Summary
Background.
Melanoma incidence has been dramatically increasing worldwide. Psoralen, a known photocarcinogen, is naturally abundant in citrus products, leading to the hypothesis that high citrus consumption may increase melanoma risk.
Objectives.
To investigate the association between total citrus consumption and melanoma risk, the association between individual citrus products and melanoma risk, and test for interactions between total citrus intake and established melanoma risk factors.
Methods.
Logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (95% CIs) for the association between citrus consumption and melanoma risk among 1,592 cases and 197,372 controls from the UK Biobank cohort. Citrus consumption data were collected via 5 rounds of 24-hour recall questionnaires. International Classification of Disease codes were used to determine melanoma outcome.
Results.
After adjusting for potential confounders, participants in the highest category of total citrus intake (>2 servings/day) had a significantly increased risk of melanoma (OR [95% CI] = 1.63 [1.24–2.12]) relative to those with no consumption. For individual citrus products, participants with the most orange and orange juice consumption (>1 serving per day) had a significantly increased melanoma risk relative to those with no consumption (ORs [95% CIs] = 1.79 [1.07–2.78] and 1.54 [1.10–2.10]), respectively. Fair/very fair-skinned participants with high citrus consumption had an even greater melanoma risk (OR [95% CI] = 1.75 [1.31–2.29]).
Conclusions.
High citrus consumption was associated with an increased risk of melanoma in a large, prospective, population-based cohort. Further validation of these findings could lead to improved melanoma prevention strategies.
Introduction
Over the last several decades, there has been a dramatic increase in melanoma incidence worldwide.1 In the United Kingdom (UK), melanoma incidence has been increasing by an average of 3% per year since the 1980s,2 and is currently the 5th most common cancer diagnosis among UK residents.3 Globally, melanoma incidence is increasing by 3–7% per year,4 a rate faster than any other cancer,1,4 suggesting a doubling in incidence every 10–20 years.5 This increasing incidence is not artifactual,6 and it is predicted to continue.2,7,8 Melanoma is also the second most diagnosed cancer among young adults,9 thus having the highest individual cost of cancer death regarding years of productive life lost.10 Because of this rapid increase in melanoma rates, primary prevention research is necessary and urgent.
Several melanoma risk factors have been established, including exposure to ultraviolet (UV) radiation from the sun;11,12 having fair skin, fair hair, light-colored eyes, or the inability to tan;12–17 the use of solariums or sunlamps;18,19 and a history of sunburn during adolescence.17,20 Psoralen, a type of furocoumarin used in photochemotherapy using oral psoralen and ultraviolet-A radiation (PUVA), is also known to be photocarcinogenic.21 Naturally occurring in nature as part of a plant’s natural defense against pathogens,22 psoralens are abundantly found in citrus products,22–24 leading to the hypothesis that citrus consumption may increase melanoma risk due to psoralen photocarcinogenicity.
This fairly new hypothesis has yielded inconsistent results. Research from the Nurses’ Health Study (NHS) and Health Professionals Follow-up Study (HPFS) found that high total citrus consumption was significantly associated with an increased risk of melanoma,25 and research from the European Prospective Investigation into Cancer and Nutrition cohort (EPIC) found high citrus fruit consumption (but not total citrus consumption) to increase melanoma risk.26 Conversely, an analysis of the Women’s Health Initiative (WHI) found no association between citrus consumption and melanoma risk save for among those with the highest consumption of citrus juice and the most time spent outdoors,27 and findings from the National Institutes of Health and the AARP (NIH-AARP) were also null.28 The only other research to investigate this association is an Italian case-control study of the Mediterranean diet, which found high citrus consumption to be a protective factor.13 This sparse, conflicting evidence represents a critical gap in melanoma knowledge and highlights the need for further investigation of citrus as a potential melanoma risk factor.
The purpose of the current study is to address these gaps in knowledge by performing the following analyses in the UK Biobank (UKBB), a large, population-based, prospective cohort: 1) investigate the association between total citrus consumption and melanoma risk, 2) investigate the association between individual citrus products and melanoma risk, and 3) test for interactions between total citrus intake and established melanoma risk factors. We believe the results of the current study will increase current knowledge and understanding of skin cancer risk factors, and, upon further validation, can serve as an empirical basis for future primary prevention interventions.
Materials and methods
Study Population
As further described elsewhere,29–31 the UKBB is a major health resource founded in the UK with the goal of improving the diagnosis, treatment, and prevention of a variety of serious illness, including cancer, heart disease, stroke, diabetes, and depression, among others. For this large, prospective cohort, a sample of 500,000+ participant volunteers aged 40–69 years old were recruited from 2006–2010. UKBB sampling covered a wide variety of socioeconomic, ethnic, and urban/rural settings to gather a diverse, representative sample and allow for associations generalizable to the UK population. At baseline, all 500,000+ UKBB participants were assessed at 22 assessment centers throughout the UK to provide physical measures, biological samples, and detailed personal information. Follow-up took place via linkage of participants’ past and future medical records, and by the completion of online questionnaires every few years. In the current analysis, we received data for 502,527 UKBB participants. After receiving the data, n = 21 participants requested to be removed from analyses. No reason for this withdrawal was provided. This left us with a sample of n = 502,506 UKBB participants.
Assessment of Citrus Consumption
In the UKBB, citrus consumption data were collected via five ‘rounds’ of 24-hour recall questionnaires. The first ‘round’ took place in the assessment center between April 2009-September 2010. The next four ‘rounds’ were administered electronically via email invitations to complete the online questionnaire. The four online ‘rounds’ were sent to participants during: February-April, 2011; June-September, 2011; October-December 2011; and April-June, 2012, respectively. A total of n=210,126 participants completed at least one 24-hour dietary recall assessment over the course of these 5 ‘rounds’ and provided citrus consumption data for the current analysis. Compared with an interviewer-administered 24-hour recall, this web-based 24-hour recall yielded Spearman’s correlation coefficients of 0.5–0.9 (mean of 0.6).32 Additionally, between the first and last dietary assessments, at least 70% of UKBB participants reported nutrient intake in the same or adjacent intake category (82% reported the same or adjacent category for fresh fruit intake).33
To measure orange consumption, participants were asked: “How may oranges (fresh, frozen, canned) did you have?”. For grapefruit consumption, participants were asked: “How many whole/servings of grapefruit (fresh, frozen, canned) did you have?”, and for satsuma consumption, participants were asked: “How many whole/servings of orange-like small fruits e.g. satsuma, clementine, mandarin (fresh, frozen, canned) did you have?”. For each of these measures, participants had the option to select: “half”, “1”, “2”, “3”, or “4+”. For orange juice and grapefruit juice intake, participants were asked: “How many glasses/cartons/250ml of pure orange (grapefruit) juice did you drink yesterday?”, with possible responses of: “half”, “1”, “2”, “3”, “4”, “5”, and “6+”. There was no available data for satsuma juice. A cumulative average of participants’ citrus consumption over the 5 ‘rounds’ of nutritional data collection was used to assess consumption of each citrus product. Participants were not required to complete all 5 ‘rounds’ of data collection to be included in the analysis. Any untaken 24-hr recall ‘rounds’ were simply treated as missing and therefore did not impact the cumulative average for citrus consumption.
