Q*1 (0) 68/ / Q2 (1 226) 45/ , / , 1.2 Q3 ( ) 49/ , / , 1.0 Q4 (815 4,451) 57/ , / , 0.7 p for trend ¼ p for trend ¼ Proportion of lifetime hours spent outdoors recreationally, 9 a.m.ÿ5 p.m., warmer months Q1 (0 <0.5) 53/ / Q2 (0.5 <0.7) 48/ , / , 3.6 Q3 (0.7 <0.9) 42/ , / , 4.6 Q4 (0.9 1) 35/ , / , 1.2 p for trend ¼ p for trend ¼ Measures of cumulative exposure: time spent outdoors, 9 a.m. to 5 p.m., warmer months Total lifetime hours Q1 (0 <6,692) 44/ / Q2 (6,692 <12,044) 42/ , / , 7.5 Q3 (12,044 <18,345) 46/ , / , 8.4 Q4 (18,345 44,437) 46/ , / , 17.3 p for trend ¼ p for trend < Table continues diagnosis of the tumor and documented the extent of actinic damage in adjacent skin tissue. This procedure included assessment for the presence of solar keratoses (yes/no) defined by the presence of atypical epithelial cells confined to the epidermis. This atypia plausibly functions as a precursor event to SCC, and, because it does not typically appear in BCC, our assessment of this trait was limited to biopsies for SCC tumors. Presence of solar keratosis was determined for 401 (83 percent) of the biopsies for SCC. The extent of solar elastosis (mild, moderate, or severe) was based on the appearance and amount of abnormal elastotic fibers that putatively result from UVR penetration. Solar elastosis was categorized as mild if single, scattered, blue-gray elastotic fibers were identified in the papillary dermis. Moderate elastosis was characterized by clumps of elastotic fibers with intervening normal papillary dermis. Severe solar elastosis

5 Measurement of Sun Exposure 723 TABLE 3. Continued Exposure to specific anatomic site Solar keratoses OR*,y ORy 95% CI* Severe solar elastosis ORy 95% CI Lifetime recreational hours Q1 (0 <4,896) 52/ / Q2 (4,896 <7,942) 60/ , / , 4.9 Q3 (7,942 <11,874) 55/ , / , 4.7 Q4 (11,874 38,565) 45/ , / , 11.0 p for trend < p for trend < Adult recreational hours Q1 (0 <2,383) 53/ / Q2 (2,383 <4,987) 56/ , / , 4.3 Q3 (4,987 <8,437) 53/ , / , 4.1 Q4 (8,437 34,737) 50/ , / , 10.2 p for trend ¼ p for trend < Childhood recreational hours Q1 (0 <1,595) 58/ / Q2 (1,595 <2,864) 54/ , / , 4.5 Q3 (2,864 <3,828) 34/ , / , 5.6 Q4 (3,828 5,104) 74/ , / , 7.4 p for trend ¼ p for trend < Last-decade hours Q1 (0 <1) 61/ / Q2 (1 <1,693) 44/ , / Q3 (1,693 <3,463) 54/ , / , 7.3 Q4 (3,463 7,291) 55/ , / , 7.9 p for trend ¼ p for trend < Adult occupational hours Q1 (0 <1) 48/ / Q2 (1 <2,178) 41/ , / , 11.3 Q3 (2,178 <6,185) 46/ , / , 11.5 Q4 (6,185 32,862) 53/ , / , 21.9 p for trend ¼ p for trend < * OR, odds ratio; CI, confidence interval; Q, quartile. y Adjusted for age, sex, and skin type (refer to the text). was characterized by replacement of the papillary dermis by clumped elastotic fibers and/or amorphous masses of elastotic material. Of the 925 biopsies overall, 82 percent were evaluable for solar elastosis. Statistical analysis We began our analyses by examining the associations between presence of solar keratosis (yes/no) and degree of histologic solar elastosis. Because so few subjects had mild elastoses, we combined the moderate and mild categories and compared them with severe elastosis. We first examined demographic factors (e.g., age and sex), pigmentary characteristics, sun sensitivity, and clinical features of the tumor (e.g., histology and anatomic site). This step included logistic regression models to estimate the odds ratio and 95 percent confidence interval associated with each factor stratified or adjusted for age and sex. Skin color measured on the colorimeter is a continuous variable, so we used quartiles of the reflectance values in the study population. We calculated sun exposure variables based on the calendar (i.e., for days worked and not at work) and interview response data. For the present analysis, variables of interest included total lifetime hours spent outdoors from 9 a.m. to 5 p.m. and between 10 a.m. and 2 p.m. during working and nonworking hours, and during the summer (i.e., warmer months). Additionally, we computed an index of intermittent sun exposure as the proportion of time spent outdoors