Ascertainment of Melanoma Cases
International Classification of Diseases (ICD) codes were used to identify melanoma outcome data, which are acquired via linkage with national registries. These registries acquire cancer diagnosis data from various sources dating back to the registries’ inception in the early 1970s. These sources include hospitals, treatment centers, nursing homes, hospices, general practices, screening programs, death certificates, and others. For participants residing in England and Wales, cancers diagnosed from 1971–1978, 1979–1994, and 1995-present are coded according to the ICD 8,9, and 10, respectively. For those residing in Scotland, cancer diagnoses through the end of 1996 and from the start of 1997 are coded according to ICD 9 and 10 classifications, respectively. ICD codes C43.0–9 were used to identify melanoma cases. A full description of the covariates used in these analyses can be found in the supplementary material.
Statistical Analysis
As done previously,25,27,28 only White/Caucasian participants were included in the analyses due to low melanoma incidence in ethnic minorities.34–37 This resulted in n=11,162 (5.3%) of the 210,126 participants with complete citrus data being excluded, leaving n=198,964 to be included in the analyses. To estimate the effect of total citrus consumption, we combined intake of all citrus products into a single composite variable with five groups: 0, >0-half, >half-1, >1–2, and > 2 servings, with 0 servings as the reference group. Consumption of each citrus product was categorized into four groups: 0, >0-half, >half-1, and >1 serving, with 0 servings as the reference group. Chi-square tests and one-way ANOVA were used to test for participant differences in categorical and continuous variables of interest, respectively, according to levels of total citrus consumption. Logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for the association between citrus consumption and melanoma risk. Three models were constructed for these analyses, one age adjusted model and two multivariable adjusted models. For the analysis of total citrus, we controlled for sex, education, income, physical activity, BMI, smoking status, alcohol intake, and total energy intake in the first multivariable adjusted model (controlling for demographic variables) and additionally for coffee intake, tanning ability, childhood sunburn occasions, natural hair color, skin color, average time spent outdoors in summer, solarium/sunlamp use, use of UV protection, and ICD-confirmed history of NMSC in the second multivariable adjusted model (controlling for demographic and sun-exposure variables). If considerable, missing data was made into a variable category, if not, it was excluded from analyses. For analyses of individual citrus products, the first multivariable adjusted model controlled for all variables in the first and second models for total citrus, and the second model additionally adjusted for consumption of the other individual citrus products. Trend tests across groups were performed by assigning median values to each category then treating them as continuous variables.
Likelihood ratio tests comparing the model with and without product terms were used to test for potential interactions between reported skin cancer risk factors and total citrus consumption. When evidence of interaction existed, odds ratios and confidence intervals were calculated for each stratum. SAS software version 9.4 (SAS Institute, Cary, NC) was used for all statistical tests, all tests were 2-tailed, with p<.05 indicating statistical significance. All data was stored using Karst, a high-performance computing cluster, at Indiana University.
Results
In the UKBB population, total citrus intake significantly varied by several demographic and sun exposure variables (Table 1). Relative to those with no total citrus consumption, participants with higher citrus intake were more likely to be older, male, have a higher education and income, less likely to smoke, and more physically active. A number of sun-exposure variables also varied by citrus intake. Relative to those with no citrus consumption, participants who consumed more citrus were less likely to use sunlamps or solariums, more likely to report sunburns during childhood, and less likely to describe themselves as olive skinned.
Table 1.
Demographic and sun-exposure characteristics of UK Biobank study participants according to daily total citrus intake.
| Total citrus intake | |||||||
|---|---|---|---|---|---|---|---|
| Total | None | >0-½ serving | >½-1 serving | >1-2 servings | >2 servings | p-value | |
| N (%) | 198,964 | 82,297 (41.4%) | 36,828 (18.5%) | 45,829 (23.0%) | 26,740 (13.4%) | 7,270 (3.7%) | |
| Age at recruitment | 56.2 (7.9) | 55.8 (8.1) | 56.3 (7.9) | 56.6 (7.8) | 56.8 (7.6) | 56.9 (7.6) | <.0001 |
| Years (mean/SD) | |||||||
| Sex | <.0001 | ||||||
| Male | 89,022 (44.7%) | 36,104 (43.9%) | 16,139 (43.8%) | 20,814 (45.4%) | 12,382 (46.3%) | 3,583 (49.3%) | |
| Female | 109,942 (55.3%) | 46,193 (56.1%) | 20,689 (56.2%) | 25,015 (54.6%) | 14,358 (53.7%) | 3,687 (50.7%) | |
| Education | <.0001 | ||||||
| College or university degree | 84,279 (42.4%) | 29,702 (36.1%) | 16,404 (44.5%) | 21,087 (46.0%) | 13,178 (49.3%) | 3,908 (53.8%) | |
| A level/AS level | 26,312 (13.2%) | 10,772 (13.1%) | 5,027 (13.7%) | 6,185 (13.5%) | 3,446 (12.9%) | 882 (12.1%) | |
| O level/GCSE | 41,543 (20.9%) | 19,095(23.2%) | 7,564 (20.5%) | 8,877 (19.4%) | 4,823 (18.0%) | 1,184 (16.3%) | |
| CSE | 8,391 (4.2%) | 4,428 (5.4%) | 1,390 (3.8%) | 1,563 (3.4%) | 831 (3.1%) | 179 (2.5%) | |
| NVQ, HND, or HNC | 10,786 (5.4%) | 5,061 (6.2%) | 1,789 (4.9%) | 2,319 (5.1%) | 1,299 (4.9%) | 318 (4.4%) | |
| Other professional degree | 9,816 (4.9%) | 3,976 (4.8%) | 1,850 (5.0%) | 2,294 (5.0%) | 1,349 (5.0%) | 347 (4.8%) | |
| Missing | 17,837 (9.0%) | 9,263 (11.3%) | 2,804 (7.6%) | 3,504 (7.7%) | 1,814 (6.8%) | 452 (6.2%) | |
| Income | <.0001 | ||||||
| Less than £30,999 | 70,866 (35.6%) | 31,524 (38.3%) | 12,648 (34.3%) | 15,654 (34.2%) | 8,782 (32.8%) | 2,258 (31.1%) | |
| £31,000 – £51,999 | 51,176 (25.7%) | 20,522 (24.9%) | 9,709 (26.4%) | 11,977 (26.1%) | 7,120 (26.6%) | 1,848 (25.4%) | |
| £52,000 and up | 57,037 (28.7%) | 21,344 (25.9%) | 10,953 (29.7%) | 13,782 (30.1%) | 8,408 (31.4%) | 2,550 (35.1%) | |
| Missing | 19,885 (10.0%) | 8,907 (10.8%) | 3,518 (9.6%) | 4,416 (9.6%) | 2,430 (9.1%) | 614 (8.5%) | |
| Tanning ability | <.0001 | ||||||
| Never tan, only burn | 34,027 (17.4%) | 14,349 (17.7%) | 6,108 (16.9%) | 7,749 (17.2%) | 4,561 (17.4%) | 1,260 (17.7%) | |
| Mildly/occasionally tanned | 43,003 (22.0%) | 17,562 (21.7%) | 8,079 (22.3%) | 10,110 (22.4%) | 5,764 (21.9%) | 1,488 (20.9%) | |
| Moderately tanned | 80,326 (41.0%) | 32,647 (40.3%) | 15,011 (41.4%) | 18,711 (41.5%) | 11,025 (41.9%) | 2,932 (41.1%) | |
| Very tanned | 38,381 (19.6%) | 16,391 (20.3%) | 7,041 (19.4%) | 8,559 (19.0%) | 4,939 (18.8%) | 1,451 (20.4%) | |
| Solarium/sunlamp use | 0.45 (4.1) | 0.54 (4.7) | 0.41 (3.5) | 0.40 (3.3) | 0.37 (4.0) | 0.31 (4.1) | <.0001 |
| Times per year (mean/SD) | |||||||
| Childhood sunburn occasions | 1.95 (6.2) | 1.89 (5.7) | 2.03 (7.9) | 1.93 (5.1) | 1.95 (3.6) | 2.28 (13.3) | <.0001 |
| Before age 15 (mean/SD) | |||||||
| Natural hair color | 0.0003 | ||||||
| Red/blonde | 30,898 (15.6%) | 12,989 (15.8%) | 5,605 (15.2%) | 7,121 (15.6%) | 4,023 (15.1%) | 1,160 (16.0%) | |
| Light brown | 80,795 (40.7%) | 33,279 (40.5%) | 15,053(40.9%) | 18,648 (40.7%) | 10,967 (41.1%) | 2,848 (39.2%) | |
| Dark brown/black | 85,366 (43.0%) | 35,142 (42.8%) | 15,837 (43.0%) | 19,688 (43.0%) | 11,502 (43.1%) | 3,197 (44.0%) | |
| Other | 1,702 (0.9%) | 777 (1.0%) | 312 (0.9%) | 334 (0.7%) | 222 (0.8%) | 57 (0.8%) | |
| Skin color | 0.0004 | ||||||
| Dark/light olive | 40,134 (20.2%) | 16,820 (20.4%) | 7,419 (20.2%) | 9,160 (20.0%) | 5,312 (19.9%) | 1,423 (18.1%) | |
| Fair | 142,125 (71.4%) | 58,381 (70.9%) | 26,343 (71.5%) | 32,886 (71.8%) | 19,302 (72.2%) | 5,213 (71.7%) | |
| Very fair | 16,705 (8.4%) | 7,096 (8.6%) | 3,066 (8.3%) | 3,783 (8.3%) | 2,126 (8.0%) | 634 (8.7%) | |
| Time spent outdoors in summer | 3.52 (2.2) | 3.57 (2.3) | 3.43 (2.1) | 3.49 (2.2) | 3.48 (2.2) | 3.51 (2.2) | <.0001 |
| Hours per day (mean/SD) | |||||||
| Use of sun/ultraviolet protection | <.0001 | ||||||
| Always | 40,618 (20.5%) | 16,705 (20.4%) | 7,285 (19.9%) | 9,462 (20.7%) | 5,601 (21.0%) | 1,565 (21.6%) | |
| Most of the time | 78,033 (39.4%) | 31,378 (38.3%) | 14,715 (40.1%) | 18,326 (40.1%) | 10,753 (40.4%) | 2,861 (39.5%) | |
| Sometimes | 65,769 (33.2%) | 27,397 (33.5%) | 12,323 (33.6%) | 15,122 (33.1%) | 8,581 (32.2%) | 2,346 (32.4%) | |
| Never/rarely | 13,651 (6.9%) | 6,368 (7.7%) | 2,366 (6.5%) | 2,744 (6.0%) | 1,701 (6.4%) | 472 (6.5%) | |
| History of NMSC | <.0001 | ||||||
| Yes | 9,873 (5.0%) | 3,790 (4.6%) | 1,928 (5.2%) | 2,396 (5.2%) | 1,427 (5.3%) | 359 (4.9%) | |
| No | 189,091 (95.0%) | 78,507 (95.4%) | 34,900 (94.8%) | 43,460 (94.8%) | 25,313 (94.7%) | 6,911 (95.1%) | |
Data shown are mean (SD) for continuous variables or N (%) for categorical variables
Some percentages do not add up to 100% due to rounding
Abbreviations: A level, advanced level; AS level, advanced subsidiary level; O level, ordinary level; GCSE, general certificate of secondary education; CSE, certificate of secondary education; NVQ, national vocational qualification; HND, higher national diploma; HNC, higher national certificate; NMSC, non-melanoma skin cancer
There were a total of 1,592 melanoma cases among the 198,964 participants included in this analysis. Total citrus consumption was significantly associated with melanoma risk, as participants consuming the most total citrus were at a significantly increased risk for melanoma (Table 2). In the fully adjusted model, the ORs (95% CIs) were 0.98 (0.82–1.16) for >0-½ a serving, 1.07 (0.91–1.25) for >½ to 1 serving, 1.13 (0.93–1.36) for >1 to 2 servings, and 1.63 (1.24–2.12) for >2 servings of total citrus per day relative to no citrus consumption (p-trend = 0.0051).
Table 2.