6 724 Karagas et al. during nonworking hours. Sun exposure, sunbathing, and sunburn variables were evaluated according to age at exposure (<15 years, 15 years). We chose the cutpoint for age at exposure based on the literature indicating that the magnitude of association between sunlight exposure and risk of SCC and BCC decreases after ages 10ÿ19 years (14 16). We further computed total hours of sunbathing and number of painful and blistering sunburns. For each of these measures, we assessed exposures specifically to the anatomic site of the tumor and evaluated them as ordinal categorical (e.g., quartiles based on the distribution in the study group) variables, and we computed a p for trend for the ordinal variables (17). For variables with strongly skewed distributions, unitary categories were applied (e.g., for number of lifetime blistering sunburns: 0, 1, 2). We assessed the potential confounding effects of histologic type (BCC, SCC), age, sun sensitivity, and other factors by using stratified analyses and logistic regression. We specifically evaluated the interaction between sex and each of the sun exposure variables but did not find evidence of one. In addition, associations did not appear to differ by histology of the tumor itself (BCC vs. SCC) for solar elastosis (and was limited to a single histology, SCC for solar keratoses); therefore, in our final models, we combined both types of tumors. RESULTS Prevalence of solar keratoses did not vary by anatomic location (table 1). Severe solar elastosis was more common in SCC than in BCC tumors (85 percent and 71 percent with severe solar elastosis, 12 percent and 23 percent with moderate, and 6 percent and 3 percent with mild, respectively, and <1 percent with none present for both types). Furthermore, severe solar elastosis occurred far more frequently in biopsies from the head and neck than from other sites (odds ratio for both sexes ¼ 11.6, 95 percent confidence interval: 7.5, 18.6) and more so in men (odds ratio ¼ 20.1, 95 percent confidence interval: 10.8, 39.7) (table 1) than women (odds ratio ¼ 6.2, 95 percent confidence interval: 3.3, 12.0) (table 1). The interaction with sex was not statistically significant, however. The prevalence of solar keratoses increased with age among women but not men (p for trend ¼ and p for trend ¼ 0.802, respectively) (table 1). The prevalence of severe elastosis increased with age among both men and women, although the magnitude of the increase appeared stronger among women (table 1). Presence of solar elastosis was inversely related to a phenotype sensitive to first summer exposure to the sun (i.e., tended to burn) (p for trend ¼ 0.018) but appeared unrelated to skin sensitivity for prolonged sun exposure (table 2). In contrast, the prevalences of solar keratoses and severe solar elastosis were the highest among those with red hair color assessed by matching to wig samples, although with limited statistical precision, and did not vary significantly by eye color (table 2). There was no evidence of effect modification by sex in these comparisons (data not shown). Measures of acute or intense exposure generally were unrelated to the prevalence of solar keratoses and severe solar elastosis (table 3). A high frequency of sunbathing with the site exposed was associated with a lower prevalence of severe solar elastosis (odds ratio ¼ 0.4, 95 percent confidence interval: 0.2, 0.7) (table 3). Adjustment for total lifetime hours spent outdoors from 9 a.m. to 5 p.m. did not appreciably alter these results (data not shown). Measures of cumulative sun exposure at the site related to an increased risk of solar keratosis or severe solar elastosis (table 3). Amount of time spent outdoors during recreational activities (nonworking hours) during the warmer months, both in childhood and adulthood, was associated with both solar keratosis and severe solar elastosis, although the p for trend for childhood exposure was statistically significant for solar elastosis only. Time spent outdoors during the last decade and time spent outdoors during occupational (working) hours were highly related to histologic evidence of severe solar elastosis and, to a lesser extent, solar keratoses (table 3). We