Melanoma risk according to frequency of total citrus consumption among study participants in the UK Biobank.
| Serving category | ||||||
|---|---|---|---|---|---|---|
| None | >0-½ serving | >½-1 serving | >1-2 servings | >2 servings | p-trend | |
| Total citrus | ||||||
| N (number of cases) | 82,297 (606) | 36,828 (277) | 45,829 (398) | 26,740 (229) | 7,270 (82) | |
| Age-adjusted OR (95% CI) | 1.00 | 1.01 (0.87–1.16) | 1.15 (1.01–1.31) | 1.13 (0.97–1.31) | 1.49 (1.17–1.86) | 0.0009 |
| Multivariable-adjusted OR1 (95% CI)1 | 1.00 | 0.98 (0.85–1.13) | 1.12 (0.98–1.27) | 1.10 (0.94–1.28) | 1.44 (1.14–1.81) | 0.0047 |
| Multivariable-adjusted OR2 (95% CI)2 | 1.00 | 0.98 (0.82–1.16) | 1.07 (0.91–1.25) | 1.13 (0.93–1.36) | 1.63 (1.24–2.12) | 0.0051 |
Note: bold font indicates statistical significance
Further adjusted for: sex (male, female), education (categorical – college or university degree; Advanced level/Advanced Subsidiary level; Ordinary level/General Certificate of Secondary Education, Certificate of Secondary Education; National Vocational Qualification, Higher National Diploma, or Higher National Certificate; other professional degree; none of the preceding), income (categorical – less than £30,999, £31,000 - £51,999, £52,000 and up, missing), physical activity (categorical – low, moderate, high, missing), BMI (continuous), smoking status (categorical – never, previous, current), alcohol intake (categorical – never/special occasions only, <3 times/week, ≥ 3 times/week), and total energy intake (quartiles – KJ).
Additionally adjusted for: coffee intake (continuous – mean cups/day), tanning ability (categorical – never tan, mildly/occasionally tan, moderately tan, very tanned), childhood sunburn occasions (continuous), natural hair color (categorical – red/blonde, light brown, dark brown/ black, other), skin color (categorical – dark/light olive, fair, very fair), average time spent outdoors in summer (continuous – hours/day), sunlamp/solarium use (continuous – times/year), sun/UV protection use (categorical – always, most of the time, sometimes, never/rarely), and history of non-melanoma skin cancer (yes, no).
Abbreviation: OR, odds ratio; CI, confidence interval
Of the individual citrus products analyzed, consumption of oranges and orange juice were independently associated with melanoma risk (Table 3). In the fully adjusted models, participants in the highest category of orange and orange juice consumption (>1 serving per day) had a significantly increased risk of melanoma relative to those with no consumption. Relative to no consumption, ORs (95% CIs) were 1.79 (1.07–2.78) and 1.54 (1.10–2.10) for the highest consumption of oranges and orange juice, respectively (p-trend = 0.043 and 0.021). No other orange or orange juice serving categories were significant in either of the multivariable-adjusted models, and no significant results were seen for any other citrus product. Respective associations for high intake of total citrus, oranges, and orange juice were larger than associations for all citrus fruit and all citrus juice analyzed separately (Supplementary Table 1). Relative to no consumption, those in the highest category (>1 serving per day) of citrus fruit intake and citrus juice intake had ORs (95% CIs) of 1.33 (1.06–1.64) and 1.43 (1.05–1.90), respectively.
Table 3.
Melanoma risk according to frequency of individual citrus product consumption among study participants in the UK Biobank.
| Serving category | |||||
|---|---|---|---|---|---|
| Citrus type | None | >0-½ serving | >½-1 serving | >1 serving | p-trend |
| Grapefruit | |||||
| N (number of cases) | 188,932 (1,502) | 8,240 (70) | 1,724 (19) | 68 (1) | |
| Age-adjusted OR (95% CI) | 1.00 | 0.99 (0.78–1.26) | 1.29 (0.79–1.98) | 1.81 (0.10–8.22) | 0.30 |
| Multivariable-adjusted OR1 (95% CI)1 | 1.00 | 1.04 (0.78–1.37) | 1.56 (0.90–2.48) | 2.49 (0.14–11.63) | 0.075 |
| Multivariable-adjusted OR2 (95% CI)2 | 1.00 | 1.04 (0.77–1.37) | 1.54 (0.89–2.45) | 2.41 (0.14–11.25) | 0.089 |
| Grapefruit juice | |||||
| N (number of cases) | 190,053 (1,521) | 6,117 (40) | 2,479 (28) | 315 (3) | |
| Age-adjusted OR (95% CI) | 1.00 | 0.81 (0.58–1.09) | 1.42 (0.95–2.02) | 1.23 (0.30–3.21) | 0.30 |
| Multivariable-adjusted OR1 (95% CI)1 | 1.00 | 0.72 (0.47–1.05) | 1.23 (0.73–1.92) | 2.07 (0.51–5.94) | 0.52 |
| Multivariable-adjusted OR2 (95% CI)2 | 1.00 | 0.71 (0.46–1.03) | 1.20 (0.71–1.89) | 1.96 (0.48–5.23) | 0.60 |
| Orange | |||||
| N (number of cases) | 162,721 (1,263) | 21,102 (195) | 13,302 (111) | 1,839 (23) | |
| Age-adjusted OR (95% CI) | 1.00 | 1.14 (0.97–1.32) | 1.00 (0.82–1.21) | 1.55 (0.99–2.29) | 0.15 |
| Multivariable-adjusted OR1 (95% CI)1 | 1.00 | 1.02 (0.83–1.23) | 1.12 (0.89–1.39) | 1.83 (1.10–2.85) | 0.030 |
| Multivariable-adjusted OR2 (95% CI)2 | 1.00 | 1.00 (0.82–1.21) | 1.11 (0.88–1.38) | 1.79 (1.07–2.78) | 0.043 |
| Orange Juice | |||||
| N (number of cases) | 126,576 (968) | 38,782 (312) | 28,976 (260) | 4,630 (52) | |
| Age-adjusted OR (95% CI) | 1.00 | 1.04 (0.92–1.19) | 1.17 (1.02–1.34) | 1.51 (1.13–1.98) | 0.0007 |
| Multivariable-adjusted OR1 (95% CI)1 | 1.00 | 0.96 (0.82–1.13) | 1.10 (0.92–1.30) | 1.57 (1.12–2.14) | 0.016 |
| Multivariable-adjusted OR2 (95% CI)2 | 1.00 | 0.96 (0.82–1.13) | 1.09 (0.92–1.29) | 1.54 (1.10–2.10) | 0.021 |
| Satsuma | |||||
| N (number of cases) | 157,266 (1,228) | 20,989 (187) | 14,592 (122) | 6,117 (55) | |
| Age-adjusted OR (95% CI) | 1.00 | 1.14 (0.97–1.32) | 1.06 (0.88–1.27) | 1.16 (0.88–1.51) | 0.17 |
| Multivariable-adjusted OR1 (95% CI)1 | 1.00 | 1.19 (0.98–1.43) | 1.04 (0.82–1.30) | 1.29 (0.92–1.74) | 0.11 |
| Multivariable-adjusted OR2 (95% CI)2 | 1.00 | 1.20 (0.99–1.45) | 1.03 (0.82–1.31) | 1.29 (0.92–1.74) | 0.11 |
Note: bold font indicates statistical significance. Four levels of citrus intake were used due to a lack of observations above the >1 serving category for consumption of grapefruit, grapefruit juice, and oranges.