did not detect any major differences in the odds ratios between men and women with either acute or chronic measures of sun exposure for either solar keratosis or severe solar elastosis; none of the interaction terms were statistically significant (data not shown). Likewise, we did not detect appreciable differences when we stratified results for solar elastosis by tumor histology (i.e., SCC vs. BCC) (data not shown). Data for some variables indicated stronger associations when restricted to exposures between 10 a.m. and 2 p.m., but the results were generally consistent with the findings for exposures between 9 a.m. and 5 p.m. (data not shown). DISCUSSION In a large, population-based study of skin cancer, we compared responses to a standardized sun exposure questionnaire, frequently used in epidemiologic studies, with histologic evidence of solar damage assessed by a single dermatopathologist. After analyzing more than 100 variables, we found that those estimating cumulative number of hours spent outdoors related to histologic evidence of solar damage (solar keratoses and solar elastosis). We restricted our analysis to estimates specific to whether the affected anatomic site was exposed to sun, eliminating potential misclassification by times when the site was covered (unexposed). In contrast to measures of cumulative exposure, measures of acute, intense exposure (e.g., sunburns or sunbathing), or a tendency to sunburn, if anything were associated with a lower prevalence of actinic damage histologically. UVR from the sun is an established cause of BCC and SCC (18, 19), but many questions remain for which valid instruments of exposure assessment are needed. Epidemiologic studies have attempted to quantify a person s lifetime sun exposure history through time-consuming personal interviews. As in our study, interviews typically question subjects about time spent outdoors over their lifetime during the day, during midday, in warmer and cooler months, and during occupational and recreational activities in addition to history of sunburns (i.e., severe sunburns or blistering sunburns). Collection of sun exposure history using this

7 Measurement of Sun Exposure 725 instrument is designed to produce quantitative information with minimal recall bias. Nonetheless, studies of melanoma indicate the potential presence of recall bias for sun sensitivity and sunbathing, although the results are not conclusive (20, 21). Furthermore, it is conceivable that the best predictors (e.g., time spent outdoors) of solar keratoses or elastosis were simply better recalled than other factors (e.g., number of severe sunburns). In general, studies on the reproducibility of standard sun exposure and sunburn history questions suggest reasonable reliability of subject responses (4, 6). In a European study (4), 90 cases and 90 controls were reinterviewed about 1 year apart. For several key sunlight-related variables (e.g., hours/week of sun exposure), the intraclass correlation coefficients were 0.6ÿ0.7 between the two interviews and did not differ by case-control status. A similar study from Australia found an intraclass correlation for time spent outdoors of 0.77 (95 percent confidence interval: 0.71, 0.83) (6). While informative, neither study had a gold standard to which exposure estimates could be compared. Clinical signs of solar skin damage, assessed by dermatologic examination, are perhaps the strongest solar-related indicators of both BCC and SCC risk, in particular solar keratoses (14, 22). However, to date, no known studies have assessed these characteristics by using histopathologic criteria; biopsies of normal skin usually cannot be obtained in epidemiologic studies because they require a trained health care provider. In our study, we used histologic evidence of actinic damage in normal skin tissue adjacent to the skin cancer as a gold standard for chronic actinic damage. While feasible, an obvious limitation of this approach is that our results were based on the biopsies of tumors from skin cancer affected persons, rather than unaffected controls, and unaffected tissue. Thus, the prevalence of actinic damage likely