Multivariable analyses were further adjusted for: sex (male, female), education (categorical – college or university degree; Advanced level/Advanced Subsidiary level; Ordinary level/General Certificate of Secondary Education, Certificate of Secondary Education; National Vocational Qualification, Higher National Diploma, or Higher National Certificate; other professional degree; none of the preceding), income (less than £30,999, £31,000 - £51,999, £52,000 and up, missing), physical activity (low, moderate, high, missing), BMI (continuous), smoking status (never, previous, current), alcohol intake (never/special occasions only, < 3 times/week, ≥ 3 times/week), coffee intake (continuous – mean cups/day), tanning ability (never tan, mildly/occasionally tan, moderately tan, very tanned), childhood sunburn occasions (continuous), natural hair color (red/blonde, light brown, dark brown/black, other), skin color (dark/light olive, fair, very fair), average time spent outdoors in summer (continuous – hours/day), sunlamp/solarium use (continuous – times/year), sun/UV protection use (always, most of the time, sometimes, never/rarely), total energy intake (quartiles – KJ), and history of non-melanoma skin cancer (yes, no).
Additionally adjusted for consumption of other individual citrus products listed in table
Abbreviation: OR, odds ratio; CI, confidence interval
We tested for interactions between total citrus consumption and other melanoma risk factors, including coffee intake, tanning ability, childhood sunburn occasions, natural hair color, skin color, average time outdoors in summer, solarium/sunlamp use, and sun/UV protection use. Before adjustment for multiple comparisons, we found statistical evidence of interaction only for skin color (p for interaction = 0.019) and then conducted stratified analyses to evaluate this potential interaction (Table 4). A linear relationship between citrus consumption and melanoma risk was observed among participants with a fair or very fair skin complexion, with a significantly increased melanoma risk associated with those in the highest category of total citrus consumption. Relative to no citrus consumption, ORs (95% CIs) of 1.05 (0.87–1.26), 1.10 (0.93–1.30), 1.17 (0.96–1.42), and 1.75 (1.31–2.29) were observed among fair/very fair participants with total citrus consumptions of >0-½, >½ to 1, >1 to 2, and >2 daily servings, respectively (p-trend = 0.0015). A decreased melanoma risk was observed among olive-skinned participants consuming half a serving of citrus per day (OR [95% CI] = 0.48 [0.23–0.89]). As citrus consumers and non-citrus consumers differed in several key demographic categories (Table 1), and since some commonly used medications are known to interact with certain citrus products,38–40 we considered that non-consumers of citrus may systematically differ to citrus consumers. Therefore, we repeated our aforementioned analyses using the lowest category of citrus consumption (>0-½ serving) as the referent to see if our results differed when only considering citrus consumers (Supplementary Tables 2–4). No appreciable differences were observed.
Table 4.
Melanoma risk according to frequency of total citrus consumption stratified by skin color among study participants in the UK Biobank.
| Total citrus serving category | ||||||
|---|---|---|---|---|---|---|
| Melanoma risk factor | None | >0-½ serving | >½-1 serving | >1-2 servings | >2 servings | p-trend |
| Skin color | ||||||
| Dark/light olive | ||||||
| N (number of cases) | 16,820 (69) | 7,419 (20) | 9,160 (37) | 5,312 (23) | 1,423 (5) | |
| Age-adjusted OR (95% CI) | 1.00 | 0.64 (0.38–1.03) | 0.93 (0.62–1.39) | 1.00 (0.61–1.58) | 0.81 (0.52–2.30) | 0.85 |
| Multivariable-adjusted OR1 (95% CI)1 | 1.00 | 0.64 (0.38–1.03) | 0.95 (0.63–1.41) | 1.02 (0.62–1.62) | 0.84 (0.29–1.89) | 0.95 |
| Multivariable-adjusted OR2 (95% CI)2 | 1.00 | 0.48 (0.23–0.89) | 0.84 (0.50–1.36) | 0.84 (0.44–1.50) | 0.68 (0.17–1.89) | 0.45 |
| Fair/very fair | ||||||
| N (number of cases) | 65,477 (537) | 29,409 (257) | 36,669 (361) | 21,428 (206) | 5,847 (77) | |
| Age-adjusted OR (95% CI) | 1.00 | 1.05 (0.90–1.22) | 1.18 (1.03–1.34) | 1.14 (0.97–1.34) | 1.56 (1.22–1.98) | 0.0005 |
| Multivariable-adjusted OR1 (95% CI)1 | 1.00 | 1.03 (0.88–1.19) | 1.14 (0.99–1.30) | 1.11 (0.94–1.30) | 1.51 (1.18–1.92) | 0.0029 |
| Multivariable-adjusted OR2 (95% CI)2 | 1.00 | 1.05 (0.87–1.26) | 1.10 (0.93–1.30) | 1.17 (0.96–1.42) | 1.75 (1.31–2.29) | 0.0015 |
Note: Skin color was the only melanoma risk factor with evidence of effect modification in our analysis. Bold font indicates statistical significance
Further adjusted for: sex (male, female), education (categorical – college or university degree; Advanced level/Advanced Subsidiary level; Ordinary level/General Certificate of Secondary Education, Certificate of Secondary Education; National Vocational Qualification, Higher National Diploma, or Higher National Certificate; other professional degree; none of the preceding), income (categorical – less than £30,999, £31,000 - £51,999, £52,000 and up, missing), physical activity (categorical – low, moderate, high, missing), BMI (continuous), smoking status (categorical – never, previous, current), alcohol intake (categorical – never/special occasions only, < 3 times/week, ≥ 3 times/week), and total energy intake (quartiles – KJ).
Additionally adjusted for: coffee intake (continuous – mean cups/day), tanning ability (categorical – never tan, mildly/occasionally tan, moderately tan, very tanned), childhood sunburn occasions (continuous), natural hair color (categorical – red/blonde, light brown, dark brown/black, other), skin color (categorical – dark/light olive, fair, very fair), average time spent outdoors in summer (continuous – hours/day), sunlamp/solarium use (continuous – times/year), sun/UV protection use (categorical – always, most of the time, sometimes, never/rarely), and history of non-melanoma skin cancer (yes, no).
Abbreviation: OR, odds ratio; CI, confidence interval
Discussion
In the current study, we found a significant association between total citrus consumption and melanoma risk after adjusting for potential confounders. This result is consistent with findings from the NHS and HPFS,25 but not with those from EPIC26, WHI,27 or NIH-AARP.28 In the NHS and HPFS study, participants with the highest total citrus consumption had a 36% increased risk of melanoma. The greater effect size in the current analysis (63% increased risk) may reflect the fact that the NHS and HPFS is comprised of health care professionals who may be more likely to work indoors and/or have greater knowledge of UV protection compared with our UKBB sample drawn from the general population. In the WHI, there was no significant association between total citrus and melanoma risk. The null finding from the WHI could potentially be explained by the fact that the WHI cohort is comprised of postmenopausal women. Since men have a greater melanoma risk,41,42 and older women are less likely than younger women to engage in certain melanoma-risk behaviors, such as indoor tanning, solarium use, and sunbathing,43–46 it is plausible to see lower risk estimates in this sample.
In EPIC, there was also no significant association between melanoma risk and total citrus intake (hazard ratio [HR] [95% CI] = 0.98 [0.83–1.15]), but there was a significant association between citrus fruit intake and melanoma risk (HR [95% CI] = 1.23 [1.02–1.48]).26 These results respectively oppose and support our current hypothesis/findings. The inconsistency for total citrus consumption between the current analysis and the EPIC study could possibly be due to study population differences. While the current study utilizes an all-British sample, EPIC combined data from several different European countries with differing levels of citrus intake as well as other possible physical or behavioral characteristics. Additionally, differing countries may have different food processing or agricultural regulations that could possibly influence furocoumarin concentrations in citrus products from country to country. There was also no significant association between citrus consumption and melanoma risk in the NIH-AARP. Although some subgroup analyses showed significant trends towards an increased melanoma risk associated with citrus consumption, none of the risk estimates were statistically significant, and neither was overall association between total citrus and melanoma risk (HR [95% CI] = 1.09 [0.86–1.39]).28 However, these findings were limited by a rough estimate of subjects’ UV exposure based on geographic residence rather than actual characteristics or behaviors, and further limited by the inability to control for key sun-exposure variables such as childhood sunburn history, skin pigmentation, and tanning ability. If such data were available, the NIH-AARP study could have provided a more accurate estimation for the association of interest.
Results of all five of these studies (the current analysis, NHS/HPFS, WHI, EPIC, and NIH-AARP) are in contrast with the findings of Fortes. et al., who reported a protective effect of citrus for melanoma risk for high citrus fruit consumption (>5 times/week) relative to low citrus fruit consumption (up to twice/week) (OR [95% CI] = 0.51 [0.32–0.80]).13 However, this small, Italian case-control study was based on a total of 304 melanoma cases and utilized hospital patients for cases and controls. Furthermore, this study did not control for several key dietary or sun-exposure variables. Because of these limitations, we believe that the results of the current study, and that of the NHS/HPFS, WHI, EPIC, and NIH-AARP analyses, provide a more accurate estimation of the association between citrus consumption and melanoma risk.
We also found that orange and orange juice consumption were independently associated with melanoma risk. These findings are consistent with results from the NHS and HPFS, which found a significant increased risk associated with orange consumption (age-adjusted model only) and a 25% increased risk associated with orange juice consumption. Unlike NHS/HPFS findings, the current study found no significant results associated with grapefruit consumption. Grapefruits are rich sources of psoralens,24 and, according to Melough et al., have an exponentially higher total furocoumarin concentration (21,858 ng/g) than oranges (0.5 ng/g), orange juice (3.2 ng/g), or satsumas (0.2 ng/g),47 so a significant association between grapefruit consumption and melanoma risk would have been plausible. However, our UKBB sample had low grapefruit consumption (5.7% of our UKBB cases reported grapefruit consumption vs. 83% of cases in the NHS/HPFS), limiting power for statistical evaluation. If consumption of grapefruit in our sample more closely resembled consumption of oranges or orange juice (20.7% and 39.2% of our UKBB cases reported orange and orange juice consumption, respectively), then our results for grapefruit consumption may have been more consistent NHS/HPFS results.25
In subgroup analyses, skin color was found to play a significant role in the relationship between total citrus consumption and melanoma risk, but only before adjustment for multiple comparisons. Participants with fair to very fair skin complexion had a citrus-associated melanoma risk that was greater than participants with olive complexions. This result is biologically plausible as fairer skin is a known melanoma risk factor12,15 due to the lack of melanin, which helps protect the skin against UV radiation.48 As psoralens and furocoumarins have known photosensitizing properties,49 it is plausible that psoralen-rich foods could magnify melanoma risk in people particularly susceptible to sun/UV damage.
As with all studies, the current analysis has limitations. Dietary data were self-reported, likely resulting in nondifferential misclassification. However, we calculated a cumulative average intake over several timepoints to minimize random error, and 82% of participants reported the same or adjacent intake category across timepoints.33 Another potential limitation of this study could be the difficulty of generalizing results from UK-based data to other populations. Different countries may have differing dietary patterns that influence the amount of citrus consumed, or may have different methods of agricultural processing that could impact the furocoumarin concentration found in citrus products. Differences in climate could be another challenge to generalizability. However, as the UK generally gets fewer hours of sunshine than many other areas with White populations, such as the United States, Australia, and elsewhere in Europe,50–53 results of our current analysis may be biased towards the null, and implications could be even greater for other populations with more sun exposure. Finally, we are limited by the inability to control for family history of melanoma. Melanoma can be heritable,54,55 so residual confounding from the inability to adjust for this factor is a limitation.
In conclusion, our current analysis, based on a large, prospective, population-based cohort, found that high citrus consumption was associated with a significantly increased risk of melanoma. Consumption of oranges and orange juice were independently associated with melanoma risk. These findings support previous evidence of the photosensitivity and photocarcinogenicity of psoralens and support the hypothesis that high consumption of psoralen-rich foods may increase melanoma risk. Although biologically plausible, further investigation is needed to confirm our findings, particularly those which support a potential effect modification by skin color. Further investigation and confirmation of these findings could lead to updated sun-exposure guidance and improved melanoma risk-reduction strategies.