exceeds what would be observed in the skin of normal controls (23) and indeed was higher in the skin adjacent to SCC than BCC tumors. However, we do not know how our findings would compare with studies that evaluated the normal skin of melanoma biopsies, for example, which appear to have far less evidence of solar damage histologically (24). Measurement of genetic alterations of UVR damage, that is, UVR-related TP53 mutations, may be useful biomarkers of sun exposure but have not yet been applied to large-scale epidemiologic investigations (25 28). Estimates of cumulative sun exposure, based on a complex series of questions, related very clearly to histologic evidence of actinic damage. Of interest, measures of cumulative exposure in childhood and in the decade prior to diagnosis were especially related to the presence of both measures. In a number of epidemiologic studies, sun exposure during childhood or the teenage years specifically related to risk of all types of skin cancer (1). In Alberta, Canada, occupational sun exposure in the 10 years prior to diagnosis was associated with an increased risk of SCC (29), suggesting an effect of recent exposure. Supporting this hypothesis are findings from a clinical trial in which recent use of sunscreens prevented SCC, but not BCC, occurrences (30). Thus, our data provide evidence that both early and recent exposures have a biologic basis in that they result in actinic damage to the skin that is discernable histologically. Our results, assuming that the findings are confirmed, suggest that questionnaire-derived measures of intense, intermittent exposure, that is, sunburns, are not linked with histologically recognizable actinic damage in skin tissue or may even be related to a diminished prevalence. Sunburns primarily affect the epidermis and therefore may not cause abnormal elastin response in the dermis; however, solar keratoses are of epidermal origin. Sunburns induce an inflammatory and apoptotic response, which in part might explain the absence of a positive association with histologic measures of chronic UVR response. Hence, a role for sunburns in the etiology of skin cancer must lie elsewhere. In conclusion, our findings based on histologic evidence of actinic damage suggest that measures of sun exposure derived from a personal interview accurately reflect a person s exposure history. Further studies incorporating molecular-genetic markers of UVR damage such as signature UVR TP53 mutations may provide additional insights into the validity of the questionnaire instruments. Ultimately, studies that investigate the full spectrum of sun exposure history variables along with molecular-genetic markers of UVR damage may help to discern the patterns of exposure that pose an excess risk and, in turn, more effectively guide preventive strategies for these common malignancies. ACKNOWLEDGMENTS This work was in part supported by the Intramural Research Program of the National Institutes of Health, National Cancer Institute, Division of Cancer Epidemiology and Genetics and by grants R01CA57494 and R01CA from the National Institutes of Health. The authors are indebted to the physicians who constitute the New Hampshire Skin Cancer Study Group and to the New Hampshire Society of Dermatology. Conflict of interest: none declared. REFERENCES 1. Karagas M, Weinstock MA, Nelson HH. Keratinocyte cancers (basal cell and squamous cell carcinomas of the skin). New York, NY: Oxford University Press, Gallagher RP, Hill GB, Bajdik CD, et al. Sunlight exposure, pigmentation factors, and risk of nonmelanocytic skin cancer. II. Squamous cell carcinoma. Arch Dermatol 1995;131: Kricker A, Armstrong BK, English DR, et al. Does intermittent sun exposure cause basal cell carcinoma? A case-control study in Western Australia. Int J Cancer 1995;60: Rosso S, Zanetti R, Martinez C, et al. The multicentre south European study Helios. II: Different sun exposure patterns in the aetiology of basal cell and squamous cell carcinomas of the skin. Br J Cancer 1996;73: van Dam RM, Huang Z, Rimm EB, et al. Risk factors for basal cell carcinoma of the skin in men: results from the Health Professionals Follow-up Study. Am J Epidemiol 1999;150:

8 726 Karagas et al. 6. English DR, Armstrong BK, Kricker A, et al. Case-control study of sun exposure and squamous cell carcinoma of the skin. Int J Cancer 1998;77: Karagas MR, Greenberg ER, Spencer SK, et al. Increase in incidence rates of basal cell and squamous cell skin cancer in New Hampshire, USA. Int J Cancer 1999;81: Kricker A, Armstrong BK, English DR, et al. Pigmentary and cutaneous risk factors for non-melanocytic skin cancer a case-control study. Int J Cancer 1991;48: English DR, Armstrong BK, Kricker A. Reproducibility of reported measurements of sun exposure in a case-control study. Cancer Epidemiol Biomarkers Prev 1998;7: Elwood JM, Gallagher RP, Hill GB, et al. Pigmentation and skin reaction to sun as risk factors for cutaneous melanoma: Western Canada Melanoma Study. Br Med J (Clin Res Ed) 1984;288: Green A, Bain C, McLennan R, et al. Risk factors for cutaneous melanoma in Queensland. Recent Results Cancer Res 1986; 102: Holman CD, Armstrong BK, Evans PR, et al. Relationship of solar keratosis and history of skin cancer to objective measures of actinic skin damage. Br J Dermatol 1984;110: Urbach F, Rose D, Bonnem M. Genetic and environmental interactions in skin carcinogenesis. In: Environment and cancer: a collection of papers presented at the 24th Annual Symposium on Fundamental Cancer Research. Baltimore, MD: Williams & Wilkins, 1972: English D, Armstrong BK, Kricker A, et al. Demographic characteristics, pigmentary and cutaneous risk factors for squamous cell carcinoma of the skin: a case-control study. Int J Cancer 1998;76: Grodstein F, Speizer FE, Hunter DJ. A prospective study of incident squamous cell carcinoma of the skin in the Nurses Health Study. J Natl Cancer Inst 1995;87: Kricker A, Armstrong BK, English DR, et al. Pigmentary and cutaneous risk factors for non-melanocytic skin cancer a case-control study. Int J Cancer 1991;48: Breslow NE, Day NE, eds. Statistical methods in cancer research. Vol 1. The analysis of case-control studies. Lyon, France: International Agency for Research on Cancer, (IARC scientific publication no. 32). 18. Solar and ultraviolet radiation. IARC monographs on the evaluation of carcinogenic risks to humans. Vol 55. Lyon, France: International Agency for Research on Cancer, 1992;55: Kricker A, Armstrong BK, English DR. Sun exposure and non-melanocytic skin cancer. Cancer Causes Control 1994; 5: Cockburn M, Hamilton A, Mack T. Recall bias in selfreported melanoma risk factors. Am J Epidemiol 2001;153: Weinstock MA, Colditz GA, Willett WC, et al. Recall (report) bias and reliability in the retrospective assessment of melanoma risk. Am J Epidemiol 1991;133: Green A, Battistutta D. Incidence and determinants of skin cancer in a high-risk Australian population. Int J Cancer 1990;46: Moon JS, Oh CH. Solar damage in skin tumors: quantification of elastotic material. Dermatology 2001;202: Berwick M, Armstrong BK, Ben-Porat L, et al. Sun exposure and mortality from melanoma. J Natl Cancer Inst 2005;97: Einspahr J, Alberts DS, Aickin M, et al. Expression of p53 protein in actinic keratosis, adjacent, normal-appearing, and non-sun-exposed human skin. Cancer Epidemiol Biomarkers Prev 1997;6: Nakazawa H, English D, Randell PL, et al. UV and skin cancer: specific p53 gene mutation in normal skin as a biologically relevant exposure measurement. Proc Natl Acad Sci U S A 1994;91: Ouhtit A, Nakazawa H, Armstrong BK, et al. UV-radiationspecific p53 mutation frequency in normal skin as a predictor of risk of basal cell carcinoma. J Natl Cancer Inst 1998;90: Ouhtit A, Ueda M, Nakazawa H, et al. Quantitative detection of ultraviolet-specific p53 mutations in normal skin from Japanese patients. Cancer Epidemiol Biomarkers Prev 1997; 6: Gallagher RP, Hill GB, Bajdik CD, et al. Sunlight exposure, pigmentary factors, and risk of nonmelanocytic skin cancer I. Basal cell carcinoma. Arch Dermatol 1995;131: Green A, Williams G, Neale R, et al. Daily sunscreen application and betacarotene supplementation in prevention of basal-cell and squamous-cell carcinomas of the skin: a randomised controlled trial. Lancet 1999;354:723 9.