Supplementary Material
What is already known about this topic?
Psoralen is a type of furocoumarin with known photosensitizing and photocarcinogenic properties, and it is naturally, abundantly found in citrus products. Since 2015, four large prospective studies have investigated the hypothesis that high citrus consumption increases melanoma risk due to psoralen photocarcinogenicity. These studies have yielded mixed results, and they were subject to limitations that could challenge external validity.
What does this study add?
In the current study, we confirm the positive association between total citrus consumption and melanoma found in a previous analysis, with consumption of oranges and orange juice being independently associated with melanoma risk. We add that the association for total citrus is modified by skin color, with greater risk among fair-skinned participants relative to olive-skinned participants. We are the first to assess this association in a British cohort.
Acknowledgements
This research was supported in part by Lilly Endowment, Inc., through its support for the Indiana University Pervasive Technology Institute, and in part by the Indiana METACyt Initiative. The Indiana METACyt Initiative at IU was also supported in part by Lilly Endowment, Inc.
This research has been conducted using the UK Biobank Resource under Application Number 49419.
Funding. This work was supported by the National Heart, Lung and Blood Institute [K01HL14033 to M.L.]; and the Eunice Kennedy Shriver National Institute of Child Health and Human Development [R03HD092854 to M.L]; and the National Cancer Institute [T32CA117865-11 to V.C.].
Footnotes
Author Disclosures. The authors have no conflicts of interest to declare.
References
- 1.Lens M, Dawes M. Global perspectives of contemporary epidemiological trends of cutaneous malignant melanoma. Br J Dermatol 2004; 150: 179–85. [DOI] [PubMed] [Google Scholar]
- 2.Whiteman DC, Green AC, Olsen CM. The growing burden of invasive melanoma: projections of incidence rates and numbers of new cases in six susceptible populations through 2031. J Invest Dermatol 2016; 136: 1161–71. [DOI] [PubMed] [Google Scholar]
- 3.Cancer Research UK. Cancer incidence for common cancers. Available at: https://www.cancerresearchuk.org/health-professional/cancer-statistics/incidence/common-cancers-compared#heading-Zero (last accessed30 December 2020).
- 4.Diepgen TL, Mahler V. The epidemiology of skin cancer. Br J Dermatol 2002; 146: 1–6. [DOI] [PubMed] [Google Scholar]
- 5.Garbe C, McLeod GRC, Buettner PG. Time trends of cutaneous melanoma in Queensland, Australia and Central Europe. Cancer: Interdisciplinary International Journal of the American Cancer Society 2000; 89: 1269–78. [DOI] [PubMed] [Google Scholar]
- 6.Shaikh WR, Dusza SW, Weinstock MA et al. Melanoma thickness and survival trends in the United States, 1989–2009. JNCI: J Natl Cancer Inst 2016; 108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Smith BD, Smith GL, Hurria A et al. Future of cancer incidence in the United States: burdens upon an aging, changing nation. J Clin Oncol 2009; 27: 2758–65. [DOI] [PubMed] [Google Scholar]
- 8.Apalla Z, Lallas A, Sotiriou E et al. Epidemiological trends in skin cancer. Dermatol Pract Concept 2017; 7: 1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Bleyer A, Viny A, Barr R. Cancer in 15-to 29-year-olds by primary site. Oncologist 2006; 11: 590–601. [DOI] [PubMed] [Google Scholar]
- 10.Hanly P, Soerjomataram I, Sharp L. Measuring the societal burden of cancer: The cost of lost productivity due to premature cancer‐related mortality in E urope. Int J Cancer 2015; 136: E136–E45. [DOI] [PubMed] [Google Scholar]
- 11.Rigel DS. Cutaneous ultraviolet exposure and its relationship to the development of skin cancer. J Am Acad Dermatol 2008; 58: S129–S32. [DOI] [PubMed] [Google Scholar]
- 12.Armstrong BK, Kricker A. The epidemiology of UV induced skin cancer. J Photochem Photobiol B 2001; 63: 8–18. [DOI] [PubMed] [Google Scholar]
- 13.Fortes C, Mastroeni S, Melchi F et al. A protective effect of the Mediterranean diet for cutaneous melanoma. Int J Epidemiol 2008; 37: 1018–29. [DOI] [PubMed] [Google Scholar]
- 14.Green A, Battistutta D. Incidence and determinants of skin cancer in a high‐risk Australian population. Int J Cancer 1990; 46: 356–61. [DOI] [PubMed] [Google Scholar]
- 15.Beral V, Evans S, Shaw H et al. Cutaneous factors related to the risk of malignant melanoma. Br J Dermatol 1983; 109: 165–72. [DOI] [PubMed] [Google Scholar]
- 16.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: 650–62. [DOI] [PubMed] [Google Scholar]
- 17.Holly EA, Aston DA, Cress RD et al. Cutaneous melanoma in women I. Exposure to sunlight, ability to tan, and other risk factors related to ultraviolet light. Am J Epidemiol 1995; 141: 923–33. [DOI] [PubMed] [Google Scholar]
- 18.Gallagher RP, Spinelli JJ, Lee TK. Tanning beds, sunlamps, and risk of cutaneous malignant melanoma. Cancer Epidemiol Biomarkers Prev 2005; 14: 562–6. [DOI] [PubMed] [Google Scholar]
- 19.Veierød MB, Adami H-O, Lund E et al. Sun and solarium exposure and melanoma risk: effects of age, pigmentary characteristics, and nevi. Cancer Epidemiol Biomarkers Prev 2010; 19: 111–20. [DOI] [PubMed] [Google Scholar]
- 20.Zanetti R, Franceschi S, Rosso S et al. Cutaneous melanoma and sunburns in childhood in a southern European population. Eur J Cancer 1992; 28: 1172–6. [DOI] [PubMed] [Google Scholar]
- 21.Ashwood-Smith M, Poulton G, Barker M et al. 5-Methoxypsoralen, an ingredient in several suntan preparations, has lethal, mutagenic and clastogenic properties. Nature 1980; 285: 407. [DOI] [PubMed] [Google Scholar]
- 22.Sayre RM, Dowdy JC. The increase in melanoma: Are dietary furocoumarins responsible? Med Hypotheses 2008; 70: 855–9. [DOI] [PubMed] [Google Scholar]
- 23.Dugrand A, Olry A, Duval T et al. Coumarin and furanocoumarin quantitation in citrus peel via ultraperformance liquid chromatography coupled with mass spectrometry (UPLC-MS). J Agric Food Chem 2013; 61: 10677–84. [DOI] [PubMed] [Google Scholar]
- 24.Dugo P, Piperno A, Romeo R et al. Determination of oxygen heterocyclic components in citrus products by HPLC with UV detection. J Agric Food Chem 2009; 57: 6543–51. [DOI] [PubMed] [Google Scholar]
- 25.Wu S, Han J, Feskanich D et al. Citrus consumption and risk of cutaneous malignant melanoma. J Clin Oncol 2015; 33: 2500. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Mahamat-Saleh Y, Cervenka I, Al-Rahmoun M et al. Citrus intake and risk of skin cancer in the European Prospective Investigation into Cancer and Nutrition cohort (EPIC). Eur J Epidemiol 2020: 1–11. [DOI] [PubMed] [Google Scholar]
- 27.Melough MM, Wu S, Li W-Q et al. Citrus Consumption and Risk of Cutaneous Malignant Melanoma in the Women’s Health Initiative. Nutr Cancer 2020; 72: 568–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Melough MM, Sakaki J, Liao LM et al. Association between Citrus Consumption and Melanoma Risk in the NIH-AARP Diet and Health Study. Nutr Cancer 2020: 1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Sudlow C, Gallacher J, Allen N et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med 2015; 12: e1001779. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.UK Biobank. UK Biobank: Rationale, design and development of a large-scale prospective resource. Available at: http://www.ukbiobank.ac.uk/key-documents/ (last accessed29 May 2019).
- 31.Allen N, Sudlow C, Downey P et al. UK Biobank: Current status and what it means for epidemiology. Health Policy Technol 2012; 1: 123–6. [Google Scholar]
- 32.Liu B, Young H, Crowe FL et al. Development and evaluation of the Oxford WebQ, a low-cost, web-based method for assessment of previous 24 h dietary intakes in large-scale prospective studies. Public Health Nutr 2011; 14: 1998–2005. [DOI] [PubMed] [Google Scholar]
- 33.Bradbury KE, Young HJ, Guo W et al. Dietary assessment in UK Biobank: an evaluation of the performance of the touchscreen dietary questionnaire. J Nutr Sci 2018; 7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Agbai ON, Buster K, Sanchez M et al. Skin cancer and photoprotection in people of color: a review and recommendations for physicians and the public. J Am Acad Dermatol 2014; 70: 748–62. [DOI] [PubMed] [Google Scholar]
- 35.Gloster HM Jr, Neal K. Skin cancer in skin of color. J Am Acad Dermatol 2006; 55: 741–60. [DOI] [PubMed] [Google Scholar]
- 36.Bradford PT. Skin cancer in skin of color. Dermatol Nurs 2009; 21: 170. [PMC free article] [PubMed] [Google Scholar]
- 37.Houghton AN, Polsky D. Focus on melanoma. Cancer Cell 2002; 2: 275–8. [DOI] [PubMed] [Google Scholar]
- 38.Bailey DG, Dresser G, Arnold JMO. Grapefruit–medication interactions: Forbidden fruit or avoidable consequences? CMAJ 2013; 185: 309–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Saito M, Hirata-Koizumi M, Matsumoto M et al. Undesirable effects of citrus juice on the pharmacokinetics of drugs. Drug Saf 2005; 28: 677–94. [DOI] [PubMed] [Google Scholar]
- 40.Bailey D, Spence J, Munoz C et al. Interaction of citrus juices with felodipine and nifedipine. The Lancet 1991; 337: 268–9. [DOI] [PubMed] [Google Scholar]
- 41.Cho E, Rosner BA, Feskanich D et al. Risk factors and individual probabilities of melanoma for whites. J Clin Oncol 2005; 23: 2669–75. [DOI] [PubMed] [Google Scholar]
- 42.Cho E, Rosner BA, Colditz GA. Risk factors for melanoma by body site. Cancer Epidemiol Biomarkers Prev 2005; 14: 1241–4. [DOI] [PubMed] [Google Scholar]
- 43.Lazovich D, Vogel RI, Weinstock MA et al. Association between indoor tanning and melanoma in younger men and women. JAMA Dermatol 2016; 152: 268–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Manne SL, Coups EJ, Jacobsen PB et al. Sun protection and sunbathing practices among at-risk family members of patients with melanoma. BMC Public Health 2011; 11: 122. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Buller DB, Cokkinides V, Hall HI et al. Prevalence of sunburn, sun protection, and indoor tanning behaviors among Americans: review from national surveys and case studies of 3 states. J Am Acad Dermatol 2011; 65: S114. e1–S. e11. [DOI] [PubMed] [Google Scholar]
- 46.Lawler SP, Kvaskoff M, DiSipio T et al. Solaria use in Queensland, Australia. Aust N Z J Public Health; 2006; 30: 479–82. [DOI] [PubMed] [Google Scholar]
- 47.Melough MM, Lee SG, Cho E et al. Identification and quantitation of furocoumarins in popularly consumed foods in the US Using QuEChERS extraction coupled with UPLC-MS/MS analysis. J Agric Food Chem 2017; 65: 5049–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Brenner M, Hearing VJ. The protective role of melanin against UV damage in human skin. Photochem Photobiol 2008; 84: 539–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Mullen MP, Pathak MA, West JD et al. Carcinogenic effects of monofunctional and bifunctional furocoumarins. Natl Cancer Inst Monogr 1984; 66: 205–10. [PubMed] [Google Scholar]
- 50.Met Office. UK temperature, rainfall and sunshine time-series. Available at: https://www.metoffice.gov.uk/climate/uk/summaries/actualmonthly (last accessed30 December 2020).
- 51.National Oceanic and Atmospheric Administration. Sunshine - Average Percent of Possible. Available at: https://www.ncdc.noaa.gov/ghcn/comparative-climatic-data (last accessed30 December 2020).
- 52.British Business Energy. Europe vs United States Sunshine Duration In Hours Per Year. Available at: https://britishbusinessenergy.co.uk/sunshine-duration/ (last accessed30 December 2020).
- 53.Australian Government Bureau of Meteorology. Sunshine: Average Daily Sunshine Hours. Available at: http://www.bom.gov.au/watl/sunshine/ (last accessed30 December 2020).
- 54.Ransohoff KJ, Wu W, Cho HG et al. Two-stage genome-wide association study identifies a novel susceptibility locus associated with melanoma. Oncotarget 2017; 8: 17586. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55.Ford D, Bliss JM, Swerdlow AJ et al. Risk of cutaneous melanoma associated with a family history of the disease. Int J Cancer 1995; 62: 377–81. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.